Author Archives: author

The Tragedy of Smoking, Alcohol, and Multiple Substance use during Pregnancy

DOI: 10.31038/AWHC.2023644

Abstract

Background: Antenatal substance use is a significant public health concern in South Africa. Information on smoking, drinking and drug use during pregnancy was collected prospectively for the Safe Passage Study of the Prenatal Alcohol in Sudden infant death syndrome and Stillbirth Network.

Objectives: Data from 4 926 pregnant women in a local community near Tygerberg Academic Hospital, were examined to determine whether associations between different substance use groups and postnatal infant outcomes at birth and 1 year were significant.

Methods: Gestational age (GA) was determined by earliest ultrasound. Maternal data were collected at enrolment or first antenatal visit. Substance use data were obtained at up to four occasions. Birthweight data were derived from medical records, and birthweight z-scores (BWZs) were specifically calculated using INTERGROWTH-21st study data. Statistical analyses were done with Statistica version 13.

Results: Women who used more substances enrolled later, were younger, and had smaller mid-upper arm circumferences (MUACs), less education and lower monthly income than women who used no substances (control group). Infants born to women who used more substances had lower GA at delivery, birthweight and BWZ than infants from the control group. At 1 year, infants born to women who used more substances had a lower weight, shorter length and smaller head circumference. Education was positively associated with all infant outcomes at birth and 1 year. MUAC was positively associated with infant BWZ, and weight and length at 1 year. Income was negatively associated with BWZ, but positively associated with all 1-year outcomes.

Conclusion: Substance use during pregnancy affects infant outcomes at birth and 1 year of age. The addictive properties of substance use make cessation difficult, so prevention strategies should be implemented long before pregnancy. Higher maternal education, associated with better infant outcomes at birth and 1 year and acting as a countermeasure to substance use, is of paramount importance.

Keywords

Smoking, Drinking, Marijuana, Methamphetamine, Pregnancy

Introduction

Substance use during pregnancy is on the increase worldwide [1-4] and is a significant public health concern [5,6]. In South Africa (SA), use of multiple substances during pregnancy is common. In a survey of 5 232 pregnant women visiting midwife obstetric units in Cape Town, it was found that 36.9% used alcohol and drugs, 34.9% alcohol only, and 1.6% drugs only [7]. Also in Cape Town, a substudy of the Safe Passage Study (SPS), on the value of maternal serum alpha-fetoprotein measurements, found that 61% of pregnant women smoked, 55% drank alcohol, and 9% and 5% used marijuana and methamphetamine, respectively [8]. Methamphetamine use in pregnancy is associated with poorer neonatal outcomes, especially decreased birthweight, head circumference and body length [9,10]. The effects of marijuana use during pregnancy are less clear, with reports ranging from no adverse effect with regard to the likelihood of prematurity or LBW [11-13] to a reduction in birthweight, length and head circumference [3] and an increase in preterm births and Growth Restriction (GR) [14,15].

The association of marijuana use with poor perinatal outcome seems to be attributable to concomitant use of tobacco and other confounding factors [12]. Perinatal outcome is particularly susceptible to socioeconomic conditions affecting lifestyle choices and behaviour [16]. Low socioeconomic status and lower educational attainment increase the risk of smoking during pregnancy significantly [17,18]. Smoking is not only associated with complications such as preterm birth, GR and stillbirth [17,19,20], but has long-term maternal implications such as lung cancer, cardiovascular and chronic respiratory disease, oral diseases and strokes, and long-term infant implications such as respiratory problems (e.g. childhood asthma), infections, obesity, cleft lip/palate, and neurodevelopmental and behavioural problems [21-24].

Interestingly, only the effect of cocaine on birthweight remained significant after adjusting for confounding variables [5]. It is important to note that very few pregnant women use methamphetamine or marijuana on their own; most of them also use nicotine or alcohol, or both. In a study of 12 069 pregnant women, it was found that 45% of marijuana users also smoked [13]. The same applied to users of methamphetamine, of whom 78.6%, 42.9% and 39.3% used tobacco, alcohol, and marijuana, respectively [2].

Of all three health-compromising behaviours, smoking, alcohol consumption and recreational drug use, cigarette smoking has been most studied and strongly implicated in reduced fetal growth [25]. Our previous finding that significantly more pregnant smokers than pregnant non-smokers engaged in heavy alcohol consumption [26] is supported by Okah et al. [27]. They found that pregnant smokers were seven times more likely than non-smokers to use alcohol and/or drugs, and that the rate of heavy smoking and moderate/heavy drinking increased with the number of health-compromising behaviours. Infants antenatally exposed to both alcohol and cigarettes had a substantially higher risk of sudden infant death syndrome compared with those who were unexposed, or exposed to alcohol or cigarettes alone [28].

As the information on smoking, drinking and drug use for the SPS was collected prospectively, this database was ideal to examine the interactions of substance use during pregnancy on infant outcome [29].

Methods

The SPS of the Prenatal Alcohol in Sudden infant death syndrome and Stillbirth (PASS) Network.was designed to investigate the role of prenatal alcohol exposure in the outcome of 12 000 pregnancies in SA (Cape Town) and the USA (Northern Plains). Women recruited included those with low-and high-risk pregnancies, with a wide range of exposures to alcohol, nicotine, marijuana and methamphetamine [29]. The present study was limited to the SA arm of the SPS, where participants were recruited at a community health centre close to Tygerberg Academic Hospital (TAH), Cape Town. Participants were enrolled between August 2007 and January 2015 and infants were followed up until the end of August 2016. Gestational Age (GA) was determined by earliest ultrasound before the second antenatal visit. Depending on the GA at enrolment, women had up to three further antenatal visits at TAH, at 20-24, 28-32 and 34-38 weeks. The revised Timeline Followback method was used at up to four occasions to obtain detailed information on drinking, cigarette smoking, and the use of marijuana, amphetamines and other substances during pregnancy [30]. Anaemia was based on laboratory results of a haemoglobin value <11 g/dL during pregnancy and obtained from Medical Chart Abstraction (MCA). Demographic and anthropometric information was obtained at enrolment or the first antenatal visit. Maternal weight was measured twice, using a regularly calibrated high-quality scale. For the Mid-Upper Arm Circumference (MUAC), the midpoint of the upper arm was first determined and then the circumference measured twice. If any two measurements differed by >1 kg (weight) or 2 mm (MUAC), a third measurement was taken and the mean of the closest two measurements used.

A pregnancy loss or fetal demise before 20 weeks, according to the US definition for SPS, was defined as a miscarriage, whereas a non-live birth at ≥20 weeks was regarded as a stillbirth [31-33]. Terminations of pregnancies after 20 weeks were done for medical reasons. Death of a liveborn infant before the age of 1 year was defined as an infant death. A social worker, employed for the SPS, was available to all women for counselling if necessary or requested.

Newborns were weighed immediately after birth and the information was entered in the maternal chart, from where it was obtained by MCA after delivery. The GA at delivery, obtained from the Electronic Data Capturing (EDC) system, together with fetal sex was used to determine birthweight z-scores (BWZs) and centiles specifically for us upon request, from the international standards of the INTERGROWTH-21st study (available for GAs from 168 to 299 days, excluding twins) [34].

The infants were seen at 1 year of age and the assessment date was adjusted for prematurity, e.g. an infant born 10 weeks (70 days) early had a required 1-year age of birth date + 365 + 70 days to birth date + 365 + 70 + 30 days at 1-year assessment. At the beginning of our study, infants born at term were required to have an age of 365-30 to 365 + 30 days at their 1-year examination, but this was soon changed to between birth date + 365 days and birth date + 365 + 30 days. Infants were weighed (1YW), and their length (1YL) and head circumference (1YHC) were measured by trained research workers according to a specific protocol. For weighing the infants, a Charder digital baby scale was used (Charder Electronic Co. Ltd, Taiwan). The child, dressed in a clean, dry diaper, with a vest during winter, was weighed to the nearest 0.1 kg. The process was repeated, and if the measurements differed by >0.2 kg, a third measurement was taken. A Seca 416 infantometer (Seca Deutchland, Germany) was used to measure the length to the nearest millimetre. The full procedure was repeated for a second measurement and if it differed by >2 mm, a third measurement was taken. A flexible tape measure was used to measure the head circumference to the nearest millimetre while the child was sitting on the mother’s lap or lying down. The tape measure was placed over the occipital protuberance at the back of the head and around to just over the supraorbital ridge and the forehead in front. The procedure was repeated, and done a third time if the first two measurements differed by >2 mm. All the measurements were entered on a specific case report form, and later on the EDC system.

To examine the effects of various combinations of exposure to nicotine, alcohol, marijuana and methamphetamine, 11 different combinations were used, namely no exposure (Control), NoDrugsDrink, NoDrugsSmoke, NoDrugsDrinkSmoke, MarSmoke, MarDrink, MarDrinkSmoke, MetDrink, MetSmoke, MetDrinkSmoke, and All (used all four substances). Since only 12 and 2 participants used only marijuana or only methamphetamine, respectively, separate groups for these drugs were not developed and they were excluded from the cohort. Outcome variables studied were BWZ, 1YW, 1YL and 1YHC. Since we, and others, have shown that MUAC, maternal education and household income play important roles in newborn and 1-year outcomes, these were used as confounders [35,36].

Statistical analyses were performed using the Statistica data analysis software system, version 13 (TIBCO Software Inc., USA). Descriptive statistics were used to describe continuous variables, which were compared between groups with analysis of variance (ANOVA). Bonferroni or least significant difference multiple comparisons identified significant differences between the means in the ANOVA. Non-parametric tests such as the Mann-Whitney U-test or the Kruskal-Wallis test compared differences between groups where responses were not normally distributed. Two-way ANOVAs were used to compare the influence of two factors on continuous response variables. The maximum likelihood χ2 test determined significance in categorical data and was used to compare the substance use groups with the Control group. Spearman correlations measured correlations between ordinal/continuous response variables. A p-value <0.05 indicated statistical significance. The three prespecified confounding variables were used in multiple regression analyses with 11 groups of smoking, drinking, marijuana and methamphetamine combinations for each of the four outcome variables to determine their association and the underlying effect of substance use.

Ethics approval for the study was obtained from the Health Research Ethics Committee of Stellenbosch University (ref. nos N06/10/210 and S19/07/119), as well as from the Western Cape Department of Health. Participants were able to withdraw at any time during the study.

Results

The full cohort consisted of 4 926 pregnant women, of whom 877 (17.8%) used no drugs, cigarettes or alcohol (Control), 825 (16.7%) used no drugs but drank (NoDrugsDrink), 862 (17.5%) used no drugs but smoked (NoDrugsSmoke), 1 801 (36.6%) used cigarettes and alcohol (NoDrugsDrinkSmoke), 64 (1.3%) used marijuana and cigarettes (MarSmoke), 27 (0.5%) used methamphetamine and cigarettes (MetSmoke), 20 (0.4%) used marijuana and alcohol (MarDrink), 11 (0.2%) used methamphetamine and alcohol (MetDrink), 274 (5.6%) used marijuana, alcohol and cigarettes (MarDrinkSmoke), 88 (1.8%) used methamphetamine, alcohol and cigarettes (MetDrinkSmoke), and 77 (1.6%) used all four substances (All). This equated to 65% of women who smoked, 63% of women who drank, 9% of women who used marijuana and 4% of women who used methamphetamine. Excluded from this cohort were twin pregnancies, withdrawals, participants lost to follow-up, women who used marijuana or methamphetamine alone or had missing substance use data, and multiple enrolments. Only the first enrolment of a participant was included in this cohort. Preterm birth (<37 weeks) and very preterm birth (<32 weeks) occurred in 598 (12.1%) and 85 (1.7%) women, respectively. Of the total cohort (4 926 women) 65 women (1.3%) were HIV positive, 1 979 (40.2%) were anaemic, 8 (0.2%) had a miscarriage, 7 (0.1%) had a termination of pregnancy, 657 (13.3%) had low-birthweight (LBW) infants who weighed <2 500 g, 840 (17.1%) had small-for-gestational-age (SGA) infants who fell below the 10th birthweight centile, 44 (0.9%) had a stillbirth, and 45 (0.9%) had an infant death.

Information on the biometric measurements and socioeconomic conditions is provided in Table 1.

Table 1: Basic descriptive statistics of all participants

Variables

Valid N Mean Median Minimum Maximum Lower quartile Upper quartile

SD

Gestational age at enrolment (days)

4 926

142 141 38* 276 105 177 49

Maternal age (years)

4 926 24.4 23 16 45 20 28

6.0

Maternal arm circumference (mm)

4 838

276 267 175 535 241 303 46

Maternal body mass index (kg/m2)

4 787 25.6 24.2 13.7 55.9 21.2 28.9

5.8

Gravidity

4 916

2.1 2 1 10 1 3 1.3

Education (years)

4 919 10.1 10 2 13 9 12

1.7

Household income (ZAR/month)

3 500

886 750 45 6 000 500 1 200 607

GA at delivery (days)

4 926 272 275 61 313 267 282

18

Birthweight (g)

4 862

3 016 3 030 190 5 740 2 700 3 380 574

Birthweight z-score

4 847 -0.34 -0.37 -6.34 4.12 -1.04 0.33

1.03

Infant age at 1 year (days)

4 500

372 369 330 475§ 366 377 17

Infant weight at 1 year (kg)

4 490 9.4 9.3 5.3 16.9 8.5 10.3

1.4

Infant length at 1 year (cm)

4 408

73.7 73.7 60.7 88.0 71.8 75.6 3.0

Infant head circumference at 1 year (cm)

4 479 46.1 46.0 41.1 54.7 45.1 47.0

1.5

SD: Standard Deviation.
*Single case that deviated from required 6 weeks, but permission obtained to keep included.
Miscarriages included.
Initial time window minimum that was corrected later.
§Time window maximum adjusted for prematurity
The only excessively large value, not removed.

Table 2 summarises the maternal biometric measurements and socioeconomic conditions that were compared for the different substance use groups. Women in the Control group enrolled the earliest for antenatal care, had the largest MUAC and BMI, and also earned the highest mean income per month. Women in the MetSmoke group enrolled the latest, had the highest gravidity without being the oldest women, had the smallest mean MUAC, had the lowest average monthly income, and had the joint lowest education together with the MarSmoke and All groups. Women in the MarDrink group had the joint lowest gravidity and the highest education. Women in the MetDrink group were the oldest and had the joint highest gravidity. Women in the MarDrinkSmoke group were the youngest, had the joint lowest gravidity, had the lowest BMI, and were significantly the most anaemic.

Table 2: Biometric measurements and socioeconomic conditions compared in different substance use groups.

Variables F

p-value

Measure Substance use group
Control

(n=877)

NoDrugsDrink

(n=825)

NoDrugs Smoke

(n=862)

NoDrugsDrinkSmoke

(n=1 801)

MarSmoke

(n=64)

MetSmoke

(n=27)

MarDrink

(n=20)

MetDrink

(n=11)

MarDrinkSmoke

(n=274)

MetDrinkSmoke

(n=88)

All

(n=77)

Gestational age at enrolment (days) <0.01* Letters d cd cd c bcd a bcd abc cd b b
Mean 137 142 142 142 145 184 151 170 141 154 157
SD 49 48 50 49 52 47 49 42 45 53 46
Maternal age (years) <0.01* Letters a bc ab cd de abcd bcde abcd e bcd bcd
Mean 25.7 24.6 24.9 24.1 22.1 24.7 21.3 26.6 20.4 23.6 23.0
SD 6.2 5.8 6.2 5.8 5.8 4.7 4.8 4.7 4.4 4.4 4.7
Maternal arm circumference (mm) <0.01* Letters a ab bc c d d abcd abcd d cd cd
Mean 286 283 276 273 253 250 265 271 253 265 260
SD 51 49 47 43 32 26 39 31 37 39 42
Body mass index (kg/m2) <0.01* Letters a a b bc d bcd abcd abcd d cd cd
Mean 26.8 26.6 25.6 25.3 23.0 23.1 24.2 25.0 22.7 23.8 23.4
SD 6.2 6.1 5.8 5.5 4.0 3.8 4.6 3.1 4.1 4.3 5.0
Gravidity <0.01* Letters abce df ab cdef cdefg acd befg abcdefg g abcdef abcdef
Mean 2.2 1.9 2.4 2.1 1.8 2.8 1.5 2.8 1.5* 2.2 2.1
SD 1.2 1.2 1.4 1.2 1.1 1.4 1.1 1.6 0.9 1.4 1.2
Education (years) <0.01* Letters b a c c d d ab abcd d d d
Mean 10.5 10.7 9.9 9.9 9.1 9.1 10.8 9.9 9.4 9.4 9.1
SD 1.7 1.6 1.7 1.7 1.6 2.0 1.4 1.7 1.5 1.5 1.6
Household income (ZAR) <0.01* Letters a ab bc c cd abcd abcd abcd d cd d
Mean 997 987 880 844 639 566 902 699 636 720 573
SD 667 597 601 586 514 483 525 296 460 515 539
Anaemia with haemoglobin <11 g/dL Compared with Control N 345 319 350 725 26 12 12 5 127 30 28
% 39.3 38.7 40.6 40.3 40.6 44.4 60.0 45.5 46.4 34.1 36.4
χ2 p-value 0.776 0.590 0.649 0.839 0.593 0.062 0.680 0.039* 0.336 0.608

Mar: Marijuana; Met: Methamphetamine; SD: Standard Deviation;
Letters=significance lettering. If the significance lettering between 2 groups have common letters (e.g. b and bcd), the groups do not differ significantly.
*Significant at p<0.05 (F or χ2).
Smallest mean value.
Largest mean value.

Infant outcomes at birth and 1 year were compared in the different substance use groups and are summarised in Table 3. Infants from the Control group were heaviest at birth, had the largest BWZ, and were joint heaviest at 1 year. Infants from the NoDrugsSmoke group were significantly more premature, with more LBW and GR (SGA), and had more deaths compared with the Control group. Infants from the NoDrugsDrink group had the highest GA at birth and were joint heaviest at 1 year, whereas infants from the MarDrink group had the largest mean length and head circumference at 1 year. Infants from the MetDrink group had the lowest mean GA (<37 weeks) and more were premature; they had the lowest birthweight, and more were stillborn. Those alive at 1 year also had the lowest mean weight, lowest mean length and lowest mean head circumference, despite their adjusted age at 1 year. The MetSmoke group had the highest significant rate of infant deaths. Infants from the MarDrinkSmoke group had the lowest BWZ and compared with the Control group had the highest proportion who had LBW and were SGA.

Table 3: Infant outcome at birth and 1 year compared in different substance use groups

Variables F p-value Continuous data measure Substance use group
Control

(n=877)

NoDrugsDrink

(n=825)

NoDrugsSmoke

(n=862)

NoDrugsDrink

Smoke

(n=1 801)

MarSmoke

(n=64)

MetSmoke

(n=27)

MarDrink

(n=20)

MetDrink

(n=11)

MarDrinkSmoke

(n=274)

MetDrinkSmoke

(n=88)

All

(n=77)

Gestational age at delivery <0.01* Letters b a cd bc cd de abcd e bcd cd d
Mean 273 275 271 272 268 265 269 255 271 269 268
SD 20 15 18 18 19 12 29 23 16 11 15
Birthweight <0.01* Letters a a b b cd abcd abc d cd bcd cd
Mean 3 131 3 111 2 994 2 976 2 818 2 913 3 029 2 564 2 851 2 932 2 812
SD 585 536 596 567 536 453 503 772 566 463 531
Birthweight z-score <0.01* Letters a b b c cd abc abcd abcd d abc cd
Mean -0.14 -0.24 -0.32 -0.43 -0.64 -0.20 -0.44 -0.21 -0.66 -0.32 -0.61
SD 1.1 1.0 1.0 1.0 0.9 1.0 1.0 1.0 0.9 0.9 0.9
Infant age at 1 year 0.02* Letters c c ab abc b b abc abc ab abc ac
Mean 371 371 373 372 376 379 375 382 373 373 369*
SD 16 15 18 17 17 16 11 35 19 16 17
Infant weight at 1 year <0.01* Letters ab a cd c bcde e abcde e e de e
Mean 9.6 9.6 9.4 9.4 9.2 8.6 9.3 8.3 9.2 9.1 8.9
SD 1.4 1.4 1.3 1.4 1.2 1.1 1.3 0.9 1.3 1.5 1.2
Infant length at 1 year <0.01* Letters a a b b bc c ab abc c c c
Mean 74.2 74.2 73.7 73.6 73.2 72.3 74.6 72.1 73.0 72.7 72.5
SD 2.9 3.0 3.0 3.0 2.6 2.6 2.8 2.4 3.1 2.7 3.3
Infant head circumference at 1 year <0.01* Letters ab b ac c cd cd abc d cd abc cd
Mean 46.2 46.2 46.0 46.0 45.7 45.6 46.4 44.9 45.9 46.0 45.7
SD 1.5 1.5 1.4 1.5 1.3 1.4 1.5 1.4 1.5 1.4 1.5
Variables χ2 pvalue Categorical data measure Substance use group
Control

(n=877)

NoDrugsDrink

(n=825)

NoDrugsSmoke

(n=862)

NoDrugsDrinkSmoke

(n=1 801)

MarSmoke

(n=64)

MetSmoke

(n=27)

MarDrink

(n=20)

MetDrink

(n=11)

MarDrinkSmoke

(n=274)

MetDrinkSmoke

(n=88)

All

(n=77)

Preterm birth <37 weeks Compared with Control N 93 68 129 224 10 4 3 5 34 15 13
% 10.6 8.2 15.0 12.4 15.6 14.8 15.0 45.5§ 12.4 17.0 16.9
χ2 p-value 0.096 0.006* 0.168 0.214 0.486 0.530 <0.001* 0.405 0.068 0.093
Very preterm birth <32 weeks Compared with Control N 16 8 18 32 2 0 1 1 6 0 1
% 1.8 1.0 2.1 1.8 3.1 0.0 5.0 9.1 2.2 0.0 1.3
χ2 p-value 0.135 0.691 0.931 0.463 0.479 0.303 0.080 0.700 0.201 0.738
Low birthweight <2 500 g Compared with Control N 87 75 123 267 10 4 4 3 56 13 15
% 9.9 9.1 14.3 14.8 15.6 14.8 20.0 27.3 20.4§ 14.8 19.5
χ2 p-value 0.560 0.005* <0.001* 0.147 0.405 0.140 0.058 <0.001* 0.154 0.009*
Growth-restricted infant <10th centile Compared with Control N 116 106 146 350 13 3 4 1 69 13 19
% 13.2 12.8 16.9 19.4 20.3 11.1 20.0 9.1 25.2§ 14.8 24.7
χ2 p-value 0.817 0.031* <0.001* 0.112 0.749 0.379 0.687 <0.001* 0.685 0.006*
Miscarriage <20 weeks Compared with Control N 3 0 1 3 0 0 1 0 0 0 0
% 0.3 0.0 0.1 0.2 0.0 0.0 5.0 0.0 0.0 0.0 0.0
χ2 p-value 0.093 0.325 0.367 0.639 0.761 0.002* 0.846 0.332 0.583 0.607
Termination of pregnancy Compared with Control N 2 2 0 3 0 0 0 0 0 0 0
% 0.2 0.2 0.0 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0
χ2 p-value 0.951 0.161 0.729 0.702 0.804 0.831 0.874 0.429 0.654 0.675
Stillbirth Compared with Control N 8 4 6 17 1 0 0 1 2 2 3
% 0.9 0.5 0.7 0.9 1.6 0.0 0.0 9.1§ 0.7 2.3 3.9
χ2 p-value 0.292 0.614 0.936 0.606 0.618 0.668 0.007* 0.777 0.230 0.019*
Infant death Compared with Control N 3 1 14 20 0 2 0 0 2 2 1
% 0.3 0.1 1.6 1.1 0.0 7.4§ 0.0 0.0 0.7 2.3 1.3
χ2 p-value 0.347 0.007* 0.043* 0.639 <0.001* 0.793 0.846 0.394 0.016* 0.213

Mar: Marijuana; Met: Methamphetamine; SD: Standard Deviation;
Letters=significance lettering. If the significance lettering between 2 groups have common letters (e.g. b and bcd), the groups do not differ significantly.
*Significant at p<0.05 (F or χ2).
Smallest mean value.
Largest mean value.
§Highest significant rate.

The maternal measures of GA at enrolment, age, MUAC and education as found in 11 substance use groups are presented in Figures 1, 2, 3 and 4, respectively. The birth outcomes of GA at delivery, birthweight and BWZ in the different substance use groups are shown in Figures 5, 6 and 7, respectively. The 1-year visit outcomes of 1YW, 1YL and 1YHC in the different substance use groups are shown in Figures 8, 9 and 10, respectively.

Table 4 summarises the multiple regression results for BWZ. There was a positive association between BWZ and MUAC for all the groups that did not use drugs. The strongest associations were in the Control and the NoDrugsDrink groups, which also had the largest MUACs. BWZ was positively associated with education in only two groups, NoDrugsDrink and MarDrink, and these two groups also had the highest education. BWZ was negatively associated with income in the MetDrink group only. In this group, a higher income was associated with a lower BWZ, whereas a lower income was associated with a higher BWZ.

Table 4: Multiple regression summary for birth outcome variable birthweight z-score

Substance use group

n MUAC Education Income
 bz p-value  bz p-value  bz

p-value

Control

646

0.31 <0.001* -0.00 0.929 0.04 0.272

NoDrugsDrink

596 0.26 <0.001* 0.10 0.021* -0.06

0.151

NoDrugsSmoke

590

0.18 <0.001* 0.01 0.898 0.02 0.664

NoDrugsDrinkSmoke

1 208 0.16 <0.001* 0.05 0.072 0.04

0.152

MarSmoke

41

0.29 0.090 -0.07 0.683 0.15 0.392

MetSmoke

16 0.45 0.244 -0.30 0.354 -0.28

0.445

MarDrink

11

-0.17 0.614 1.04 0.046* -0.66 0.214

MetDrink

7 -0.05 0.865 0.62 0.058 -0.98

0.043*

MarDrinkSmoke

154

0.10 0.221 -0.02 0.788 0.04 0.683

MetDrinkSmoke

56 0.18 0.182 0.09 0.505 0.16

0.239

All

50

0.16 0.291 0.03 0.856 -0.04

0.830

MUAC: Mid-Upper Arm Circumference; Mar: Marijuana; Met: Methamphetamine.
*Significant at p<0.05; bz: Standardized Regression Coefficient.

Table 5 summarises the multiple regression results for 1YW. There was a positive association between infant weight at 1 year and MUAC for all the groups that did not use drugs, apart from the All group. Mothers in the All group had 4th-lowest MUAC, that was associated with the 3rd-lowest weight at 1 year. There was also a positive association between 1-year weight of infant and education of mother for the Control, NoDrugsDrink, NoDrugsSmoke, NoDrugsDrinkSmoke, MarDrinkSmoke and MetDrinkSmoke groups. There was a positive association between 1-year weight of infant and income of mother in the NoDrugsDrinkSmoke group. These mothers earned the 5th-highest income and had infants with the 3rd-largest weights at 1 year.

Table 5: Multiple regression summary for outcome variable infant weight at 1 year

Substance use group

n MUAC Education Income
bz p-value bz p-value bz

p-value

Control

608

0.14 <0.001* 0.12 0.003* 0.02 0.668

NoDrugsDrink

550 0.13 0.003* 0.15 0.001* -0.00

0.916

NoDrugsSmoke

538

0.11 0.013* 0.09 0.047* 0.09 0.060

NoDrugsDrinkSmoke

1 109 0.14 <0.001* 0.12 <0.001* 0.11

<0.001*

MarSmoke

36

0.21 0.236 0.13 0.489 0.23 0.225

MetSmoke

14 0.56 0.178 -0.34 0.313 0.15

0.726

MarDrink

10

-0.40 0.347 0.73 0.184 -0.47 0.426

MetDrink

6 0.42 0.692 0.15 0.837 0.66

0.564

MarDrinkSmoke

145

0.08 0.319 0.21 0.015* 0.12 0.180

MetDrinkSmoke

48 0.08 0.573 0.42 0.003* 0.09

0.515

ALL

43

0.35 0.029* -0.07 0.707 0.06

0.718

MUAC: Mid-Upper Arm Circumference; Mar: marijuana; Met: Methamphetamine.
*Significant at p<0.05; bz: Standardized Regression Coefficient.

