Type 2 diabetes is responsible for a staggering morbidity and mortality burden. As of 2015, 9.4% percent of the United States population has diabetes; 95% of these individuals have type 2 diabetes [1]. Type 2 diabetes is associated with microvascular complications such as retinopathy, neuropathy and nephropathy, and with higher risk of macrovascular complications, including coronary artery disease, peripheral arterial disease, and stroke. Currently, type 2 diabetes is the seventh-leading cause of death in the United States [1].

Furthermore, profound racial and ethnic disparities exist in type 2 diabetes morbidity and mortality in the United States, particularly for African-Americans. Currently, 12.7% of African-Americans have type 2 diabetes [1]. African-Americans are less likely to have controlled HbA1c than non-Hispanic whites [2], are also more likely to develop retinopathy and nephropathy [3], and more likely to be hospitalized with diabetes-related complications [4]. African-Americans with type 2 diabetes also report higher levels of diabetes-related distress than non-Hispanic whites [5]. Ultimately, African-Americans have the highest diabetes-related mortality rates of any racial or ethnic group in the United States [3].

These Type 2 diabetes disparities result from a complex mix of factors. Low birth-weight and maternal-fetal stress are more common in African-American children and increase the risk of developing type 2 diabetes [6]. Higher type 2 diabetes prevalence and poorer HbA1c control may result from ethnic differences in obesity rates, body fat distribution, and glucose metabolism [6]. Cultural food practices and customs may also pose a challenge to diabetes management, such as consumption of breaded and fried meats and simple carbohydrates [7]. Additionally, African-Americans are disproportionately affected by socioeconomic factors such as poverty, poorer quality housing, lack of neighborhood spaces for physical activity, and limited access to healthy food [8]. Health care access barriers and lower quality of care also contribute to poorer diabetes outcomes in African-Americans, as can patient-provider racial discordance, perceived racial bias in medical encounters, and resulting patient mistrust in healthcare providers and systems [9].

Self-management of type 2 diabetes requires regular blood glucose monitoring, management of diet, physical activity, medications, and ongoing medical care. A key goal of diabetes self-management is the control of hemoglobin A1c (HbA1c), which is a measure of average blood glucose over several months. Poorly-controlled HbA1c is associated with microvascular and macrovascular complications [1]. The demands of managing this complex illness also affect many dimensions of quality of life (QOL), which encompasses physical, emotional and social well-being. Individuals with diabetes report lower QOL than individuals without chronic illnesses [10]. Contributors to lower QOL include diabetes-related distress; in the recent Diabetes Attitudes, Wishes and Needs second (DAWN2) study, 44.6% of those with type 2 diabetes reported distress regarding hypoglycemic events, physical health, emotional well-being, and financial strain [11]. In turn, lower QOL affects the ability to manage HbA1c and other diabetes care activities [12].

Recognizing the many challenges of managing type 2 diabetes, the American Diabetes Association (ADA) recommends that all individuals receive diabetes self-management education (DSME) at the time of a type 2 diabetes diagnosis, as well as ongoing self-management support as needed [13]. The goal of DSME is to increase an individual’s self-efficacy to manage diet, physical activity, glucose monitoring, stress management, and other necessary skills and behaviors for successful diabetes outcomes [13]. Meta-analyses have established the impact of DSME on glycemic control and QOL. In a 2002 meta-analysis, DSME participants demonstrated reductions of 0.76% in hemoglobin A1c (HbA1c) at immediate follow-up, with reductions in HbA1c attenuating to 0.24% at follow-up points 4 or more months post-intervention. The authors found three interventions measuring QOL, two of which showed QOL improvements in DSME participants; they did not combine these studies in a meta-analysis [14]. A more recent meta-analysis of group DSME programs showed HbA1c declines of 0.44% six months post-intervention, and 0.46% at 12 months. Three studies were eligible for a QOL meta-analysis; QOL changes were not significant, but the authors stipulated that the heterogeneity of the included studies was high [15]. However, neither meta-analysis examined outcomes by racial/ethnic group.

