DOI: 10.31038/EDMJ.2018247

Abstract

Very few studies have been done on oral health in the elderly and in particular the impact of type 2 diabetes on the dental appliance in the elderly. This is a feasibility study, with the use of a questionnaire, conducted in consultation with geriatric physicians, diabetologists and dentists.

Keywords

Periodontal Health, Type 2 diabetes, Elderly, Nursing Homes

Letter to the Editor

In 2014, an international project entitled “Factors Influencing the Oral Health of Elderly Diabetics 65 Years of Age and Older” was born. It is the fruit of collaboration between the University Hospital of Rouen in France and the University of Fortaleza in Brazil. The goal of the project was to deepen and broaden our current understanding of Type 2 Diabetes, and to evaluate the factors influencing the oral condition of elderly diabetics over 65 years of age.

It was an epidemiological study, prospective and non-interventional, carried out over a period of 9 months, between March 2014 and November 2015 at the Saint-Julien site at the University Hospital of Rouen, France.

The subjects were 78 patients with type 2 diabetes, predominantly males (43 subjects, 55.1%), with an average age of 80.8 +/- 8.0 years (80.7 +/- 8.1 for the females, 80.9 +/- for the men), with the extremes in age being 65 and 98, and a median age of 81. There are no significant differences between the control of diabetes and variables like age, gender, BMI, tobacco use and prosthetic parameters. A trend is apparent between inclusion links and poor glycemic control (p=0.07) variables: 80% of patients in nursing homes are poorly controlled versus only 46.2% of home patients. 62% of patients presenting an HbA1c<=7.5% have more than 4 Functional Units (FUs), versus 47% of patients with poor glycemic control. We observe that a higher percentage of poorly controlled patients do not present a functional pair (32% versus 17% of controlled subjects). Out of 59 patients with teeth, we have 16 patients with a partial set of teeth whose HbA1c is unknown, 24 with teeth and an HbA1c<=7.5%, and 19 with teeth and an HbA1c>7.5%. We observe that sextant scores 3, X and 2 occur primarily among those patients with an HbA1c <=7.5%, representing respectively 31.25%, 27.08% and 24.31% of total sextants. These correspond to testing <3 mm (score 2); from 3.5 to 5.5 mm
(score 3) or sextants with either 0 teeth or 1 tooth. Diabetic patients with poorly controlled diabetes (>7.5%) predominantly present a score of X (44.74%), followed by scores of 3 (21.93%) and 2 (18.42%). A score of 4, representing the most advanced conditions of poor periodontal health, is found in 6.25% of controlled patients, versus 9.65% of poorly controlled patients. Out of 42 diabetic patients with teeth, only 9 patients present a plaque index <25%. The average plaque index for those patients presenting an HbA1c<=7.5% is 58.12%, versus 69.10% for those patients with an HbA1c>7.5%.

This indicates that the majority of sextants examined that present a score of 0 or 1 occur in controlled diabetic patients (11.1% versus 5.2%). A score of 4, representing the most advanced periodontal disease, is primarily found in poorly controlled diabetic patients (9.5% versus 6.2%). However, there is a wider gap between the two controlled versus poorly controlled groups in the number of edentulous sextants (score X), which consequently influences the number of functional units and the Masticatory Coefficient (MC) (p=0.0011). Many studies have reported high numbers of missing teeth among diabetic patients, but in most cases the accounts show that the difference observed is not significant [1].

In the case of Functional Units (FU) number <=4, with a majority of sextants edentulous (non rehabilitated) is discernable (60%), while in more balanced dental situations (FU>4) the sextants with teeth presenting scores of 2 and 3 are more common (26.2% and 40.4% respectively). Similarly, situations in which CM>40 are primarily present in edentulous sextants (46.2%), versus sextants presenting scores of 2 and 3 (27.7% and 51.1% respectively) for situations in which CM>=40.

Conservation of a minimum of 20 teeth is necessary in order to maintain proper masticatory function. However, in our group, over ¾ of patients have fewer than 20 teeth in their mouths [2]. We observe a weak masticatory coefficient among our type 2 diabetic patients with teeth. 89% of patients have a masticatory coefficient under 60, and it is exacerbated among the poorly controlled patients. However the lower masticatory coefficient and low number of remaining teeth could be related to undernourishment or malnutrition [3–4]. Evaluation of the group of patients with teeth shows a clear prosthetic need among elderly diabetics. 50.8% of these patients need maxillary and/or mandibular rehabilitation, either due to a complete lack of rehabilitation or to an inadequate prosthesis or prostheses.

This is an ongoing study.

Reference

1. Taylor  Gw,  Manz  MC,  Borgnakke  WS (2004)  Diabetes,  periodontal  diseases,  dental  caries,  and  tooth  loss :  a  review  of  the  literature. Comprend Contin Educ Dent 25: 179–184, 186–188, 190 [crossref]
2. Krall E, Hayes C, Garcia R (1998) How dentition status and masticatory function affect nutrient intake. J Am Dent Assoc 129: 1261–1269 [crossref]
3. Ervin  RB,  Dye  BA (2012) Number  of natural  and  prosthetic  teeth  impact  nutrient  intakes  of  older  adults  in  the  United  States. Gerodontology 29:  693–702 [crossref]
4. Savoca  MR,  Arcury  TA,  Leng  X,  Chen  H,  Bell  RA, et al (2010)  Severe  tooth  loss  in  older  adults  as  a  key  indicator  of compromised dietary quality. Public Health Nutr 13: 466–474 [crossref]

DOI: 10.31038/EDMJ.2018246

Abstract

Objective

Obesity rates are increasing in women of child bearing age but it is unclear if the recognised risk of stillbirth from obesity is increasing over time. This study aimed to describe the association between stillbirth and maternal obesity and explore temporal trends.

Study design

Singleton stillbirth cases (1995–2012) in Fife, Scotland were used in a case control study matching for maternal age, parity and prematurity. Maternal height, weight at booking and relevant maternal and infant confounders were obtained for 271 stillbirths and 976 controls. Analysis included conditional logistic regression analyses using WHO categories on body mass index (BMI). Data were subdivided into 3 epochs to investigate temporal trends. The main outcome measures were multivariate odds ratio (OR) and confidence interval (CI) for stillbirth associated with BMI categories and change between epochs in OR for a stillbirth associated with maternal obesity.