Table 6 summarises the multiple regression results for 1YL. A positive association between infant length at 1 year and MUAC was only found for the Control and NoDrugsSmoke groups. The Control group had the largest MUACs, which was associated with the tallest infants at 1 year, whereas the NoDrugsSmoke group had significantly smaller MUACs and significantly shorter infants at 1 year when compared with the Control group. Infant length at 1 year was also positively associated with education of mothers in the Control, NoDrugsDrink, NoDrugsSmoke, NoDrugsDrinkSmoke and MetDrinkSmoke groups. Education was highest in the Control and NoDrugsDrink groups, with the tallest infants at 1 year, and lowest in smoking plus drug use groups, and these infants were also significantly shorter at 1 year, as seen in the MetDrinkSmoke group. There was a positive association between 1-year length of infant and income of mother for the NoDrugsSmoke and NoDrugsDrinkSmoke groups. Those who had a higher income in these groups had taller infants at 1 year.

Table 6: Multiple regression summary of outcome variable infant length at 1 year

Substance use group

n MUAC Education Income
bz p-value bz p-value bz

p-value

Control

592

0.13 0.001* 0.10 0.014* 0.02 0.557

NoDrugsDrink

545

0.06

0.144 0.18 <0.001* 0.07

0.095

NoDrugsSmoke

522

0.09 0.045* 0.10 0.035* 0.12 0.011*

NoDrugsDrinkSmoke

1 089 0.05 0.104 0.14 <0.001* 0.14

<0.001*

MarSmoke

35

-0.05 0.765 0.33 0.073 0.24 0.193

MetSmoke

14 0.44 0.313 -0.44 0.223 0.16

0.721

MarDrink

10

-0.24 0.582 0.78 0.181 -0.51 0.424

MetDrink

6 0.37 0.742 0.34 0.670 0.08

0.943

MarDrinkSmoke

143

0.08 0.362 0.16 0.058 0.16 0.075

MetDrinkSmoke

47 0.21 0.096 0.45 0.001* 0.18

0.159

All

42

0.19 0.233 0.29 0.113 -0.05

0.786

MUAC: Mid-Upper Arm Circumference; Mar: Marijuana; Met: Methamphetamine.
*Significant at p<0.05; bz: Standardized Regression Coefficient.

Table 7 summarises the multiple regression results for 1YHC. Infant head circumference at 1 year was not associated with MUAC, but was positively associated with maternal education for the Control, NoDrugsDrink, NoDrugsDrinkSmoke, MarDrinkSmoke and MetDrinkSmoke groups. Higher education was associated with larger head circumferences and lower education was associated with smaller head circumferences in these groups. In the NoDrugsDrinkSmoke group, head circumference of the infant at 1 year was positively associated with income. Those who had a higher income in this group also had infants with larger head circumference at 1 year.

Table 7: Multiple regression summary for infant head circumference outcome at 1 year

Substance use group

n

MUAC Education Income

bz

p-value

bz

p-value

bz

p-value

Control

604

0.07 0.077 0.10 0.013* 0.04 0.378

NoDrugsDrink

549 0.01 0.730 0.13 0.003* -0.01

0.823

NoDrugsSmoke

536

0.00 0.930 0.08 0.090 0.09 0.060

NoDrugsDrinkSmoke

1 105 0.03 0.391 0.11 0.001* 0.08

0.015*

MarSmoke

37

0.12 0.519 0.15 0.438 -0.11 0.576

MetSmoke

15 0.45 0.304 -0.30 0.391 -0.25

0.580

MarDrink

10

-0.01 0.977 0.80 0.181 -0.44 0.498

MetDrink

6 0.75 0.278 0.50 0.303 0.93

0.223

MarDrinkSmoke

145

-0.02 0.775 0.23 0.007* 0.07 0.404

MetDrinkSmoke

47 0.03 0.823 0.66 <0.001* 0.14

0.201

All

43

0.28 0.089 -0.05 0.801 -0.11

0.563

MUAC: Mid-Upper Arm Circumference; Mar: Marijuana; Met: Methamphetamine.
*Significant at p<0.05; bz: Standardized Regression Coefficient.

Discussion

Maternal Measures and Trends

We found a significant trend in the GA at enrolment, when women booked for antenatal care, from the earliest GA in women who took no substances to a later GA in those who used all substances, but the MetSmoke and MetDrink groups enrolled even later (Figure 1). The finding of McCalla et al. [36] that, although recreational drug users had a wide range of social problems that compromised fetal growth and development and were in greater need of prenatal care, they were less likely to make use of antenatal care services, supports our finding.

FIG 1

Figure 1: Gestational age at enrolment compared among different substance groups

There was also a trend in maternal age (Figure 2), with the oldest women in the Control group to the youngest in the All group, except for the MarSmoke, MarDrink and MarDrinkSmoke groups. Women who used marijuana were the youngest. Our finding that marijuana users are young is in agreement with other researchers [3,13,14].

FIG 2

Figure 2: Maternal age compared among different substance groups

The trend in MUAC (Figure 3), from no substance users to users of all substances, was significantly smaller MUACs, but MUACs were even smaller in the MarSmoke, MetSmoke and MarDrinkSmoke groups. Our finding that women who smoked, whether combined with drugs, alcohol or not, had significantly smaller MUACs, has been confirmed by two previous studies [26,35]. The reduced MUAC, associated with cigarette smoking and indicating poorer nutritional status, was associated with an increased risk of spontaneous preterm birth as well as a lower infant BWZ [26,35].

FIG 3

Figure 3: Maternal arm circumference compared among different substance groups

The trend in education (Figure 4) and income (Table 2) from Control to All was lower education and lower income with more substances used. Women who smoked, in any combination, all had significantly lower education when compared with the Control group or drinkers only. Numerous studies that have reported on the association of cigarette smoking with a lower level of education [37-41] and income [39-43] support our finding. Compared with the women in the Control group, women in the NoDrugsDrink group had a higher education, and women who drank combined with marijuana or methamphetamine, but did not smoke, did not differ significantly. Woman in the NoDrugsDrink, MarDrink and Control groups had the highest mean education, ranging from 10.5 to 10.8 years. This finding is validated by research by Patrick et al., [40] who reported that young adults with the highest family education and income were most prone to alcohol and marijuana use, and by Rees [44], who found little evidence that drinking affected educational attainment.

FIG 4

Figure 4: Education compared among different substance groups

Birth Outcomes and Trends

Gestation at delivery declined as the number of substances increased, although this did not apply to alcohol use alone. Compared with the Control group, GA at delivery was significantly lower for methamphetamine users and for smoking on its own or in combination with marijuana, while it was significantly higher for the NoDrugsDrink group (Figure 5), with the highest mean GA of 39 weeks and 2 days. There was no significant difference between the Control and NoDrugsDrinkSmoke, MarDrink or MarDrinkSmoke groups. Our previous study also found that alcohol use alone was associated with a higher GA, while alcohol seemed to counteract the negative association of smoking with GA [26], and lends support to our findings. The highest significant difference in GA was found when we compared the NoDrugsDrink group (highest GA) with the MetDrink group (lowest GA). This suggests a combined effect of methamphetamine and alcohol on GA. Not only did the MetDrink group have the most preterm births, but it also had the highest significant rate of stillbirths, despite being such a small group. Our results endorse the findings by other researchers that methamphetamine was associated with a lower GA at birth [9,45-47] and with preterm birth [46-48]. However, according to England et al., [49] little is known about the co-use of other substances by women who drink during pregnancy. It appears that the combined effect of methamphetamine and alcohol on GA has not been reported previously. It is interesting that Sowell et al. [50] found that brain morphology was affected in children with prenatal methamphetamine and alcohol exposure above and beyond the effects of alcohol exposure alone, suggesting a synergistic effect between methamphetamine and alcohol.

FIG 5

Figure 5: Gestational age at delivery compared among different substance groups

The trend in birthweight from Control to All was lower birthweight with more substances used (Figure 6). Okah et al. [27] reported that women with alcohol and/or drug use during pregnancy did not appear to be at greater risk of giving birth to a term LBW infant than women who reported abstinence. However, the addition of smoking to either behaviour produces placental vasoconstriction that will decrease oxygen delivery to the fetus, limit fetal growth [51], and increase the risk of LBW by 2-to 4-fold. Gibson et al. [52] found that infants born to smokers had lower birthweights and were more prone to GR. These reports support our findings of significantly more infants with LBW in the smoking groups (NoDrugsSmoke, NoDrugsDrinkSmoke, MarDrinkSmoke and All) and of the non-smoking groups (all but one) being the only groups with a mean birthweight >3 000 g (Table 3). The MetDrink group, being the exception, had the lowest mean birthweight and also the lowest mean GA at delivery (<37 weeks), with 45.5% of infants being preterm. Many researchers have found that methamphetamine was associated with lower birthweight [10,45,53,54], and Black et al. [55] found antenatal drug use to increase the risk of LBW infants above that related to cigarette smoking. Odendaal et al. [56] and Jackson et al. [57] reported that the combined use of cigarettes and alcohol during pregnancy had a synergistic effect for LBW and GR, which also concurs with our findings.

FIG 6

Figure 6: Birthweight compared among different substance groups

The trend in BWZ from Control to All was lower BWZs with more substances used. The lowest BWZs were associated with marijuana and smoking, but not methamphetamine (Figure 7). Significant GR was detected in the infants from the smoking groups (NoDrugsSmoke, NoDrugsDrinkSmoke, MarDrinkSmoke, and All), with >25% of the MarDrinkSmoke group being affected. El Marroun et al. [58] reported that marijuana use during pregnancy resulted in more pronounced GR than tobacco use, while Sturrock et al. [59] also found that cigarette smoking was associated with a lower BWZ, but that women who both smoked and used marijuana during pregnancy had infants with a lower BWZ than those who used cigarettes alone. Spinillo et al. [60] reported on fetal GR among women who smoked throughout pregnancy, while Hayatbakhsh et al. [61], after controlling for smoking, alcohol consumption and other drugs, showed that marijuana use in pregnancy was associated with SGA infants with lower BWZs. The abovementioned researchers all validate our findings.

FIG 7

Figure 7: Birthweight Z-score compared among different substance groups

One-year Outcomes and Trends

The trend in infant weight from Control to All was lower infant weight at 1 year with more substances used (Figure 8). The lowest weights were in the methamphetamine-using groups, especially the MetDrink group, which had the lowest mean weight, with the most preterm births and infant ages adjusted for prematurity, and the MetSmoke group. In previous studies, weight and growth were reported as significantly decreased in methamphetamine-exposed children at ages 1-4 years [54,62,63], which endorses our results.

FIG 8

Figure 8: Infant one-year weight compared among different substance groups

The trend in infant length from Control to All was shorter infant length at 1 year with more substances used (Figure 9). Smoking only, or smoking combined with drugs and/or alcohol, was associated with significantly shorter infants at 1 year. Many studies have shown a long-term negative effect of maternal smoking during pregnancy on height of infants, from birth to adolescence [64-70], which supports our finding. Zabaneh et al. [71], Smith et al. [63] and Eriksson et al. [62] reported decreased height velocity throughout the first 3 years of life in methamphetamine-exposed children, corroborating our findings that infants from the MetDrink and MetSmoke groups, although adjusted for prematurity, had the shortest and second-shortest mean length at 1 year, respectively (Table 3).

FIG 9

Figure 9: Infant one-year length compared among different substance groups

The trend in infant head circumference from Control to All was a smaller infant head circumference at 1 year with more substances used. The smallest head circumferences were in the MetDrink group, despite adjustment for prematurity (Figure 10). Other researchers have found that infants prenatally exposed to methamphetamine tended to show a significantly smaller head circumference at birth or 1 year [54,62,72,73], supporting our findings.

FIG 10

Figure 10: Infant one-year head circumference compared among different substance groups

Effects of Combined Drug Use, Smoking and Drinking on Maternal Measures, Birth and 1-year Outcome

Many significant differences were found when the MarDrinkSmoke and MetDrinkSmoke groups, who used three substances, were compared with the Control group. Women using three substances (methamphetamine or marijuana with smoking and drinking) were younger, had a smaller MUAC, lower education and smaller income, and had infants with lower birthweight, 1-year weight and 1-year height than those from the Control group. These results are supported by the findings of other researchers [2,13]. Although polysubstance use in pregnancy is common [74], there is little information available, and the full range of substance combinations and their health impacts remain incompletely understood [75]. Alcohol, tobacco and drug co-use during pregnancy is particularly problematic and compounds the adverse effects on fetal growth [55,75,76].

Women in the methamphetamine three-substance (MetDrinkSmoke) group enrolled much later and had a lower GA at birth than Controls. They were also older than marijuana users but younger than abstainers. Smith et al. [2] found that infants exposed to methamphetamine or tobacco during pregnancy were 3.5 times or 2 times more likely, respectively, to be SGA compared with unexposed infants, suggesting more GR if the infant was exposed to methamphetamine and smoking. GR together with our finding of lower GA in the MetDrinkSmoke group (17% preterm births, which was second highest after the 45.5% in the MetDrink group) supports the association of methamphetamine with preterm birth.

Women in the marijuana three-substance (MarDrinkSmoke) group were much younger (also younger than methamphetamine users), had lower gravidity, were significantly more anaemic, had infants with a lower BWZ and smaller head circumference, and had more LBW and SGA infants when compared with the Control group. Interestingly, Chabarria et al. [13] and Grzeskowiak et al. [77] reported decreased head circumference at birth to be associated with maternal marijuana use combined with smoking, or independent of tobacco use, respectively. This may help explain the association found between MarDrinkSmoke and smaller head circumference of infants at 1 year in our study. Although we agree with others that marijuana use in pregnancy is harmful to the fetus in that it was associated with low infant birthweight [3,13,77] and SGA infants [14,78,79], our findings support those of Conner et al. [12] and Forray et al. [74], who reported that the association between maternal marijuana use and adverse outcomes appears to be attributable to comorbid substance use. Our findings are consistent with many reports of marijuana users being younger [75], of lower parity, better educated, and more likely to use alcohol, cigarettes and hard drugs [3,13,14]. However, we found no direct association between marijuana use and spontaneous preterm birth, as others have reported [13,14].

Confounders

Our finding that a larger MUAC, indicative of better nutritional status, was associated with a higher BWZ was supported by Smith et al. [2], who found that lower maternal weight gain during pregnancy was more likely to result in an SGA infant. A larger MUAC was also associated with a taller, heavier infant at 1 year.

Higher education was positively associated with outcomes at birth (BWZ) and all outcomes at 1 year, resulting in a larger infant who weighed more, was taller and had a larger head circumference. Numerous researchers have reported a strong inverse relationship between education and cigarette smoking [37-41,80] and drug use [81,82]. By decreasing substance use, academic outcomes may improve, and therefore also birth and 1-year outcomes.

Higher income was associated with a lower BWZ, perhaps suggesting more methamphetamine and alcohol use while pregnant, but was also associated with a larger infant at 1 year who weighed more, was taller and had a larger head circumference.

Study Strengths and Limitations

SPS was a unique, large study performed in population groups with similar socioeconomic circumstances and known to have a high incidence of antenatal substance use. A wealth of maternal, fetal and infant data were collected prospectively over a 9-year period. Substance use exposure data were collected on up to four occasions throughout pregnancy, and infant assessments were done at up to three time points throughout the first year of life. All measurements were taken twice, and we used validated recognised instruments and adjusted 1-year infant age for prematurity.

Limitations include that despite this being a large study with a high incidence of substance use, the small numbers in certain substance use groups limit the strength of the findings. Substance use was self-reported and may therefore be under-reported. Although we have detailed smoking and drinking exposure continuous data, drug information was not quantified, limiting us to nominal (yes or no) data for the various substances used.

Conclusion

The tragedy of substance use during pregnancy not only affects maternal and fetal health during pregnancy, but also infant growth and wellbeing at 1 year of age. Given that these substances are modifiable risk factors [28], and that detailed information on the preventable adverse effects of smoking and drinking during pregnancy was not effective in the population studied [83], it is clearly a major public health problem. The co-use of methamphetamine and alcohol (smallest group) seemed to have a confounding negative association with infant birth and 1-year outcomes, but reasons for this remain unknown. The addictive properties of substance use make cessation difficult, so prevention strategies should rather be addressed. As the prevalence of tobacco use among 13-15-year-old females in SA was 20% in 2002 [21], prevention strategies should be implemented long before pregnancy in order to limit the uptake of addictive substance use among young women. Higher maternal education, associated with better infant outcomes at birth and 1 year and acting as a countermeasure to substance use, is of paramount importance.

Acknowledgements

We wish to thank the South African Journal for permission to publish this manuscript (S Afr Med J 2022; 112(8) 526-538). We thank the personnel of the SA arm of the SPS for their outstanding work, which included the recruitment of 7 060 pregnant women and the collection and capturing of valuable information at up to seven assessment time points per participant.

Author Contributions

Concept and Design: LB, HO; Acquisition, Statistical analysis, or Interpretation of data: LB, PS, DN, MP, HO; Drafting of the manuscript: LB, HO; Editing, Revising, or Proofreading of manuscript: LB, PS, DN, MP, HO. Authors have nothing to declare.

Funding

The study was funded by the National Institute on Alcohol Abuse and Alcoholism, Eunice Kennedy Shriver National Institute of Child Health and Human Development, and National Institute on Deafness and Other Communication Disorders (ref. nos U01 HD055154, U01 HD045935, U01 HD055155, U01 HD045991 and U01 AA016501. The funding body had no role in conducting the research or writing the article.

References

  1. Kuczkowski KM (2007) The effects of drug abuse on pregnancy. Curr Opin Obstet Gynecol 19: 578-585. [crossref]
  2. Smith LM, LaGasse LL, Derauf C, Grant P, Shah R, et al. (2006) The infant development, environment, and lifestyle study: Effects of prenatal methamphetamine exposure, polydrug exposure, and poverty on intrauterine growth. Pediatrics 118: 1149-1156. [crossref]
  3. Fergusson DM, Horwood LJ, Northstone K (2002) Maternal use of cannabis and pregnancy outcome. Int J Obstet Gynaecol 109: 21-27. [crossref]
  4. Terplan M, Smith EJ, Kozloski MJ, Pollack HA (2009) Methamphetamine use among pregnant women. Obstet Gynecol 113: 1285-1291. [crossref]
  5. Schempf AH, Strobino DM (2008) Illicit drug use and adverse birth outcomes: Is it drugs or context? J Urban Health 85: 858-873. [crossref]
  6. Godleski SA, Shisler S, Eiden RD, Huestis MA (2018) Co-use of tobacco and marijuana during pregnancy: Pathways to externalizing behavior problems in early childhood. Neurotoxicol Teratol 69: 39-48. [crossref]
  7. Petersen Williams P, Jordaan E, Mathews C, Lombard C, Parry CD (2014) Alcohol and other drug use during pregnancy among women attending midwife obstetric units in the Cape Metropole, South Africa. Adv Prev Med 2014: [crossref]
  8. Odendaal H, Geerts L, Nel D, Brink LT, Hitchcock E, et al. (2018) Effects of alcohol, cigarettes, methamphetamine and marijuana exposure during pregnancy on maternal serum alpha-fetoprotein levels at 20-24 weeks’ gestation. J Pediatr Neonatal Care 8: 73-82. [crossref]
  9. Kalaitzopoulos DR, Chatzistergiou K, Amylidi AL, Kokkinidis, Goulis DG (2018) Effect of methamphetamine hydrochloride on pregnancy outcome: A systematic review and meta-analysis. J Addict Med 12: 220-226. [crossref]
  10. Gabrhelík R, Skurtveit S, Nechanská B, Handal M, Mahic M, et al. (2021) Prenatal methamphetamine exposure and adverse neonatal outcomes: A nationwide cohort study. Eur Addict Res 27: 97-106. [crossref]
  11. Bada HS, Das A, Bauer CR, Shankaran S, Lester BM, et al. (2005) Low birth weight and preterm births: Etiologic fraction attributable to prenatal drug exposure. J Perinatol 25: 631-637. [crossref]
  12. Conner SN, Bedell V, Lipsey K, Macones GA, Cahill AG, et al. (2016) Maternal marijuana use and adverse neonatal outcomes: A systematic review and meta-analysis. Obstet Gynecol 128: 713-723. [crossref]
  13. Chabarria KC, Racusin DA, Antony KM, Kahr M, Suter MA, et al. (2016) Marijuana use and its effects in pregnancy. Am J Obstet Gynecol 215: 506.e1-506.e7. [crossref]
  14. Luke S, Hutcheon J, Kendall T (2019) Cannabis use in pregnancy in British Columbia and selected birth outcomes. J Obstet Gynaecol Can 41: 1311-1317. [crossref]
  15. Metz TD, Borgelt LM (2018) Marijuana use in pregnancy and while breastfeeding. Obstet Gynecol 132: 1198-1210. [crossref]
  16. Joseph KS, Liston RM, Dodds L, Dahlgren L, Allen AC (2007) Socioeconomic status and perinatal outcomes in a setting with universal access to essential health care services. Can Med Assoc J 177: 583-590. [crossref]
  17. Mund M, Louwen F, Klingelhoefer D, Gerber A (2013) Smoking and pregnancy-a review on the first major environmental risk factor of the unborn. Int J Environ Res Public Health 10: 6485-6499. [crossref]
  18. Chamberlain C, O’Mara-Eves A, Porter J, Coleman T, Perlen SM, et al. (2017) Psychosocial interventions for supporting women to stop smoking in pregnancy. Cochrane Database Syst Rev 2: CD001055. [crossref]
  19. Salihu HM, Wilson RE (2007) Epidemiology of prenatal smoking and perinatal outcomes. Early Hum Dev 83: 713-720. [crossref]
  20. Baba S, Wikström AK, Stephansson O, et al. (2012) Influence of smoking and snuff cessation on risk of preterm birth. Eur J Epidemiol 27: 297-304.
  21. World Health Organization. Report on the Global Tobacco Epidemic, 2008: The MPOWER Package. MPOWER Packag., 1-342.
  22. Wehby G L, Prater K, McCarthy AM, Castilla EE, Murray JC, et (2011) The impact of maternal smoking during pregnancy on early child neurodevelopment. J Hum Cap 5: 207-254. [crossref]
  23. Banderali G, Martelli A, Landi M, Moretti F, Betti F, et al. (2015) Short and long term health effects of parental tobacco smoking during pregnancy and lactation: A descriptive review. J Transl Med 13: 1-7. [crossref]
  24. Shuffrey LC, Myers MM, Isler JR, Lucchini M, Sania A, et al. (2020) Association between prenatal exposure to alcohol and tobacco and neonatal brain activity: Results from the Safe Passage Study. JAMA Netw Open 3: e204714. [crossref]
  25. Abraham M, Alramadhan S, Iniguez C, Duijts L, Jaddoe VW, et al. (2017) A systematic review of maternal smoking during pregnancy and fetal measurements with meta-analysis. PLoS ONE 12: e0170946. [crossref]
  26. Brink LT, Nel DG, Hall DR, Odendaal HJ (2021) The intricate interactions between maternal smoking and drinking during pregnancy and birthweight Z-scores of preterm births. J Womens Health Care Manag 2: 47275/2692-0948-121. [crossref]
  27. Okah FA, Cai J, Hoff GL (2005) Term-gestation low birth weight and health-compromising behaviors during pregnancy. Obstet Gynecol 105: 543-550. [crossref]
  28. Elliott AJ, Kinney HC, Haynes RL, Dempers JD, Wright C, et al. (2020) Concurrent prenatal drinking and smoking increases risk for SIDS: Safe Passage Study report. EClinMed 19: [crossref]
  29. Dukes KA, Burd L, Elliott AJ, Fifer WP, Folkerth RD, et al. (2014) The Safe Passage Study: Design, methods, recruitment, and follow-up approach. Paediatr Perinat Epidemiol 28: 455-465. [crossref]
  30. Dukes K, Tripp T, Petersen J, Robinson F, Odendaal H, et al. (2017) A modified Timeline Followback assessment to capture alcohol exposure in pregnant women: Application in the Safe Passage Study. Alcohol 62: 17-27. [crossref]
  31. Odendaal HJ, Elliott A, Kinney HC, Human M, Gaspar D, et al. (2011) Consent for autopsy research for unexpected death in early life. Obstet Gynecol 117: 167-171. [crossref]
  32. Odendaal H, Wright C, Brink L, Schubert P, Geldenhuys E, et al. (2019) Association of late second trimester miscarriages with placental histology and autopsy findings. Eur J Obstet Gynecol Reprod Biol 243: 32-35. [crossref]
  33. Odendaal H, Dukes KA, Elliott AJ, Willinger M, Sullivan LM, et al. (2021) Association of prenatal exposure to maternal drinking and smoking with the risk of stillbirth. JAMA Netw Open 4: e2121726. [crossref]
  34. Villar J, Papageorghiou AT, Pang R, Ohuma EO, Cheikh Ismail L, et al. (2014) The likeness of fetal growth and newborn size across non-isolated populations in the INTERGROWTH-21st project: The fetal growth longitudinal study and newborn cross-sectional study. Lancet Diabetes Endocrinol 2: 781-792. [crossref]
  35. Brink LT, Nel DG, Hall DR, Odendaal HJ (2020) Association of socioeconomic status and clinical and demographic conditions with the prevalence of preterm birth. Int J Gynaecol Obstet 149: 359-369. [crossref]
  36. McCalla S, Minkoff HL, Feldman J, et al. (1991) The biologic and social consequences of perinatal cocaine use in an inner-city population: Results of an anonymous cross-sectional study. Am J Obstet Gynecol 164: 625-630.
  37. Gilman SE, Martin LT, Abrams DB, et al. (2008) Educational attainment and cigarette smoking: A causal association? Int J Epidemiol 37: 615-624.
  38. Dhavan R, Melissa S H, Perry CL, Arora M, Reddy KS (2010) Is tobacco use associated with academic failure among government school students in urban India? J Sch Health 80: 552-560. [crossref]
  39. Escobedo LG, Peddicord JP (1996) Smoking prevalence in US birth cohorts: The influence of gender and education. Am J Public Health 86: 231-236. [crossref]
  40. Patrick ME, Wightman P, Schoeni RF, Schulenberg JE (2012) Socioeconomic status and substance use among young adults: A comparison across constructs and drugs. J Stud Alcohol Drugs 73: 772-782. [crossref]
  41. Zhu BP, Giovino GA, Mowery PD, Eriksen MP (1996) The relationship between cigarette smoking and education revisited: Implications for categorizing persons’ educational status. Am J Public Health 86: 1582-1589. [crossref]
  42. Farmer S, Hanratty B (2012) The relationship between subjective wellbeing, low income and substance use among schoolchildren in the north west of England: A cross-sectional study. J Public Health (Oxf) 34: 512-522. [crossref]
  43. Addiction Center. Addiction and Low-income Americans. (accessed June 2022). https: //www.addictioncenter.com/addiction/low-income-americans/
  44. Rees DI (2019) Does substance use affect academic performance? IZA World Labor 1-10.
  45. Wright TE, Schuetter R, Tellei J, Sauvage L (2015) Methamphetamines and pregnancy outcomes. J Addict Med 9: 111-117. [crossref]
  46. Premchit S, Orungrote N, Prommas S, Smanchat B, Bhamarapravatana K, et al. (2021) Maternal and neonatal complications of methamphetamine use during pregnancy. Obstet Gynecol Int 2021: 8814168. [crossref]
  47. Phupong V, Darojn D (2007) Amphetamine abuse in pregnancy: The impact on obstetric outcome. Arch Gynecol Obstet 276: 167-170.
  48. Gorman M, Orme K, Nguyen N, Kent EJ 3rd, Caughey AB (2014) Outcomes in pregnancies complicated by methamphetamine use: A retrospective cohort study. Am J Obstet Gynecol 210: S309. [crossref]
  49. England LJ, Bennett C, Denny CH, Honein MA, Gilboa SM, et al. (2020) Alcohol use and co-use of other substances among pregnant females aged 12-44 years-United States, 2015-2018. MMWR Morb Mortal Wkly Rep 69: 1009-1014. [crossref]
  50. Sowell ER, Leow AD, Bookheimer SY, Smith LM, O’Connor MJ, et al. (2010) Differentiating prenatal exposure to methamphetamine and alcohol versus alcohol and not methamphetamine using tensor-based brain morphometry and discriminant analysis. J Neurosci 30: 3876-3885. [crossref]
  51. Carter RC, Wainwright H, Molteno CD, Georgieff MK, Dodge NC, et al. (2016) Alcohol, methamphetamine, and marijuana exposure have distinct effects on the human placenta. Alcohol Clin Exp Res 40: 753-764. [crossref]
  52. Gibson G, Baghurst P, Colley D (1983) Maternal alcohol, tobacco and cannabis consumption and the outcome of pregnancy. Aust N Z J Obstet Gynaecol 23: 15-19. [crossref]
  53. Smith LM, Yonekura ML, Wallace T, Berman N, Kuo J, et al. (2003) Effects of prenatal methamphetamine exposure on fetal growth and drug withdrawal symptoms in infants born at term. J Dev Behav Pediatr 24: 17-23. [crossref]
  54. Wouldes TA, LaGasse LL, Huestis MA Dellagrotta S, Dansereau LM, et al. (2014) Prenatal methamphetamine exposure and neurodevelopmental outcomes in children from 1 to 3 years. Neurotoxicol Teratol 42: 77-84. [crossref]
  55. Black M, Bhattacharya S, Fairley T, Campbell DM, Shetty A (2013) Outcomes of pregnancy in women using illegal drugs and in women who smoke cigarettes. Acta Obstet Gynecol Scand 92: 47-52. [crossref]
  56. Odendaal HJ, Steyn DW, Elliott A, Burd L (2009) Combined effects of cigarette smoking and alcohol consumption on perinatal outcome. Gynecol Obstet Invest 67: 1-8. [crossref]
  57. Jackson DJ, Batiste E, Rendall-Mkosi K (2007) Effect of smoking and alcohol use during pregnancy on the occurrence of low birthweight in a farming region in South Africa. Paediatr Perinat Epidemiol 21: 432-440. [crossref]
  58. El Marroun H, Tiemeier H, Steegers EAP, Jaddoe VW, Hofman A, et al. (2009) Intrauterine cannabis exposure affects fetal growth trajectories: The Generation R Study. J Am Acad Child Adolesc Psychiatry 48: 1173-1181. [crossref]
  59. Sturrock S, Williams E, Ambulkar H, Dassios T, Greenough A (2020) Maternal smoking and cannabis use during pregnancy and infant outcomes. J Perinat Med 48: 168-172. [crossref]
  60. Spinillo A, Capuzzo E, Piazzi G, Baltaro F, Iasci A, et al. (1996) Effect measures for behavioral factors adversely affecting fetal growth. Am J Perinatol 13: 119-123. [crossref]
  61. Hayatbakhsh MR, Flenady VJ, Gibbons KS, Kingsbury AM, et al. (2012) Birth outcomes associated with cannabis use before and during pregnancy. Pediatr Res 71: 215-219. [crossref]
  62. Eriksson M, Jonsson B, Steneroth G, Zetterström R (1994) Cross-sectional growth of children whose mothers abused amphetamines during pregnancy. Acta Paediatr 83: 612-617. [crossref]
  63. Smith LM, Diaz S, LaGasse LL, Wouldes T, Derauf C, et al. (2015) Developmental and behavioral consequences of prenatal methamphetamine exposure: A review of the Infant Development, Environment, and Lifestyle (IDEAL) Study. Neurotoxicol Teratol 51: 35-44. [crossref]
  64. Muraro AP, Gonçalves-Silva RMV, Moreira NF, Ferreira MG, Nunes-Freitas AL, et al. (2014) Effect of tobacco smoke exposure during pregnancy and preschool age on growth from birth to adolescence: A cohort study. BMC Pediatr 14: 1-9. [crossref]
  65. Karvonen M, Saari A, Sund R, Sankilampi U (2021) Maternal smoking during pregnancy and offspring head growth in comparison to height and weight growth up to 6 years of age: A longitudinal study. Clin Epidemiol 13: 959-970. [crossref]
  66. Matijasevich A, Brion MJ, Menezes AM, Barros AJ, Santos IS, et al. (2011) Maternal smoking during pregnancy and offspring growth in childhood: 1993 and 2004 Pelotas cohort studies. Arch Dis Child 96: 519-525. [crossref]
  67. Howe LD, Matijasevich A, Tilling K, Brion MJ, Leary SD, et al. (2012) Maternal smoking during pregnancy and offspring trajectories of height and adiposity: Comparing maternal and paternal associations. Int J Epidemiol 41: 722-732. [crossref]
  68. Leary S, Smith GD, Ness A (2006) Smoking during pregnancy and components of stature in offspring. Am J Hum Biol 18: 502-512. [crossref]
  69. Koshy G, Delpisheh A, Brabin BJ (2011) Dose response association of pregnancy cigarette smoke exposure, childhood stature, overweight and obesity. Eur J Public Health 21: 286-291. [crossref]
  70. Gigante DP, Horta BL, Lima RC, Barros FC, Victora CG (2006) Early life factors are determinants of female height at age 19 years in a population-based birth cohort (Pelotas, Brazil). J Nutr 136: 473-478. [crossref]
  71. Zabaneh R, Smith LM, Lagasse LL, Derauf C, Newman E, et al. (2012) The effects of prenatal methamphetamine exposure on childhood growth patterns from birth to 3 years of age. Am J Perinatol 29: 203-210. [crossref]
  72. LaGasse LL, Wouldes T, Newman E, Smith LM, Shah RZ, et al. (2011) Prenatal methamphetamine exposure and neonatal neurobehavioral outcome in the USA and New Zealand. Neurotoxicol Teratol 33: 166-175. [crossref]
  73. Zhang Y, Gong F, Liu P, He Y, Wang H (2021) Effects of prenatal methamphetamine exposure on birth outcomes, brain structure, and neurodevelopmental outcomes. Dev Neurosci 43: 271-280. [crossref]
  74. Forray A. Substance use during pregnancy [version 1; referees: 2 approved]. F1000Research,2016,5(F1000 Faculty Rev): 887.
  75. Qato DM, Zhang C, Gandhi AB, Coleman-Cowger VH, Simoni-Wastila L (2020) Co-use of alcohol, tobacco, and licit and illicit controlled substances among pregnant and non-pregnant women in the United States: Findings from 2006 to 2014 National Survey on Drug Use and Health (NSDUH) data. Drug Alcohol Depend 206: 107729. [crossref]
  76. Cornelius MD, Day NL (2000) The effects of tobacco use during and after pregnancy on exposed children. Alcohol Res Health 24: 242-249. [crossref]
  77. Grzeskowiak LE, Grieger JA, Andraweera P, Knight EJ, Leemaqz S, et al. (2020) The deleterious effects of cannabis during pregnancy on neonatal outcomes. Med J Aust 212: 519-524. [crossref]
  78. Janisse JJ, Bailey BA, Ager J, Sokol RJ (2014) Alcohol, tobacco, cocaine, and marijuana use: Relative contributions to preterm delivery and fetal growth restriction. Subst Abus 35: 60-67. [crossref]
  79. Hatch EE, Bracken MB (1986) Effect of marijuana use in pregnancy on fetal growth. Am J Epidemiol 124: 986-993.
  80. Wagenknecht L, Perkins L, Cutter G, Sidney S, Burke GL, et al. (1990) Cigarette smoking behavior is strongly related to educational status: The CARDIA study. Prev Med 19: 158-169. [crossref]
  81. Fothergill KE, Ensminger ME, Green KM, Crum RM, Robertson J, et al. (2008) The impact of early school behavior and educational achievement on adult drug use disorders: A prospective study. Drug Alcohol Depend 92: 191-199. [crossref]
  82. King KM, Meehan BT, Trim RS, Chassin L (2006) Substance use and academic outcomes: Synthesizing findings and future directions. Addiction 101: 1688-1689. [crossref]
  83. Odendaal HJ, Brink LT, Nel DG, Carstens E, De Jager M, et al. (2020) Smoking and drinking habits of women in subsequent pregnancies after specific advice about the dangers of these exposures during pregnancy. S Afr Med J 110: 1100-1104. [crossref]

