Quality improvement strategies at primary care level to reduce inequalities in diabetes care: an equity-oriented systematic review

Quality improvement strategies at primary care level to reduce inequalities in diabetes care: an... Background: There is evidence that disparities exist in diabetes prevalence, access to diabetes care, diabetes- related complications, and the quality of diabetes care. A wide range of interventions has been implemented and evaluated to improve diabetes care. We aimed to review trials of quality improvement (QI) interventions aimed to reduce health inequities among people with diabetes in primary care and to explore the extent to which experimental studies addressed and reported equity issues. Methods: Pubmed, EMBASE, CINAHL, and the Cochrane Library were searched to identify randomized controlled studies published between January 2005 and May 2016. We adopted the PROGRESS Plus framework, as a tool to explore differential effects of QI interventions across sociodemographic and economic factors. Results: From 1903 references fifty-eight randomized trials met the inclusion criteria (with 17.786 participants), mostly carried out in USA. The methodological quality was good for all studies. Almost all studies reported the age, gender/sex and race distribution of study participants. The majority of trials additionally used at least one further PROGRESS-Plus factor at baseline, with education being the most commonly used, followed by income (55%). Large variation was observed between these studies for type of interventions, target populations, and outcomes evaluated. Few studies examined differential intervention effects by PROGRESS-plus factors. Existing evidence suggests that some QI intervention delivered in primary care can improve diabetes-related health outcomes in social disadvantaged population subgroups such as ethnic minorities. However, we found very few studies comparing health outcomes between population subgroups and reporting differential effect estimates of QI interventions. Conclusions: This review provides evidence that QI interventions for people with diabetes is feasible to implement and highly acceptable. However, more research is needed to understand their effective components as well as the adoption of an equity-oriented approach in conducting primary studies. Moreover, a wider variety of socio-economic characteristics such as social capital, place of residence, occupation, education, and religion should be addressed. Keywords: Type 2 diabetes, Quality improvement strategies, Equity, Systematic review * Correspondence: s.vecchi@deplazio.it Department of Epidemiology, Lazio Region- ASL Rome1, Rome, Italy Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 2 of 18 Background Methods Diabetes is a complex, chronic disease recognized as an For the purpose of the review, a “socially disadvantaged important cause of premature death and disability [1] group” is defined by differences that place the group at dis- and disproportionately affects socially and economically tinct levels in a social hierarchy. To explicitly consider disadvantaged populations [2–4]. According the Na- health equity and to capture characteristics possibly indicat- tional Institute for Health and Care Excellence guide- ing disadvantaged status, we adopted the PROGRESS-Plus lines [5], patients with type 2 diabetes should receive a framework recommended by the Campbell and Cochrane clear gamut of care to be provided by primary care pro- Equity Methods Group and the Cochrane Public Health viders. Annual routine monitoring of health indicators Group to identify studies with a focus on reducing health such as urinary albumin, BMI, cholesterol, blood creatin- inequalities [11]. PROGRESS-Plus stands for place of resi- ine, HbA1c and BP measured, eyes and feet examined dence, race/ethnicity/culture/language, occupation, gender/ and a smoking review, forms a major part of patient dia- sex, religion, socioeconomic status and social capital. This betes care. In addition patients should expect to receive systematic review was conducted in accordance with an evidenced-based education and access to specialist PRISMA-E 2012 (Preferred Reporting Items for Systematic healthcare professionals including ophthalmologists, po- Reviews and Meta-Analyses, Equity 2012 Extension), a vali- diatrists and dieticians. dated tool to improve both the reporting and conducting of Quality of care among diabetic patient can be influenced equity focused systematic reviews, were upheld in this re- by a range of factors that has been already described. Pre- view [12]. vious systematic reviews showed that low individual socio-economic status and residential area deprivation are Data sources and searches often associated with both worse process indicators and We searched all relevant biomedical databases such