Results

Risk of stillbirth was positively related to BMI category even after adjusting for known confounders. The relationship was particularly marked in nulliparous women where the adjusted OR of stillbirth was 4.1 (95% CI 1.9 – 9.0) when BMI was 35-<40 and 8.0 (95% CI 2.9 – 21.9) if BMI was > = 40. The proportion of stillbirth cases with a maternal BMI > = 35 increased from 5.6% in 1995 – 2000 to 18.9% in 2001–2006 and to 23.1% in 2007 – 2012 (P = 0.001, test for trend). There were no equivalent trends in the controls. The difference in mean BMI between cases and controls increased from 1 kg/m² (SE 0.6) in 1995–2000 to 3 kg/m² (SE 0.8) in 2007–2012. The OR of a stillbirth associated with a BMI above the 90^th percentile of the population distribution increased between epochs from 1.21 (P = 0.60) in 1995–2000 to 3.30 (P<0.001) in 2007–2012.

Conclusions

Maternal obesity is associated with an increased risk of stillbirth, particularly marked in nulliparous women with a BMI >35. The relative importance of maternal obesity as a risk factor for stillbirth may be increasing with time which has implications for future clinical practice and when interpreting the impact of the various interventions being developed to reduce stillbirth rates.

Keywords

Stillbirth, Maternal Obesity, Body Mass Index, Pregnancy, Temporal Trend.

Introduction

The prevalence of obesity amongst women of child-bearing age in the UK has shown a rising trend over the past 3 decades, in keeping with trends in other developed countries [1]. Studies of pregnant women at booking in England [2] and Scotland [3,4] have reported a doubling in the prevalence of obesity (defined as a body mass index, BMI of > = 30 kg/m²) between 1990 and the mid-2000s. Maternal obesity is associated with adverse pregnancy outcomes of which one is stillbirth [5–10]. Previous studies have suggested a strong link between pre-pregnancy obesity and the risk of stillbirth [11–13]. Furthermore, one study has suggested that stillbirths in obese women (compared with those in women of ideal body mass) more often occur at term or post-term, and are more likely to be classified as ‘unexplained’ [13]. It has been suggested that maternal obesity may be one of the most important modifiable risk factors associated with stillbirth [12] having a large population attributable risk relative to other features [8–10]. In Scotland the rate of stillbirths has remained stubbornly high [14] despite efforts to reduce it and the increasing prevalence of obesity in Scotland [15] and, in particular amongst pregnant women in the UK [2–4], may be a contributory factor. The purpose of the present study was to determine the magnitude of the association between maternal obesity and stillbirth in Scotland and to assess if its relative importance as a feature in stillbirths is increasing with time.

Materials and Methods

We used data from Fife (a county in east central Scotland with a stable population of about 360,000 and about 3,800 births per year) because BMI has been routinely measured at booking since the early 1990s. Stillbirths between 1995 and 2008 inclusive were identified from the Fife Birth Register. A stillbirth was defined as a baby delivered with no signs of life after 24 completed weeks of gestation [5]. An initial analysis of these data proved insufficient to meet the study’s required statistical power so we identified further cases from 2009 to March 31^st 2012 from a list of all Fife births from the Information Services Division (ISD) of the NHS National Services Scotland maternity inpatient and day case records (SMR02) [16]. The obstetric records of stillbirths were reviewed (TM, AL) to exclude multiple pregnancies, women from ethnic minorities (<1% of total births) and cases with a listed cause of the stillbirth (for example, fetal anomaly, maternal medical disorders, maternal infection). For the remaining cases we used the respective registers to identify two control pregnancies on either side of the date of birth (+/- 7 days) of the stillborn infant. Women with multiple pregnancies and those from ethnic minorities were excluded from the control group. Pregnancies were matched for:

1. Maternal age (+/- 5 years), [but for teenage mothers we sought controls aged as close as possible]
2. Prematurity (term or preterm) where term was defined as 37+ weeks gestation. For preterm infants matching was done +/- 28 days of the date of birth.
3. Parity (nulliparous or parous, where nulliparous was defined as a woman who had never given birth to an infant capable of survival).

Power calculations suggested we needed 260 cases, each matched to 4 controls to have an 80% chance of detecting an odds ratio of 1.5 (at 5% two-tailed significance) for a stillbirth associated with a BMI of > 30 kg/m², assuming the prevalence of a BMI >30 in the control sample was 30% [17]. The number of stillbirths recorded on the two registers (1995–2012) was 335 cases, of which at least 300 (about 90%) were expected to be singletons. Data for the cases between 1995 and 2008 were collected from obstetric notes at booking and at delivery. Data for cases between 2009 and 2012 were taken from the SMR02 records that had been recorded from data submitted by the Fife Health Board from information held in the women’s obstetric notes.

We recorded the following maternal factors: age, height, weight (at booking), smoking history, details of past obstetric history (including history of previous stillbirth) and postcode, from which we derived the Scottish Multiple Index of Deprivation quintile [18]. Height and weight were used to calculate Body Mass Index (BMI = weight / (height²), kg / m²) which was categorised according to the World Health Organisation’s definitions (<18.5 underweight, 18.5–24.9 ideal range, 25–29.9 overweight or pre-obese, 30–34.9 obese class 1, 35–39.9 obese class 2, 40 or more obese class 3) [19].

The infant factors recorded were gestational age, mode of delivery, gender, birth weight and its percentile for gestational age based on population birth records from Scotland adjusted for infant gender and maternal parity [20]. Small for Gestational Age (SGA) was defined as a birth weight below the 10^th percentile (<-1.282 Z-score) of the normal distribution and Large for Gestational Age (LGA) was defined as a birth weight above the 90^th percentile (>1.282 Z-score).

Data were analysed with SPSS (version 20) using t-tests, comparison of proportions (Chi-square), analysis of variance and conditional logistic regression analyses. The 5% level was accepted as indicating statistical significance. The association between maternal BMI at booking and stillbirth was assessed in a univariate analysis and in multivariate analyses stratified by parity after adjusting for known confounders. Odds Ratios (OR) and their 95% Confidence Intervals (CI) were calculated for categorical variables, including BMI categorised into subgroups.

Results

Three hundred and thirty five stillbirths were identified, of which 33 were excluded (23 multiple pregnancies or births to women of ethnic minority, 6 with missing case notes and 4 with case notes but a missing BMI). Of the 302 remaining cases a further 31 were excluded (30 where there was a fetal abnormality or maternal infection listed as the cause and 1 for which the stillbirth was a consequence of a road traffic accident). Hence there were 271 stillbirths. For each of these cases we obtained 4 controls for 199 cases, 3 controls for 37 cases, 2 controls for 34 cases and 1 control for 1 case (976 controls in total).