Increasingly, attention has been paid to the differential impact of DSME in racial and ethnic minority groups—including African-Americans--and development of DSME that is culturally-adapted for the language, beliefs, values, and customs of particular groups. In their DSME position statement the (ADA), the American Association of Diabetes Educators (AADE) and Academy of Nutrition and Dietetics call for DSME that addresses a patient’s “cultural needs,” [13] and the AADE lists provision of “culturally competent supportive care across the lifespan” as a competency for diabetes educators [16]. Nam et al.’s 2012 meta-analysis of 12 culturally-tailored DSME interventions—four of which targeted African-Americans—showed an effect size of − 0.29 on HbA1c [17], indicating a small effect. A 2014 meta-analysis of the impact of DSME on HbA1c in ethnic minorities found an overall 0.31% HbA1c reduction in the 39 included studies; 33% of these studies included African-Americans [18]. However, these meta-analyses did not explore HbA1c results for African-Americans separately, nor did they examine QOL as an outcome.

Despite the higher type 2 diabetes morbidity and mortality burden in African-Americans, no systematic reviews or meta-analyses have specifically analyzed the impact of DSME on two critical measures--HbA1c and QOL-in this population. Further, none have examined whether certain DSME characteristics, such as number of contact hours or culturally-adapted interventions, might result in better outcomes for African-Americans. The purpose of this systematic review and meta-analysis is to examine the impact of DSME in African-American adults with type 2 diabetes mellitus on HbA1c and QOL. Subgroup analyses also examined the impact of several DSME characteristics, including cultural adaptations, on HbA1c.

The systematic review and meta-analysis study protocol was developed prospectively and reported using Preferred Reporting for Systematic Review and Meta Analyses (PRISMA) guidelines [19]. The systematic review and meta-analysis procedures used were developed in consultation with the Cochrane Handbook for Systematic Reviews of Interventions [20]. The protocol was registered at the international prospective register of systematic reviews (PROSPERO) (ID: CRD42017057282) [21].

Fig. 1 shows the PRISMA diagram for the study selection process. A total of 352 citations were retrieved from OVID MEDLINE, Ovid Eric, PsycINFO, Scopus, CINAHL EBSCO, Cochrane Central Register of Controlled Trials, grey literature, and hand searches. After removing duplicates, 279 abstracts remained. After abstract review, 44 articles were selected for full-text review. Ultimately, 14 of those 44 articles were eligible for inclusion in the systematic review; all were from the peer-reviewed literature [29–42].

DSME interventions were heterogeneous in terms of setting, structure, content, contact hours, and provider type. Settings included primary care offices, hospitals, community health centers, diabetes education centers, churches, and patient homes. Seven provided individual DSME, [32, 33, 35, 36, 38, 40, 42] four provided group DSME, [29, 30, 37, 41] and three utilized both individual and group sessions [31, 34, 39]. The number of contact hours varied from 4 to 27; in two studies the contact hours were not specified [36, 39]. In half of the interventions, DSME was delivered by one type of health professional, most commonly a diabetes educator or nurse educator; other studies utilized combinations of diabetes educators, nurse case managers, registered dieticians, pharmacists, peer educators, and community health workers. Attrition was addressed in all but one study [41]; the mean attrition rate was 22.1%.

Twelve of the fourteen interventions were described by their authors as culturally-adapted for African-Americans [29–33, 35–41]. Seven of these authors provided further descriptions of their cultural adaptations, which consisted primarily of incorporation of African-American dietary preferences in nutrition education and/or use of race-concordant diabetes educators, peer educators, or community health workers.

All studies included in the systematic review measured change in HbA1c % as an outcome; 10 studies compared changes in HbA1c for intervention participants versus usual care [29, 31–39]. Of these 10 studies, five reported HbA1c changes favoring the intervention group [32–35, 39]. Eight studies reported pre and post HbA1c means and standard deviations for both the intervention and control groups and were therefore eligible for the HbA1c meta-analysis [29–32, 34, 36, 38, 39].

Eight studies included in the systematic review measured QOL as an outcome, with several using multiple QOL assessment tools. QOL measures included Mental Well Being and Social Well Being, [36] Psychological Distress Scale, [37] Healthy Days Measure Scale, [37] the 12-Item Short-Form Survey, (SF-12) [34] the Problem Areas in Diabetes Survey (PADS), [41, 42] the Diabetes Care Profile (DCP), [30] the Diabetes Empowerment Scale Short-Form (DESSF), [30] Diabetes Attitude Scale, [30] Diabetes Symptom Distress Scale, [40] and Quality of Life in Diabetes [40]. Five of these studies compared changes in QOL for intervention participants compared to usual care [30, 33, 39, 40, 42]. Four reported statistically significant improvements in intervention participants’ QOL compared to usual care, including improved physical, [33] mental, [33, 39, 40] and social well-being [40] and improved psychosocial adjustment [42]. Only two studies included pre- and post-intervention QOL means and standard deviations, and these studies used different QOL tools. Use of the standardized mean difference for comparing different patient-reported outcomes such as QOL score in a meta-analysis is cautioned against because the responsiveness of different QOL instruments to change may vary dramatically [43]. Therefore, QOL results were ultimately not combined in a meta-analysis.