The Function of Supercritical Fluids for the Solvus Formation and Enrichment of Critical Elements

DOI: 10.31038/GEMS.2023582

Abstract

The opening of a solvus curve around their critical point starts with a singularity. At this singularity, an element distribution forms, described by a Lorentzian-type curve, showing the highest concentration of some trace and primary elements at the critical point, corresponding to the most elevated temperatures, and with cooling, this Lorentzian curve opens and snuggles up to the solvus curve. This combination is strong proof of supercritical transition to critical and under-critical conditions. Furthermore, we show here that remnants of an older mineralization are present, demonstrating that the current picture must not be correct. Supercritical fluids can have a significant influence on mineralization as a whole.

Keywords

Supercritical fluids, Element enrichment, Solvus- and Lorentzian-curves, Tin deposit Ehrenfriedersdorf

Introduction

Since the beginning of the studies on melt inclusion in pegmatite quartz from the tin deposit Ehrenfriedersdorf, particularly on the pegmatites from the Sauberg mine, in the years around the turn of the century, we often found very water-rich melt inclusions. Water content and temperature appear to be a characteristic relationship from the beginning: a characteristic solvus curve [1]. After that, such curves were also found for many other pegmatites and evolved granites worldwide (Figure 1) [2].

fig 1

Figure 1: Pseudo-binary solvus curve for 19 different evolved granites and pegmatites worldwide. Note: each point represents the arithmetic mean of measurement on up to 100 melt inclusions. Values at T/TC (in °C)=1.0 correspond to the solvus curve’s critical point. The abscissa (analytically determined water content) first approximates the melt density. Note here that the point scattering is, in a first approximation, the result of the complex interaction of volatiles (H2O, F, H3BO3), which sometimes work additively.

For the origin of such curves, a clear answer could not given. At this time, the necessary analytical technique was still in it’s infancy-however, the evidence of the characteristic relationship between water content and temperature increases significantly year to year. Applying reduced parameters (T/TC) displays relatively good comparability for different granite and pegmatite systems (Figure 1) see also Figure 2 in Thomas and Davidson, 2015) [3]. So, a more universal relationship is probably. However, demonstrating such curves was the first step to solving this puzzle because a deeper origin is behind the solvus curves (temperature versus water concentration). Here, we explicitly use the water concentration because the density decreases and is not steady with the increase in the melt’s water content (especially at or near the solvus crest). We have shown [4] that at the solvus crest, the concentration of volatiles can obtain extreme values far away from the solvus crest water concentration (including F, B, C, P, and high concentrations of alkali elements, Be, Sn, and others.

fig 2

Figure 2: Lorentzian distribution of P versus H2O concentration of water-rich melt inclusions in pegmatite quartz from the Sauberg mine, Ehrenfriedersdorf. Note: each point represents the data mean on 5 to 10 melt inclusions.

Key Observations

Starting with the work on the enrichment of Be in granite-pegmatite systems in 2011 [5], it would, from case to case, be evident that the highest element enrichment is related to the solvus crest-the start point for the opening the solvus with the drop in temperature. According to our findings of diagnostic HP-HT minerals from mantle depths transported by supercritical fluids, it would be more evident that the solvus crest is the locus where the supercritical fluid changes into the critical/under critical state. Therefore, here, we mainly found the highest enrichment of the most scarce elements (Be, Cs, Zn, Sn, and many others). To our surprise, the strongly enriched trace and main elements show a characteristic Lorentzian distribution related to the water concentration, with the maximum corresponding to the solvus crest [2]. Some elements show a Gaussian distribution caused by overlapping two or more Lorentzian curves of different anion species. Mathematically speaking, the critical point here is a physicochemical singularity. During the transition from supercritical to critical and hydrothermal conditions, unusual processes are far from equilibrium, such as moissanite and beryl’s synchronous growth [6]. Typically, for technical processes, moissanite grows at temperatures above 1000°C. Also, typical rare elements in granitic systems show extreme enrichment in part. One such process is the excessive element enrichment, such as zinc [7]: 75.000 to 85.000 ppm Zn in a fluid inclusion trapped near the solvus crest with a water concentration of about 30%. Table 1 shows the extraordinarily enriched elements and their relationship to the Clarke values.

Table 1: Extreme trace element enrichment in syngenetic fluid inclusions trapped at or near the solvus crest [7].

Element (in ppm)

Melt inclusion 1 Melt inclusion 2 Clarke (ppm) Enrichment

Rb

12.300 15.700 200

61.5-78.5

Cs

22.700

20.000 5 4540-4000

Zn

80.000 45.000 50

1600-900

Cd

200

580 0.1 2000-5800

Sn

240 2.200 3

80-733

Sb

900

740 0.26 3462-2846

Pb

1.790 1.430 20

89.5-71.5

Clarke, according to Rösler and Lange (1975) [8]

At room temperature, Zn is, according to Raman spectroscopy, present as a well-transportable potassium tetrachlorocomplex: K2[ZnCl4]. Another unusual observation at the beginning of our research was the finding of extreme P-rich melt inclusions in pegmatite quartz Qu8 with 50.7 ± 3.5% P2O5 [8,9]. Because the phosphorus here was easily water-soluble, we were cautious during sample cleaning before any analytical studies. Later, we found melt inclusion in pegmatite quartz from the Sauberg mine with high P2O5 concentrations. Near the solvus crest, phosphorus shows strong enrichment in the form of the Lorentzian distribution (Figure 2). That is significantly higher than London (1998) [10] showed for P-rich peraluminous granites. This author interpreted higher P values as local build-up in boundary layers. Here, we tell another story: Enrichment of P by the transition of supercritical fluids into undercritical fluids during the interaction of the first one with the granites. Similar results arise also for the sulfate-anion and other elements.

In different publications, the authors [2,4,11] have shown that many trace and main elements are Lorentzian distributed. The proof was not straightforward because many pieces of evidence are nearly invisible because of the extreme postmagmatic hydrothermal intensity in the region, which blurs the primary processes. The remnants of strongly peralkaline mineralization prove such changes. The first author found strong peralkaline melts in the Ehrenfriedersdorf pegmatite Qu8, indicated by nepheline in quartz (Figure 3) [12]. That is an extraordinary observation because, usually, nepheline and quartz exclude themselves.

fig 3

Figure 3: Nepheline in pegmatite quartz (Qu8) from Ehrenfriedersdorf, Sauberg mine, Germany. The nepheline is composed of nepheline [(Na, K)AlSiO4] and kalsilite [KAlSiO4]. Between the nepheline and quartz is a tiny K-feldspar rim.

Table 2 gives the microprobe results on nepheline and feldspar. The nepheline aggregate (Figure 3) resembles a corroded crystal according to the form and feldspar rim. However, most nepheline crystals in this sample are in K-feldspar. That means that the nepheline points to very different rock chemistry before the reworking of the whole deposit by intensive hydrothermal activity. Non-consideration can lead to entirely wrong conclusions. Figure 4 shows another nepheline crystal in K-feldspar.

Table 2: Shows the composition of nepheline and K-feldspar by EMP analyses

Nepheline

K-feldspar-rim

SiO2

45.8 ± 1.4

68.5

Al2O3

31.8 ± 0.4

16.6

FeO

 0.3 ± 0.1

 0.4

Na2O

17.1 ± 0.1

 3.2

K2O

 3.8 ± 0.3

 9.1

Rb2O

 0.2 ± 0.04

 0.2

P2O5

~0.02

 0.02

Sum

99.02

98.0

n

10

1

XNe

0.86

XKls

0.14

XOr

0.65

XAb

0.35

fig 4

Figure 4: Nepheline (Ne) and Kalsilite (Kls) in K-feldspar (Kfs) in pegmatite Qu8. Quartz (Qtz) is secondary.

After re-homogenization, in the same sample are larger aggregates of silicate glass (~ 5mm) with a very low ASI (aluminum saturation index): ASI = 0.418 ± 0.009. The origin of this rock (now glass) is unclear. In the glass are small remnants of albite, K-feldspar, and nepheline. Table 3 shows the composition of this glass, determined with the microprobe SX50.

Maybe this rock (glass) is the reaction product of the supercritical fluid with the granitic rock in the reaction room (in the crust). At this time, maybe till 2012, the sporadically found high concentration of some elements (also tin) was explained more by chance (see also the discussion by London and Evensen 2002) [13]. We have had no explanation for the nepheline and glass with the low ASI values. However, from then on, we found indications of regularities in the appearance of many trace elements. A prerequisite for this systematic study was the water determination of the melt inclusion in question. Surprising was the Lorentzian distribution of about 20 trace elements with the water content of the corresponding melt inclusions [2].

Table 3: Composition of the re-homogenized silicate glass (700°C, 3.0 kbar) in the sample with nepheline in quartz and K-feldspar.

 

Mean

± 1σ

SiO2

71.39

0.25

TiO2

~0.01

<0.02

Al2O3

7.17

0.13

P2O5

0.05

0.02

FeO

0.21

0.04

MnO

<0.03

CaO

0.00

MgO

<0.01

Na2O

7.64

0.10

K2O

4.14

0.04

Rb2O

0.15

0.01

Cs2O

<0.04

F

0.00

Cl

<0.01

H2O

9.00

0.25

Sum

99.85

ASI

0.42

0.02

Mean from 72 determinations, σ: Standard deviation, ASI: Aluminum Saturation Index, water is the difference to 100% and corresponds to the Raman spectrometric determined water content.

Interpretation

The combination of the pseudo-binary melt-water solvus with the extreme element enrichment in the shape of the typical Lorentzian distribution is, together with the HP-HT mineral relicts (diamond, graphite, moissanite, reidite, coesite, and others), a solid-proof of the interaction of supercritical fluids coming from mantle depths with rocks in the upper crust. Furthermore, the extraordinary element parageneses in the Ehrenfriedersdorf case imply an older subducted deposit in the mantle region already postulated by Schütze et al. (1983) [14] according to isotope, element-geochemical and radio-geochronological studies.

Our latest studies, which brought unambiguous proofs of the interaction between supercritical fluids coming from mantle depths and granitic rocks in the upper crust, the following points are essential:

  1. The most spherical crystals, often tiny, of diamond, graphite, moissanite, and others in crustal minerals prove that a supercritical fluid transported these crystals very fast from mantle depths to the crust.
  2. All such minerals are spherical, which means that during transport from the mantle to the crust, elements of these crystals go into the supercritical fluid by partial solution.
  3. The “chemical way” from mantle to crust differs from point to point. By that, the composition of the supercritical fluid is also not uniform.
  4. The transition from the supercritical state into the critical and under-critical state is related to the processes far from equilibrium.
  5. The fast-moving supercritical fluids are also not in an energetic equilibrium with the surroundings, bringing much water and energy into the crustal region.
  6. In this highly excited state, processes happen that are conventionally not feasible (moissanite whiskers grow in beryl at significantly lower temperatures and pressures.
  7. The decisive element enrichment in the Lorentzian distribution shows impressively that enrichment processes occur, which, under hydrothermal conditions, are impossible.

Discussion

Apart from our work, other colleagues have not found such correlations (a combination of natural solvus curves and Lorentzian element distributions). Here, we see a beautiful field of activity that can bring us completely new views into geological processes related to supercritical fluids. An excellent experimental starting point demonstrates [15]. For example, Zn is highly soluble in high-temperature fluids. Therefore, a longer transportway is possible. That means, not far from the Ehrenfriedersdorf tin deposit, there is also a Zn deposit possible. In reverse, a larger Sn deposit is also possible under the famous Pb-Zn deposit in the Freiberg area because we have found there spherical remnants of REE-rich tveitite-Y crystals [Ca14Y5F43], typical for evolved tin-granites from Zinnwald, E-Erzgebirge.

Acknowledgment

We thank the many people who contributed to the supercritical fluids work over the last 25 years. A special thanks go to James (Jim) D. Webster (1955-2019), who initiated the intense work on evolved granites and pegmatites.

References

  1. Thomas R, Webster JD, Heinrich W (2000) Melt inclusions in pegmatite quartz: complete miscibility between silicate melts and hydrous fluids at low pressure. Mineral. Petrol. 139: 394-401.
  2. Thomas R, Davidson P, Rericha A, Voznyak DK (2022) Water-rich melt inclusions as “frozen” samples of the supercritical state in granites and pegmatites reveal extreme element enrichment resulting under non-equilibrium conditions. Mineralogical Journal (Ukraine) 44: 3-15.
  3. Thomas R, Davidson P (2015) Comment on “A petrologic assessment of internal zonation in granitic pegmatites” by David London (2014). Lithos 212-215: 462-468.
  4. Thomas R, Rericha A (2023) The Variscan tin deposit Ehrenfriedersdorf, Germany: To the solubility of tin in supercritical and near-supercritical fluids. Geology, Earth and Marine Sciences 5: 1-5.
  5. Thomas R, Webster JD, Davidson P (2011) Be-daughter minerals in fluid and melt inclusions: implications for the enrichment of Be in granite-pegmatite systems. Mineral. Petrol. 161: 483-495.
  6. Thomas R, Recknagel U, Rericha A (2023a) A moissanite-diamond-graphite paragenesis in a small beryl-quartz vein related to the Variscan tin-mineralization of the Ehrenfriedersdorf deposit, Germany. Aspects in Mining & Mineral Sciences 11: 1310-1319.
  7. Borisova AY, Thomas R, Salvi S, Candaudap F, Lanzanova A, et al. (2012) Tin and associated metal and metalloid geochemistry by femtosecond LA-ICP-QMS microanalysis of pegmatite-leucogranite melt and fluid inclusions: new evidence for melt-melt-fluid immiscibility. Mineralogical Magazine 76: 91-113.
  8. Rösler HJ, Lange H (1975) Geochemische Tabellen. Second Edition. VEB Deutscher Verlag fürGrundstoffindustrie Leipzig, pg: 675.
  9. Thomas R, Webster JD, Rhede D (1998) Strong phosphorus enrichment in a pegmatite-forming melt. Acta Universitatis Carolinae-Geologica. 42: 150-164.
  10. London D (1998) Phosphorus-rich peraluminous granites. Acta Universitatis Carolinae-Geologica 42: 64-68.
  11. Thomas R, Davidson P, Appel K (2019) The enhanced element enrichment in the supercritical states of granite-pegmatite systems. Acta Geochim 38: 335-349.
  12. Thomas R, Webster J.D., Rhede D, Seifert W, Rickers K, et al. (2006) The transition from peraluminous to peralkaline granitic melts: Evidence from melt inclusions and accessory minerals. Lithos 91: 137-149.
  13. London D, Evensen JM (2002) Beryllium in silicic magmas and the origin of beryl-bearing pegmatites. In: Beryllium-Mineralogy, Petrology, and Geochemistry. Edited by E.S. Grew. Reviews in Mineralogy & Geochemistry 8: 445-486.
  14. Schütze H, Stiehl G, Wetzel K, Beuge P, Haberland R, et al. (1983) Isotopen-und elementgeochemische sowie radiogeochronologische Aussagen zur Herkunft des Ehrenfriedersdorfer Granits-Ableitung erster Modellvorstellungen. ZFI-Mitteilungen 76: 232-254.
  15. Sun Y, Liu X, Lu X (2023) Structures and transport properties of supercritical SiO2-H2O and NaAlSi3O8-H2O fluids. American Mineralogist in press.

How does the Language Prediction Model, the ChatGPT Evaluate Negative Emotions?

DOI: 10.31038/JCRM.2023634

Summary

The Generative AI (Artificial Intelligence: AI), the ChatGPT (Generative Pretrained Transformer: ChatGPT), is a language prediction model that generates sentences based on word frequencies and interrelationships. In this study, we evaluated how the ChatGPT, a generative AI, performs in cognitive conflicts (conflicts) between healthcare professionals and patients/families encountered in healthcare settings, using dialogue transcripts of licensed medical mediations (with joint decision making), in which the ChatGPT is said to convey limitations and misinformation regarding negative emotions. We report the results of our study of the ChatGPT’s negative emotion evaluation, comparing it with human evaluations.

Abstract

We investigated that how the ChatGPT (Version 3.5), a generative AI, evaluated negative emotions in narratives of cognitive discrepancies (conflicts) between medical professionals and patients/families encountered in the medical field. As a result, negative emotion evaluation by the ChatGPT did not reach the level that people do. It can be inferred that there are limitations to negative emotion evaluation by the ChatGPT at this time.

Introduction

ChatGPT (Generative Pretrained Transformer: ChatGPT) is an artificial intelligence (AI), neural network-based language prediction. It is one of the models that generate sentences based on word frequencies and their interrelationships. This is said to cause so-called hallucination (hallucination), which is the conveyance of incorrect information, due to the limitations of human emotion processing that depends on context and situation [1,2]. Conflicts between medical professionals and patients/families encountered in the medical field are always accompanied by negative emotions. How does the ChatGPT, a generative AI, evaluate such negative emotions? There do not seem to be any evaluation reports on this issue. Therefore, we conducted a human evaluation of the ChatGPT’s verbal assessment of negative emotions using recorded dialogue data from past medical mediations (a concept with a dialogue process involving collaborative decision making [3] and investigated the rate of agreement.

Case Presentation

The purpose of the study was to determine whether “The ChatGPT (Version3.5), a generative AI, can capture negative emotions from dialogue narratives.”

The overall flow of the research methodology is shown in Figure 1. The evaluation period was from August 31, 2023 to September 30, 2023. The subjects were the Ethics Committee and the record language of the medical mediator of the first complaint claim submitted with the permission and consent of the patient’s family, among the previously resolved complaint cases, as shown in Table 1.

fig 1

Figure 1: Research Methods

Table 1: Complaint status of subject cases

Case

Patient narratives of situation content

1

The patient’s daughter and the outpatient attending physician talk about the delay in seeing the patient for inappropriate medical care.

2

Patient speaks up to health care provider about his dissatisfaction with the treating technician.

3

The head technician and nurse talk about the policy for dealing with problems between patients and technicians in the department in charge.

4

The patient’s son and daughter-in-law have doubts about the medical personnel’s handling of the sudden change.

5

This is a scene in which a patient and a medical professional are discussing a treatment plan for a nerve palsy that has appeared since the surgery.

6

A patient who was seen for abdominal pain is misdiagnosed initially and speaks with the corresponding physician.

7

A family member of a patient who has accidentally swallowed a partial denture is talking with a medical professional, a nurse.

8

Patient speaks of how nurses treated him during his hospitalization.

9

The patient is just after surgery. He discusses the response of the resident in charge of the patient.

10

A bereaved family member who lost a patient tells the story.

11

A bereaved family member who lost a patient suddenly told the story.

For the content of the ChatGPT instructions for the generated AI, prompts were created based on the Fukatsu-style questioning technique created by Fukatsu [4], as shown in Figure 2. For the constraints and emotions in the prompts, nine emotions [5] were used, including “joy,” “expectation,” “anger,” “disgust,” “sadness,” “surprise,” “fear,” “trust,” and “anxiety,” referring to Pulchick’s Emotion Circle, and responses were also required to provide reasons for their emotion evaluations. Responses were also asked for reasons for their own emotion ratings. The ChatGPT used for the evaluation was Version 3.5, which was set to not perform the learning function. The prompts were then entered and implemented on two PCs, and the interpersonal emotions of the narratives were estimated for three participants (an ethics mediator and two male and two female medical students in their 20s each, before they left for clinical practice).

fig 2

Figure 2: Instructions to the generating AI (ChatGPT)

For the evaluation of the generated AI ChatGPT emotion output, examples of output results are shown in Table 2; the output results of the 11 cases were individually evaluated by the three aforementioned prompt implementers using the published rating criteria [5,6] scale. The respective emotional evaluation items and evaluation methods are shown in Figure 3. The two evaluations were then aggregated to conduct an overall evaluation of ChatGPT’s emotion evaluation status [7].

Table 2: Generative AI (ChatGPT) answers to the narrative of Case 1

Narrative of a patient’s family

Answer 1

Answer 2

 

My father had been suffering from severe constipation for a long time. It took days from the time he was admitted to the hospital until he was

diagnosed, and in the end, the disease was found to be in such an advanced stage, but my father trusted the doctor like a god and had been coming to the hospital for a long time, but he was betrayed. I’m sure my father must be holding a grudge against him now.

#Emotions Emotions felt from the patient’s family narrative are “anger” and “sadness”.

#ReasonThe family of the patient is angry because their father was suffering from severe constipation. They express anger that the attending physician instructed them that over-the-counter laxatives were fine and did not understand their father’s suffering and did not take appropriate action. They are also surprised and saddened that the attending physician did not know that the patient’s wife was accompanying him. He also expresses disappointment and sadness that it took days to diagnose the disease and that it was discovered in an advanced state. He also feels disgust and sadness at the loss of trust in the doctor, as he did not receive proper medical care, despite the fact that his father had trusted the doctor and had visited him for a long time.

#Emotion: Anger Reason: The patient’s daughter is angry that her father has been suffering from severe constipation for a long time and her doctor did not provide proper medical care, only recommending over-the-counter laxatives.

She is also angry that the attending physician did not know that the patient’s wife was accompanying her and did not provide an

appropriate explanation, and regrets that she should have been more proactive.