as worse intermediate outcomes among patients with type 2 Pubmed, EMBASE, CINAHL, and the Cochrane Li- diabetes [6]. These differences are present even in coun- brary for relevant published RCTs and cluster-RCTs tries with a significant level of economic development that published in English. We limited the search from 1 have a universal health care system. Moreover, disparities January 2005 to 31 May 2016. A combination of MeSH in diabetes care exist among racial or ethnic minority terms and keywords were chosen to reflect selection cri- groups, independent of economic status [7]. teria tailored to each database. Details of the full search To improve diabetes care, it might be important to focus strategy for PubMed are included in supplemental mater- on quality management (QM), especially because the com- ial (Additional file 1). In addition, we scanned the refer- plexity of healthcare system and patients complexities has ence lists of relevant reviews to track relevant RCTs. dramatically increased. QM comprises procedures to moni- tor, assess, and enhance the quality of care. In the last years many countries have developed quality improvement inter- Study selection ventions (QI) to improve both patient outcomes and the Two authors (NT, AMB) independently screened all title quality of diabetes care [8, 9]. A meta-analysis of studies in- and abstracts of all studies obtained from electronic vestigating QI strategies [10] found that interventions target- searches. For studies meeting the inclusion criteria, we ing the entire system of disease management (team changes, retrieved full texts and the same authors independently case management, promotion of self-management) along evaluated them for inclusion. Any disagreements were with patient-mediated QI activities were important compo- resolved through consensus or in discussion with the ex- nents of strategies to improve diabetes care. However, the tended authorial group. studies included in this review were targeted to the general We used the “population, intervention, comparison, population, irrespective of socio-demographic characteristics outcome, setting” (PICOS) logic to guide the systematic or socio-economic status. review (Additional file 2). We included randomized con- Acknowledging the existence of such disparities, our trolled trials (RCTs) and cluster-randomized trials, aims are to: a) describe the extent to which effects on so- evaluating all QI interventions designed to improve cial inequalities are considered in randomized controlled health outcomes in social disadvantaged people with trials (RCTs) evaluating the effects of QI interventions to type 2 diabetes and designed to reduce inequalities in improve quality of diabetes care and b) synthesize evi- diabetes care. We considered studies that reported quan- dence on the effectiveness of QI strategies to reduce titative estimates of total effect of treatment and differ- health inequities in diabetes care in the primary care set- ential effects for the PROGRESS-Plus factors. ting. We conducted an equity-oriented systematic review We used the Agency for Healthcare Research and including RCTs only, using an international taxonomy of Quality [13] taxonomy to identify QI strategies QI interventions, and assessing the quality of included (Additional file 3). QI strategies can be delivered to studies with a methodological rating tool. specific levels of influence: Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 3 of 18 Patient level (e.g. patient education, patient and subjective outcome measures. We defined clinical reminders, or promotion of self-management); and laboratory measures, process indicators, diabetes Health care provider level (e.g. electronic medical complications, hospital admissions, emergency admis- record reminders, audit & feedback, cultural sions and all-cause mortality as objective outcome mea- competency training); sures. We defined measures of self-management/ Health care system level (e.g. change in the health adherence to recommendations as subjective outcome system structure or delivery, adjusting roles of care measures. With respect to missing data, we judged indi- team members, nurse care management model). vidual trials at high risk of bias if data from more than 10% of participants were not available. We used the Data extraction and quality assessment quality criteria for descriptive purposes only to highlight Two authors independently extracted data (NT, SV), and differences between studies. We used RevMan 2014 soft- disagreements were resolved by discussion. Data from ware [15] to generate figures related to risk of bias. multiple publications of the same study was considered as a single study. A data extraction form was designed to Data synthesis document the following study details: trials characteris- We synthesized findings from