Characteristics of cases and controls are given in Table 1. Matching of cases and controls for maternal age, parity and prematurity was satisfactory. In univariate analyses significant terms associated with an increased risk of stillbirth were maternal weight, BMI, birth weight and Small for Gestational Age (SGA). For SGA the unadjusted Odds Ratio (OR) for a stillbirth was 3.8 (95% CI 2.7 to 5.3). For BMI categories the linear test for trend was significant at P < 0.001, (ANOVA, F = 15.57).

Table 1. Characteristics of stillbirth cases and controls. Values are numbers (percentages) unless otherwise specified. Matching variables are in bold.

* Chi-square test (proportions) or unmatched t-test (continuous data).
** n=206 cases, 717 controls
† SGA below 10^th percentile (< -1.282 Z-score birth weight)
‡ LGA above 90^th percentile (> 1.282 Z-score birth weight)

A conditional logistic regression analysis including the maternal and fetal terms identified increased maternal BMI categories and an SGA infant as associated with stillbirth. The risk of stillbirth from a raised BMI was apparently greater in nulliparous than in parous women (Table 2, Figure 1) though the difference between them in odds ratios per BMI category was not statistically significant.

Figure 1. Adjusted odds ratio for stillbirth and BMI in Parous and Nulliparous women (adjusted, where relevant, for smoking status, primigravida, deprivation, fetal sex, SGA, LGA and history of previous stillbirth, previous spontaneous miscarriages and previous therapeutic terminations. For details see Table 2)

Table 2. Conditional logistic regression analysis of maternal and fetal factors and their association with stillbirth.

aOR adjusted odds ratio
† SGA below 10^th percentile (< –1.282 Z–score birth weight)
†† LGA above 10^th percentile (> +1.282 Z-score birth weight)

Data were split into three epochs (1995–2000, 2001–2006 and 2007–2012). The percentage of stillbirths with a maternal BMI of 30 or more did not differ significantly between periods but that where BMI was 35 or more, and where BMI was 40 or more did show a significant trend over time (P = 0.001 and P = 0.007, respectively). These patterns were not apparent in the controls (Table 3). The difference in mean BMI between stillbirth and controls also increased with time (Table 4). The Standard Deviation (SD) in BMI increased over time, particularly amongst the stillbirth cases (Table 4). To adjust for this change we pooled the cases and controls in each epoch and expressed each woman’s BMI as a Z-score. We then used logistic regression to calculate the odds ratio per unit Z-score (i.e. per SD) in BMI for each epoch and calculated the OR of a stillbirth for a woman with a BMI above the 90^th percentile of the distribution (+ 1.282 SD). The OR of a stillbirth for a woman in this category increased over time from 1.21 in 1995–2000 (P = 0.60) to 3.30 in 2007–2012 (P < 0.001, Table 5). These ORs were little affected by also allowing for parity and SGA in a logistic regression.

Table 3. Percentage of stillbirth and controls with an increased maternal BMI (kg/m²) by time periods (1995 – 2012)

Table 4. Difference in mean BMI (kg/m²) between stillbirth and controls by time period

SD: standard deviation, SE: standard error

Table 5. Difference in odds ratio per standard deviation in BMI (kg/m²) by time period.

* Chi-square test

Discussion

We posed two questions. The first was to determine the magnitude of the association between maternal obesity and stillbirths. Our study confirmed the work of many others that maternal obesity is an important, independent risk factor associated with stillbirth [8–13]. The trend in increased risk of stillbirth was positively related to BMI category with an apparent stronger association in nulliparous than parous women, which also confirms the work of others [12]. The second question concerned the change in maternal BMI and apparent risk of stillbirth with time. The proportion of stillbirth cases with a maternal BMI > = 35 increased from 5.6% in 1995 – 2000 to 18.9% in 2001–2006 and to 23.1% in 2007 – 2012 (P = 0.001, test for trend). A similar, significant trend was noted for BMI > = 40 (P = 0.007). In comparison, there were no apparent equivalent trends in the controls. The mean and variance of maternal BMI of stillbirth cases also increased over time with the mean difference in BMI between stillbirth cases and their matched controls also increasing. An analysis adjusting for the increase in population variance revealed an apparent increase over time in the risk of a stillbirth from a raised BMI. The relative increase in maternal BMI may partly explain the apparent stubborn stability in the rate of stillbirths in Scotland over the past 2 decades which has remained at about 4–5 / 1000 total births [21].

This study used the WHO classification of obesity and benefited from a population-based source of BMI data that had been routinely collected from measured height and weight at booking since the early 1990s. This enabled us to look at trends over time using data of sufficient statistical power to detect change. Practice in recording BMI has varied in other health boards in Scotland and maternal BMI was not routinely available from the SMR02 records (ISD) before about 2003. We had sought 4 controls per case but were only able to achieve this with 73% (199/271) of stillbirths, though, despite this, our cases were appropriately matched. In addition, we excluded stillbirths with congenital abnormalities as these have been related to maternal obesity in a meta-analysis of 18 studies [22]. We did have history of diabetes and gestational diabetes for the 1995–2008 dataset but not for the SMR02 source (ISD data). However, a preliminary analysis of the 1995–2008 data did not reveal a trend in risk of stillbirth associated with diabetes though the overall prevalence was low (results not shown). Women with a raised BMI are at greater risk of gestational diabetes and hypertensive disorders of pregnancy [23] though, in a large, case-control study the association of maternal obesity and risk of stillbirth was not explained by diabetes or hypertension [9].

In estimating birth weight for gestational age we used population-based data from Scotland (1998–2003) that provided estimates adjusted for fetal sex and parity [20]. The mean Z-score and standard deviation of the 976 control pregnancies was -0.003 and 1.02, respectively, which were appropriate for a reference population where the expected mean and standard deviation would be 0 and 1, respectively. Our findings with respect to stillbirth risk, SGA and raised BMI may have differed had we used fully customised percentile charts for fetal size [24], particularly as 60% of pregnancies were preterm (<37 weeks) when the apparent differences between population based and customised percentiles in determining stillbirth risk in a fetus classified as SGA are larger [25].