He also attributes his anger to the fact that it took so long from the time he was admitted to the hospital until he was diagnosed, and that his disease had progressed.

My father trusted his doctor like a god, and the fact that his trust was

betrayed suggests that he may also be angry with his doctor.

fig 3

Figure 3: Evaluation items and methods for the emotional results of the generated AI (ChatGPT)

Results

Table 3 shows the aggregate results of the emotion judgments of the ChatGPT by the generated AI; the total percentage of 100% agreement and 75% agreement ranged from 63.6% to 75.0% for questions 1 through 5, respectively. The reason statements evaluated in Question 2 were quotes from the narrative and interpreted feelings. All of the responses that were interpreted as emotions were inconclusive. Next, the context of the emotional discrepancy was indicated as inadequate or excessive. The most common emotion perceived as excessive was anger. There were also many responses that listed only one emotion.

Table 3: Rater ratings of generated AI (ChatGPT) emotional output results

Question

1 2 3 4

5

100%

7(15.9) 6(13.6) 23(52.3) 22(55.0) 19(43.2)
75% 21(47.7) 22(50.0) 10(22.7) 6(13.6)

14(31.8)

50%

11(25.0) 11(25.0) 10(22.7) 10(22.7) 8(18.2)
25% 1(2.3) 1(2.3) 1(2.3) 5(11.4)

3(6.8)

0%

4(9.1) 4(9.1) 0(0.0) 1(2.3) 0(0.0)

Total number of responses

44(100.0) 44(100.0) 44(100.0) 44(100.0)

44(100.0)

1.       Is the sentiment consistent with the response regarding “feelings”?

2.       Is the reason for the response regarding “emotion” appropriate?

3.       Is the rationale vague in response to the “emotion” response?

4.       In response to the “emotion” response, is the emotion expressed in the supporting reasons consistent with the “emotion”?

5.       For the response regarding “emotion,” is the emotion expressed in the supporting reasons appropriate?

Next, the following characteristics of the responses were noted. First, (a) In the case of simple structures such as a single emotion, the emotion was appropriately captured. Second, (b) when emotions were mixed in a complex way, the rating of agreement decreased. Also, (c) in the case where the patient died, mixed responses were generated without distinguishing between past emotions and emotions at the time of the narrative. Furthermore, in the case of (d) where the interest (expression of interest, desire, and values: hereafter, interest) changed during the course of the narrative, the respondent responded to the emotion by addressing only the first half of the interest and ignoring the interest after the change.

Discussion

Table 3 shows that the total percentage of 100% agreement and 75% agreement ranged from 63.6% to 75.0%. 75% agreement was adopted because the three raters were two men and two women in their 20s with limited emotional and life experience and a medical mediator who had gained the patients’ trust and supported collaborative decision making during actual interviews with the patients. Since the hypothesis was that the results would “accurately capture negative emotions,” the 100% agreement between the emotions responded to by the ChatGPT and the emotions responded to by the raters was low (15.9% and 13.6%), if 100% agreement is considered “accurate” in the hypothesis, the 100% agreement between the emotions (Question 1) and their rationale (Question 2) was low (15.9% and 13.6%). The results showed that the AI was unable to accurately assess negative emotions in the items that must be given the most weight in human emotion assessment. This indicates that emotion evaluation based on language alone is limited or impossible, considering that humans evaluate the emotions of others by synthesizing the situation, context, and nonverbal messages and matching them with their own interests. The definition of accurate should have been clarified in order to refine the evaluation.

As shown in Figure 3, “anger” was frequently over-rated. We considered that this was caused by grasping only the final emotion, “anger,” and ignoring the primary emotion that caused the anger.

Next, for results (a)-(d), we considered that the emotions in the language of narration can be accurately taken, but not in the area of judging by context.

The following points are necessary to improve the agreement of ChatGPT’s emotion judgments with human evaluations. For example, parameters such as environment, atmosphere, facial expressions, and tone of voice, which are quasi-linguistic and non-linguistic. It is necessary to add these elements as linguistic information. The emotional evaluation of ChatGPT, a generative AI, was limited to the age of the evaluator and the number of evaluators. For more accurate evaluation, it is necessary to add parameters such as the age of the evaluators, the number of evaluators, and their expertise. It is also important to clarify the type of linguistic information.

Conclusion

The negative sentiment evaluation of the generative AI was only partially affirmed. The emotion evaluation of the ChatGPT of the generated AI based solely on linguistic information at the time of this study is limited. At present, it is difficult to accurately identify emotions in detail.

Conflicts of Interest

There are no corporate or other COI relationships that should be disclosed.

References

  1. Japanese Ministry of Education, Culture, Sports, Science and Technology: Tentative Guidelines for the Use of Generative AI at the Primary and Secondary Education <https://www.mext.go.jp/content/20230710-mxt_shuukyo02-000030823_003. pdf> (see 2023/10/01).
  2. Ministry of Education, Culture, Sports, Science and Technology of Japan: Handling of teaching and learning aspects of generative AI in universities and technical colleges (Notification) <https://www.mext.go.jp/content/20230714-mxt_ senmon01-000030762_1.pdf> (see 2023/10/01).
  3. Yoshitaka Wada, Toshimi Nakanishi: Medical mediation: A narrative approach to conflict management. Seigne Ltd., 2011 Japanese
  4. Takayuki Fukatsu: ChatGPT Work Art Revolution, Weekly Toyo Keizai, 2023/4/22, Toyo Keizai Shinposha, 42-43.
  5. Robert Plutchik (2001) The Nature of American Scientist 89: 344-350.
  6. Kaarre J, Feldt R, Keeling LE, Dadoo S, Zsidai B, et (2023) Exploring the potential of ChatGPT as a supplementary tool for providing orthopaedic information. Knee Surg Sports Traumatol Arthrosc 31: 5190-5198. [crossref]
  7. Robert C Benirschke, Joshua Wodskow, Kritika Prasai, Andrew Freeman, John M Lee, et (2023) Assessment of a large language model’s utility in helping pathology professionals answer general knowledge pathology questions, American Journal of Clinical Pathology aqad106. [crossref]

Impact of the COVID-19 Pandemic on Breast Incidence and Stage during the COVID-19 Pandemic in the Netherlands, and a Comparison with Other Countries

DOI: 10.31038/AWHC.2023643

Short Report

In the Netherlands, the first COVID-19 patient was confirmed on February 27th, 2020. Thereafter, the number of infected patients quickly increased, just as the number of hospitalized COVID-19 patients. In March, 2020, the first policy measures were taken to prevent the spread of the virus. Those measures included, amongst others, the recommendation to keep 1.5 meters distances and for elderly and vulnerable people to stay at home. Additionally, the Dutch national breast cancer screening program was suspended in week 12 of 2020. In week 26 of 2020 it was resumed at 40% capacity, which was slowly increased to a capacity of 100% in the spring of 2021. These policy measures and decisions in healthcare were made without, or with little, prior knowledge of the consequences on breast cancer detection and care. Therefore, this short report aimed to give an overview of the effect of the COVID-19 pandemic on breast cancer incidence and stage in the Netherlands, and compared those results with other countries.

First results of the effect of the COVID-19 pandemic on breast cancer detection showed a decrease in breast cancer incidence (expressed as the number of breast cancer diagnoses per 100,000 women) in women diagnosed with breast cancer in weeks 2-17 of 2020, compared with women diagnosed in weeks 2-17 of 2018 or 2019. This effect was seen in all age groups and tumor stages, except stage IV [1-3]. As expected, the incidence of DCIS and stage I tumors decreased to the largest extent, as these are the tumors mainly detected by the breast cancer screening program.

When we focus on women in the screening age, 50-74 years, 67% fewer screen-detected tumors were diagnosed in weeks 9-35 of 2020, compared to week 9-35 of 2018/2019 [4]. The incidence of screen-detected tumors was significantly lower in all age groups during weeks 14-35 of 2020, and the incidence of all tumor stages, except stage IV, was significantly lower during weeks 14-25. During weeks 26-35 the incidence of DCIS and stage I-II tumors stayed significantly lower. Less pronounced effects were observed for the incidence of clinically-detected breast cancer. Compared to weeks 9-35 of 2018/2019, 7% fewer clinically-detected breast tumors were diagnosed in weeks 9-35 of 2020. The incidence decreased in all age groups in weeks 12-16. Incidence of stage I-II tumors was significantly lower in weeks 12-13 and incidence of DCIS and stage I-III tumors was significantly lower during weeks 14-16.

Follow-up research of our group investigated the effect of the pandemic on breast cancer incidence and stage during January 2020 till December 2021 [5]. This study showed that the incidence was significantly lower in women eligible as well as not eligible for screening (i.e., those aged <50 and >74 years). This suggests that the decrease in incidence was caused by both the suspension of the screening program and the reluctance of patients to visit the general practitioner. During the second wave, i.e., October 2020-April 2021, the incidence of clinically-detected tumors was significantly higher in Dutch women aged 50-74 years [5]. This suggests that the method of detection in some women changed from screen-detected to clinically-detected. Additionally, a small and temporary increase in the incidence of stage IV tumors was seen in Dutch women aged 50-69 years. However, it is unclear whether this increase is due to the COVID-19 pandemic or other factors. The increased incidence could also be a result of the increase in the usage of improved diagnostic methods, such as the PET-CT scan which is highly accurate in detecting distant metastases compared to conventional methods [6].

Comparable to our studies, studies from other countries also showed a decrease in both the absolute number of breast cancer patients [7-14] and in the crude breast cancer incidence rate [15,16] at the beginning of the pandemic. The largest decrease in breast cancer incidence was seen in women in the screening age groups [3,4,8,16]. Part of the decrease in breast cancer incidence can be explained by the suspension of the national breast cancer screening program. Many countries had to suspend their screening program to reduce the pressure on healthcare [17]. A previous meta-analysis showed a 41% decline in mammogram rates between 2019 to 2020 according to data of three registry-based studies, and a 53% decline based on data of ten non-registry-based studies [18]. Another part of the decrease in breast cancer incidence could be explained by a decline in the number of women visiting the General Practitioner (GP) due to fear of contracting the virus or overburding the healthcare system [1,2]. The decrease in the number of women visiting the GP in the Netherlands did not differ by age group [2].

Studies from Norway and New Zealand only showed a minimal decrease in the number of breast cancer diagnoses during the start of the pandemic [5,19]. Both Norway and New Zealand had a low COVID-19 infection rate and low COVID-19 death rates compared to other countries [20-25]. This indicates that the stable incidence in Norway and New Zealand could be due to the low severity of the pandemic, resulting in a minimal decrease of breast cancer diagnoses.

Comparable to our results, some studies from other countries showed that breast cancer incidence quite quickly reached pre-COVID levels after the first wave [7-9,14-16]. However, a couple of countries/regions had more difficulties in reaching pre-COVID incidence levels. These include Italy [10], Hungary [11], the United States [12], and Bavaria [13]. The level of political regulation and the number of COVID-19 infections or deaths were comparable between those four countries/regions and other countries [26-28]. Hence, this probably does not explain the difference in incidence. One reason for the decreased incidence in Italy could be that Italian women were still hesitant to visit screening after the end of the first wave [29]. A Italian study showed a 20% decrease in the number of women attending screening between October-December 2020, compared to the same period in 2019, while the number of women invited reached pre-COVID levels [29]. The decrease in incidence in Hungary might be explained by the relatively high number of COVID-19 patients in the hospitals, compared to other countries [30]. Also, the breast cancer screening program was suspended a second time in April 2021. The decrease in Bavaria (Germany) could have been caused by a relatively high number of patients at the Intensive Care Units (ICU) in Germany during the second wave, compared to other countries [31]. A negative association between the number of patients at the ICU and the diagnostic capacity at the oncological care was found in Germany [32]. The potential cause for the decrease in incidence found in the study from the United States is unknown. These cross-country comparisons show that the cause for the decline in incidence varies from country to country.

In the Netherlands, the maximum allowed screening interval between two invitations increased from two to three years in November 2020. The increase in the screening interval was both due to the COVID-19 pandemic and due to a shortage in mammography technologists. As a result, the mean screening interval was 32.2 months in 2021 [33]. This increased screening interval probably caused the method of detection in some women to change from screen-detected to clinically-detected, as a significant higher number of women were diagnosed with a clinically-detected cancer during October 2020-April 2021 compared to the same period in 2017-2019 [5]. A Dutch modelling study showed that a three-months suspension of the screening program, without catch-up, might already cause a 19% increase in the number of interval tumors detected between the last and first screening after interruption, compared to no suspension [34].

The majority of studies on tumor stage investigated whether the proportion of women diagnosed with a certain stage tumor changed during the pandemic [10,21-25]. However, as the suspension of the breast cancer screening program mainly led to a decrease in the incidence of DCIS and stage I tumors it was expected that a lower proportion of women would be diagnosed with these tumors, and that a higher proportion would be diagnosed with late-stage tumors. It would have given more insight if these studies investigated the effect of the pandemic on the incidence of breast cancer by tumor stage, as we did in our studies.

In our studies we did not adjust for the aging of the population or the increase in risk factors associated with breast cancer, while those factors might have led to an increase in the number of cancer patients. However, as the crude breast cancer incidence rate stayed rather constant in the Netherlands during the seven years before the pandemic (2013-2019) [35], and the study period used in our studies is relatively small, it is not expected that this influenced the results.

This report showed the effect of the COVID-19 pandemic on breast cancer incidence and tumor stage in the Netherlands, and a comparison with other countries. More studies on the effect of the COVID-19 pandemic on breast cancer incidence, both in total and per tumor stage, are needed to determine the association between delays in diagnosis and tumor stage.

References

  1. Quinn-Scoggins HD, Cannings-John R, Moriarty Y, Whitelock V, Whitaker KL, et al. (2021) Cancer symptom experience and help-seeking behaviour during the COVID-19 pandemic in the UK: a cross-sectional population survey. BMJ open 11: e053095. [crossref]
  2. Grant MP, Helsper CW, Stellato R, van Erp N, van Asselt KM, et al. (2022) The Impact of the COVID Pandemic on the Incidence of Presentations with Cancer-Related Symptoms in Primary Care. Cancers (Basel) 14: 5353.
  3. Eijkelboom AH, de Munck L, Peeters M-JTV, Broeders MJ, Strobbe LJ, et al. (2021) Impact of the COVID-19 pandemic on diagnosis, stage, and initial treatment of breast cancer in the Netherlands: a population-based study. J Hematol Oncol 14: 1-12. [crossref]
  4. Eijkelboom AH, de Munck L, Lobbes MBI, van Gils CH, Wesseling J, et al. (2021) Impact of the suspension and restart of the Dutch breast cancer screening program on breast cancer incidence and stage during the COVID-19 pandemic. Prev Med 151: 106602. [crossref]
  5. Eijkelboom AH, de Munck L, Larsen M, Bijlsma MJ, Tjan-Heijnen VC, et al. (2023) Impact of the COVID-19 pandemic on breast cancer incidence and tumor stage in the Netherlands and Norway: A population-based study. Cancer Epidemiol 87: 102481.
  6. Paydary K, Seraj SM, Zadeh MZ, Emamzadehfard S, Shamchi SP, et al. (2019) The evolving role of FDG-PET/CT in the diagnosis, staging, and treatment of breast cancer. Mol Imaging Biol 21: 1-10. [crossref]
  7. Gurney JK, Millar E, Dunn A, Pirie R, Mako M, et al. (2021) The impact of the COVID-19 pandemic on cancer diagnosis and service access in New Zealand-a country pursuing COVID-19 elimination. The Lancet Regional Health-Western Pacific 10: 100127. [crossref]
  8. Peacock HM, Tambuyzer T, Verdoodt F, Calay F, Poirel HA, et al. (2021) Decline and incomplete recovery in cancer diagnoses during the COVID-19 pandemic in Belgium: a year-long, population-level analysis. ESMO open 6: 100197. [crossref]
  9. Ramanakumar AV, Annie B, Frederic L, Christine B, Cathy R, et al. (2022) Evaluating the impact of COVID‐19 on cancer declarations in Quebec, Canada. Cancer Medicine 2022. [crossref]
  10. Mentrasti G, Cantini L, Vici P, D’Ostilio N, La Verde N, et al. (2022) Rising incidence of late stage breast cancer after COVID-19 outbreak. Real-world data from the Italian COVID-DELAY study. The Breast 65: 164-171. [crossref]
  11. Elek P, Csanádi M, Fadgyas-Freyler P, Gervai N, Oross-Bécsi R, et al. (2022) Heterogeneous impact of the COVID-19 pandemic on lung, colorectal and breast cancer incidence in Hungary: results from time series and panel data models. BMJ open 12: e061941. [crossref]
  12. Drescher CW, Bograd AJ, Chang SC, Weerasinghe RK, Vita A, et al. (2022) Cancer case trends following the onset of the COVID‐19 pandemic: A community‐based observational study with extended follow‐up. Cancer 128: 1475-1482. [crossref]
  13. Voigtländer S, Hakimhashemi A, Grundmann N, Radespiel-Tröger M, Inwald EC, et al. (2023) Impact of the COVID-19 pandemic on reported cancer diagnoses in Bavaria, Germany. J Cancer Res Clin Oncol 2023: 1-11. [crossref]
  14. Vrdoljak E, Balja MP, Marušić Z, Avirović M, Blažičević V, et al. (2021) COVID‐19 Pandemic Effects on Breast Cancer Diagnosis in Croatia: A Population‐and Registry‐Based Study. The oncologist 26: e1156-e1160. [crossref]
  15. Caswell-Jin JL, Shafaee MN, Xiao L, Liu M, John EM, et al. (2022) Breast cancer diagnosis and treatment during the COVID-19 pandemic in a nationwide, insured population. Breast Cancer Res Treat 194: 475-482. [crossref]
  16. Greene G, Griffiths R, Han J, Akbari A, Jones M, et al. (2022) Impact of the SARS-CoV-2 pandemic on female breast, colorectal and non-small cell lung cancer incidence, stage and healthcare pathway to diagnosis during 2020 in Wales, UK, using a national cancer clinical record system. Br J Cancer 2022: 1-11. [crossref]
  17. Perin DMP, Elfström KM, Bulliard J-L, Burón A, Campbell C, et al. (2021) Early assessment of the first wave of the COVID-19 pandemic on cancer screening services: The International Cancer Screening Network COVID-19 survey. Prev Med 151: 106642. [crossref]
  18. Ng JS, Hamilton DG (2022) Assessing the impact of the COVID-19 pandemic on breast cancer screening and diagnosis rates: A rapid review and meta-analysis. J Med Screen 29: 209-218. [crossref]
  19. Mitchell H, Mclean J, Gavin AT, Visser O, Millar E, et al. (2023) Impact of COVID-19 control on lung, breast, and colorectal pathological cancer diagnoses. A comparison between the Netherlands, Aotearoa New Zealand, and Northern Ireland. BMC Cancer 23: 1-8. [crossref]
  20. Our World in Data. Total confirmed COVID-19 deaths and cases per million people. https://ourworldindata.org/grapher/total-covid-cases-deaths-per-million?time=2020-01-04..2021-12-31&facet=metric&country=NOR~NLD. Accessed 22-09-2023.
  21. Rocha AFBM, Freitas-Junior R, Ferreira GLR, Rodrigues DCN, Rahal RMS (2023) COVID-19 and Breast Cancer in Brazil. International Journal of Public Health 2023: 48.
  22. Kang Y-J, Baek JM, Kim Y-S, Jeon YW, Yoo T-K, et al. (2021) Impact of the COVID-19 Pandemic on the Diagnosis and Surgery of Breast Cancer: A Multi-Institutional Study. J Breast Cancer 24: 491. [crossref]
  23. Purushotham A, Roberts G, Haire K, Dodkins J, Harvey-Jones E, et al. (2021) The impact of national non-pharmaceutical interventions (‘lockdowns’) on the presentation of cancer patients. ecancermedicalscience 15: 1180. [crossref]
  24. Li T, Nickel B, Ngo P, McFadden K, Brennan M, et al. (2023) A systematic review of the impact of the COVID-19 pandemic on breast cancer screening and diagnosis. The Breast 2023. [crossref]
  25. Knoll K, Reiser E, Leitner K, Kögl J, Ebner C, et al. (2022) The impact of COVID-19 pandemic on the rate of newly diagnosed gynecological and breast cancers: a tertiary center perspective. Arch Gynecol Obstet 2022: 1-9. [crossref]
  26. Our World in Data. COVID-19 Containment and Health Index. https://ourworldindata.org/grapher/covid-containment-and-health-index?tab=chart&time=earliest..2021-02-28&country=USA~ITA~NLD~HUN~BEL~DEU~CAN~NOR~NZL~HRV~GBR. Accessed 05-25-2023.
  27. Our World in Data. Biweekly confirmed COVID-19 deaths per million people. https://ourworldindata.org/grapher/biweekly-covid-deaths-per-million-people?tab=chart&time=2020-01-15..2021-03-08&country=USA~GBR~ITA~NLD~NOR~NZL~HRV~HUN~BEL~DEU. Accessed 30-05-2023.
  28. Our world in Data. Biweekly confirmed COVID-19 cases per million people. https://ourworldindata.org/grapher/biweekly-covid-cases-per-million-people?tab=chart&time=2020-01-15..2021-12-31&country=FRA~DEU~ITA~USA~NOR~NLD~GBR~HRV~HUN~CAN~NZL~BEL. Accessed 26-06-2023.
  29. Rossi PG, Carrozzi G, Falini P, Sampaolo L, Gorini G, et al. (2023) The impact of the COVID-19 pandemic on Italian population-based cancer screening activities and test coverage: Results from national cross-sectional repeated surveys in 2020. Elife 12: e81804. [crossref]
  30. The New York Times. Europe’s Deadly Second Wave: How Did It Happen Again? https://www.nytimes.com/interactive/2020/12/04/world/europe/europe-covid-deaths.html. Accessed 25-05-2023.
  31. Our World in Data. Number of COVID-19 patients in ICU per million. https://ourworldindata.org/grapher/covid-icu-patients-per-million?country=CAN~DEU~NLD~ITA~USA~Wales. Accessed 05-25-2023.
  32. Arndt V, Doege D, Fröhling S, Albers P, Algül H, et al. (2023) Cancer care in German centers of excellence during the first 2 years of the COVID-19 pandemic. J Cancer Res Clin Oncol 149: 913-919. [crossref]
  33. Netherlands Comprehensive cancer organisation (IKNL). national monitoring of the breast cancer screening programme in the Netherlands 2020/2021. 2022.
  34. Poelhekken K, Greuter MJ, de Munck L, Siesling S, Brokken FB, et al. (2023) Long-term effects of the interruption of the Dutch breast cancer screening program due to COVID-19: A modelling study. Prev Med 166: 107376. [crossref]
  35. Netherlands comprehensive cancer organisation (IKNL). NCR Data. https://nkr-cijfers.iknl.nl/#/viewer/665a7495-c6fd-415c-9e62-afd630859b20. Accessed 22-02-2023.

Impact of BRCA1 and BRCA2 Gene Mutations in Prostate Cancer in Thies

DOI: 10.31038/MGJ.2023611

Abstract

Prostate cancer (CaP) is a public health problem among men worldwide, particularly those over the age of 50, and its incidence continues to rise. Despite improvements in early detection methods, a large proportion of patients succumb to the disease. Studies have shown that men with BRCA1/BRCA2 gene mutations in prostate cancer are likely to have more severe disease and a poorer prognosis. A BRCA2 gene mutation is known to confer the highest risk of prostate cancer (8.6 times in men ≤ 65 years of age) while BRCA1 presents an increased risk, albeit to a lesser extent (3.5 times); making BRCA genes a conceivable genomic biomarker for prostate cancer risk. It is in this context that we will examine the impact of BRCA1/BRCA2 gene mutations in prostate cancer in Thiès. Our study is conducted between January 2020 and December 2022 with 59 patients diagnosed with a prostate tumor in the urology department of the Thiès regional hospital and the Saint Jean de Dieu hospital in Thiès. The variables studied were age, PSA levels, Gleason score and histological grades. Total DNA from prostate tissue was extracted using the Qiagen protocol (Qiagen Dneasy Tissue Kit) and the three primers for the BRCA1-185delAG, BRCA1-5382insC and BRCA2-6174delT genes were amplified. The results indicate a frequency of 62.71% of patients diagnosed with prostate cancer versus 37.29% with the lesion of benign prostatic hyperplasia. BRCA1-5382insC and BRCA2-6174delT mutations showed higher frequencies (2-3 fold) in patients with CaP than in those with the BPH lesion, with 62.7% vs. 37.3% and 65.1% vs. 34.9% respectively. Gleason score 8 was more represented with a rate of 44% corresponding to grade IV according to WHO-ISUP 2016. However, individuals carrying mutations (BRCA1-5382insC; BRCA2-6174DelT) could be associated with a higher risk of prostate cancer, and are also likely to have a poor survival rate.

Keywords

Prostate cancer; Mutations; BRCA1, BRCA2

Introduction

Prostate cancer (CaP) is the second most common cancer diagnosis in men (14.1%) and the fifth leading cause of death (6.8%) worldwide in 2020 [1]. Every year, Africa records around 1.1 million new cases of cancer and up to 700000 deaths from the disease [2]. Many men with prostate cancer are diagnosed by a biopsy and analysis of the prostate, a prostate specific antigen (PSA) test and a digital rectal examination. Risk factors for prostate cancer include family risk, ethnicity, age, obesity and other environmental factors [3]. Demographic expansion and improved life expectancy worldwide are expected to contribute to an increase in the number of cases of CaP [4]. making it a major global health problem. Prostate cancer is a heterogeneous disease, both epidemiologically and genetically. The interaction between genetics, environmental and social influences results in lower estimates of prostate cancer survival rates by race, which explains the differences observed in the epidemiology of prostate cancer in different countries [3]. There is documented evidence of a genetic contribution to prostate cancer. Hereditary prostate cancer and genetic predisposition to prostate cancer have been studied for years. One of the most predisposing genetic risk factors for prostate cancer is family inheritance. Twin studies and epidemiological studies have both demonstrated the role of heredity in CaP [5]. Many researchers have investigated the possible role of genetic variations in androgen biosynthesis and metabolism, as well as the role of androgens [6,7]. Genomics research has identified molecular processes that lead to certain cancerous developments, such as chromosomal rearrangements [3]. Although new treatments have emerged in the last decade, prostate cancer is still a major source of cancer deaths in men [8]. Advanced age is the main risk factor, with more than three-quarters of CaP detections made in men over 65 [9]. Prostate cancer susceptibility genes are genes involved in the androgen pathway and testosterone metabolism. The development of the prostate epithelium and prostate cancer cells depends on the androgen receptor and testosterone signalling pathway [10]. The identification of cancer biomarkers and the targeting of specific genetic mutations can be used for the targeted treatment of prostate cancer. Biomarkers that can be used for targeted therapy include tumour biomarkers, DNA biomarkers and general biomarkers [11]. Family history and genetic predisposition such as BRCA1/BRCA2 pathogenic variants have also been identified as important risk factors [12,13]. It is known that a mutation in the BRCA2 gene confers the highest risk of prostate cancer in men (8.6 times higher in men aged 65 years, while BRCA1 shows increased risk, although to a lesser extent (3.5 times) [14]. These genes have attracted much attention from researchers, but their role in the clinical assessment and treatment of prostate cancer remains complex. The aim of this study is to examine the impact of BRCA1/BRCA2 gene mutations in prostate cancer in Thiès.

Materials and Methods

This study covers 59 patients with prostate tumours. These patients were recruited from the urology department of the Thiès regional hospital and the Saint Jean de Dieu hospital in Thiès between January 2021 and December 2022. Inclusion criteria were a suspicious digital rectal examination (DRE) with a PSA level greater than 4 ng/ml, followed by biopsies for histopathological diagnosis. After review in accordance with the rules laid down by Senegal’s National Health Research Ethics Committee (SNHREC) and in compliance with the procedures established by Cheikh Anta Diop University in Dakar (UCAD) for all research involving human participants, ethical approval was obtained for this study. The objectives of the study, the protocol, the benefits and the confidentiality criteria were explained to each patient to give them the opportunity to accept or refuse to take part. In the case of acceptance, a duly completed and signed informed consent form was required for admission to the study. For data collection, we collected demographic data (surname, first name, age, ethnicity, reason for consultation), PSA levels and medical history from routine family files.