the included studies by tics; participants (total number at baseline, age range, intervention level (patients, health care provider, and gender, clinical features); type of intervention and com- health care system). The wide variety of interventions parator; clinical and no clinical outcomes; timing; risk of (in terms of mode of delivery, frequency and duration of bias; study results. For continuous outcomes, we ex- follow up assessment) and population groups considered tracted the mean change from baseline (with the stand- in the included studies did not allow for a meaningful ard deviation) and the mean difference, if available, with meta-analysis to be conducted. We summarized results the corresponding 95% confidence interval (CIs). Relative using narrative methods. We described in more detail risk (RR), and absolute risk differences, with the corre- studies reporting differences in QI interventions effects sponding 95% CI, was extracted for binary primary out- across subgroups. comes. If studies reported data for more than one time point, we extracted data for the longest-term outcomes. Results Baseline population characteristics relevant for ad- The search strategy generated 1903 citations after remov- dressing potential issues in health equity were extracted ing duplicates. Upon reviewing titles and abstracts, we re- using the PROGRESS-Plus framework. We extracted trieved full text articles for 247 studies that were screened data on outcome assessed, according to whether by two authors independently (NT, AMB). We excluded PROGRESS-Plus factors were considered as control vari- 189 trials. Most common reasons for exclusion were not ables (e.g., by adjusting in regression analyses) and the addressing a socially disadvantaged group, an evaluation methods utilized to investigate differential effects (strati- of primary prevention intervention, and being conducted fied analysis or modification/interaction analysis). We in a setting other than primary care. Fifty-eight RCTs met also extracted details on the duration of intervention, eligibility criteria. PRISMA Flow Diagram Fig. 1 shows the duration of follow up, health professional group in- details of study selection process. volved, details of the strategy being implemented (i.e. modality, delivery format). Overview of the included studies Two authors independently assessed risk of bias of in- A substantial synthesis of the characteristics of all 58 cluded studies using the Cochrane ‘Risk of bias’ tool for studies included in this review is reported in Table 1. RCTs [14]. We considering the following domains: se- Overall the majority of studies (n = 54) used a parallel quence generation, allocation concealment, blinding of RCT design while four trials were cluster RCTs [16–19]. participants and personnel, blinding of outcome assess- Follow-up periods varied in duration from less than 1 ment, incomplete data, selective reporting, and other month to 5 years, with the majority lasting 6 to biases. For each domain, risk of bias was classified as 12 months. Most of trials were conducted in the USA “high,”“low,” or “unclear”. Since we included (n = 47); the remaining studies were carried out in cluster-randomized controlled trials, additional items Canada [20], Asia [21], the United Kingdom [16], New were considered: (1) recruitment bias: did recruitment of Zealand [22], Australia [19], Trinidad and American diabetes patients take place before or after Samoa [18, 23]. randomization of the clusters?, (2) did the intervention Almost all studies reported the age, gender/sex and and control group differ in baseline characteristics?, (3) race distribution of study participants. The majority of did any of the clusters drop out during follow-up, (4) studies additionally used at least one further was clustering accounted for in the statistical analyses? PROGRESS-Plus factor for the description of partici- We investigated detection bias separately for objective pants’ baseline characteristics. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 4 of 18 Records identified through database Additional records identified through searching other sources (n =2931) (n = 4 ) Records after duplicates removed (n =1903) Records excluded on title Records screened and abstracts (n = 1903) (n = 1656) Full-text articles assessed for Full-text articles excluded eligibility (n = 189) (n = 247) -167 not addressing social disadvantaged people -1 no outcome defined in the inclusion criteria Studies included in -21 no RCT qualitative synthesis (n =58 ) Studies evaluating differential intervention effects by PROGRESS-Plus factors (n =7 ) Fig. 1 PRISMA 2009 Flow Diagram. Study selection process Among these, education was the most commonly re- delayed intervention or no intervention. Health profes- ported factor (n = 45), followed by income (n = 32). sionals who