For most types of bias assessed, the risk of bias was low for the majority of studies included in the systematic review. Risk of selection bias due to random sequence generation was high for two studies and unclear for three; bias sue to allocation concealment was unclear for 10 studies. One study had unclear performance bias, all studies had low detection bias, one had unclear reporting bias, and none had other biases detected. The complete risk of bias ratings for included articles may be found in Additional file 2.

Given that the majority of included studies were randomized-controlled trials, the overall evidence was initially assessed as high-quality per GRADE criteria. One point was deducted from the evidence quality for heterogeneity of the study findings. As noted earlier, due to the small number of studies publication bias could not be assessed. Ultimately, the quality of the evidence was graded as moderate.

This meta-analysis found no significant impact of DSME on HbA1c in African-American DSME participants. This finding contrasts with prior DSME meta-analyses that have found HbA1c reductions ranging from 0.44–0.76% in the general population [14, 15] to 0.31% in DSME targeted at ethnic minorities [18]. The subgroup analysis of < 10 versus ≥10 contact hours also contrasts with a prior meta-analysis [14]; although Ricci-Cabello et al. found that effects did not vary by contact hours or intervention intensity [18]. The similarity of HbA1c outcomes for individual versus group DSME is consistent with prior meta-analyses [14, 18]. Likewise, the variation in DSME settings, delivery methods, intensity and contact hours is similar to the findings of other DSME meta-analyses [14, 18]. The high heterogeneity of HbA1c changes (I² = 92%) may be a result of the substantial variations in these intervention characteristics.

The smaller number of DSME interventions measuring QOL relative to HbA1c is consistent with prior DSME meta-analyses, which have found a greater focus on surrogate outcomes such as HbA1c rather than patient-reported outcomes such as QOL [14]. It is promising that four of the five studies measuring QOL found statistically significant improvements in participants’ QOL versus controls; however, no studies explained if the statistically significant differences in QOL scores translated into clinically meaningful QOL improvements for patients.

The variety of QOL scales used likely reflects that QOL is a complex construct without a universal definition; however, this variety and the small number of DSME studies measuring QOL hampers the ability to compare findings across studies. In order to better understand the impact of DSME on QOL, more DSME studies should include QOL measures, which would allow for eventual pooling of studies using the same/similar QOL measures in meta-analyses. Future DSME research could also examine with relationship of QOL and potential moderators such as self-efficacy and social support [44].

Notably, in a subgroup analysis culturally-adapted DSME interventions did not yield better HbA1c results than non-culturally-adapted DSME. Prior DSME meta-analyses have not compared culturally to non-culturally tailored DSME. In Nam et al.’s meta-analysis of culturally-tailored DSME in ethnic minorities, the authors noted that more research was needed to determine the most effective culturally-tailored elements for various racial and ethnic groups [17]. Similarly, for a number of the studies in our meta-analysis it was difficult to ascertain the types of cultural adaptations made, and it was unclear whether certain features—such as use of race-concordant educators or recipe modifications—had a greater impact than others. More detailed guidelines are needed for the development and evaluation of culturally-adapted DSME in specific populations. Future research should also more rigorously assess approaches to cultural adaptations of DSME for African-Americans and the relative effectiveness of various culturally-tailored approaches. Furthermore, the included studies in this meta-analysis did not explicitly address social and systems-level contributors to diabetes disparities in African-Americans, such as socioeconomic status, racial discrimination or mistrust in the medical system. When developing DSME in the United States and globally, educators should be sensitive to the experiences of marginalized groups and how these experiences can impact diabetes self-management [45].

Significant disparities remain in type 2 diabetes prevalence and outcomes among African-Americans, and DSME is recommended as part of standard type 2 diabetes care. Our study adds to the body of knowledge of the impact of DSME in African-Americans by showing a non-significant impact on HbA1c in African-American participants. The high levels of heterogeneity in the HbA1c findings--as evidenced by wide CIs and I² values—demonstrate a need for more rigorously-designed DSME trials for African-Americans and further research to understand what DSME intervention characteristics, if any, consistently contribute to improved HbA1c in this population. Finally, the smaller number of interventions measuring QOL indicates the need for greater prioritization of QOL and other patient-important outcomes in future DSME research among African-Americans.