DNA Extraction and Amplification of the BRCA1 and BRCA2 Genes

Total DNA from each sample was extracted using the Qiagen protocol (Qiagen Dneasy Tissue Kit). DNA quality was checked by electrophoretic migration on a 1.5% agarose gel. For a given gene, the conditions for DNA amplification are the same whatever the pathology and for both tumour tissues and controls. PCR amplification conditions included a 1st step of a 12 minute of initial denaturation at a temperature of 95°C, followed by a 2nd step consisting of 35 cycles of 15 seconds of denaturation and hybridization at 94°C and 57°C respectively, primer elongation at 72°C/1 minute, and a 3rd step: final elongation or polymerization at 72°C for 5 min. PCR products were checked by electrophoretic migration on 1.5% agarose gel from 5 μl of amplicons. The size of each amplified gene was estimated using a 500 bp SmartLadder size marker.

The primer sequences and corresponding amplicon sizes are shown in Table 1.

Three founder mutations in the BRCA1 and BRCA2 genes were identified for PCR: 185delAG in exon 2 and 5382insC in exon 20 of the BRCA1 gene, and 6174delT in exon 11 of the BRCA2 gene [15-17]. Germline mutations in the BRCA1 and BRCA2 genes have been reported in several studies of different ethnic populations [16,18,19]. For each mutation, three primers (one common, one specific for the mutant and one specific for the wild-type allele) were used. The competing mutant and wild-type primers were designed to differ in size by 20 bp, allowing easy detection of the PCR products by routine electrophoresis. Both the mutant (long) and wild-type (short) primers contain a mismatched base sequence near the 3′ end. The long (mutant) primer also incorporates two additional mismatched bases at two contiguous positions corresponding to the 5′ end of the short (wild-type) primer. During the final cycles of the PCR reaction, heteroduplexes can be formed from the short and long products, but the contiguous mutagenic sequences in the long product prevent the short product from being filled in using the long strand as a template. If a mutation is present in one of the alleles, two bands will be present. PCR conditions were optimised for each primer pair and applied uniformly to all samples. Amplifications were performed in a reaction volume of 25 μl. The composition of the reaction mixture is given in Table 2.

Table 1: Primers used

Primers

Primers sequences Amplicon size

BRCA1-del185AG

Foward  5′ggttggcagcaatatgtgaa 3′
Reverse wild 5′gctgacttaccagatgggactctc 3′ 335pb
Reverse mutant 5′cccaaattaatacactcttgtcgtgacttaccagatgggacagta 3 ′ 354pb

BRCA1-5382insC

Foward wild 5′aaagcgagcaagagaatcgca 3′ 271pb
Foward mutant 5′aatcgaagaaaccaccaaagtccttagcgagcaagagaatcacc3′ 295pb
Reverse 5′gacgggaatccaaattacacag 3′

BRCA2-6174delT

Foward wild 5′gtgggatttttagcacagctagt 3′ 151pb
Foward mutant 5′cagtctcatctgcaaatacttcagggatttttagcacagcatgg 3′ 171pb
Reverse 5′agctggtctgaatgttcgttact 3′

Table 2: Composition of the PCR reaction medium for each gene

Volume to be sampled for a PCR with a reaction volume of 25 μl.

Reagents

Gènes amplifiés
BRCA1-185delAG BRCA1-5382insC

BRCA2-6174delT

Water

8,75 μl

8,75 μl 8,25 μl

Master mix

12,5 μl 12,5 μl

12,5 μl

Fw

0,25 μl

0,25 μl 0,25 μl

Fm

0,25 μl 0,25 μl

0,25 μl

R

0,25 μl

0,25 μl 0,25 μl

Mgcl2

1 μl 1 μl

1,5 μl

Results and Discussion

Results

For 59 patients recruited, 37 (62.71%) were diagnosed with prostate cancer (CaP) and 22 (37.29%) with benign prostatic hyperplasia (BPH). With regard to the BRCA1 (185delAG and 5382insC) and BRCA2-6174delT mutations, the frequency of BRCA1-185delAG mutations in patients with CaP was 40% compared with 60% in those with a BPH lesion, indicating that this mutation shows no significant difference in men with CaP and probably does not contribute to the incidence of this cancer. However, the other two BRCA1-5382insC and BRCA2-6174delT mutations showed higher frequencies (2 to 3 times) in patients with CaP than in those with the BPH lesion, with respectively 62.7% versus 37.3% and 65.1% versus 34.9%. For individuals with adenocarcinoma of the prostate, most cases had a Gleason score greater than or equal to 7 (87%); with 13% of individuals having a Gleason score equal to 6. Gleason scores for prostate tumours were classified into subgroups <7 and ≥7. This threshold was chosen based on clinical experience and previous literature suggesting that the clinical outcome for prostate cancer of Gleason score 7 is more similar to that of Gleason score 8 to 10 than for Gleason score <7 disease2 [20]. Table 3 shows the association between Gleason scores and BRCA1/BRCA2 mutations. Individuals with cape with BRCA1/2 germline mutations were more frequently associated with Gleason score ≥ 8, at stage T3/T4. BRCA1-5382insC and BRCA2-6174delT mutation carriers conferred a 2 to 3-fold increased risk of high-grade prostate cancer. Although the BRCA1-185delAG mutation has not been associated with prostate cancer, it may be associated with high Gleason score tumours. These results must be carefully taken into account in genetic counselling.

Table 3: Association between Gleason scores and BRCA1 /BRCA2 mutations

Gleason score/BRCA mutations

BRCA1-185delAG BRCA1-5382insC BRCA2-6174delT
N individuals N individuals

N individuals

Gleason score 6

0

5 4

Gleason score 7

4 12

9

Gleason score 8

6

13 11

Gleason score 9

0 2

2

Discussion

When analysing the genetics of CaP, it is essential to distinguish between localised, high-risk and metastatic disease. Firstly, due to the widespread adoption of PSA, the majority of new CaP diagnoses are low-grade localised disease with an excellent prognosis. These diagnoses are clinically distinct from the comparatively fewer diagnoses of advanced metastatic CaP [21] which are known to have the potential for a poor outcome. Several studies have shown that the genomic/genetic landscape of metastatic castration-resistant CaP (mCRPC) is different from that of localised [22,23]. It is difficult to obtain meaningful clinical predictions by examining CaP as a whole, given the great clinicopathological heterogeneity of the disease. This can be illustrated by germline mutations in BRCA2 which have been underestimated as a driver of hereditary prostate cancers. Genomic profiling of CaP was initially extrapolated from material acquired during unselected prostatectomies and genetic abnormalities were therefore considered rare [24]. As a result, verification bias prevented reporting the true prevalence of pathogenic genetic mutations in advanced metastatic cape. This work was designed to assess the impact of BRCA 1 and BRCA 2 mutations in prostate cancer in the Thiès region with the association of the three founder mutations BRCA1-185delAG and BRCA1-5382insC and BRCA2-6174delT. This study revealed that the highest frequency of BRCA1 mutations in CaP patients was BRCA1-5382insC (62%) followed by BRCA1 185-delAG (40%). The frequency of BRCA1 mutations in patients with a BPH lesion was 60% for BRCA1-185delAG and 38% for BRCA1-5382insC. In addition, the global BRCA2-6174delT mutation was identified in 65.1% of patients with cape versus 34.9% of those with a BPH lesion. These results suggest that the BRCA2-6174delT and BRCA1-5382insC mutations are likely to contribute to the incidence of prostate cancer in the Thiès region, which is not the case for the BRCA185-delAG mutation, which shows no significant difference in patients with CaP. Our results are comparable to those of Gallagher et al. in 2010 [24] and Agalliu et al. in 2009 [25] where they found mutation frequencies for the BRCA1-5382insC and BRCA2-6174delT genes to be largely predominant in individuals with CaP. Studies of breast cancer by Abou El Naga et al. reported contradictory results, with the two mutations (BRCA1-5382insC and BRCA2-6174delT) showing higher frequencies in healthy controls than in breast cancer patients [26]. In addition, we found that the risk of prostate cancer associated with carrying these mutations was higher in men diagnosed at an older age (65 or over) and in particular in those with the BRCA2-6174delT and BRCA1-5382insC mutations.

A number of previous studies have examined the associations between these BRCA1/BRCA2 mutations and prostate cancer [17,28-30]. Struewing et al. [27] estimated a lifetime risk of CaP of 16% for BRCA1/BRCA2 mutation carriers and 3.8% for non-carriers. Our results reported that BRCA2-6174delT and BRCA1-5382insC mutation carriers had two to three times the risk of prostate cancer, and as indicated here, the BRCA1-185delAG mutation was not associated with prostate cancer. Our results contradict those of Giusti et al. [30] who found the BRCA1-5382insC mutation not to be associated with prostate cancer. The absence of a detectable effect for the BRCA1-185delAG mutation could be linked to its low prevalence in the population, or to the effects of allelic heterogeneity. In support of a role for prostate cancer-associated BRCA2 mutations, studies of breast and/or ovarian cancer families harbouring disease-associated BRCA2 mutations have reported that male family members carrying such mutations have an increased risk of prostate cancer [31-33]. A Finnish study [34] of breast and/or ovarian cancer families also reported a 5-fold increase in the risk of prostate cancer in men carrying BRCA2 protein-truncating mutations. First-degree male relatives of breast cancer patients with protein-truncating BRCA2 mutations had a 4.8% risk of prostate cancer [35]. In 2012, studies by Castro et al. [36] reported that BRCA2 mutation status was found to be an independent predictor of median cause-specific survival. Interestingly, the non-carrier group also had a poorer outcome than other sporadic CaP series, suggesting that a family history of breast cancer could somehow affect the prognosis of prostate cancer patients.

In the present study, a major proportion of mutation carriers had a Gleason score ≥ 7 (87%); our results are similar to those of Gallagher et al. in 2010 [24] where 85% of mutation carriers had Gleason disease ≥7. Our results were striking, with 22 of 26 (84.6%) BRCA2-6174delT mutation carriers and 27 of 32 BRCA1-5382insC mutation carriers (84.3%) showing Gleason disease ≥7, representing a group with an aggressive phenotype and confirming this association reported by Agalliu et al. [25]. However, individuals carrying these two mutations may be associated with a higher risk of prostate cancer and are also likely to have poor survival as reported by Edwards et al. in 2010 [37]. This is also reflected in the study by Kote-Jarai et al. where the proportion of high grade CaP (Gleason score ≥ 8) was 63% significantly elevated [14].The study by Gallagher et al. reported that BRCA2 mutation carriers had an increased risk of CAP and a higher histological grade and that BRCA1 or BRCA2 mutations were associated with a more aggressive clinical course [24] results confirmed by studies by Castro et al. in 2013 in a large retrospective cohort [38].

Conclusion

Prostate cancer is the second most common cancer in men worldwide and is a complex heterogeneous disease with high heritability. Our results showed that BRCA2-6174delT and BRCA1-5382insC mutations are strongly associated with a very aggressive form of prostate cancer. Molecular characterisation of CaP patients should be systematically integrated into healthcare structures in order to select patients who are more likely to respond to targeted agents. In addition, in the event of a family history of hereditary breast cancer (± hereditary ovarian cancer), it is recommended that the patient be referred to an oncogenetic consultation to look for a mutation in the BRCA1 and BRCA2 genes. In the case of aggressive prostate cancer (high Gleason score or locally advanced or metastatic stage) in a patient under the age of 50, it is recommended that the patient be referred to an oncogenetic consultation to look for a mutation in the BRCA2 and HOXB13 genes (level of evidence 2a) [39]. Further clinical trials would be needed to assess the impact of genomic nuances in reducing the morbidity and mortality prevalent with prostate cancer.

Acknowledgement

The authors are very grateful to the patients who participated in this study. We are extremely grateful to Dr Modou Faye who helped for sample collection. Also Pr SEMBENE, head of the genomics laboratory for all the molecular studies carried out and Pr Tonleu Linda Bentefouet, head of the cytological and pathological anatomy unit for the histopathological diagnoses.

Conflict of Interest

The authors have declared no conflicts of interest.

References

  1. Sung H Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. (2021) Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 71: 209-249. [crossref]
  2. Santé OMdl Cancer. 2022. Available from: https: //www.who.int/fr/news-room/fact-sheets/detail/cancer.
  3. Hjelmborg JB, Scheike T, Holst K, Skytthe A, Penney KL, et al. (2014) The Heritability of Prostate Cancer in the Nordic Twin Study of Cancer. Cancer Epidemiol. Biomark Prev 3: 2303-2310. [crossref]
  4. Ferlay J, Ervik M, Lam F (2018) Global Cancer Observatory: Cancer Today; International Agency for Research on Cancer: Lyon, France.
  5. Termini D, Hartogh DJD, Jaglanian A, Tsiani E (2020) Curcumin against Prostate Cancer: Current Evidence. Biomolecules 10. [crossref]
  6. Wen S, Chang HC, Tian J, Shang Z, et al. (2015) Stromal Androgen Receptor Roles in the Development of Normal Prostate, Benign Prostate Hyperplasia, and Prostate Cancer. Am J Pathol 185: 293-301. [crossref]
  7. Cittadini A, Isidori AM, Salzano A (2021) Testosterone therapy and cardiovascular diseases. Cardiovasc Res 118: 2039-2057. [crossref]
  8. Rawla P (2019) Epidemiology of Prostate Cancer. World J Oncol 10: 63-89. [crossref]
  9. Parkin DM, Bray F, Ferlay J, Pisani P (2005) Global cancer statistics, 2002. CA Cancer J Clin 55: 74-108. [crossref]
  10. Bluemn EG, Nelson PS (2012) The androgen/androgen receptor axis in prostate cancer. Curr Opin Oncol 24: 251-257. [crossref]
  11. Ziaran S, Novakova ZV, Böhmer D, DanišoviˇcL (2015) Biomarkers for determination prostate cancer: Implication for diagnosis and prognosis. Neoplasma 62: 683-691. [crossref]
  12. Patel AR, Klein EA (2009) Risk factors for prostate cancer. Nat Clin Pract Urol 6: 87-95.
  13. Messina C, Cattrini C, Soldato D, Vallome G, Caffo O, et al. (2020) BRCA Mutations in Prostate Cancer: Prognostic and Predictive Implications. J Oncol 2020: 4986365. [crossref]
  14. Kote-Jarai Z, Leongamornlert D, Saunders E, Tymrakiewicz M, Castro E, et al. (2011) BRCA2 is a moderate penetrance gene contributing to young-onset prostate cancer: Implications for genetic testing in prostate cancer patients. Br J Cancer 105: 1230-1234. [crossref]
  15. Ewald IP, Izetti P, Vargas FR, Moreira MA, Moreira AS, et al. (2011) Prevalence of the BRCA1 founder mutation c.5266dupin Brazilian individuals at-risk for the hereditary breast and ovarian cancer syndrome. Hered. Cancer Clin Pract 9.
  16. Abeliovich D, Kaduri L, Lerer I, Weinberg N, Amir G, et al. (1997) The founder mutations 185delAG and 5382insC in BRCA1 and 6174delT in BRCA2 appear in 60% of ovarian cancer and 30% of earlyonset breast cancer patients among Ashkenazi women. Am J Hum Genet 60: 505-514. [crossref]
  17. Ford D, Easton DF, Stratton M, Narod S, Goldgar D, et al. (1998) Mar. Genetic heterogeneity and penetrance analysis of the BRCA1 and BRCA2 genes in breast cancer families. The breast cancer linkage consortium. Am J Hum Genet 62: 676-689. [crossref]
  18. Neuhausen SL, Ozcelik H, Southey MC, John EM, Godwin AK, et al. (2009) Breast Cancer Family Registry. BRCA1 and BRCA2 mutation carriers in the Breast Cancer Family Registry: an open resource for collaborative research. Breast Cancer Res Treat 116: 379-386. [crossref]
  19. Hartwig M, Janiszewska H, Bąk A, Pilarska M, Heise M, et al. (2013) Prevalence of the BRCA168_69delAG (BIC: 185delAG) mutation in women with breast cancer from north-central Poland and a review of the literature on other regions of the country. Contemp Oncol (Pozn) 17: 34-37.
  20. Thompson KE, Hernandez J, Canby-Hagino ED, Troyer D, Thompson IM (2005) Prognostic features in men who died of prostate cancer. J Urol 174: 553-556; discussion 56. [crossref]
  21. NCCN Clinical Practice Guidelines in Oncology. Prostate Cancer Version 2.2031; National Comprehensive Cancer Network: Plymouth Meeting, PA, USA, 2021.
  22. Robinson D, Van Allen EM, Wu YM, Schultz N, Lonigro RJ, et al. (2015) Integrative clinical genomics of advanced prostate cancer. Cell 161: 1215-1228. [Crossref]
  23. Network CGAR (2015) The Molecular Taxonomy of Primary Prostate Cancer. Cell 163: 1011-1025. [crossref]
  24. Gallagher DJ, Gaudet MM, Pal P, Kirchhoff T, Balistreri L, et al. (2010) Germline BRCA mutations denote a clinicopathologic subset of prostate cancer. Clin Cancer Res 16: 2115-2121. [crossref]
  25. IlirAgalliu, Robert Gern, Suzanne Leanza, Robert DB (2009) Associations of High-Grade Prostate Cancer with BRCA1 and BRCA2 Clin Cancer Res15 February 1, 2009.
  26. Amoura Abou-El-Naga, Ahmed Shaban, Hayam Ghazy, Afaf Elsaid, Rami Elshazli, Ahmad Settin (2018): Frequency of BRCA1 (185delAG and 5382insC) and BRCA2 (6174delT) mutations in Egyptian women with breast cancer compared to healthy controls, Meta Gene 15: 35-41.
  27. Struewing JP, Hartge P,Wacholder S, et al. (1997) The risk of cancer associated with specific mutations of BRCA1 and BRCA2 among Ashkenazi Jews. NEngl J Med 336: 1401-1408. [crossref]
  28. Kirchhoff T, Kauff ND, Mitra N, et al. (2004) BRCA mutations and risk of prostate cancer in Ashkenazi Jews. Clin Cancer Res ;10: 2918-21. [crossref]
  29. Vazina A, BanielJ, ShtrikerA, et al. (1999) The prevalence of the major Jewish mutations BRCA1 and BRCA2 in prostate cancer patients in Israel. J Urol 161: 55-56. [crossref]
  30. Giusti RM, Rutter JL, Duray PH, et al. (2003) A twofold Iicrease in Brca mutation related prostate cancer among Ashkenazi Israelisisn ot associated with distinctive histopathology. J Med Genet 40: 787-792. [crossref]
  31. The Breast Cancer Linkage Consortium. Cancer risks in BRCA2 mutation carriers (1999). J Natl Cancer Inst 91: 1310-1316.
  32. Van Asperen CJ, Brohet RM, Meijers-Heijboer EJ, Hoogerbrugge N,Verhoef S (2005) Cancer risks in BRCA2 families: estimates for sites other than breast and ovary. JMed Genet 42: 711-719. [crossref]
  33. Thompson D, Easton DF (2002) Breast Cancer Linkage Consortium. Cancer incidence in BRCA1mutation carriers. JNatl Cancer Inst 94: 1358-1365. [crossref]
  34. Eerola H, Pukkala E, Pyrhonen S, et al. (2001) Risk of cancer in BRCA1 and BRCA2 mutation positive and-negative breast cancer families (Finland). Cancer Causes Control 12: 739-746. [crossref]
  35. Tulinius H, Olafsdottir GH, Sigvaldason H, et al. (2002) The effect of a single BRCA2 mutation on cancer in Iceland. JMed Genet 39: 457-462. [crossref]
  36. Elena Castro, Rosalind Eeles (2012) The role of BRCA1 and BRCA2 in prostate cancer. Asian Journal of Andrology 14: 409-414. [crossref]
  37. Edwards SM, Evans DG, Hope Q, Norman AR, Barbachano Y, et al. (2010) Prostate cancer in BRCA2 germline mutation carriers is associated with poorer prognosis. Br J Cancer 103: 918-924. [crossref]
  38. Elena Castro, Chee Goh, David Olmos, Ed Saunders, Daniel Leongamornlert, et al. (2013) Germline BRCA Mutations Are Associated With Higher Risk of Nodal Involvement, Distant Metastasis, and Poor Survival Outcomes in Prostate Cancer. J Clin Oncol 31: 1748-1757. © by American Society of Clinical Oncology.
  39. Rozeta F, Hennequin C, Beauval JB, Beuzeboc P, Cormier L, et al. (2018-2020) Recommandations françaises du comité de cancérologie de l’AFU-Actualisation.

Function Inspired Structures of Proto-Ribosome and the First Aminoacyl-tRNA Synthetase: Origin of life in the water of the Earth (III)

DOI: 10.31038/GEMS.2023581

Abstract

The first protein must produce at random processes. However, it is difficult to replicate correct protein without any control. The activities of control must be different from activities to be control. It is known that organisms replicate proteins via ribosomes by using genetic information. The mechanism that had replicated proteins naturally is a bridge between living organism and nonliving organism. We can make assumptions about structure of proto-ribosome on the base of its functions. That is, a proto-ribosome would have sandwiched L-type mRNA and D-type tRNA in part of a phospholipid bilayer. This structure can be used to estimate the initial processes of replicated proteins and the initial formation of aminoacyl-tRNA synthetases.

Keywords

Enzyme world, Ribosome, DNA, mRNA, tRNA, Aminoacyl-tRNA synthetase

Introduction

The first life had formed in non-extreme environment on the Earth because the first cell with gene system must have naturally formed. However, many of traditional studies of the origin life have been carried out on extremophiles [1], because most of professional researchers must acquire budgets, they have been studied by the acquired budget. The organisms had been born, then evolved by adaptations in its environment. By the results of evolution, extremophiles can live only in extreme environment [2]. On the other hand, there are numerous descriptions on molecular biology [3]. Especially ongoing progress in the structural biology is giving a physico-chemical basis that explains facets about tRNA [4]. Replication of protein is controlled via the informational media of double helix of DNA discovered by Watson and Click [5]. Karasawa reported that the replication processes of DNA are possible to reveal based on the structure of DNA [6,7]. The initial process of protein replication can be revealed based on the structure of the proto-ribosome which includes mRNA and tRNA in a part of a phospholipid bilayer. Here, the mRNA is left-handed (L-type) chirality, while the tRNA is right-handed (D-type). Although these two strands with different chirality enter a plathome of processing of central part of the double-layer, those strands never coalesce. The double-layered helical structures are indispensable in the ribosome-translation machinery.

Preparations

Formation of mRNA, tRNA and DNA

Organic molecules such as hydrocarbons were accumulated on the surface of water, and macroscopic boundary conditions formed a membrane [6]. When amino acid molecules adhered to the membrane, those molecules were formed molecular structures possessing with the function of enzymes. The first organization of life had formed in the world of enzymes. Current protein is replicated in a ribosome via short-lived mRNA and tRNA. Those mRNA and tRNA are produced from a replicated DNA [7]. Since RNA is a tool to deal with genetic information, the first life should be discussed in the real world of enzymes instead of the informational world of RNA.

Matching Processes between mRNA and tRNA

When amino acids adhere a membrane, conformation of a part of membrane is modified. The changed conformation of the membrane includes information on the amino acid sequence of a protein. Even though a proto-DNA is formed by simplifications from the membrane by exclusion of the protein, it possesses information of size and segmentation on the amino acid. Such information is used for the first matching processes between template of mRNA and matching objects of tRNA. Subsequent evolutions, pairing relationships for the matching was established by complementary base pairs of codons and anticodons. So, intermittently fixing of a segment of mRNA for a specified amino acid in a protein, each tRNA is shifted along the mRNA in order to looking for the partner of hydrogen bonds.

Leading Strand and Lagging Strand of DNA

Since chirality of mRNA is L-type but tRNA is D-type, two kinds of RNA do not merge. It is known that the chirality of biomolecule in the Earth, amino acids are L-type, and sugars are D-type. So, the membrane adhered with protein and bases has L-type chirality, but D-type of helical structure will be formed due to antagonistic and organized movements of interconnected helix structures [7]. That is, alternate rotations around X axis through the center of tetrahedron units changes the shape of tetrahedral unit projected in the X-Y plane from square to trapezoidal. When a pair of atoms of tetrahedron located at the end of the long and short site vibrate up-and-down movements, inner and outer in the tetrahedron’s vertices vibrate opposite directions. Under an assumption of such antagonistic movements, leading strand and lagging strand of DNA are synthesized simultaneously at each segmentation of the amino acid. Since each amino acid has individual size, the segment of constituent of ribosome for an amino acid is the same. tRNA is formed by a single D-type of lagging strand with base due to chirality difference between amino acid and lagging strand. So, the leading strand replicates continuously, whereas the lagging strand replicates discontinuously forming short fragments. Since bases of two RNAs touch via hydrogen bonds, tRNA is possible to move independently from mRNA. Here, the complementary anticodon of tRNA is vertical flip symmetry of corresponding amino acid of mRNA.

Results

Formation of Proto-ribosome in a Phospholipid Bilayer

Phospholipid contains a chiral center at C2 position of a glyceryl moiety [7]. Twisted phospholipids laterally interlock, and the interlock induces systematic motions due to systematic thermal vibrations of atoms [8]. A double layer sandwiched between two layers of hydrophilic heads spontaneously forms. If one of the layers in part of the bilayer is rotated by 180°, the progress of the helix is changed from the output side to the input side at the center of the bilayer. This bilayer, despite having two chirality centers, provides a one-directional screw movement by the one-directional rotation. When only mRNA enters the bilayer, it passes through the bilayer. However, if tRNA enters from the other side of the mRNA, both strands come into conflict at the central portion owing to chirality [7]. mRNA and the series of tRNA’s are sandwiched between a protein and a series of amino acids with base pairs facing each other at the center. Figure 1 is an illustration of a structure of proto-ribosome and its constituents proposed in this paper.

FIG 1

Figure 1: A structure of proto-ribosome and with its constituents

Prospect of the Protein Replication: Evolutions of Gene System on Chain Reactions

When a biological reaction is performed, a new reaction occurs due to change of the situation caused by the reaction. A chain reaction will continue to circulate if it forms a loop. A relationship of “from demand to the supply” will be included in those chain reactions. In various chain reactions, protein molecules that express repeated chain reactions will be formed and the enzymes will be formed. The chain reactions those support the survival of life will be incorporated into genes system in the form of long DNA. Then, Prokaryote have evolved to Eukaryote by formation of a nucleus of the cell in order to memorize very long DNA.

Discussions

Molecular Mechanisms Underlying Ribosome Dynamics

A step of protein replication is proceeded by amino acid unit at platform of a ribosome. The triplet base pairs in a DNA for each amino acid are formed via pattern matching on hydrogen bond between codon of mRNA and anticodon of tRNA. The complementally base pairs are adenine with thymine (A-T) and cytosine with guanine (C-G) for each base step. Incidentally, an aminoacyl-tRNA synthetase makes linkage between the triplet code and an amino acid by the direct attachment of an amino acid and corresponding tRNA [9]. Tamura, Schimmel reported about non-enzymatic aminoacylation of an RNA minihelix [10]. Karasawa proposes a functional model of aminoacyl-tRNA synthetase that comes from a cover around the functional model of aminoacyl-tRNA as shown in Figure 1. The linkage between amino acid and the triplet is carried out by an aminoacyl-tRNA synthetase. The proto-DNA forms a unique conformation when interacting with amino acids. The proto-DNA must possess information on amino acid. The enzyme binds with specific molecules, resulting in a conformational change, and carries out function of the catalyst. However, even the base sequence of tRNA has been revealed, understanding the molecular mechanisms underlying tRNA dynamics is yet challenging [11].

Conclusions

The author proposes that the first life should be discussed in the real world of enzymes instead of the informational world of RNA. Over the course of evolution, if a new mechanism is added alongside conventional mechanisms that functioning during life activities, the new mechanism must coexist with the conventional system. Eventually, the new system that successes to survive will remain, and unnecessary system will disappear. The research based on the current system is difficult to reveal the disappeared structures. The proto-ribosome was estimated based on the necessity that shifts tRNA reversely for mRNA and confirms the matched amino acid sequences. We can describe fundamental functions of ribosome by assuming such simple initial structure and its environments. The bottom-up approaches based on acceptable assumptions are useful to reveal initial processes of protein replications. However, there is a gap between the proposed model and current nucleotide sequence models in the molecular biology. It is known that the evolution of living organisms has influenced the Earth’s atmosphere and geology. It is another desire of the author that the proposed functional models will become bridges between living organisms and the field of geology of the Earth.