Two studies in Scotland have reported an increase in mean maternal BMI of about 1–2 kg/m² per decade [3,4]. The percentage of pregnant women in Glasgow with a BMI > = 30 at booking increased from 9.4% in 1990 to 18.9% in 2002/04 [3]. This reflects trends in other parts of the UK where the percentage of pregnant women with a BMI > = 30 was 7.6% in 1989 and 15.6% in 2007 [2]. A recent UK study of adverse pregnancy outcomes associated with obesity in nulliparous women reported an increased risk of stillbirth with raised maternal BMI whereby the occurrence of stillbirth in 3102 women with a BMI 20-<25 equated to 1 stillbirth in 443 pregnancies and in 105 women with a BMI >40 to 1 stillbirth in 26 pregnancies [26]. These findings, and our data, raise concerns over future stillbirth rates in the face of a growing epidemic in population obesity in Scotland [15] particularly amongst teenage girls.

Recommendations on the management of obese pregnant women have been made for the UK [27] and, more recently, by the European Board and College of Obstetrics and Gynaecology (EBCOG) which has published standards of care for obstetric and neonatal services to be adopted across Europe [28]. It is recommended that women of child-bearing age with a BMI >30 should have access to services offering advice on weight management pre-conception and be warned of the obstetric risks associated with a raised BMI. Such advice should also be given during antenatal care. However, evidence suggests compliance with providing advice in both circumstances is poor [29]. The EBCOG report recommends all units use multidisciplinary input to develop clear policies and protocols for the care of women with a BMI>30 who should be risk assessed at each antenatal visit. The UK guidelines recommend additional scans in the third trimester for evidence of, amongst other conditions, growth restriction that is associated with about half of all stillbirths [30, 31]. However, sonographic visualisation can be compromised as a result of obesity [32], which in itself could reduce the potential for detection of growth restriction and congenital abnormalities [33]. Finally, it has been suggested that obese women may be less perceptive of reduced fetal movement than women of a healthy weight [12].

In Fife, pregnant women are offered advice and guidance on healthy lifestyle choices. Those overweight or obese are invited to join classes on weight management and offered one-to-one advice from a dietician though anecdotal evidence suggests uptake is inconsistent. Information is given regarding the maximum recommended weight gain throughout pregnancy depending on their BMI. However, the increased risk of stillbirth is apparently related to BMI during early pregnancy rather than to weight gain during it [11, 34] which reinforces the need for greater pre-conception weight management.

What of studies seeking the views of women? In a recent survey of 428 overweight and obese pregnant women in Fife 81% were concerned about their current weight but 39% were unconcerned about potential weight gain during their pregnancy [35]. Of 252 women who were in their second or subsequent pregnancy 47% had failed to return to the pre-pregnancy weight associated with their previous pregnancy. Inter-pregnancy weight gain can increase the risk of stillbirth in a subsequent pregnancy [36, 37]. In a recent qualitative study in the UK 40 women (32 currently pregnant) were asked about public health messages on the risks of stillbirth [38]. In general, they were resistant to the increased weight message, which they considered was not modifiable as ‘all pregnant women are overweight’.

In conclusion, this study has confirmed the association between maternal obesity and an increased risk of stillbirth with a particularly high risk in nulliparous women with a BMI > = 35. The relationship follows a ‘dose-response’ that adds to the body of evidence suggesting a causal link between stillbirth and excess weight in pregnancy per se [39]. The relative importance of maternal obesity as a risk factor, and putative cause of stillbirth (and other adverse pregnancy outcomes) may be increasing with time. This needs to be acknowledged when designing interventions to reduce the incidence of stillbirths. Our findings may be relevant for other populations given the recently published global estimates of obesity in women of child-bearing age [1]. Concerted efforts raising public awareness of the pregnancy-related consequences of obesity are urgently needed though presenting the message represents a challenge.

Contribution to authorship

TM and lC conceived the study with advice from HC. IC obtained the funding. TM, AL reviewed the obstetric notes and contributed to the literature review. AL collected and prepared the data for analysis. DJC advised on the study design, analysed and interpreted the data, contributed to the literature review and wrote the initial draft of the manuscript. All authors contributed to the manuscript and approved the final draft.

Ethics Approval

This was a secondary analysis of data currently available to NHS Fife (SMR02 and obstetric case records). Caldicott approval was obtained to access the obstetric records.

Funding

The study was funded from an NHS Fife research fund.

Acknowledgements

We are grateful to Mr Bryan Archibald of the Public Health Department, NHS Fife for extracting the data for 2009–2012 from the SMR02 data provided routinely from the Information Services Division, NHS Scotland.