participated in studies included physicians, Twenty-six studies considered at least one PROGRESS-Plus specialist nurses, social workers, dietitians, diabetes educa- factor as control variable when measuring intervention ef- tors, community health workers, general practitioners, fects (e.g., by adjusting in multivariate analyses). Again, age practice nurses and home care nurses. (n = 23) and gender/sex (n = 20) were the factors most com- The majority of trials (96%) provided data on change monly controlled for, followed by education (n = 9). Seven in HbA1c. Thirty-seven trials (63%) reported BMI out- (12%) trials used at least one PROGRESS-Plus factors for come; blood pressure and cholesterol data in 38 and 30 examining differential intervention effects, and gender, age, trials, respectively. Process measures including diabetic race and education were those most often considered. foot exam, dilated eye exam and attendance at office ap- Detailed descriptions of the QI interventions were not pointments were seldom reported. always clearly provided in the trials. In order of frequency, For secondary outcomes, data were available for were twenty-nine studies (50%) focused on interventions patient-reported measures including diet and physical activ- delivered at the patient level [17, 20, 21, 24–27, 29, 32–39, ity (n = 28) using a considerable variety of instruments. 41, 42, 55, 61–70], and twenty-six at the health care Medication adherence and home glucose monitoring were organization level (45%) [16, 18, 19, 22, 23, 28, 30, 31, 40, measured less consistently (in 17 and 15 studies, respectively) 45–54, 56–60, 72, 73]. The remaining three studies (5%) as were diabetes complications and hospital admissions. [43, 44, 71] described interventions at the provider level. A detailed description of trials characteristics and In the majority of studies comparators were “usual” or intervention components by intervention level is pre- “standard” care (69%), five studies reported waiting list, sented in Additional file 4. Eligibility Screening Identification Included Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 5 of 18 Table 1 Synthesis of the characteristics of the included studies by level of intervention and PROGRESS factors Level of intervention Patient level Provider level Health care systems level Total QI strategies Total of studies 29 3 26 58 N % N% N % N% Sample characteristics Age 55.13 - 55.37 - 53.82 - 55.06 - Sex, female (%) 64.05 58.69 57.84 60.20 Baseline HgA1c (%; mmol/mol) 8.88; 74 7.0–11.8; 53–105 9.53; 81 8.1–12.05; 31–109 8.51; 70 7.6–10.5; 60–91 8.88; 74 7.0–12.05; 53–109 Progress factors reported at baseline Place of residence 29 50 3 5.2 26 44.8 58 - Race/ethnicity 26 49.1 3 5.7 24 45.3 53 - Occupation 12 54.5 – - 10 45.5 22 - Gender/sex 24 46.2 3 5.8 25 48.1 52 - Religion – - – - – - – - Education 26 57.7 1 2.2 18 40.1 45 - Socioeconomic status (SES) – - – - – - – - Income 20 62.5 – - 12 37.5 32 - Social capital 10 62.5 - - 6 37.5 16 - Age 28 50.0 3 5.4 25 44.7 56 - Disability – - – - – - – - Sexual orientation – - – - – - – - Study characteristics Year of publication 2005–2010 11 19 2 3.5 11 19 24 41.4 2011–2016 18 31 1 1.7 15 25.9 34 58.6 Study location North America 25 86.2 3 100 22 85 47 UK 1 3.4 – - 1 3.8 2 3.4 Australia –– – - 2 - 2 3.4 Asia 3 10.4 – - 1 7.7 4 6.9 Duration of study (months) 10 3–26 4.5 0.25–36 12 6–60 8.9 0.25–60 Average sample size (range) 190 (56–526) 1573 (182–4138) 290 (65–1665) 684 (50–4138) Risk of bias in included studies studies blinded providers [20, 21, 25–30]. For studies A summary of ‘risk of bias’ for each study and compara- reporting objective outcomes with standardized collec- tive data across the studies is reported in Figs. 2 and 3 . tion methods (e.g. automated blood test), we assigned a All studies were described as individual RCT (n = 54) or low risk of detection bias (79%), as knowledge of treat- cluster-RCTs (n = 4). None of the randomized studies ment assignment was considered unlikely to affect the had uniformly low risk of bias. The allocation sequence outcome. Twenty-eight studies reporting subjective out- was adequately reported in 48% of the studies (28/58), comes, those that used self-reported measures (i.e. ques- with random number tables or a computer-generated tionnaire on dietary habits or physical activities) were at randomized list as the most commonly used methods. high risk of bias due to the lack of blinding of outcome One study was categorized as high risk due to the use of assessment (24 studies). In the remaining 30 studies, in- a gender-based randomization procedure [24]. Most dependent research personnel who were not involved in RCTs (40/58) did not describe or described in sufficient the intervention performed outcome assessments, which detail the allocation concealment to allow a judgment we evaluated as low risk of detection bias. and were evaluated to be at unclear risk of bias. Thirty studies were at low risk of incomplete outcome In the majority of the trials, all participants were aware data due to a low attrition rate (< 10%) or an of the treatment they were receiving, and only eight intention-to-treat (ITT) analysis for primary outcomes. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 6 of 18 Fig. 2 Risk of bias graph Thirteen studies were at high risk of bias because a high with problem-solving training sessions [29], was effective proportion of participants were lost to follow-up or were in improving glycemic control (MD = − 0.72, 95% CI − missing outcome measurements. Selective reporting bias 1.42 to − 0.01, p = 0.02, n = 56). was difficult to detect in most studies because published Two studies (n = 265) showed an improvements in gly- protocols were often unavailable. Most trials reported all cemic control as measured by HbA1c (8.2% ± 0.4 vs 8.6% outcomes. One study [30] collected a large quantity of ±0.3, p = 0.004 and 7.6 ± 1.8 vs 8.2 ± 2.5; p = 0.006, respect- baseline data but did not adequately describe follow-up ively), comparing behavioral education programs via tele- data. One paper [31] did not report some subjective health [33] or using a computerized self-management measures listed in the published protocol. Risk of con- program [26]vs standard care. tamination was high in most of the studies because pa- Berry et al. [17] reported a greater improvement in tients receiving interventions and those receiving usual HbA1c levels in low-income participants receiving ses- care or other interventions were seen within the same sions led by a multidisciplinary team than in the control health center. Among cluster RCTs, three accounted for group (7.6% vs 9.3%; p = 0.001, n = 80). the effects of clustering in their results analysis. One study [21] found that an education program with incentives and self-monitoring devices produced a sig- Study evaluating the effect of QI strategies by nificant reduction in HbA1c (7.29% ±0.58 vs 7.73% intervention level (n = 51) ±0.57; p < 0.05, n = 132). Patient level Philis-Tsimikas et al. [34] did not report difference be- More than half (n = 17) of the studies showed significant tween groups but a significant decrease of HbA1c from effect in at least one of the outcomes considered in this baseline to follow-up (− 1.5%, p < 0.01) was observed in review; most (n = 11) of these interventions include the experimental group. group education sessions or visits and principles of Finally, two trials [35, 36] did not find a significant de- self-management. crease in HbA1c in the study population, but reported a Twenty-seven out of 29 trials reported data on gly- positive association for a subgroup of participants. cemic control measured as HbA1c level. Ten studies re- Brown et al. [35](n = 460) found that for those who ported an improvement in HbA1c levels in the attended ≥50% of the self-management patient education experimental group compared to the control group. sessions, the reduction of HbA1c was − 0.6% for the An education program based on telephone calls [32] “compressed” group and − 1.7% for the “extended” was found to be associated with a decrease in HbA1c group. In Gerber et al. [36](n = 244), the intervention both in the unadjusted (− 0.23 ± 0.11% vs 0.13 ± 0.13%, resulted in significant improvement in HbA1c among p <0.04, n = 526) and adjusted analysis (MD = 0.40, low–health literacy subjects with poor glycemic control. 95% CI 0.10–0.70; p = 0.009). Eighteen trials reported data on change in BMI, three Rosal et al. [27] evaluated a nutritionist or health found a significant improvement in the experimental educator-led self-management education program sup- group. ported by counseling and a self-monitoring device. The Anderson-Loftin et al. [37] reported that the group ex- study showed a difference between groups in HbA1c level posed to the dietary self-management intervention had a at 4 months (MD = − 0.53, 95% CI-0.92 to − 0.14; decrease in BMI while the control group showed an in- p > 0.008, n = 252) but not sustained at 12 months. crease in BMI control group (− 0.81 kg/m2 vs + 0,57 Kg/ An intensive training group intervention addressing m ; p =0.009, n = 97). Tang et al. [38] reported a decrease both diabetes and cardiovascular diseases, combined in BMI in the intervention group receiving behavioral Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 7 of 18 support delivered by a peer leader compared with the con- trol group; the benefit was observed at different follow-up times and maintained at the longest one (15 months) (MD = − 0.8 Kg/m 95CI%-1.6 to − 0.1; p =0.032, n =106). Toobert et al. [39] showed a significant difference in BMI (MD of − 0.40 Kg/m ; p <0.05, n =280) in an underserved and high-risk Latino