References

  1. Merino N, Heidi SA, Diana PB, Jayme FB, Michael LW, et al. (2019) Living at the Extremes: Extremophiles and the Limits of Life in a Planetary Context, Front Microbiol. [crossref]
  2. von Hegner I (2021) Extreme exoworlds and the extremophile paradox, arXiv.org | Cornell University Library, September.
  3. Watson JD, et al. (2004) Molecular biology of the Gene, 5th Ed. Benjamin Cummings.
  4. Nakanishi K, Nureki O (2005) Recent progress of structural biology of tRNA processing and modification. Mol Cells 19: 157-166. [crossref]
  5. Watson JD, Click FH (1953) Genetical Implications of the Structure of Deoxyribonucleic Acid, Nature 171: 964-967.
  6. Karasawa S (2023) Origin of life in the water of the Earth. Geology, Earth and Marine Science 5: 1-7.
  7. Karasawa S (2023) Initial processes on replication of DNA by interactions of helical structured molecules-Origin of life in the water of the Earth (Ⅱ). Geology, Earth and Marine Sciences 5: 1-7.
  8. MacKenzie LE, Stachelek P (2021) The twists and turns of chiral chemistry. Nature Chemistry 13: 521-522.
  9. Ibba M, Soll D (2000) Aminoacyl-tRNA synthesis. Annual Review of Biochemistry 69: 617-650.
  10. Tamura K, Schimmel P (2004) Non-enzymatic aminoacylation of an RNA minihelix with an aminoacyl phosphate oligonucleotide. Nucleic Acids Symposium Series 48: 269-270.
  11. Giege R, Frank J, Joern P, Peter S, Claude S, et al. (2012) Structure of transfer RNAs: Similarity and variability. WIREs RNA 3: 37-61. [crossref]

Predictors of Listening and Reading Comprehension in Arabic as the First Language and Listening and Comprehension in English as a Foreign Language: Two Different Orthographies

DOI: 10.31038/ASMHS.2023722

Abstract

Both listening and reading comprehension have not been sufficiently studied in the context of Arabic and English language orthographies. The main goal of the current research was to investigate how listening and reading comprehension among native Arabic speakers predict the use of Arabic and English language orthographies in learning English as a foreign language (FL) in Israel. The Arab minority in Israel learn three languages: Arabic as their first language (L1), Hebrew as the language of the state of Israel (L2), and English as a foreign language (FL). Consequently, the dissimilarity between Arabic and English orthographies poses several challenges in learning English as a foreign language among Israeli Arab students. Arabic and English are alphabetical writing systems but represent different orthographies. A total of 100 Arabic-speaking high school students were asked to administer a set of phonological, linguistic, and cognitive scales. The results of the present study indicated that predictors of listening and reading comprehension in Arabic and predictors of listening and comprehension in English are different. Another finding of the present study was that predictors in Arabic language predict listening and reading comprehension in English. However, the Arabic skills predictors for listening comprehension are different than the Arabic skills predictors for reading comprehension. The results are discussed in light of previous findings in the literature as well as in relation to the different orthographies of Arabic and English languages.

Keywords

Reading comprehension, Listening comprehension, Two different orthographies, First and foreign languages

Introduction

The present research examines the Arabic and English language orthographies and their effect on listening and reading comprehension among Arab learners of English as a foreign language in Israel. Arabic and English are alphabetical language systems but differ in many aspects. This research aims to focus on the difference between Arabic and English orthographies, which lies within the relationship between the sound and letters also known as orthographic depth. Therefore, this research aims to clarify how listening and reading comprehension among Arabic speakers predict the use of these two abilities in English as a foreign language. The novel aspect of this research is examining listening and reading comprehension among Arabic learners of English in relation to their different orthographies in one study. It is expected that the characteristics of English and Arabic orthographies may not affect linguistic skills, as learners with different orthographic backgrounds adapt distinct linguistic skills to learn the language. Additionally, the learners’ performance in listening and reading comprehension may be lower in English compared to Arabic, considering the distinct characteristics of Arabic and English orthographies.

By studying Arabic and English language orthographies among Arabic learners in Israel, we can better understand the process of how learners with different orthographic backgrounds learn a language. In this paper, the uniqueness of the Arabic and the English language orthographies and their features and role in listening and reading comprehension will be discussed in depth. Subsequently, the research methodology and procedure will be presented. Finally, the findings will be analyzed followed by the discussion, conclusions, and limitations of the study.

Literature Review

Few studies have investigated the role of listening and reading comprehension in the first language as a predictor of linguistic skills in another language [1,2]. Listening comprehension and reading comprehension are similar skills but constitute two distinct forms of comprehension involving different cognitive processes [3-7]. Listening comprehension depends on an understanding of a spoken language while reading comprehension depends on an understanding of a written language [8]. While reading comprehension depends on decoding skills, which are predicted by phonological awareness, listening comprehension depends on word processing and is predicted by vocabulary knowledge [9]. Aspects of language processing and skills, namely Arabic orthography, English orthography, reading comprehension, listening comprehension, transfer of skills, phonological decoding skills, spelling, morphology and vocabulary, syntax, speed of processing, orthographic knowledge, working memory, attitudes and language, will be deeply explored in order to investigate predictors of listening and reading comprehension in both Arabic and English, two different orthographies.

Arabic Orthography

The Arabic language is a Semitic language, and it has 28 letters, including six vowels: three short vowels and three long vowels [10]. The short vowels are presented by diacritic marks above or beneath the letter [11], while long vowels are represented by the following letters: “alef,” “ya,” and “waw” [12]. Short vowels in Arabic add phonological information for word decoding and thus contribute to understanding texts [13,14]. In other words, the short vowels contribute to the correct pronunciation of Arabic sounds and letters [15]. This language is read and written from right to left and is regarded as shallow orthography that exhibits a predictable relationship between letters and sounds [16]. As a result, the development of literacy skills such as reading and spelling are learned quicker than in inconsistent orthographies [17]. In addition, the Arabic language is regarded as deep orthography when unvowelized [14,18,19]. In Arabic, short vowelization contributes to reading comprehension of less complex texts, such as informative and narrative texts and newspaper articles among readers of different ages and levels [20,21] reviewed the role of vowelization in Arabic and showed that short vowelization improves reading accuracy and reading comprehension. Other studies by [13,16,22,23] proved that Arabic orthography poses an intensive visual burden and slows down reading and can be considered as deep orthography. Therefore, reading in Arabic depends on the visual orthographic representation of a word. Furthermore, Arab readers after the fourth grade are expected to learn and read unvowelized texts in which they tend to rely on sentence context [13].

It is interesting to note the phenomenon of diglossia in the Arabic language, which refers to the use of two versions: spoken Arabic and literary Arabic [13,24,25]. Spoken Arabic has various dialects, and native Arabic speakers use spoken Arabic in everyday life, but literary Arabic is used in education, writing, the Qur’an, and literature [13,22,25,26]. Spoken and literary Arabic vary considerably in terms of phonology, vocabulary, syntax, and grammar and thus affect language development and reading acquisition [11,14,15,29]. This phenomenon also affects reading and writing in Arabic as it has negative effects on phonological awareness that is associated with reading and spelling acquisition [26].

English Orthography

The English language has 26 letters, including 5 vowels and 21 consonants, and it is read and written from left to right [10]. The vowels in English are letters within the words and represent the sounds [28]. Some letters are written in a word but not pronounced as they are spelled, which is more common in English than in Arabic [26,29]. English is also an alphabetical system characterized by an indirect relationship between a letter and its sound [16]. Therefore, English is regarded as deep orthography, in which the same letter may have various sounds written differently [10,30]. For example, the grapheme-phoneme “gh” is pronounced differently in the following words “ghost,” “light,” and “tough.” As a result, the rate of reading development in English is low among children in comparison with more consistent orthographies [30,31].

Second and foreign language learners tend to use L1 linguistic skills and learning strategies to learn another language according to the language linguistic skills transfer [28,32]. Although English and Arabic are alphabetical systems, the degree of dissimilarity of their orthographic representation affects learning English as a foreign language by Arabic-speaking learners [14]. According to the orthographic depth hypothesis, in deep language orthography, semantic context (lexical) is used for word recognition, while shallow language orthography relies on the phonology for decoding words [11,10]. In the case of the Arabic language, which is a shallow orthography, Arabic-speaking learners of English rely on consonants in word decoding for word recognition [33]. However, in the English language, the corresponding spelling-to-sound is inconsistent, and thus learners must rely more on activating both phonological and orthographic processes in learning and reading [34]. A more recent study by [8] shows that in deep orthographies, word recognition is the main predictor of reading comprehension, especially in an early stage of reading development, while in shallow orthographies listening comprehension is more significant in reading comprehension in different phases of reading development.

Reading Comprehension

Scholars have widely investigated reading comprehension in first and second languages [8,15,11,35,36]. Numerous studies explored reading comprehension and demonstrated that it is a determinant of reading accuracy [8,10]. Reading comprehension of written texts is the outcome of decoding and listening comprehension [1,14,36,37]. It also refers to the ability to extract meaning from written representations of the language in order to construct new meanings by activating previous knowledge [15,25]. The two foundational skills of reading comprehension are word recognition, which refers to the ability to read individual words accurately, and listening comprehension [8]. In a more recent study, word recognition highly contributes to reading comprehension among beginner readers, and listening comprehension appears strongly to correlates with advanced readers [36]. Previous studies also found that vocabulary contributes to reading comprehension [35,37].

It is important to note that these skills of reading comprehension may function differently according to the orthographic features of the language. In shallow orthographies, learners depend more on phonology and the high consistency of the sound and letters in word recognition and word learning [11,38]. In contrast, in deep orthographies, learners must learn the complexity of the low consistency of the sound and letters in word recognition [11,38]. In addition, word recognition appears to be a predictor of reading comprehension in deep orthographies while listening comprehension appears to be a determinant predictor of reading comprehension in shallow orthographies [8]. In deep alphabetic orthographies, learners depend on increased word processing skills, such as phonetic awareness, letter-to-sound relationship, and visual representation of the word in ESL reading comprehension compared to less deep alphabetic orthographies [38].

Listening Comprehension

Listening comprehension is critical for language acquisition and reading comprehension and also has a significant role in the communication and language learning process [18,34,39]. However, unlike reading comprehension, listening comprehension has failed to attract the attention of language researchers [7,34], despite the rise of audio-visual platforms, such as TV and computers among children and adolescents. Listening comprehension refers to the ability to construct the meaning of spoken language and relate it to previous knowledge [8]. The literature review shows that the process of listening comprehension involves several key components, including word recognition, syntax, vocabulary, speed of talk, and previous knowledge that impact listening comprehension [18]. Unlike reading comprehension, the process of listening comprehension poses a major challenge for learners, as it requires rapid processing of meaning and linguistic skills, including syntax and lexical, due to the transient and temporary nature of the spoken text [36,40]. The contribution of listening comprehension to reading comprehension increases with age while decoding decreases as readers become increasingly proficient in decoding as they get older [14,34,35]. Vocabulary is the most important component of listening comprehension [7,35].

Little research has been done on listening comprehension in L1 and its role in predicting later language skills in a second language [34,36]. [39] has suggested that pre-listening activities and repetitive listening to a passage enhance listening comprehension and contribute to language proficiency among learners of Arabic as a foreign language. Because Arabic is a diglossic language, listening comprehension is reportedly affected by the spoken language as it relies on oral language [14]. Additionally, short vowels in Arabic contribute to listening comprehension across all grade levels [18]. However, the role of spoken language and listening comprehension across a variety of languages is still unclear [8].

A closer look at the literature on listening comprehension and reading comprehension, however, reveals a number of gaps and shortcomings. Therefore, this research addresses the need for an in-depth examination of the predictors of listening and reading comprehension in different orthographies among Arabic-speaking learners of English as a FL.

Transfer of Skills

Cummins’ (1991) linguistic interdependence hypothesis suggests a relationship between the first language and the learning of the second language. This relationship is indicated in the transfer of language skills, including phonological awareness, word recognition, reading comprehension, and other linguistic skills between languages [31]. This hypothesis depends on language-independent skills, such as phonology, morphology, and syntax that transfer across languages [11,34]. Studies on the role of listening comprehension and the transfer of this skill from L1 to reading comprehension in L2 are rare however [34]. Most studies have focused on language linguistics skills transfer, including reading, spelling, and phonological awareness and have neglected oral language skills transfer [2,16,34,41]. The transfer of skills across languages, either first to second\foreign language or in the opposite direction has been widely investigated by researchers [11,16,27]. In a study that investigated the transferability of phonological awareness in opposite direction, from L2 to L1, it was found that improvement in linguistic and meta-linguistic skills in a second language positively influenced similar skills in L1 [41]. It is interesting to relate that the transfer of linguistic and language skills is affected by the degree of similarity between the two languages [11,34,38]. As a result, a high degree of similarity between L1 and L2 enables the transfer of language skills, while distinct orthographic backgrounds of L1 and L2 affect decoding efficiency and word learning [42]. Despite decades of research, this issue continues to be debated regarding the transfer of linguistic skills across different languages [11,15,27,28,41].

Phonological Decoding Skills

Phonological awareness is one of the most researched basic meta-cognitive linguistic skills in language learning, and it refers to the awareness of the sound structures of language units [24,41]. Most studies, early as well as current, have demonstrated the importance of phonological awareness in predicting high performance in the reading process across languages [11,15,24]. A study of reading-disabled Arabic-English bilingual Canadian children showed that they scored higher scores in phonological tests than monolingual English-speaking children, despite the different nature of the two orthographies [28]. However, some studies have assumed that the degree of similarity of orthographic systems between L1 and L2 influence the learners’ performance in phonological decoding and word learning in L2 [42]. In another study, it was noted that differences in the orthographic background between L1 and L2 might affect phonological processing and word decoding in L2 [2]. In a recent study, an intervention program in English for improving linguistic and meta-linguistic skills in Hebrew as L1 and English as FL indicated a significant improvement in most skills, including phonological awareness from pre-intervention to post-intervention among three groups of readers: dyslexic, poor, and normal readers [10]. In short, phonological awareness is a cross-language skill; with increased exposure and practice of the target language, learners may be able to become more skilled in phonological skills [2,11].

Spelling

Spelling skills refer to the visual representation of a spoken language and depend on other language skills, such as phonology, orthography, and morphology of the target language [38,43]. Spelling involves two significant processes: phonological awareness and alphabetic knowledge that influence the ability in learning to read, write, and spell [2,44]. In a study by [33], Japanese poor readers revealed high performances in recognition tasks compared to Russian readers, due to their L1 experience with kanji (logographic Japanese writing system). Likewise, [28] proved that spelling is a cross-language correlated skill among Arabic-English disabled bilingual readers. The varying degree of orthography between languages affects the development of reading and spelling, as deep orthographies impede spelling skills compared to shallow orthographies [43]. The orthographic features of the Arabic language pose difficulties in spelling and reading among learners considering the diglossic situation and when Arabic is unvowelized [26]. More current research also has indicated that spelling needs direct instruction in the target language, supporting the fact that each language has its own orthographic rules and system [27]. Therefore, it is critical in this study to thoroughly examine spelling skills in both the Arabic and English languages.

Morphology and Vocabulary

English and Arabic have different morphological systems and features [21,22]. Arabic is a Semitic language characterized by root and pattern in which vowels determines the meaning of the word [12,16]. In contrast, English is characterized by semantic stem and morphemes (prefixes and suffixes), which allows the recognition of the meaning of the word [10]. Shallow orthographies are often complicated morphological systems while deep orthographies are less complicated [15,11]. As a result, it is assumed that it should be easier to transfer morphological skills from complex morphological systems like Arabic compared to English but the studies have proved that morphological skills are cross-language transferable [15,11]. In addition, vocabulary contributes significantly to both reading comprehension and listening comprehension [7,35]. The processing of information in listening comprehension is less focused on the vocabulary as the spoken text has a limited time frame, in contrast to reading comprehension, in which the text is permanent, allowing the reader to decode the meaning of the written text [40]. It is critical to deeply explore the role of morphology and vocabulary in predicting listening and reading comprehension both in Arabic and English and their influence among adult learners.

Syntax

Language syntax refers to the understanding of sentence structures and word order [28,35,45]. Moreover, syntax is a transferable linguistic skill across languages and contributes to improvement in both languages despite their different orthographies [11,15,41]. In a recent study, it was reported that syntax relates to reading comprehension and improvement in syntax contributes positively to reading comprehension among L2 learners of English [35]. Additional studies to better understand the role of syntax skills in language learning are needed.

Speed of Processing

The varying degree of orthography between languages affects both reading speed and comprehension [41]. In a study by [45], differences in reading were apparent among Hebrew L1 speakers when reading Hebrew and English passages. Despite Hebrew being their L1, they were more proficient in reading in English. In another study, [21,22] demonstrated that the complexity of Arabic orthography affects word identification and letter processing when compared to Hebrew. Thus, the participants demonstrated low performance in Arabic tasks in comparison to similar tasks in Hebrew. Other findings suggest that language orthography is not a direct indicator of reading comprehension levels and that reading speed is not significant in the prediction of reading comprehension. For example, a study by [6] revealed that oral comprehension is the strongest predictor of reading comprehension rather than reading speed and accuracy in Italian. Likewise, [8] suggest that listening comprehension was the best indicator of reading comprehension in comparison to oral reading fluency and word recognition in European Portuguese. Therefore, other components rather than speed seem to increase reading comprehension levels. However, the research is limited on the role of speed and its contribution to reading comprehension and listening comprehension. Thus, the speed of processing issue is still debatable and future research should be conducted.

Orthographic Knowledge

The ability to recognize written orthographic symbols of a language helps in understanding written texts or in identifying words known as orthographic knowledge [10,11,15,27]. Both Arabic and English are alphabetical orthographies, but they vary considerably in the consistency of the relationship between letters and sounds [5,29]. This variance refers to the consistency between the graphic symbol and the sound of each writing system of a language [10,29]. In other words, this variance is known as orthographic depth and has an impact on the process of reading and learning a language [5,10,11,14,30]. Orthographic depth represents the degree of sound-symbol correspondence [38]. Arabic seems to be more consistent and it is considered shallow orthography when vowelized and deep orthography when unvowelized [11,16], while English seems to be the least consistent alphabetic writing system and is considered deep orthography [16,28,30].

A large number of existing studies have examined the cross-language transfer of orthographic skills [10,11,15,16,27,28]. However, the existing research remains inconclusive on whether orthographic skills are cross-language or language-specific. Previous studies have shown that bilingual Arabic and English-speaking children performed better in pseudo-word reading and spelling tasks than monolingual English speaking children, which reflects a positive transfer from Arabic to English [28]. Recent work has proved that orthographic skills are cross-language transferable from English to Hebrew after an intervention program in English [31]. However, other studies have found that orthographic knowledge is language-specific and thus unlikely to be transferred from one language to another [16,41]. Another study also reported that orthographic knowledge is language specific and cannot be transferred from L2 to L1 among poor readers in the experiential group [11]. In addition, no significant improvement in orthographic skills both in Arabic and Hebrew was apparent after the intervention program in English L3 [15]. A more recent study confirmed that orthographic differences in languages influence the acquisition of linguistic and meta-linguistic skills, as orthographic knowledge did not show significant improvement in Hebrew following an intervention program for EFL among dyslexic, poor, and normal readers in the experimental group [10]. Therefore, it is crucial to investigate orthographic skills more extensively in order to understand their contribution and transferability across languages.

Working Memory

According to [28,45], working memory in reading refers to the capacity to retain information within the short-term memory storage while processing information. The study of bilingual Arabic-English Canadian children by [28] found that working memory is correlated across languages despite different orthographies. Furthermore, a study by [46] revealed that basic visual memory is significant in the process of reading Arabic as well as Hebrew, but not English. More recent research suggests that shallow and deep orthographies impact the performance of reading and writing as shallow orthographies are less memory dependent [43]. According to [3], short vowels in Arabic add phonology to words, which help in saving information in working memory to understand written or read aloud texts. This approach remains briefly addressed in the literature. Therefore, more research is needed on working memory skills and their contribution to learning a language in different orthographies.

Attitudes and Language Acquisition

The educational system in Israel requires the Arab minority to learn three languages in schools: Arabic as their first language, Hebrew as a second language, and English as a foreign language [16,15]. Attitudes toward language acquisition and its effect on learning Hebrew and English among the Arab minority in Israel have been widely investigated by [22,47]. Abu Rabia showed that instrumental and integrative attitudes motivate learners to learn the target language. The instrumental attitude suggests that learners learn the target language for practical reasons such as academic studies, while the integrative attitude suggests that learners identify with the target-language group and are willing to learn its language [22,47]. Many researchers claim that positive attitudes toward a language can facilitate the learning of the target language [48,49]. In addition, teachers play a significant role in developing positive attitudes toward that target language [49]. Other studies have suggested both motivation and attitude in language learning have a direct effect on learning a language [49,50]. Further research has pointed out that two types of motivation—integrative motivation and instrumental motivation—promote successful learning of the target language with emphasis on integrative motivation in maintaining long-term success [51,50]. This study is intended to examine attitudes toward learning English as a foreign language, among Arab high school students in Israel. Investigating this variable can contribute significantly to the researched topic.

Research Questions

This study has two research questions:

  1. How are listening and reading comprehension predictors in Arabic distinct from comprehension predictors in English?
  2. To what degree do language skills in Arabic predict listening and reading comprehension in English?

Both research questions are based on the literature review and the need to investigate the issue of listening and reading comprehension among native Arabic speakers in language acquisition.

Research Hypotheses

  1. Orthographic differences between Arabic and English suggest distinct predictors between listening and reading comprehension in Arabic as the first language and listening and reading comprehension in English as a foreign language.
  2. Considering the characteristics of English and Arabic orthographies, listening and reading comprehension in deep orthographies will predict lower performances.

Methodology

This section addresses the methods and research tools used. This study used the quantitative approach. The quantitative research paradigm therefore assumes that knowledge is “there,” waiting to be revealed, and it is the role of researchers to be “objective” and not to allow their attitudes, values, and beliefs to affect the research process. Epistemologically quantitative research is deductive and confirmatory (Friedman, 2013).

Research Participants

The study sample included 100 intermediate-advanced high school students (50 male students and 50 female students) aged 16-17. The selection criterion was based on a random sample from classrooms doing the four and five level unit English Bagrut (Israel’s high school matriculation exams). The sample was recruited from a high school in the northern district of Israel. All students were native Arabic speakers learning in an Arabic school, in which the teaching language is Arabic, Hebrew is learned as a second language, and English as a third language.

Research Tools

The following tests and tasks were administered (see Appendix A). All of the tests were built especially for this study except for the working memory test in Arabic and vocabulary test in English and Arabic.

1. Phonological Awareness Test in English and Arabic: Both versions were adapted from Morais, Cary, Alegria, and Bertelson (1979). Phonological awareness was tested by phoneme deletion. The participants were presented with 20 items and then asked to delete a phoneme from the beginning, the middle, or the end of a word. For example: repeat the word /jump/ without /j/. The percentage of correct responses out of the total was calculated for each participant.

2. Word Identification Test in English and Arabic: The English version was a subtest of the Woodcock Reading Mastery Test created by Woodcock (1973). Each participant was asked to read 40 words aloud in the English version and 30 words in the Arabic version (based on Arabic readers used in Arab high schools in Israel), which were listed in increasing order of difficulty. The percentage of correct responses out of the total was calculated for each participant.

3. Working Memory Test in English and Arabic: Both versions were adapted from the English version developed by Siegel and Ryan (1989). Participants were asked to fill in the missing words in sentences presented orally. Then they had to remember the words and repeat the missing words from that set in the correct order. There were a total of 12 sets of sentences (lengths ranging from 2 to 5 sentences). Examples are: In summer it is very ______, People go to see monkeys in a _______. The percentage of correct responses out of the total was calculated for each participant.

4. Orthographic Knowledge Test in English and Arabic: Both versions were adapted from the English version developed by Olson, Kliegel, Davidson, and Foltz (1985). The participants were presented with 20 homophonic pairs of words and in each pair, they were asked to mark the corrected spelled word. Examples of these pairs include /all-oll/. The percentage of correct responses out of the total was calculated for each participant.

5. Morphological Knowledge Test in English and Arabic: Both versions were inspired by Kahn-Horwitz (2006). The participants were presented with 10 sentences, and they were asked to fill in the missing word in the sentence from a list of words from the same family of words beneath each sentence in each version. For example, the boy eats (quick/ quickly/quickable). The percentage of correct responses out of the total was calculated for each participant.

6. Syntactic Judgment Test in English and Arabic: The tests appeared in Author and Sanitsky (2010) and Santiskty (2013). The participants were presented with 10 sentences and were asked to decide whether the sentence’s syntax was correct or not. For example, the girl reading a book. The percentage of correct responses out of the total was calculated for each participant.

7. Spelling Test in English and Arabic: Both versions are adapted from the English WRAT-R spelling tests (Jastak & Wilkinson, 1984). This skill was tested through dictation. The participants were asked to write down 40 words correctly in increasing levels of difficulty as they were read aloud. The list of words in English were taken from Band III core word list in the Israeli English Curriculum, such as the word /natural/. The list of words in Arabic were selected from the Arabic readers used in Arab high schools in Israel, such as the word /مبادئ/. The percentage of correct responses out of the total was calculated for each participant.

8. Reading Comprehension Test in English and Arabic: The tests appeared in Author and Sanitsky, (2010) and Sanitisky (2013). The participants were asked to read a text that was followed by 10 multiple choice comprehension questions. In the English version, the text was taken from the Bagrut examination in English (Module E) for four and five unit levels. In the Arabic version, the text was taken from the Bagrut examination in Arabic. The percentage of correct responses out of the total was calculated for each participant.

9. Listening comprehension Test in English and Arabic: Both versions were adapted from [34]. The listening comprehension in English was taken from the Bagrut examination in English (Module E) for four and five unit levels in English. The listening comprehension in Arabic was taken from a recorded TV news interview in Arabic. In both versions, the participants answered 10 multiple choice questions after they listened to passages. The percentage of correct responses out of the total was calculated for each participant.

10. Vocabulary Test in English and Arabic: Both versions were adapted from the English version of The Peabody Picture Vocabulary Test, (3rd edition) developed by Dunn and Dunn (1997). The vocabulary task measured receptive vocabulary. The students were shown four different pictures and were asked to point at the one matching the target word. For example, when the students heard the word bell, they had to point at the picture that matches the word they heard. The percentage of correct responses out of the total was calculated for each participant.

11. Speed Test in English and Arabic (for listening): The test consisted of three parts with three different texts in each part. The first part was quick reading, the second part was medium-paced reading, and the third was slow reading. In each part, the students were asked to answer 10 multiple-choice questions. In the English version, the audios were taken from the Bagrut examination in English (Module E) for four and five unit levels. In the Arabic version, the audio was taken from a recorded TV news interview in Arabic The percentage of correct responses out of the total was calculated for each participant.

12. Attitudes toward the Language and Target Group: The test appeared in [23] The participants were asked to answer a questionnaire which tested a few constructs, such as integrative motivation: indoor integrativeness and outdoor integrativeness, instrumental motivation, and attitude toward the learning situation in class. Statements were asked about each construct and participants were expected to answer on a five levels Likert scale of 1 (strongly disagree) and 5 (strongly agree). The reliability was reported α= .86 which indicates a high reliability.

Research Procedure

The participants were tested in the school that they attended during the school day. Participants were tested individually and collectively by the researcher in quiet rooms. Some tests were carried out collectively like orthographic knowledge, morphological knowledge, syntactic judgment, spelling, reading comprehension, listening comprehension, speed, attitudes toward the language and target group, while other tests were carried out individually. All instructions were given in Arabic, L1. The testing was held in two sessions for each student. Each session lasted between 50-60 minutes. In each session, all tests were given in one of the two languages. The tests were administered in the following order: phonological awareness, word identification, working memory, orthographic knowledge, morphological knowledge, syntactic judgment, spelling, reading comprehension, listening comprehension, vocabulary, speed, and attitudes toward the language and target group.

Data Analysis

Descriptive statistics were calculated for all variables involved in this study (averages, standard deviations, ranges, in addition to Skewness and Kurtosis indices). In the next stage, correlations indices (Pearson’s correlation coefficient) within and across languages were conducted to analyze the relationships between variables in this study. Next, an analysis of variance using ANOVA/ Multivariate linear regression analysis was performed to determine whether there were significant differences between Arabic and English tests and which variables were important predictors of listening and reading comprehension in Arabic as the first language and listening for comprehension in English as a Foreign Language.

Results

In the first stage, the descriptive values will be given to the research variables, and in the second stage, the research hypotheses will be answered. The descriptive statistics for all variables involved in this study are presented in Table 1.

Table 1: ANOVA Test and Descriptive statistics between research groups: dyslexic, chronological age-matched group and reading level matched group.