References

1. Ng M, Fleming T, Robinson M, Thomson B, Graetz N, et al. (2014) Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 384: 766–781
2. Heslehurst N, Rankin J, Wilkinson JR, Summerbell CD (2010) A nationally representative study of maternal obesity in England, UK: trends in incidence and demographic inequalities in 619 323 births, 1989–2007. Int J Obes 34: 420–428.
3. Kanagalingam MG, Forouhi NG, Greer JA, Sattar N (2005) Changes in booking body mass index over a decade: retrospective analysis from a Glasgow Maternity Hospital. BJOG: Int J Obstet Gynaecol 112: 1431–1433.
4. Bhattacharya S, Campbell DM, Liston WA, Bhattacharya S (2007) Effect of body mass index on pregnancy outcomes in nulliparous women delivering singleton babies. BMC Public Health 7: 168.
5. Confidential Enquiry into Maternal and Child Health (CEMACH) (2008) Perinatal Mortality 2006: England, Wales and Northern Ireland. CEMACH.
6. Fretts RC (2005) Etiology and prevention of stillbirth. Am J Obstet Gynecol 193: 1923–1935. [crossref]
7. Smith GC, Fretts RC (2007) Stillbirth. Lancet 370: 1715–1725. [crossref]
8. Flenady V, Koopmans L, Middleton P, Frøen JF, Smith GC, et al. (2011) Major risk factors for stillbirth in high-income countries: a systematic review and meta-analysis. Lancet 377: 1331–1340. [crossref]
9. Stillbirth Collaborative Research Network Writing Group (2011) Association between stillbirth and risk factors known at pregnancy confirmation. JAMA 306: 2469–2479.
10. Gardosi J, Madurasinghe V, Williams M, Malik A, Francis A (2013) Maternal and fetal risk factors for stillbirth: population based study. BMJ 346: 108. [crossref]
11. Nohr EA, Bech BH, Davies MJ, Frydenberg M, Henriksen TB (2005) Prepregnancy obesity and fetal death: a study within the Danish National Birth Cohort. Obstet Gynecol 106: 250–259.
12. Cnattingius S, Bergström R, Lipworth L, Kramer MS (1998) Prepregnancy weight and the risk of adverse pregnancy outcomes. N Engl J Med 338: 147–152. [crossref]
13. Kristensen J, Vestergaard M, Wisborg K, Kesmodel U, Secher NJ (2005) Pre-pregnancy weight and the risk of stillbirth and neonatal death. BJOG 112: 403–408. [crossref]
14. Information Services Division, NHS Scotland (2009) Trends in perinatal mortality in Scotland: a review over 30 years. Edinburgh, 2009.
15. Obesity in Scotland. An epidemiological briefing. Public Health Observatory, Scotland. NHS National Health Services Scotland, 2007 (www.scotpho.org.uk)
16. Information Services Division of the NHS National Services Scotland maternity inpatient and day case records (SMR02).
17. Kirkwood BR (1988) Essentials of Medical Statistics. Blackwell Scientific Publications. London.
18. Scottish Index of Multiple Deprivation (2012) www.scotland.gov.uk/Topics/Statistics/SIMD/.
19. World Health Organisation (1997) Obesity: preventing and managing the global epidemic. Report of a WHO consultation on obesity, 1997. WHO, Geneva, 1998.
20. Bonellie S, Chalmers J, Gray R, Greer I, Jarvis S, Williams C (2008) Centile charts for birthweight for gestational age for Scottish singleton births. BMC Pregnancy and Childbirth 8: 5.
21. Healthcare Improvement Scotland. Scottish Perinatal and Infant Mortality and Morbidity Report 2012. Edinburgh, March 2014.
22. Stothard KJ, Tennant PW, Bell R, Rankin J (2009) Maternal overweight and obesity and the risk of congenital anomalies: a systematic review and meta-analysis. JAMA 301: 636–650. [crossref]
23. Scott-Pillai R, Spence D, Cardwell CR, Hunter A, Holmes VA (2013) The impact of body mass index on maternal and neonatal outcomes: a retrospective study in a UK obstetric population, Ȁ4–2011. BJOG 120: 932–939.
24. Gardosi J, Clausson B, Francis A (2009) The value of customised centiles in assessing perinatal mortality risk associated with parity and maternal size. BJOG 116: 1356–1363. [crossref]
25. Hutcheon JA, Zhang X, Cnattingius S, Kramer MS, Platt RW (2008) Customised birth weight percentiles: does adjusting for maternal characteristics matter? BJOG 115: 1397–1404.
26. Mantakas A, Farrell T (2010) The influence of increasing BMI in nulliparous women on pregnancy outcome. Eur J Obstet Gynecol Reprod Biol 153: 43–46. [crossref]
27. Centre for Maternal and Child Enquiries & Royal College of Obstetricians and Gynaecologists. CMACE & RCOG Joint Guideline. Management of women with obesity in pregnancy. London, 2010.
28. European Board and College of Obstetrics and Gynaecology (2014) Standards of Care for Women’s Health in Europe: Obstetric and Neonatal Services 2014.
29. Centre for Maternal and Child Enquiries (2010) Maternal obesity in the UK: findings from a national project. London, 2010
30. Reddy UM1 (2007) Prediction and prevention of recurrent stillbirth. Obstet Gynecol 110: 1151–1164. [crossref]
31. Vais A, Kean L (2012) Stillbirth-is it a preventable public health problem in the 21st century? Obstet Gynaecol Reprod Med 22: 129–134
32. Hendler I, Blackwell SC, Bujold E, Treadwell MC, Wolfe HM, et al. (2004) The impact of maternal obesity on midtrimester sonographic visualisation of fetal cardia and craniospinal structures. Int J Obesity 28: 1607–1611
33. Best KE, Tennant PW, Bell R, Rankin J (2012) Impact of maternal body mass index on the antenatal detection of congenital anomalies. BJOG 119: 1503–1511. [crossref]
34. Stephansson O, Dickman PW, Johansson A, Cnattingius S (2001) Maternal weight, pregnancy weight gain, and risk of antepartum stillbirth. Am J Obstet Gynecol; 184: 463–469.
35. Leslie WS, Gibson A, Hankey CR (2013) Prevention and management of excessive gestational weight gain: a survey of overweight and obese pregnant women. BMC Pregnancy Childbirth 13: 10. [crossref]
36. Villamor E, Gnattingius S (2006) Interpregnancy weight change and risk of adverse pregnancy outcomes: a population-based study. Lancet 368: 1164–1170.
37. Whiteman VE, Crisan L, McIntosh C, Alio AP, Duan J, et al. (2011) Interpregnancy body mass index changes and risk of stillbirth. Gynecol Obstet Invest 72: 192–195. [crossref]
38. Stillbirth and Neonatal Death Charity (SANDS). Research Report: Stillbirth Public Health Messages, November 2013.
39. Salihu HM (2011) Maternal obesity and stillbirth. Semin Perinatol 35: 340–344. [crossref]

DOI: 10.31038/EDMJ.2018245

Abstract

Objective: To provide program methodology and outcomes data identifying the impact of clinical pharmacy services provided to patients with diabetes mellitus.

Design: Prospective pilot study.

Patients: Adult patients with diabetes mellitus identified by a member of the primary care team were referred to the pharmacist-led disease state management program, a patient-centered medication therapy management (MTM) program developed through university collaboration with a local Federally Qualified Health Center.

Interventions: Pharmacist-delivered disease state management and medication therapy management across three or more face-to-face encounters over the course of six months.

Main outcome measures: Clinical outcomes were followed for 6 months from the time of referral and enrollment into the program. The primary diabetes endpoint, glycosylated hemoglobin, and patient-reported experience with care were collected at baseline and the end of the study. Clinical pharmacists documented the content of clinical visits, including the number of visits per patient, duration of encounters, number and proportion of identified medication therapy problems, and the number and proportion of associated interventions to optimize pharmacotherapy.

Results: Glycosylated hemoglobin was significantly reduced versus baseline at the 6-month assessment in both the intent-to-treat (−2.7%; P < 0.0001) and the per-protocol groups (−3.0%; P < 0.0001). Patient-reported satisfaction with care was higher for the pharmacists as compared to the primary care providers with significantly more patients rating the care received from the pharmacist as excellent (P = 0.001). The pharmacists completed 158 visits, identifying and resolving an average of 7.7 medication therapy problems for each subject included in the analysis.

Conclusion: In this model of MTM, the clinical pharmacists were able to identify and resolve interventions which subsequently resulted in statistically significant reductions observed in the primary diabetes endpoint and high levels of satisfaction with care.