population treated with a long-term multiple-behavior-change program. Fifteen of the 26 studies examining healthcare inter- ventions in diabetes care considered blood pressure among the outcomes. Two studies showed differences favoring the experimental intervention. In the study con- ducted by Hill-Briggs et al. [29], participants receiving a self-management training adapted for low literacy expe- rienced an individual improvement in DBP and SBP (median reduction = − 7.17 mmHg, n = 8, median reduc- tion of − 14.67 mmHg, n = 9, respectively). Tang et al. [40] also reported a greater reduction in the group that re- ceived a combination of self-management and peer support interventions than the control group, both in SBP (MD = − 10.0 mmHg (95% CI -17.6 to − 2.4, p = 0.01) and DBP (MD = − 8.3 mmHg (95% CI -13.2 to − 3.4, p =0 .001). A significant improvement (p < 0.001) in hypertension in both groups was found by Shahid et al. [24](n = 440) but between-group differences were not reported. Eighteen studies reported data on diet adherence. Seven studies [22, 25, 31, 35, 39, 44, 51] observed be- tween group differences although using different instru- ments and scales. Anderson-Loftin et al. [37] used the Food Habits Questionnaire (FHQ) adapted for southern African Americans to measure dietary pattern. The intervention was a patient education program delivered by nurse case manager with nutrition focus combined with support groups, and weekly telephone follow-up. The authors re- ported a significant improvement in the experimental group with a decrease in high-fat diet while the control group continued previous high-fat dietary behaviors (MD =0.2 points, p = 0.005). One trial [20] used the Summary of Diabetes Self-care Activities Questionnaire (SDCA) to assess the nutrition adherence in Canadian Portuguese-speaking adults. There was an improvement in self-reported nutrition ad- herence at 3 months in favor of the experimental inter- vention (MD = 0.42 ± 0.14, p < 0.05, n = 87). Negarandeh [41] evaluated patient education program based on different format (Pictorial or teach back strategy) compared to usual care. Adherence to dietary pattern was measured through a self-structured nine-item scale. The score improved in all study participants (n =130) in follow up measurements but the improvement was more pro- Fig. 3 Risk of bias summary nounced for the intervention groups than the control group (p < 0.05). The mean difference between groups was − 2.24 (95% CI- 2.67 to-1.81) for the Pictorial format Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 8 of 18 group, and − 2.52 (95% CI:-2.95 to − 2.09) for the Teach strategies including interventions such as individualized back format group. case management activities [23], and culturally tailored A culturally tailored self-management intervention counseling delivered by a CHW [46, 47, 49] and/or adapted for a low income Latino group [27], improved NCM [18, 45], and promotoras [50]. Three RCTs in- the quality of diet as measured by the Alternative cluded additional activities, in particular home visits to Healthy Eating Index. Significant between group differ- support patient’s progress [30, 47, 48]. ences were found at 12 months (MD = 2.83 95% CI 0.58 Seven studies found a significantly greater reduction in to 5.08, p = 0.014, n = 252). HbA1c levels in the experimental group between base- A similar intervention was evaluated by Shahid et al. line and follow up. One study [51] evaluating individual [24] among people residing in rural areas in Pakistan. In culturally tailored care provided by NCM and CHW the intervention group there was a significant increase compared to minimal care, showed a significant decrease in the proportion of participants compliant to the diet in HBA1c levels. The effect was significant only in the plan (17.3% at baseline to 43.6% at follow up, p < 0.01) group of participants receiving a higher number of home while in the control group there was no significant in- visits (− 0.68% vs 0.43%, p = 0.03, n = 522). Another study crease (13.6% at baseline to 15.9% follow up, p = 0.522). conducted with Korean Americans immigrants [52] Weinstein’s trial [42] assessed fruit and vegetable con- found that a culturally tailored program including sumption self-reported daily following brief educational psycho-behavioral education, home glucose monitoring intervention. At 12 weeks, the percentage of participants with tele-transmission, and bilingual nurse telephone who reported ever purchasing from a produce market counseling, was associated with a greater improvement increased significantly in the intervention group (81% vs in HbA1c values (− 1.3% vs − 0.4%; p = 0.01, n = 79). 