A-Normal Readers-CA B-Dyslexic Students C-RL Controls F η²
M S.D. M S.D. M S.D.
Reading ability

 

reading vowelized words 33.60 4.40 12.50 3.22 11.23 2.67 385.005*** .898

reading un-vowelized words

32.97 4.95 14.33 3.01 12.47 2.08 305.120*** .875
  Working Memory

 

12.63 1.94 4.17 1.23 3.48 1.33 328.911*** 0.884
  Orthographic processing

 

orthographic processing 89.00 7.38 72.13 7.62 64.38 10.04 66.320*** 0.607
  orthographic processing time (Sec.) 399.40 68.94 451.53 87.60 465.21 70.76 6.155*** 0.125
  Phonological awareness

 

phonological awareness 1 13.79 3.49 4.27 2.57 4.45 1.45 122.356*** 0.744
  phonological awareness 2 time 146.71 51.16 158.50 28.98 161.55 29.82 1.222 0.028
  phonological awareness 3 13.32 4.94 4.93 1.91 4.62 2.50 62.046*** 0.596
morphological awareness

 

29.45 9.62 11.27 8.22 7.64 3.23 70.778*** 0.622
morphological judgment 19.73 0.64 16.53 2.74 16.69 2.85 18.265*** 0.298
  Pseudo word decoding

 

pseudo word decoding 21.27 2.70 9.93 3.02 9.69 2.47 173.721*** 0.802
  pseudo decoding time (sec.) 96.57 33.03 158.33 27.62 162.93 22.30 51.999*** .0547
  Spelling 20.43 2.81 11.43 4.41 11.41 5.20 44.792*** 0.510
  RAN Errors

 

RAN numbers L2 Error 0.03 0.18 0.21 0.56 0.10 0.31 1.536 0.035
  RAN letters L5 Error 0.37 0.85 0.62 0.82 0.66 1.42 0.649 0.015
  RAN objects L7 Error 0.37 0.85 0.79 0.86 0.66 0.72 2.116 0.047
  RAN colors L9 Error 0.53 0.73 0.86 1.03 0.86 1.19 1.076 0.025
  RAN Time

 

RAN numbers L2 Error 28.77 9.19 29.00 5.50 32.55 4.28 2.956 0.064
  RAN letters L5 Error 36.47 8.65 40.03 7.70 42.10 10.11 3.062 0.066
  RAN objects L7 Error 47.20 8.06 44.77 6.01 47.45 5.87 1.444 0.032
  RAN colors L9 Error 47.83 10.61 50.23 8.36 53.62 7.89 3.041 0.066

***p<.001

Testing the Study’s Hypotheses

Hypothesis 1

The first hypothesis was that orthographic differences between Arabic and English would suggest distinct predictors between listening and reading comprehension in Arabic as the first language and listening and reading comprehension in English as a foreign language. To test this hypothesis, in the first stage, we examined the correlation between the variables by using the Pearson coefficient. The coefficient correlation (Table 2) was calculated between the scores of reading comprehension, and listening comprehension with all the other linguistic skills, each separately, once within the Arabic language and another within the English language. The results indicate a statistically significant difference between the two languages.

Table 2: ANOVA Test and Descriptive statistics between level’s RAN tasks

numbers letters objects colors F η²
M

 

S.D. M

 

S.D. M

 

S.D. M

 

S.D.
RAN errors by task type.

 

RAN Error .11 .38 .55 1.06 .60 .82 .75 1.00 11.545*** 0.120
RAN time by task type.

 

RAN Time 30.08 6.83 39.51 9.07 46.46 6.76 50.53 9.25 239.362*** 0.736

***p<.001

Table 2 shows the correlation coefficient between reading comprehension and listening comprehension, with Arabic/English Skills (variables), within every language. It shows that reading comprehension skills in both Arabic and English had a positive and significant correlation and had weak-medium intensity (.22, .35, respectively), with the phonological awareness skills. Also, listening comprehension skills in both Arabic and English were positively and significantly correlated and had weak-medium intensity (.26, .31, respectively), with the phonological awareness skills.

Reading comprehension in Arabic had no significant correlation with orthographic knowledge in Arabic, while the reading comprehension in English had a positive correlation, with weak intensity (.25), with orthographic knowledge in English. In contrast, listening comprehension skill in both Arabic and English had a positive and significant correlation and a weak-medium intensity (.22, .4, respectively), with the orthographic knowledge skills.

Reading comprehension in Arabic had no significant correlation with morphological knowledge in Arabic, while the reading comprehension in English had a positive correlation, with medium intensity (.53), with morphological knowledge in English. In contrast, the listening comprehension skills in both Arabic and English had a positive and significant correlation and a weak-medium intensity (.20, .52, respectively), with the morphological knowledge skills.

Reading comprehension skills in both Arabic and English had a positive and significant correlation and has weak-medium intensity (.25, .51, respectively), with the syntactic judgment skills. In contrast, listening comprehension in Arabic had no significant correlation with syntactic judgment in Arabic, while listening comprehension in English had a positive correlation, with weak intensity (.25), with syntactic judgment in English.

Reading comprehension skills in both Arabic and English had a positive and significant correlation and has weak-medium intensity (.24, .51, respectively), with the spelling skills. Also, listening comprehension skills in both Arabic and English was positively and significantly correlated and had weak-medium intensity (.21, .37, respectively), with the spelling skill.

Reading comprehension skills in both Arabic and English had no significant correlation with vocabulary skills. In contrast, listening comprehension in Arabic had no significant correlation with vocabulary in Arabic, while listening comprehension in English had a positive correlation, with weak intensity (.25), with vocabulary in English.

Reading comprehension skills in both Arabic and English had a positive and significant correlation and a weak-medium intensity (.22, .55, respectively), with the speed (0.75) skills. Also, listening comprehension skills in both Arabic and English had a positive and significant correlation and strong intensity (.74, .94, respectively), with the speed (0.75) skills.

Reading comprehension in Arabic had no significant correlation with speed (1.25) in Arabic, while reading comprehension in English had a positive correlation, with medium intensity (.44) with speed (1.25) in English. In contrast, the listening comprehension skills in both Arabic and English had a positive and significant correlation and a strong intensity (.89, .92, respectively), with the speed (1.25) skill.

Lastly, we should mention that listening and reading comprehension skills in both Arabic and English did not have any significant correlation with the word identification skills, working memory skills, and attitudes toward the English language.

In the second stage, the relationship between the variables was tested by adjusting multiple linear regression using the Stepwise method, to predict reading comprehension and listening comprehension skills in Arabic/English by skills in Arabic/English, respectively, as predictors. In other words, this stage sought to predict reading comprehension and listening comprehension skills in Arabic based on the other skills in the Arabic language, and then reading comprehension and listening comprehension skills in English based on the other skills in the English language as predictors.

The first dependent variable examined was reading comprehension. Table 3 contains the parameter estimates of the predictor’s variables—all the significant linguistic skills within every language.

Table 3: ANOVA Test and Descriptive statistics for reading time between research groups: dyslexic, chronological age matched group and reading level matched group.

A-Normal Readers-CA B-Dyslexic Students C-RL Controls F η²
M

 

S.D. M

 

S.D. M

 

S.D.
Speed of processing in reading (Time).

 

orthographic awareness E time (Sec.) 399.00 70.79 451.53 87.60 465.21 70.76 5.855*** 0.122
phonological awareness F time 146.71 51.16 158.50 28.98 161.55 29.82 1.222 0.028
pseudo decoding time (sec.) 96.68 33.98 158.33 27.62 162.93 22.30 48.661*** .0537
average time (sec.) 214.13 30.44 256.12 34.64 263.23 32.33 18.873*** 0.310

Table 3 shows that Phonological awareness and spelling skills in Arabic were significant (p<0.05) predictors of reading comprehension in Arabic. The predictors positively affected Reading comprehension in Arabic and were able to explain about 17% of the variance of reading comprehension in Arabic.

In Contrast, speed (0.75), syntactic judgment, and spelling skills in English were significant (p<0.05) predictors for reading comprehension in English. The predictors positively affected reading comprehension in English and were able to explain about 50% of the variance of reading comprehension in English.

The second dependent variable examined was listening comprehension. Table 3 contains the parameter estimates of the predictor’s variables—all the significant linguistic skills within every language.

Table 3 shows that Phonological awareness and orthographic knowledge skills in Arabic were significant (p<0.05) predictors for listening comprehension in Arabic. Both positively affected the listening comprehension in Arabic and were able to explain about 10% of the variance of listening comprehension in Arabic.

In contrast, morphological knowledge, orthographic knowledge, word identification, and spelling skills in English were significant (p<0.05) predictors for listening comprehension in English. All positively affected listening comprehension in English and were able to explain about 36% of the variance of listening comprehension in English.

In conclusion, a comparison of the predictors for reading comprehension in Table 3 shows that only the predictor of spelling was common between the two languages, while the other predictors for reading comprehension were different.

A comparison of the predictors for listening comprehension in Table 3 shows that only the predictor of orthographic knowledge is common between the two languages, while the other predictors for listening comprehension are not identical.

Accordingly, the first hypothesis is partially confirmed.

Hypothesis 2

The second hypothesis was that the characteristics of English and Arabic orthographies would predict lower performances in listening and reading comprehension in deep orthographies.

To examine this hypothesis, in the first stage, the correlation between the variables was tested using the Pearson coefficient. The coefficient correlation (Table 4) was calculated between the scores of reading comprehension, and listening comprehension in English, with the Arabic language skills. The results indicate that most skills in Arabic are statistically significance in terms of both reading comprehension and listening comprehension in English, except for the predictor of orthographic knowledge in Arabic.

Table 4: ANOVA Test and Descriptive statistics for general reading ability between research groups: dyslexic, chronological age matched group and reading level matched group.

A-Normal Readers-CA B-Dyslexic Students C-RL Controls F η²
M

 

S.D. M

 

S.D. M

 

S.D.
General reading ability scores.

 

reading ability-general (0-106) 90.13 10.47 35.13 5.93 31.73 4.18 595.839*** .932
working memory-general 12.63 1.94 4.17 1.23 3.48 1.33 328.911*** 0.884
orthographic awareness general 89.00 7.38 72.13 7.62 64.38 10.04 66.320*** 0.607
phonological awareness general 27.11 7.61 9.20 3.80 9.07 2.70 117.764*** 0.737
morphological awareness general 49.18 10.03 27.80 9.46 24.33 5.17 74.046*** 0.633
pseudo word decoding 21.27 2.70 9.93 3.02 9.69 2.47 173.721*** 0.802
spelling 20.43 2.81 11.43 4.41 11.41 5.20 44.792*** 0.510

Table 4 shows the correlation coefficient between Reading comprehension and listening comprehension in English, with Arabic skills (variables). Table 4 shows that reading comprehension and listening comprehension skills in English had a positive and significant correlation and a medium intensity (.33, .31, respectively), with the phonological awareness skills.

Reading comprehension in English had no significant correlation with orthographic knowledge in Arabic while listening comprehension in English had a positive correlation and a weak intensity (.25), with orthographic knowledge in Arabic.

Also, reading comprehension and listening comprehension skills in English had a positive and significant correlation and a medium intensity with syntactic judgment (.36, .28, respectively), spelling (.41, .24, respectively), reading comprehension (.67, .45, respectively), listening comprehension (.39, .46, respectively), speed (0.75) (.39, .36, respectively), speed (1.25) (.32, .41, respectively).

Lastly, we should mention that the reading comprehension and listening comprehension skills in English had no significant correlation with any variables of word identification, working memory, morphological knowledge, and vocabulary.

In the second stage, to examine this hypothesis, we used adjusting multiple linear regression using the Stepwise method, to predict English language skills (reading comprehension and listening comprehension) by language skills (variables) in Arabic as predictors.

The first dependent variable examined was reading comprehension. Table 5 contains the parameter estimates of the predictor’s variables—all the significant linguistic skills in the Arabic language.

Table 5: Pearson correlation results between the variables measured in the study

1 2 3 4 5 6 7
1. phonolory_reading_ability_general
2. working_memory_general .938**
3. orthographic_awareness_general .770** .705**
4. phonological_awareness_general .894** .836** .664**
5. morphological_awareness_general .821** .800** .650** .756**
6. pseudo_word_decoding_general .927** .861** .758** .853** .782**
7. Spelling_general .753** .692** .816** .716** .643** .784**
8. RAN_general -.268* -.203 -.282** -.285** -.224* -.275** -.356**

**p<0.01 *p<0.05

Table 5 shows that reading comprehension, speed (0.75), spelling, and syntactic judgment skills in Arabic were significant (p<0.05) predictors of reading comprehension in English. The predictors positively affected reading comprehension in English and were able to explain about 58% of the variance of reading comprehension in English.

The second dependent variable examined is listening comprehension. Table 5 contains the parameter estimates of the predictor’s variables—all the significant linguistic skills in the Arabic language.

Table 5 shows that listening comprehension and reading comprehension skills in Arabic were significant (p<0.05) predictors of listening comprehension in English. The predictors positively affected listening comprehension in English and were able to explain about 34% of the variance of listening comprehension in English.

In conclusion, the Arabic skills predictors for reading comprehension in English (Table 5) are reading comprehension, speed (0.75), spelling, and syntactic judgment skills. In contrast, the Arabic skills predictors for listening comprehension in English (Table 5) are listening comprehension and reading comprehension skills. Accordingly, the second hypothesis is confirmed.

In summary, the results showed that the first hypothesis was partially confirmed. While it was hypothesized that orthographic differences between Arabic and English suggested distinct predictors between listening and reading comprehension in Arabic and listening and reading comprehension in English, spelling was a common predictor between the two languages in reading comprehension, and orthographic knowledge was a common predictor between the two languages in listening comprehension. The second hypothesis was confirmed: Considering the characteristics of English and Arabic orthographies, listening and reading comprehension in deep orthographies predicted lower performances.

Discussion

The main goal of the current research was to investigate how listening and reading comprehension among native Arabic speakers predict the use of Arabic and English language orthographies in learning English as a foreign language (FL) in Israel. The novel aspect of this research is examining listening and reading comprehension among Arabic learners of English in relation to their different orthographies in a single study. The existing research has many problems in representing predictors of listening and reading comprehension in Arabic as the first language and English as a foreign language concerning their distinct orthographies. Previous studies showed mixed results concerning the role of language orthography in listening and reading comprehension. In an attempt to fill the gap, this study presents an analysis of significant differences between Arabic and English tests and the variables that are important predictors of listening and reading comprehension in Arabic as the first language and listening for comprehension in English as a foreign language.

According to the research findings, the first research hypothesis was partially confirmed, meaning that that only the predictor of spelling is common between the two languages, and the other predictors for reading comprehension are different. Phonological awareness in Arabic is a significant predictor of reading comprehension in Arabic, while speed (0.75) and syntactic judgment in English are significant predictors of reading comprehension in English. These findings are in line with previous studies that reported that in shallow orthographies, learners depend more on phonology in word decoding and word learning (Author et al., 2013; 10; Jiang, 2017). In addition, the development of literacy skills such as reading and spelling are learned more quickly in consistent orthographies when compared to inconsistent orthographies (Caravolas et al., 2012). Regarding the predictors of reading comprehension in English, speed (0.75) indicates a significant predictor of reading comprehension in English, suggesting that slower reading in listening comprehension contributes to better performance in reading comprehension in English. Regarding syntactic judgment, the findings are in accordance with previous studies that suggest that syntax relates to reading comprehension, and improvement in syntax contributes positively to reading comprehension among L2 learners of English (Gottardo et al., 2018).

Furthermore, only the predictor of orthographic knowledge is common between the two languages, while the other predictors for listening comprehension are not identical. Phonological awareness in Arabic predicts listening comprehension in Arabic, while morphological knowledge, word identification, and spelling predict listening comprehension in English. This aligns with the previous studies that showed that the varying degree of orthography between languages affects the development of reading and spelling, as deep orthographies impede spelling skills compared to shallow orthographies (Andreou, 2016). In addition, word recognition highly contributes to reading comprehension among beginner readers while listening comprehension appears to be strongly related to advanced readers (Babayiğit & Shapiro, 2020). Furthermore, shallow orthographies are often complicated morphological systems while deep orthographies are less complicated (Author & Shakkour, 2014; Author et al., 2013). These findings on the differences between Arabic and English predictors of listening and reading comprehension in Arabic and listening and reading comprehension in English shed light on the fact that learners with different orthographic backgrounds adapt distinct linguistic skills to learn the language. In addition, the transfer of linguistic and language skills is affected by the degree of similarity between the two languages (Author et al., 2013; 34,38]. Therefore, the results of the current study prove that distinct orthographic backgrounds of L1 and L2 influence the use of language skills in listening and reading comprehension in both languages.

The second research hypothesis was fully confirmed, meaning that the Arabic skills predictors for reading comprehension in English are different from the Arabic skills predictors for listening comprehension in English. Reading comprehension, speed (0.75), spelling, and syntactic judgment skills in Arabic are significant predictors of reading comprehension in English. However, listening comprehension and reading comprehension skills in Arabic are significant predictors of listening comprehension in English. These results are in line with previous studies that showed that listening comprehension and reading comprehension are similar skills but constitute two distinct forms of comprehension involving different cognitive processes (Author 2019a; Diakidoy et al., 2005; Taha, 2016; Tobia & Bonifacci, 2015; Wolf et al., 2019). Moreover, the Arabic predictors of listening and reading comprehension in English demonstrate that due to the orthographic background differences between Arabic and English, learners had to use distinct skills in their L1 in order to perform well in English.

Conclusion

The present study explored to what extent first language skills predict listening and reading comprehension in English and to what degree the predictors of listening and reading comprehension in Arabic are distinct from listening and reading comprehension predictors in English. The findings of the study proved that the performance of listening and reading comprehension in the two languages is affected by the different orthographic system of each language. Additionally, these findings add to the body of literature of high school Arabic learners of English as a foreign language.

The present study has some significant empirical and instructional recommendations regarding the teaching and learning of English as a foreign language for Arabic speakers. Firstly, the different orthographies of the Arabic and English languages should be taken into consideration in teaching English as foreign language among native speakers of Arabic. Secondly, teaching Arabic should focus on phonological awareness, orthographic knowledge, and spelling skills for better performance in listening and reading comprehension in Arabic. Thirdly, teaching English should focus on speed, syntactic judgment, morphological knowledge, word identification, orthographic knowledge and spelling skills for better performance in listening and reading comprehension in English. It was crucial to conduct such research as limited attention has been given to listening and reading comprehension in language acquisition in the context of their different orthographies. The main contributions of this study are understanding the difficulties that students might face in learning Arabic and English as high school learners and reconsidering the teaching of Arabic and English in relation to their different orthographies.

Limitations and Future Research

The present study has few limitations that should be taken into consideration and in interpreting the results. One of the limitations of this study is the sample prevents generalization. The sample concerns a specific group of a limited number of students belonging to a certain age group and background. Another limitation of this study is the use of constructed assessments for assessing the language skills in Arabic and English, as standardized measures for assessing these skills among high school aged native Arabic speakers’ learners of English as a third language are unavailable. Despite these limitations, the present study expands the existing knowledge in the field of language education. Future studies should include a larger sample of participants from different backgrounds and ages. Further, validated measures for assessing language skills should be included.

References

  1. Edele A, Stanat P (2016) The role of first-language listening comprehension in second-language reading comprehension. Journal of Educational Psychology 108(2): 163-180.
  2. Sparks RL, Patton J, Ganschow L, Humbach N, et al. (2008) Early first-language reading and spelling skills predict later second-language reading and spelling skills. Journal of Educational Psychology 100(1): 162-174.
  3. Author S (2019a)
  4. Diakidoy IAN, Stylianou P, Karefillidou C, Papageorgiou P (2005) The relationship between listening and reading comprehension of different types of text at increasing grade levels. Reading Psychology 26(1): 55-80.
  5. Taha H (2016) Deep and shallow in Arabic orthography: New evidence from reading performance of elementary school native Arab readers. Writing Systems Research 8(2): 133-142.
  6. Tobia V, Bonifacci P (2015) The simple view of reading in a transparent orthography: The stronger role of oral comprehension. Reading and Writing 28(7): 939-957.
  7. Wolf MC, Muijselaar MM, Boonstra AM, de Bree EH (2019) The relationship between reading and listening comprehension: shared and modality-specific components. Reading and Writing 32(7): 1747-1767.
  8. Cadime I, Rodrigues B, Santos S, Viana FL, et al. (2017) The role of word recognition, oral reading fluency and listening comprehension in the simple view of reading: a study in an intermediate depth orthography. Reading and Writing 30(3): 591-611.
  9. Tobia V, Bonifacci P (2015) The simple view of reading in a transparent orthography: The stronger role of oral comprehension. Reading and Writing 28(7): 939-957.
  10. Feder L, Author S (2020) An examination of differences in linguistic and meta-linguistic skills in English (FL) and Hebrew (L1): English intervention program for dyslexic, poor and normal readers. The Journal of Educational Research 113(3): 226-243.
  11. Author S, Shakkour W, Siegel L (2013)
  12. Author S, Hijjazi E (2020)
  13. Author S (2019b)
  14. Asadi IA, Khateb A, Shany M (2017) How simple is reading in Arabic? A cross‐sectional investigation of reading comprehension from first to sixth grade. Journal of Research in Reading 40(S1): S1-S22.
  15. Author S, Shakkour W (2014)
  16. Author S, Siegel LS (2003)
  17. Caravolas M, Lervåg A, Mousikou P, Efrim C, et al. (2012) Common patterns of prediction of literacy development in different alphabetic orthographies. Psychological Science 23(6): 678-686.
  18. Author S (2019a)
  19. Author S (2019b)
  20. Author S, Hijjazi E (2020)
  21. Ibrahim R, Eviatar Z, Aharon-Peretz J (2002) The characteristics of Arabic orthography slow its processing. Neuropsychology 16(3): 322-326.
  22. Author S (1998)
  23. Taha H (2016) Deep and shallow in Arabic orthography: New evidence from reading performance of elementary school native Arab readers. Writing Systems Research 8(2): 133-142.
  24. Asadi IA, Author S (2019) The impact of the position of phonemes and lexical status on phonological awareness in the diglossic Arabic language. Journal of Psycholinguistic Research 48(5): 1051-1062.
  25. Al-Janaideh R, Hipfner-Boucher K, Cleave P, Chen X (2021) Contributions of code-based and oral language skills to Arabic and English reading comprehension in Arabic-English bilinguals in the elementary school years. International Journal of Bilingual Education and Bilingualism 25(7): 1-16.
  26. Hussien AM (2014) The effect of learning English (L2) on learning of Arabic literacy (L1) in the primary school. International Education Studies 7(3): 88-98.
  27. Andreou G, Segklia M (2019) Cross‐linguistic skills transfer from the second/foreign language to the first among students with learning disabilities after an intervention program in the second language. Creative Education 10(5): 1023-1036.
  28. Author S, Siegel LS (2002)
  29. Miller RT (2019) English orthography and reading. In J.I. Liontas (Ed. ): The TESOL Encyclopedia of English Language Teaching 1(7)
  30. Caravolas M, Lervåg A, Defior S, et al. (2013) Different patterns, but equivalent predictors, of growth in reading in consistent and inconsistent orthographies. Psychological Science 24(8): 1398-1407.
  31. Feder L, Author S (2022) Cognitive Retroactive Transfer of Language Skills from English as a Foreign Language to Hebrew as the First Language. Journal of Learning Disabilities 55(3): 213-228.
  32. Author S, Siegel LS (2002)
  33. Wade‐Woolley L (1999) First language influences on second language word reading: All roads lead to Rome. Language Learning 49(3): 447-471.
  34. Edele A, Stanat P (2016) The role of first-language listening comprehension in second-language reading comprehension. Journal of Educational Psychology 108(2): 163-180.
  35. Gottardo A, Mirza A, Koh PW, et al. (2018) Unpacking listening comprehension: The role of vocabulary, morphological awareness, and syntactic knowledge in reading comprehension. Reading and Writing 31(8): 1741-1764.
  36. Babayiğit S, Shapiro L (2020) Component skills that underpin listening comprehension and reading comprehension in learners with English as first and additional language. Journal of Research in Reading 43(1): 78-97.
  37. Goodwin AP, Petscher Y, Tock J (2020) Morphological supports: Investigating differences in how morphological knowledge supports reading comprehension for middle school students with limited reading vocabulary. Language, Speech, and Hearing Services in Schools 51(3): 589-602.
  38. Jiang X (2017) Lower-level processing skills in English-as-a-second-language reading comprehension: Possible influence of first language orthography. Studies in English Language Teaching 5(3): 2329-3111.
  39. Elkhafaifi H (2005) The effect of pre listening activities on listening comprehension in Arabic learners. Foreign Language Annals 38(4): 505-513.
  40. van Zeeland H, Schmitt N (2013) Lexical coverage in L1 and L2 listening comprehension: The same or different from reading comprehension? Applied Linguistics 34(4): 457-479.
  41. Author S, Bluestein-Danon D (2012)
  42. Hamada M, Koda K (2008) Influence of first language orthographic experience on second language decoding and word learning. Language Learning 58(1): 1-31.
  43. Andreou G (2016) Dyslexia and spelling in two different orthographies (Greek vs. English): a linguistic analysis. Journal of Advances in Linguistics 7(1): 1172-1192.
  44. Sipra M (2013) Impact of English orthography on L2 acquisition. English Language Teaching 6(3): 116-124.
  45. Author S (2)
  46. Author S, Share D, Mansour MS (2003)
  47. Author S (2003)
  48. Elyildirim S, Ashton S (2006) Creating positive attitudes towards English as a foreign language, English Teaching Forum 44(4): 2-21.
  49. Gömleksiz MN (2010) An evaluation of students’ attitudes toward English language learning in terms of several variables. Procedia-Social and Behavioral Sciences 9: 913-918.
  50. Oroujlou N, Vahedi M (2011) Motivation, attitude, and language learning. Procedia-Social and Behavioral Sciences 29: 994-1000.
  51. Ghazvini SD, Khajehpour M (2011) Attitudes and motivation in learning English as second language in high school students. Procedia-social and Behavioral Sciences 15: 1209-1213.

Advancing Mind Genomics: Using AI (Artificial Intelligence) to Generate Topics, Questions, Messages as Answers, and ‘Synthesized’ Minds (Respondents)

DOI: 10.31038/PSYJ.2023581

Abstract

AI created a resource bank of statements about what a doctor might say to a child in order to deal with the child’s obesity. After the AI generated messages were developed, 16 of the messages (elements) were selected and combined into vignettes according to an underlying experimental design, whose specific combinations differed for each ‘respondent’. Each set of 24 vignettes comprised a stand-alone set of combinations and were evaluated by AI prompted to act as a specific person in the medical world (receptionist, doctor with 10 years of experience, nurse practitioner). Deconstruction of the ratings by regression showed the contribution of each AI created message to the rating scale. The coefficients ‘made sense’ when the regression was done according to ‘WHO’ the AI was defined to be. Further clustering the coefficients across the respondents revealed two clearly different mind-sets. The systematic approach using AI as both the provider of ideas and the evaluator of these ideas presents a new vista for learning about how to communicate with people, using technology to dramatically accelerate and fine-tune that learning.

Introduction

It is hard to overestimate the excitement with which AI, artificial intelligence, has been greeted and adopted, especially since the introduction of Chat GPT, and associated technology [1]. One can scarcely read of any topic of huma endeavor without one or another pundit bringing up the impact of AI for that field. This paper looks at the potential of AI to synthesize ‘respondents’, with the goal of accelerating the learning of professionals who want to learn to counsel people in nutritional health. The use of synthesized people, personas, is not new, and has been a topic of interest for some decades now [2-4]. What is new is the vision of moving beyond personas derived from large scale studies to one-off studies created entirely by synthetic means.

This paper combines AI with Mind Genomics to develop a new system for training and education. The underlying vision is to accelerate knowledge development by having the AI provide topic-relevant ideas (Idea Coach in Mind Genomics), and then ‘personas’ created by artificial intelligence, based upon combinations of features of the way people think, who the people are etc.. The raw material for these personas come from ‘self-profiling classification questions’ that a researcher might ask a human respondent. In short, the paper presents a Mind Genomics study, with the materials, from beginning end under the control of the machine, not the researcher. All the researcher does is select ideas at the very start of the study, these ideas later being tested in the study itself.

The process of Mind Genomics begins with a series of questions, those questions telling a story, and then for each question provide four different answers. The actual experiment consists of combining these answers into small combinations of 2-4 answers, at most one answer from a question, but often no answer from a question. The 24 combinations, called vignettes, are evaluated by respondents. Each respondent evaluates a set of 24 different vignettes, the uniqueness of the sets of 24 vignettes ensured by a permutation algorithm [5]. The analysis, done by OLS, ordinary least squares regression, shows through the coefficients of a linear equation the degree to which the 16 elements, viz., answers to the question, drive the respondent’s rating.