Background

The pharmacist’s role in direct patient care is ever evolving. Traditionally, the pharmacist’s ability to provide care to patients has been associated with the provision of a medication product. However, the past several decades have witnessed a significant expansion in how pharmacists may provide care. Unlike other healthcare providers, pharmacists are unique in their extensive training in performing medication therapy management (MTM) as part of their clinical service, whereby MTM is defined by the American Pharmacists Association (APhA) [1] as “a distinct service or group of services that optimize therapeutic outcomes for individual patients.” The goal of MTM is to promote safe and effective medication use via collaboration among pharmacists and other healthcare professionals in order to achieve optimal patient outcomes. As described by the joint framework developed by the APhA and the National Association of Chain Drug Stores Foundation, [2] MTM services incorporate five core elements:

1. A medication therapy review
2. A personal medical record
3. A medication-related action plan
4. Intervention and/or referral, and
5. Documentation and follow up.

There are a large number of publications in the literature that support the role of a pharmacist in disease state management via MTM services. However, there is a relative scarcity of structured studies evaluating the potential impact of MTM in the primary care medical home. In a review of randomized-controlled trials evaluating MTM services in this setting, it was determined that the most appropriate delivery methods were those that resulted in efficient implementation of recommended interventions [3]. In those studies that demonstrated positive impact, the practice settings were associated with an educational institution and promoted pharmacist autonomy, such as through the use of collaborative practice agreements authorizing pharmacists to directly adjust medications without need to consult the primary care provider (PCP) first. Study authors concluded that in order for MTM to be successful in the primary care medical home and to benefit patient care, it should be delivered discriminately to patients with a specific therapeutic problem as determined by the PCP. Also, MTM services should involve timely communication with PCPs to discuss therapeutic problems and routine patient follow up should be incorporated into the model to support adherence to changes in therapy.

In order to add to the evidence supporting the role of clinical pharmacists in delivering MTM in the primary care collaborative practice setting and to expand our current knowledge of the role of the pharmacist in the indigent population, a descriptive report of MTM was conducted at the FOCUS Clinic in Newark, New Jersey [4]. Over the course of 8 months, there were 313 documented patient-centered pharmacist interventions (PCPIs) affecting 69 unique patients across a range of chronic illnesses, such as hypertension, diabetes mellitus, dyslipidemia, and asthma.

While this report provides preliminary evidence supporting the role of a pharmacist in the management of a variety of disease states, it does not consider clinical outcome measures. Therefore, in 2017 we began a prospective pilot study to assess the impact of adding a pharmacist to the healthcare team in a Federally Qualified Health Center (FQHC). Through this study, pharmacists provide MTM services as suggested by Kucukarslan et al [3]. in the context of a structured study protocol that quantifies MTPs and interventions while also documenting impact on clinical outcome measures. This report presents interim data demonstrating the impact of chronic disease state management by clinical pharmacists on patients with diabetes mellitus at Henry J. Austin Health Center (HJAHC) in Trenton, New Jersey.

Methods

Between March 21, 2017 and March 21, 2018, 50 patients were referred to and enrolled into the study for management of diabetes mellitus by a clinical pharmacist. To be included in the study, patients may be referred to the pharmacist by any member of the healthcare team. In instances when the patient is referred by someone other than the PCP, the PCP is contacted to ensure that the PCP agrees with and supports the referral. If the PCP does not agree with the referral, the patient is not enrolled. Patients may be enrolled if they are 18 years old or older with at least one chronic medical condition in which pharmacotherapy is indicated and the total duration of pharmacotherapy is expected to continue for at least 6 months after referral. The disease state may be newly diagnosed or pre-existing. Patients are excluded from participation if they are less than 18 years old, pregnant, or referred for a disease state that is managed by an outside provider, is not chronic, or is one in which pharmacotherapy is not indicated and/or the total duration of pharmacotherapy is not expected to continue for at least 6 months after referral. Patients are administratively withdrawn from the study under the following circumstances:

• The patient is no longer being managed by a HJAHC PCP for the referred medical condition;
• The patient misses three appointments during the study period;
• The pharmacist is unable to schedule an appointment within 30 days of a missed appointment;
• The patient does not complete a minimum of three visits within the 6 month study period or the third appointment does not occur at least 30 days prior to the ideal study close out date; or
• The patient becomes pregnant at any point during the course of the study.

Patients who are seen by an outside provider for management of the referred disease state are excluded as the pharmacists cannot work directly with that provider and will not be able to make interventions within a team-based setting. As the purpose of this study is to assess the impact a pharmacist can make on chronic disease states over a 6 month period of time through MTM, patients who do not have at least one chronic medical condition that is expected to last for at least six months and requires pharmacotherapy are excluded. Any patient who fails to adhere to the study protocol is withdrawn to allow the pharmacists to enroll additional patients who may benefit from the intervention. Pediatric patients and/or pregnant patients are excluded due to variable disease course throughout the six month period of time and potential to create a non-homogenous sample. Upon enrollment, the pharmacists collect demographic information, clinical information, including the primary outcome measure, the patient’s self-rated health, which is reported on a 5-point Likert scale, healthcare utilization over the 6 months prior to enrollment, including visits to the PCP, emergency departments visits, and hospital admissions, and the patient’s satisfaction with care as it relates to his or her PCP measured by a seven question patient-reported survey.

Patients meet with clinical pharmacists for a minimum of three study visits over the course of 6 months. Visits include a comprehensive medication review and MTM, with a specific focus on improving clinical outcomes for the referred disease state. Clinical pharmacists document all clinical measures, vital signs, and contents of the encounters in HJAHC’s Electronic Medical Record (EMR) system to make this information available to all HJAHC team members. Recommendations are also communicated to PCPs verbally or via electronic notifications within the EMR. Patients complete routine blood work as it relates to their medical conditions. Meetings with the clinical pharmacists are conducted independently or in conjunction with the PCP or other healthcare professional visits. The following data is collected at each visit:

1. Date of study visit,
2. Duration of study visit,
3. Number and type of MTPs identified by the pharmacist, and
4. Number and type of PCPIs.

Upon completion of the study, patients undergo a final study evaluation. In addition to collection of the patient’s self-rated health and healthcare utilization, the patient completes a satisfaction survey regarding care received from the pharmacist.

Outcomes

The primary outcome of this interim report is to evaluate the change in glycosylated hemoglobin (A1C) from baseline to end of study for those patients referred to the pharmacist for chronic disease state management of diabetes mellitus.

Secondarily, this report evaluates the content of the interaction between clinical pharmacists and ambulatory care patients by reporting on the frequency and duration of clinical pharmacy appointments, total number and type of MTPs identified and associated disease states, and total number and type of pharmacist-initiated interventions. Additionally, patient satisfaction with care is compared between care received from the PCP and care received from the pharmacist.

Statistical Analysis

The intention-to-treat (ITT) analysis for the primary outcome includes all the participants who had at least one additional A1C measurement after the baseline assessment and within six months of enrollment. The per-protocol (PP) population includes all participants who adhered adequately to the study protocol, which is defined as follows. The patients completed at least three visits with the clinical pharmacist with the third visit conducted no later than 30 days prior to the ideal end of study date. The patients continued to receive treatment for diabetes from a HJAHC PCP throughout the duration of the study. Patients did not miss three or more scheduled appointments with the pharmacists, and in the event that the patient did miss an appointment, that appointment was rescheduled within 30 days of the missed appointment. Patients did not become pregnant at any time during the trial.

Change in A1C is compared from baseline to end of study using two-sample T-test assuming unequal variances for the ITT analysis and a paired T-test for the PP analysis. Differences in patient satisfaction with care between care received from the PCP and the pharmacist are compared using a chi-square test. Only patients who completed both baseline and final provider satisfaction surveys are included in the analysis of patients’ satisfaction with care. Statistical analysis is performed in Microsoft Excel. Descriptive statistics are reported for the MTM analysis.

The point of care (POC) A1C machine used to measure the A1C is unable to detect the exact value for any A1C that is greater than 15%. Therefore, the A1C value is reported as 15% for any patient who experienced an undetectable A1C greater than 15% at baseline.

Results

Participant Population

Fifty patients were referred and enrolled into the study between March 21, 2017 and March 21, 2018 for management of diabetes mellitus. Of those patients, 47 were referred for type 2 diabetes mellitus (T2DM), one was referred for type 1 diabetes mellitus, one was referred for gestational diabetes, and one was referred for pre-diabetes. Thirty-one patients were referred by a nurse practitioner, 18 by a physician, and one by a registered dietician. Of the 50 patients referred and enrolled into the study, 27 patients were included in the ITT population and 15 were included in the PP analysis. All patients included in either analysis were referred for T2DM. Demographic information for the patients included in the ITT analysis is summarized in (Table 1).

Table 1. Demographic Data

Of the 27 patients with at least one additional A1C result prior to the end of the study period, twelve patients were not included in the PP analysis. Three patients were lost to follow up, three patients did not reschedule a missed appointment within 30 days, two patients did not complete three visits with the pharmacists within the study period, two patients transferred care to an outside facility, one patient missed three appointments within the study period, and one patient was inappropriately enrolled into the study for diabetes management as the patient was referred for medication reconciliation. Four of the 27 patients included in the ITT analysis were newly diagnosed with T2DM at the time of referral.

Impact on A1C

Impact on A1C for the ITT and PP groups are summarized in Figures 1a and 1b. For the 27 subjects included in the ITT analysis, the mean baseline and final A1C results were 11.1% and 8.4%, respectively, resulting in a mean reduction in A1C of 2.7% (P=0.00003). For the 15 subjects included in the PP analysis, the mean baseline A1C was 10.9% and the mean final A1C was 7.9% to provide a mean reduction of 3% (P = 0.00007). (Figure 1a, 1b).

Figure 1a. Impact of Pharmacist Management of Diabetes on A1C for Intent-To-Treat Population

Figure 1b. Impact of Pharmacist Management of Diabetes on A1C for Per Protocol Population

Medication Therapy Management

Overall, the pharmacists conducted 158 visits with those patients included in the ITT analysis. During these visits, the pharmacists identified a total of 209 MTPs and associated PCPIs (Table 2).

Table 2. Summary of Medication Therapy Management Visits

The most common disease states requiring pharmacist intervention were T2DM (61.2%), hypertension (11.5%), dyslipidemia (5.3%), and vitamin D deficiency (3.3%). The most common MTPs identified by the pharmacists included an indication for additional medication therapy (29.7%), medication nonadherence (26.8%), sub-therapeutic medication dosage (20.5%), and presence of an adverse drug reaction (10.0%). Common associated PCPIs included counseling the patient or caregiver (35.4%), initiating therapy (30.1%), increasing dosage (20.6%), and discontinuing therapy (7.2%).

(Table 3) provides a summary of the number of patients for whom diabetic medication therapy required adjustment and the types of interventions made.

Table 3. Patients Requiring Adjustment of Diabetes Medication Therapy

(Table 4) highlights the number of patients for whom other interventions were recommended by the pharmacy team with relative frequency.

Table 4. Patients Requiring Other Interventions

Patient Satisfaction with Care

Eleven patients finished the study and successfully completed both pre-study and post-study provider satisfaction surveys. Two of the 11 patients (18.2%) reported an excellent experience with the PCP whereas 10 of the 11 patients (90.9%) reported an excellent experience with the pharmacist (P = 0.001). A comprehensive listing of all evaluated satisfaction criteria is listed in (Table 5).

Table 5. Patient Experience with Medical Provider Survey

Discussion

It is estimated that approximately 30.3 million people of all ages have diabetes in the United States, with higher prevalence noted among Asians, non-Hispanic blacks, and Hispanics as compared to their non-Hispanic white counterparts. Prevalence of diabetes also varies significantly by education level, a marker for socioeconomic status, with higher rates among those of lower academic achievement. Uncontrolled diabetes is associated with a number of significant complications, including cardiovascular disease, extremity amputation, and diabetic ketoacidosis. As such, diabetes is a leading cause of death in the United States, ranking seventh in 2015, and results in significant healthcare costs. Total healthcare costs associated with diabetes in 2012 was 245 billion US dollars, and after adjusting for age group and sex, average medical expenditures among patients with diabetes were about 2.3 times higher than expenditures for those without diabetes [5].

Given the significant morbidity and mortality associated with diabetes, it is of paramount importance to explore novel methods of delivering healthcare services to this vulnerable population. One of the challenges facing the nation’s ability to adequately manage diabetes is the projected physician shortage over the next decade. It is estimated that due to population growth, an increase in the number of aging Americans, and retirement of practicing physicians, the United States could experience a shortage of up to 120,000 physicians by the year 2030. More specifically, the United States may experience a deficit of between 14,800 and 49,800 primary care physicians by the same year [6].