During the evolution of Mind Genomics, a process taking place for more than 30 years, since its introduction in 1993 [6-8], the consistently hardest part of any effort, basic research or applied research, was developing the questions, and to a far lesser degree coming up with the answers to the questions. More than one study ended up being aborted simply because the researcher could not generate the four questions which tell the story. Once the four questions were developed, for the most part, researchers were able to push through to the four answer for that question. A solution to the issue of frustration in the creation of questions and answers emerged with the incorporation of AI into the process.

A third problem, motivating people, respondents, to participate, proved to be simple to solve because of the emergence of companies which provided paid panelists. Money solved the problem of participation, if not motivation. This paper will deal with the introduction of synthesized respondents, to reduce cost, and to facilitate new types of systematized investigations not possible before.

Mind Genomics requires systematized thinking about a topic, an ability that all too often needs to be learned through coaching. It was to provide this coaching that the original AI was introduced in mind-year, 2023, in the form of Idea Coach. This paper focuses on the further introduction of AI ‘synthesized respondents’, in an effort to make the Mind Genomics process a streamlined one, from beginning to end, appropriate for teaching as well as for practical application.

Explicating the Process-Part 1-Setting Up the Study Using AI

Figure 1 (Panel A) shows the request by the Mind Genomics set-up ‘template’ for the four questions, each to be addressed by four answers. It is the development of these four questions which become a difficult hurdle. The creation of the Idea Coach allows this topic to be addressed. Figure 1 (Panel B) shows the rectangle where the researcher can write out the question. The question posed to AI through Idea Coach is very simple: I am a doctor treating obesity in children. How do I talk to the children to make them understand. One could further fine-tune the Idea Coach by telling it to explicate the question in discussion form, as well as to provide questions of length 15 words or less, and questions understandable to a 12-year-old. Those statements become part of the query. The actual process is made as simple as possible so that the effort focuses on the topic.

fig 1

Figure 1: The templated screens requesting the researcher to provide four questions. Panel A shows the request for four questions which tell a story Panel B shows the rectangle inside which the researcher can describe the topic, and from which the Idea Coach returns with 15 questions. The actual text of the request is: I am a doctor treating obesity in children. How do I talk to the children to make them understand?

The actual short description appears as the topic in Table 1. The Idea Coach was run three times to generate a reservoir of questions. The Idea Coach is typically run 5-10 times, providing information which ends up teaching the researcher. The Idea Coach is also run several times for each of the four questions. The combination of different sets of 15 questions for a topic description and different sets of 15 answers for each question provides a unique resource booklet on the topic. Each iteration in Idea Coach lasts about 20 seconds, so that in five minutes one can produce 15 sets of 15 questions each. In the end, only four questions will be chosen.

The results from the first 15 questions appear in Table 1. The questions emerging from the AI embedded in Idea Coach are easily understood by a human being.

Table 1: Questions emerging from the Idea Coach which address the topic. The topic is provided by the researcher

tab 1

After the researcher receives the various sets of 15 questions, the next task is for the researcher to provide four questions. These questions can be taken from those suggested by the AI, either ‘as is’ or edited to tailor the ‘language’ and ‘style’ of the answer. The researcher may also contribute questions. Quite often the questions shown in Table 1 have to be modified, not so much for the respondent who never sees the questions, but rather for the AI to provide the proper format of the answer.

Table 2 shows the modified question, edited by the researcher, and submitted to the Idea Coach. In turn, for each question in Table 2, Idea Coach returned with sets of 15 answers, formatted in the way request by the researcher. Table 2 also shows the four answers to each question, these answers having been returned by AI (Mind Genomics’ Idea Coach), and then slightly edited to make them flow more easily. The rationale was to generate answers that could be given to the Idea Coach programmed to act as a human respondent.

Table 2: The four questions and the four answers to each question. The questions and the answers were edited slightly to make them understandable to human beings, but the language and meaning was maintained.

tab 2

Running the Study and Preparing the Data for Regression Modeling and for Clustering

Following the creation of the edited answers, the Mind Genomics platform was instructed to run a study with 301 synthetic respondents. The program was instructed to create a panel comprising approximately equal numbers of the three types of synthetic respondents defined by their job (receptionist, doctor with 10-years-experience, nurse practitioner). Then the Mind Genomics program presented 24 vignettes to the synthetic respondent, defining WHO the respondent is, defining the SCALE (Table 3), and then one vignette at a time to the AI ‘respondent’ with the request to rate the vignette on the two-sided scale by choosing one rating point. This means that the AI had to consider the vignette from the complex of how the patient would feel and how the doctor would feel. The Mind Genomics platform recorded the information about the respondent, the vignette, the rating, and the time elapsed for the synthesized respondent to rate the vignette.

Table 3: The preliminary question for assignment into respondent job (top) and then the ‘two-sided rating scale (bottom).

tab 3

Once the data was collected for a ‘respondent’ the AI returned the raw data to the Mind Genomics platform to create the database shown as an Excel file in Figure 2. Each respondent generates 24 rows of data, one row for each of the 24 vignettes. Figure 2 is divided into several sections, representing different aspects of the data and of the analysis. As we follow the structure of the file, we must keep in mind that as the data is being transferred to the database, there are preparatory transformations occurring in ‘real time,’ these transformation necessary for the analysis.

  1. Row-for the entire database. There are 301 respondents, each with 24 rows of data, generating 7224 rows, each row corresponding to a specific respondent and a specific vignette.
  2. Study and panelist number.
  3. The AI assignment of the respondent to one of three groups (receptionist, doctor with 10-years-experience, nurse practitioner by the membership of the synthesized respondent into one of two mind-sets, and then into one of three mind-sets.
  4. Columns showing test order and ‘half’ defined as 1 for test order 1-12, and 2 for test order 13-24, respectively.
  5. The 16 elements, A1-D4, coded as ‘1’ when present in the vignette, coded as ‘0’ when absent from the vignette.
  6. Rating, new variables R54x (patient motivated), R12x (patient not motivated), R52x (medical person feels the right thing was said), R41x (medical person thinks the wrong things were said), R3x (cannot answer), and RT (response time to the nearest 10th of a second. The new variables were created by adding the appropriate variables. In no case does the creation of a new variable produce any number other than 0 or 1. After these variables were created a vanishingly small random number was added to each new variable (viz., R54x, R12x, R52x, R41x, R3) to ensure that these newly created variables possessed minimal variability as required for analysis by regression.

fig 2

Figure 2: Example of database showing the actual information generated by the AI, and the set of responses to each vignette.

Using Regression Modeling to Link the 16 Elements to the Newly Created Binary Variables

The first analysis looks at the average ratings of the newly created binary variables (e.g., R54x, Patient Motivated). The question is whether the instruction to the AI to assume a certain persona (e.g., receptionist) has an effect on the average ratings across all the vignettes in which the AI assumed that it was a ‘receptionist’. Table 4 shows that the average ratings for a specific binary variable are quite similar across the three different ‘AI personas’. Our first conclusion leaves us with the concern that AI may not be able to perform the task in a way which makes sense.

Table 4: Average rating for five binary variables and ‘response time’ for three respondent personas created by AI and set to the task of rating the vignette. The data come from the averages of the newly created binary variables. Each average comes from approximately 2400 numbers, assuming that each synthesized persona comprised 100 respondents, each respondent evaluating 24 vignettes.

tab 4

The final analysis for this first run used OLS (ordinary least squares regression, to relate the 16 elements to the binary transformer rating. The equation is: Transformed Binary Variable=k1(A1) + k2(A2) + … k16(D4). That is, the regression model uses the data to show how the presence or absence of the element drives the binary variable. The least-squares regression can be run for the total panel, and for all respondents defined by the AI synthesizer as being ‘receptionist,’ ‘doctor with 10-years-experience’, or nurse practitioner.

Table 5 shows the regression coefficients, computed as if the ratings and therefore the binary variables (e.g., R54x) came from people. Relevant coefficients are shown in shaded cells. There may be a number of stories within the data, but no simple organizing principle. Experience with ratings generated by people produce the same story, namely some strong performers but not many, and stories that could be told, but seem too isolated.

Table 5: Regression coefficients for the key groups (receptionist, doctor with 10-years-experience, and nurse practitioner. The regression coefficients were estimated by OLS regression based upon the ratings of the vignette assigned by artificial intelligence.

tab 5(1)

tab 5(2)

A better way to look for ‘humanness’ in the data may be to search for strong-performing elements for each. As a further step, one might present these strong performing elements to others, and ask them how they feel about the motivating power of the elements, or how they feel about the elements in terms of want might be expected from someone in the medical profession. As a first approximation, there are no strong surprises in Table 6, which shows the strongest elements for each newly created binary variable (e.g., R54x, Patient Motivated), when the AI took on three of the personas.

Table 6: Strongest performing elements for three AI personas (receptionist, doctor, nurse practitioner) on four newly created binary variables.

tab 6

The final set of analyses, clustering, divides the respondents into mutually exclusive and exhaustive groups, called ‘clusters’ by statisticians, ‘segments’ by consumer researchers, and ‘mind-sets’ in the language of Mind Genomics. The underlying notion, foundational for Mind Genomics, is that ‘people’ differ systematically in the way they make decisions about the world of the everyday. The plethora of different products, services, even layouts of towns and the styles of houses and their decorations clearly announce these differences in judgment. Rather than considering this person-to-person variability to be an unwanted secondary factor, noise, in an otherwise simple world, Mind Genomics looks for organizing principles, so-called mind-sets. These mind-sets are derived empirically from a study of the differences among people in how those people evaluate the world of the everyday, at the granular level, not the 20,000-foot level. A variety of papers have been published on the use of Mind Genomics to identify these mind-sets in various situations and for various products.

The question for this study is whether mind-sets can be uncovered when we deal with synthetic respondents. The process to discover these mind-sets will certainly work with the data generated by AI because the input data needed to create the mind-sets is simply the set of coefficients, one per respondent, as shown in Table 5, specifically in this study for coefficients where R54x (patient motivated) is the dependent variable.

The process for creating mind-sets follows strict mathematical rules. The clustering is totally insensitive to the ‘meaning’ of the elements and does not care from where the coefficients came Mind-sets emerge after the researcher develops the equation for each of synthesized respondent, puts the data into a matrix of 301 rows (one row for each synthesized respondent, 16 columns (one column for each of the 16 coefficients from regression). The k-means clustering program [9] then creates groups of synthesized respondents whose patterns are similar across the 16 coefficients.

The process of creating the mind-sets is straightforward, using clustering. The process follows these steps:

  1. Create the basic equation for each respondent: R54x=k1A1 + k2A2 … k16 The equation has 16 coefficients, estimated from the 24 cases generated for each synthetic respondent.
  2. Estimate the ‘distance’ or ‘dissimilarity’ between each pair of the 301 respondents by computing the Pearson Correlation (R). Then create the new distance parameter, (1-R). The value (1-R) is 0 when the two sets of coefficients are parallel, meaning that the two items, our synthetic respondents, show identical patterns. The value (1-R) is 2 when the two sets of coefficients are opposite.
  3. Use clustering to assign respondents to one of two clusters, or later one of three clusters. The criterion is that the variability within a cluster should be ‘small’ because the respondents show similar patterns of coefficients. In contrast, the variability across the centroids of the clusters should be large because the clusters are different groups.
  4. Once the respondents are assigned to the appropriate cluster (which of the two clusters, which of the three clusters), recompute the equation using the data from all respondents in a cluster.

Table 7 shows the performance of the elements for the Total Panel, for the two mind-sets, and for the three mind-sets. Just looking at the Table shows a number of shaded cells with coefficients of 21 or higher. High coefficients by themselves do not suffice, however. Rather, the coefficients must ‘tell a story’, and allow for interpretation.

The first clustering creating two mind-sets produces easily interpreted mind-sets.

Table 7: Coefficients emerging from clustering the 301 synthetic respondents into two mind-sets and then three mind-sets, based on the pattern of the coefficients for R54x (Motivates).

tab 7

Mind-Set 1 of 2-Focus on Activity and Vitality

We can find fun ways to stay active and make healthy choices together.

Let’s talk about ways to keep your body strong and full of energy

Mind-Set 2 of 2-Focus on What Specifics to Do

Here is part of our program designed for you Motivate with positive and encouraging words.

Here is part of our program designed for you Involve the child in grocery shopping and meal planning.

Here is part of our program designed for you: Share inspiring success stories of people who achieved their weight loss goals.

In contrast, the three mind-set-solution produces elements with higher coefficients, but the underlying pattern is hard to interpret.

Our final analysis considers the performance of the elements using the IDT, Index of Divergent Thought [10]. The IDT attempts to quantify the degree to which the clustering generates truly different groups of people based upon the coefficients. Table 8 shows the computations. Optimal levels of the IDT are found in the range of 70-75. Higher values of the IDT (e.g., 80 or higher) mean that there are many high coefficients but not dramatically patterns of coefficients across the mind-sets Lower values of the IDT (e.g., 60 or lower) mean that there are many low coefficients, and once again the pattern of differences in coefficients across mind-sets is simply not dramatic.

For this study, and for others not reported here, the pattern of low values for the IDT continues to emerge when we deal with synthetic respondents. Here it is 47. Simply stated, the synthetic respondents may ‘work’ but do not yet produce dramatically stronger performance. As a starting point, however, it is gratifying to see that initial exploration into the AI does produce interpretable mind-sets.

Table 8: Calculation of the IDT, Index of Divergent Thought, for the synthesized data. The IDT is based on clustering coefficients estimated for R54x, patient is motivated.

tab 8

Discussion and Conclusions

The emerging interest in AI generated respondents, so-called synthetic respondents, provides a new area of opportunity for the equally emerging science of Mind Genomics. As shown here, it is straightforward to craft a series of prompts to AI for a specific topic, these prompts being a description of ‘WHO’ the person is (our three medical professionals), how the person is to ‘JUDGE’ (the rating scale), and finally ‘WHAT’ the person is to judge (the vignette).

The Mind Genomics task is difficult for people. It presents combinations of elements which may or may not fit together, but which at least do not contradict each other. Yet, human beings can do it. What is remarkable is that AI can do the task, perhaps not as well as people because of the lower coefficients, but nonetheless AI can do the job.

What is most remarkable is that AI with synthetic respondents can deal with the two-sided scale, one side for motivating or not motivating the patient, the second side for the right versus the wrong thing said.

‘Looking to the meaning of the data, we can focus on both the philosophical issue of the ‘Turing Test’ and the use of the approach to create a body of knowledge for teaching. The issue of the Turing Test is a well known one in philosophy. Quite simply, can we create a test which to figure out whether a machine is a machine or a human being. The data here suggest that the Mind Genomics process can be reasonably mimicked by a machine, with the answers ‘making sense.’

Following quickly on the heels of the foregoing question is whether or not the Mind Genomics system can be engineering to become a teaching/learning system, wherein we design the persona to have a variety of emotional and other issues, and then evaluate the response of alternative descriptions of treatments. The sheer number of papers dealing with this issue of learning about interactions between groups of people is heartening [11-13]. This second avenue is likely to be the one more interesting, and ultimately more fruitful to researchers, to philosophers and to society alike.

Acknowledgment

The senior author, HRM, dedicates this paper to the memory of a mentor whom he never met, but who was instrumental in his thinking since 1965. The mentor is the late Harvard University professor, Anthony Gervin Oettinger of blessed memory, who laid the foundation of HRM’s interest in artificial intelligence. It was Professor Oettinger who planted the seeds of this paper almost six decades ago, in his offer to have the author participate in the 1960’s Harvard project, TACT, Technical Aids to Creative Thought. Thank you Tony. This paper is for you.

References

  1. Dave T, Athaluri SA, Singh S (2023) ChatGPT in medicine: an overview of its applications, advantages, limitations, future prospects, and ethical considerations. Frontiers in Artificial Intelligence 6: p.1169595. [crossref]
  2. Campbell RT, Hudson CM (1985) Synthetic cohorts from panel surveys: An approach to studying rare events. Research on Aging 7: 81-93. [crossref]
  3. Dang HAH, Dabalen AL (2019) Is poverty in Africa mostly chronic or transient? Evidence from synthetic panel data. The Journal of Development Studies 55: 1527-1547.
  4. Dang HAH, Lanjouw PF (2018) Poverty dynamics in India between 2004 and 2012: Insights from longitudinal analysis using synthetic panel data. Economic Development and Cultural Change 67: 131-170.
  5. Gofman A, Moskowitz H (2010) Isomorphic permuted experimental designs and their application in conjoint analysis. Journal of Sensory Studies 25: 127-145.
  6. Moskowitz HR (2012) ‘Mind genomics’: The experimental, inductive science of the ordinary, and its application to aspects of food and feeding. Physiology & Behavior 107: 606-613.
  7. Moskowitz HR, Gofman A (2007) Selling Blue Elephants: How to Make New Products That People Want Before They Even Know They Want Them. Pearson Education.
  8. Moskowitz HR, Martin D (1993) How Computer Aided Design and Presentation of Concepts Speeds up the Product Development Process. Proceedings of the ESOMAR Congress Copenhagen.
  9. Likas A, Vlassis N, Verbeek JJ (2003) The global k-means clustering algorithm. Pattern Recognition 36: 451-461. [crossref]
  10. Todri A, Papajorgji P, Moskowitz H, Scalera F (2020) Perceptions regarding distance learning in higher education, smoothing the transition. Contemporary Educational Technology 13: p.ep287.
  11. Berry DC, Michas IC, Gillie T, Forster M (1997) What do patients want to know about their medicines, and what do doctors want to tell them?: A comparative study. Psychology and Health 12: 467-480.
  12. Collins J, Farrall E, Turnbull DA, Hetzel DJ, Holtmann G, et al. (2009) Do we know what patients want? The doctor-patient communication gap in functional gastrointestinal disorders. Clinical Gastroenterology and Hepatology 7: 1252-1254. [crossref]
  13. Kay M, Mitchell G, Clavarino A (2010) What do doctors want? A consultation method when the patient is a doctor. Australian Journal of Primary Health 16: 52-59. [crossref]

A Brief Commentary on Improving the Quality of Healthcare Delivery Practices Using the Soft Skills of Communication

DOI: 10.31038/IJNM.2023442

 

“Communication is life and healthcare is where it begins”

The influence of rising fears and anxieties in healthcare and its impact on the shape of the future of educational curriculum cannot be ignored. We are experiencing a new epoch in which the concept of absolute truth is gradually becoming as subjective as our various individualist viewpoints. Now more than ever before there is a dire need for humanity to constructively address the various factors that may play minor or major roles in the influence of healthcare delivery. Cultural and behavioral viewpoints, as well as affinities are examples of such influences. It’s fair to mention that my previous message in a 2017 publication “The Patient in Room 1B: Confronting our Fears to Build Trust” was a peek into the future which has gradually unfolded to reveal the diverse aspects and challenges that are encountered in healthcare delivery practices across the nation.

Check out “Patient in Room 1B LinkedIn article:

https://bit.ly/3G0Gqy9

In recent years, the word “culture” has become more or less popular, depending on one’s perspective of its influence in healthcare and overall existence dispositions. In addition, the current world of pandemic and post pandemic infections have ushered in new approaches to clinical/ nursing education which not only address care management for acute illnesses but seek to uncover effective approaches for healthcare providers to efficiently manage patient diverse behaviors, attitudes, and rising fears/anxieties brought on by past and current life experiences and their impact on perceived economic and political atmosphere. The use of virtual healthcare technology has increasingly surged, and the expanded use of audio, video and other electronic communications devices has offered some much-needed relief from individuals’ anxieties and the pandemic stressors by enabling patients to easily and yet discreetly connect with their healthcare practitioners via mobile health apps, health information systems etc.

However, while these technological advancements are impressive, they are insufficient to address the unique approach to care delivery which explores the application of such disciplines as psychology and the observation of patient behavioral traits, cognitive biases, and the philosophy of language and its attempt to assist the patient achieve a healthy mental and physical balance. Patient’s distinct personalities, experiences, and backgrounds affect their outlook and mental status. Tailoring an individual care approach is necessary for optimal care delivery. There are lingering questions that continue to pervade various communities in an era where many people are not only dealing with the devastating effects of a pandemic, but also have differing perspectives on the best ways to help stabilize the country’s political, economic, educational, and health structures. As truth increasingly becomes as subjective as our diverse personal perceptions, active attempts to create a common understanding through deliberate interactive means are more important than ever. In recent times, many critical race theory arguments have erupted due to differing individual experiences and perspectives. Some suggest that students may experience feelings of discomfort, guilt, misery, or psychological stress because of a race-based curriculum while others refute such perspectives and emphasize the need for full disclosure of events of historical practices and events. Many more reveal that there are more progressive and uniting strategies to introduce students to knowledge that allows them to think critically about diversity, equality, and discrimination. There are claims that such an approach is undoubtedly possible when instructional materials are presented as human behavior and human attribute-based materials rather than race-based materials. In my earlier published book, “Tips For Effective Communication: A vital Tool for Trust Development in Healthcare” and my recent publication, “Think Communicate, Heal”, I discussed the most effective ways to attain educational success by implementing an educational curriculum that will aid in the expansion of clinical reasoning in order to close the gaps that have been identified in existing tangible educational formats. This will enable students in the healthcare industry to increase their perceptual grasp of clinical intuitions based on a variety of patient behaviors (other than race or ethnicity) to provide more efficient care. Unintended outcomes of educational tools that focus mainly on historical accounts of race and diversity may include growing reservations and a desire for vengeance across various communities. This defeats the goal of shining a good light on solidarity in the face of changing perspectives, ideas, and viewpoints. Due to underlying fears, a purposeful shift in attention to understanding human qualities and behaviors would be a more efficient technique for success in our educational institutions. This is because these human attributes are universal and occur across all diverse communities. There is no color to the human emotions or fears, anger, elation, feelings of exclusion or inclusion etc. More nursing/medical educational resources for expanding the reasoning to bridge the gaps noted in current forms of concrete educational formats are recommended. It will help practitioners and students improve their perceptual grasp of diverse clinical situations, to achieve a more trusting environment resulting in better efficiency for care delivery practices.

Nonye Tochi Aghanya is a Family Nurse Practitioner with current practice in retail clinic setting. She’s maintained interactions with patients in diverse healthcare settings for over 32 years. An author of various publications, more information about her work can be found on her website www.nonyetochi.com

Methods of Zootechnical Evaluation of the Queens

DOI: 10.31038/IJVB.2023713

 

In the genetic selection and improvement programs of the reproductive animals, an assessment must be made of the quantitative or measurable characteristics whose hereditary differences are transmitted from generation to generation by the same mechanisms of the genes responsible for the qualitative differences [1]. A reproductive animal can be evaluated through the analysis of several aspects: assessment of its genetic potential at the request of the qualities of its ancestors (pedigree), by that of its half-siblings (collaterals), by that of its descendants (testing), or through their own genetic material (studying their genome); and it can also be evaluated through its productive performance and its phenotype [2]. Despite the fact that honey bees were the first intensively breeding animals, after the silkworm Bombyx mori L [3] and that the first research carried out on intensive livestock farming was directed at beekeeping [4] there have not set criteria, patterns and parameters to describe the queens bees phenotype and neither the reproductive performance, as has been done in other livestock farms with their breeding animals. All of which bring about a great void in the queens evaluation. The races of bees have been differentiated based on biometric methods and behavioral characteristics [5]. Biometric measurements have to do with the width of the thorax and abdominal segments; the length of the tongue, legs and wings; the color of the first segment dorsal abdomen; the length of the tongue; the hairy covering and the wings nervation [6,7]. Thus we see that the biometric methods have an eminently entomological approach and basically oriented to the bee workers, not giving importance to the zootechnical-reproductive characteristics of queens. In the queens has been studied: the relationship of weight with the number of ovarioles [8]; how the fecundity of queens is influenced by their weight [9]; the relationship among the weight of the queens at birth with the number of ovarioles and the volume of the spermatheca [10]; how the selection for the width of the abdomen of the queens improves some production characteristics [11]; the correlation among the genetic and phenotype parameters and the weight, width and length of the abdomen [12]. From a zootechnical point of view, of queens in general it has only been said that those with large abdomen, rounded flanks that gradually thin out and have uniform color are good for laying eggs, although sometimes a queen with the characteristics described above is not necessarily a good egg-laying queen [13]. The criteria used to assess the behavioral characteristics of the queens are inferred from the behavior of the colonies (of their progeny) the ability to winter, the degree of docility, the tranquility on the combs when the hives are been inspecting, the non-willingness to swarm, or that if it is a good honey-producing colony queen [13]. There have been indicated of the queens imprecise aptitudes, such as: whether she has filled three or four combs with brood or whether the egg laying is concentric and concentrated and with brood of similar ages, she is a good queen; that if she is erratic in her movements it is not desirable; that if she lays continuously, producing brood through out the season and into late fall, she is a good queen [13]. Hence, the absence of criteria, patterns, parameters, indicators and zootechnical characteristics of phenotypic and reproductive value to evaluate queens creates a great void in the exploitation of bees as livestock animals. Consequently, it is necessary to establish practical and easy-to-apply zootechnical methods with the following objectives:

  1. Provide the beekeeper with some parameters and patterns that, at the time of the inspection of the hives, allow him to zootechnically describe the queens, in a technical, truthful and fast way; relying on the observation of its most outstanding phenotypical aspects.
  2. Make available a technique for quantifying the queen egg laying, based on biological criteria that are easy to understand and apply, with which the beekeeper can evaluate the reproductive behavior of queens within the conception of a farm animal.

References

  1. Shrode RR (1980) Sección IV. Herencia y mejora animal; La herencia y su forma de actuar. En, Curso de zootecnia, editorial Acribia. Zaragoza. España, pp. 253-358.
  2. Sañudo, C Sánchez, C Marcén JM (2009) Capítulo 8. Variación morfológica en bovino lechero. En, Valoración morfológica de los animales domésticos. Realización: SEZ. Coordinador: Carlos Sañudo Astiz. Edita Ministerio de Medio Ambiente y Medio Rural y Marino. Secretaría General Técnica. Centro de Publicaciones, pp. 235-269.
  3. Borror DJ Triplehorn, Ch.A Johnson NF (1989) An introduction to the study of insects. Sext Editions Saunders College Publishing. Harcourt Brace Jovanovich College Publishers. Printed in the United States of America. Library of Congress, pp. 588-664.
  4. Shrimpton DH (1970) Sección II. Especial. La investigación en relación con la ganadería intensiva. En, Zootecnia intensiva, editorial Acribia. Zaragoza. España, pp. 371-613.
  5. Ruttner, F (2015) Races of bees. In, The hive and the honey bee. Editorial Dadant & Sons. Inc, Hamilton, IL, (6234)1, USA, pp 47-70.
  6. Daly HV Balling SS (1978) Identification of africanized honey bees in the western hemisphere by discriminat analysis. Journal of the Kansas Entomological Society. 51(4): 857-869.
  7. Rinderer TE Sylvester HA Collins AM Pesante, D (1986) Identification of africanized and european honeybees: Effects of nurse bee genotype and comb size. Bulletin of the Entomological Society of America (32): 150-152.
  8. Hoopingarner, R Farrar CL (1959) Genetic control of size in queen honeybees Econ. Ent. 52: 547-548.
  9. Boch, R Jamieson CA (1960) Relation of body weight to fecundity in queen honeybees in The Canadian Entomol. V 92, 9: 700-701.
  10. Corbella, E Gonçalves LS (1982) Relationship between weight at emergence, number of ovarioles and spermathecal volume of Africanized honey bee queens (Apis mellifera). Rev. Bras. Genet 4, 835-840.
  11. Costa FM (2005) Estimativas de parâmetros genéticos e fenotípicos para peso e medidas morfométricas em rainhas Apis mellifera 39 f. Dissertação (Mestrado em Zootecnia)-Universidade Estadual de Maringá, Maringá.
  12. Halak AL 2012 Parámetros e correlacóes genéticas e fenotípicas para peso e medidas morfométricas em rainhas. Apis mellifera Dissertação apresentada, como parte das exigências para a obtenção do título de Mestre em Zootecnia, no Programa de Pós-Graduação em Zootecnia.
  13. Cale GH Sr Banker, R Powers, J 2015 Management of the hive for the production of honey. In, The hive and the honey bee, Editorial Dadant & Sons. Inc, Hamilton, IL, 62341, USA, pp. 463-531.