When considering this looming barrier, it is necessary to reflect upon how other members of the healthcare team may complement the physician and increase access to care. One of the methods of addressing this need is via disease state management by clinical pharmacists. This study adds to the growing body of literature to support the role of pharmacist-provided disease state management in the primary care setting, particularly as it relates to the management and control of T2DM.

Because patients included in the analysis were referred from PCPs, baseline A1C results are representative of level of disease state control when managed by the PCP alone, with the exception of four patients who were newly diagnosed with diabetes at the time of referral. Final study results are representative of the care provided by the clinical pharmacist via direct collaboration with the PCP. Mean reductions in A1C of 2.7% and 3% in the ITT and PP analyses, respectively, show that clinical outcomes can be improved when patients receive additional care from the clinical pharmacist in an FQHC. It is possible that the impact on A1C was even larger than reported as the baseline A1C values for two of the patients was not precise. For these two patients, the POC result was > 15% and the baseline result with labeled as 15% per protocol. By listing this number as 15%, the baseline A1C value is underestimated. Additionally, provider satisfaction survey results show that patients report receiving a high level of care from clinical pharmacists. With the exception of goal-setting and listening carefully to the patient, results of which were high for both PCPs and pharmacists, results for the pharmacists were statistically significantly higher than for the PCP regarding all other criteria. Though the focus of this interim analysis is management of diabetes mellitus, results from the MTM analysis indicate that pharmacists are able to manage a number of interrelated disease states. This demonstrates the comprehensive nature of disease state management.

Limitations

While results are compelling, one key limitation of this study is the lack of an active control arm. Without such, it may be argued that the improvement in outcomes is not solely due to the study intervention. This is particularly true for those patients who were newly diagnosed. While the baseline A1C may serve as a historic control for those patients with preexisting diabetes, this is not the case for patients who are new to treatment. To confirm that the results are truly due to pharmacist-delivered disease state management, a more robust study including an active control arm is advised.

Fortunately, similar studies have been conducted in patients with T2DM in which a control arm was included. For example, Brummel AR, et al [6]. published a study in which patients were invited to receive MTM services from pharmacists across a large healthcare organization. In this evaluation, patients who had diabetes and received MTM services were compared to a random sample of patients with diabetes who did not receive the service but were eligible to do so. While the magnitude of A1C reduction was not as large as what was seen in our study, when considering difference-in-differences, the results remained statistically significant. Results from this study indicate that even when a control arm is included, disease state management impacts clinical outcome measures.

Another limitation of our study was the lack of an economical evaluation. A key challenge when implementing pharmacy within the primary care medical home is the questionable cost-effectiveness of such an intervention. Considering that pharmacists are not recognized as medical providers in all states, reimbursement from third party payers for clinical pharmacy services is limited. However, a number of studies have been conducted that indicate pharmacist-led pharmacotherapy management of T2DM can improve clinical outcomes and result in a reduction in overall costs associated with the disease [7–10].

Finally, the number of patients included in our study was relatively small. As nearly 50% of patients were lost to follow up, this could have self-selected for the more adherent patients to be included in the analysis. Nonetheless, our results indicate statistically significant improvement in A1C. A larger sample size would be advised in order to determine if there is impact on healthcare utilization, which may offer additional insight into potential cost-savings. As this is an interim analysis, it is anticipated that the sample size will continue to grow, allowing for a more robust evaluation regarding cost-effectiveness. In addition, as the study progresses, a comparison arm may be included to control for potential confounders.

Conclusion

In summary, our analyses demonstrate that clinical pharmacists can improve clinical outcomes in patients with diabetes in an FQHC and that patients report high levels of satisfaction in the care that they receive. Therefore, primary care settings should consider methods of incorporating clinical pharmacist services into the healthcare model to increase patient access to MTM.

Acknowledgments

The authors would like to thank the following individuals for providing valuable assistance with data collection:

Dana Chippi, PharmD Candidate, 2019

Jinsu Im, PharmD Candidate, 2019

Young Kim, PharmD Candidate, 2019

Affan Aamir, PharmD Candidate, 2021

Nandini Patel, PharmD Candidate, 2020

Author Disclosure Statement

Drs. McCarthy and Bateman are employees of the Ernest Mario School of Pharmacy at Rutgers, the State University of New Jersey and practice clinical pharmacy at HJAHC. Drs. McCarthy and Bateman disclose no conflicts of interest in the research, authorship, and/or publication of this article.

References

1. Medication Therapy Management (MTM) (2018) American Pharmacists Association Foundation website. https://www.aphafoundation.org/medication-therapy-management. Accessed September 30, 2018.
2. American Pharmacists Association; National Association of Chain Drug Stores Foundation (2008) Medication therapy management in pharmacy practice: core elements of an MTM service model (version 2.0). J Am Pharm Assoc (2003) 48: 341–353.
3. Kucukarslan SN, Hagan AM, Shimp LA, Gaither CA, Lewis NJ (2011) Integrating medication therapy management in the primary care medical home: A review of randomized controlled trials. Am J Health Syst Pharm 68: 335–345.
4. McCarthy C, Gunkel K, Willard S, Findley PA, Wagner M (2014) Impact of neuropsychiatric pharmacists on an urban interdisciplinary primary care collaborative practice [abstract]. Journal of Pharmacy Practice 27: 318.
5. Centers for Disease Control and Prevention (2017) National Diabetes Statistics Report, 2017. Atlanta, GA: Centers for Disease Control and Prevention, U.S. Dept of Health and Human Services; 2017.
6. Association of American Medical Colleges (2018) New Research Shows Increasing Physician Shortages in Both Primary and Specialty Care. Published April 11, 2018. Accessed September 30, 2018.
7. Monte SV, Slazak EM, Albanese NP, Adelman M, Rao G, et al. (2009) Clinical and economic impact of a diabetes clinical pharmacy service program in a university and primary care-based collaboration model. J Am Pharm Assoc (2003) 49: 200–208.
8. Fera T, Bluml BM, Ellis WM (2009) Diabetes Ten City Challenge: final economic and clinical results. J Am Pharm Assoc (2003) 49: 383–391. [crossref]
9. Brummel A, Lustig A, Westrich K, Evans MA, Plank GS, et al. (2014) Best practices: improving patient outcomes and costs in an ACO through comprehensive medication therapy management. J Manag Care Spec Pharm. 20: 1152–1158.
10. Jameson JP, Baty PJ (2010) Pharmacist collaborative management of poorly controlled diabetes mellitus: a randomized controlled trial. Am J Manag Care 16: 250–255.