48%; p = 0.003, n = 79). Moreover, there was an overall A study conducted in a rural setting [53], showed an im- decrease of the percentage of participants reporting diffi- provement in HbA1c levels among patients exposed to culty affording fresh fruits and vegetables (55% vs 74% at diabetes education with interactive online sessions, deliv- baseline, p = 0.008). This decrease was not significantly ered by a multidisciplinary team (0.7 ± 1.3% vs 0.1 ± 1.0%; different between arms. p < 0.03 after adjustment for baseline HbA1c, n =95). Toobert et al. [39] reported the percent of calories A significant decrease of HbA1c was observed follow- from saturated fat measured using a food frequency ing a case management program delivered by a CHW questionnaire following a culturally adapted Mediterra- with the support of a clinical outreach team that in- nean lifestyle intervention. He found an improvement of cluded home visits [19](− 1.0% vs − 0.2%, p = 0.02, n = 0.33 points at the 24-month follow-up. 233). Lujan et al. [54] tested the effectiveness of a multi-component education program led by promotoras Provider level showing a mean change of HbA1c in the intervention Two studies evaluating reminder and reminder+feed- group significantly greater than that of the control group back interventions [43, 44] showed an improvement in at 6 months (p < 0.001, n = 149). glycemic control (HbA1c) compared to the usual care or A multicenter study [55] considered a composite out- no intervention group (0.6% vs 0.2%, p < 0.02, n = 399; come measure based on the achievement of target values MD = − 0.80 p < 0.001, n = 2046, respectively). Both of for HbA1c, SBP, and LDL. Participants assigned to the these interventions utilized computerized systems to intervention arm (health coaching group) showed higher produce physician reminders. One study [43] found an proportions of people reaching all clinical goals (46.4% improvement for LDL cholesterol for all intervention vs 34.3%, p = 0.02, n = 389) compared to usual care. arms, with the greater change observed in the reminders A study evaluated an education program [56] supervised +feedback group (− 18 mg/dl). No studies reported dif- by a nurse specifically trained for case management (DPP ferences between intervention and control arms for Lifestyle Program) where participants in the experimental blood pressure and BMI. group also received an evidence-based medication algo- rithm. The authors observed a significant improvement in Health care system level HbA1c levels in the experimental group compared to the The majority of studies that evaluated interventions tar- control (− 1.87% ± 0.81 vs − 0.54% ± 0.55; p = 0.011). How- geting the health care system (n = 20), showed significant ever, no information on sample size and participant char- effect in at least one of the outcomes considered in this acteristics were reported. review. Significant differences in blood pressure were found As far HbA1c, nine studies reported a significant re- between groups in three studies [16, 22, 45]. A difference duction of HbA1c values [18, 23, 30, 45–50] with a in means of change from baseline in diastolic blood mean difference ranging from − 0.29% to − 0.8%. The pressure significantly favored the intervention in a multi- studies considered a range of health care system-based center study [16] where participants received intensive Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 9 of 18 disease management led by practice nurse supported by Babamoto et al. [58] found that the proportion of patients link workers and a diabetes specialist (adjusted MD = − consuming two or more servings of fruits and vegetables 1.91 mmHg; p < 0.001, n = 1486). In the study of Hotu et daily increased significantly in the CHW and case manage- al. [22], Maori and Pacific patients with diabetes and ment groups but not in the standard provider care group. chronic kidney diseases who received twelve months of Patients’ self-reported intake of fatty foods decreased sig- home visits by a nurse, achieved a significant lower sys- nificantly from 29 to 16% (p < 0.05) in the CHW group but tolic blood pressure compared to usual care group remained unchanged in the other groups. (149 mmHg vs 140 mmHg; p < 0.05, n = 55). In a Cramer et al. [56] used the Dietary Questionnaire to long-term follow-up study [45] (60 months, n = 1665), a measure eating habits and observed a significant im- significant reduction in SBP (MD = − 4.32 mmHg, 95% provement in the experimental group compared with CI -6.72 to − 1.92] and DPB (MD = − 2.63 mmHg, 95% the usual care group (p < 0.001). Lynch et al. [59] also CI -3.74 to − 1.52] was detected among ethnically di- observed a significant increase in the number of days verse, medically underserved patients receiving a following a general and specific diet among participants self-management intervention with the support of home receiving a culturally-oriented self-management program telemedicine and a nurse case manager. (MD = 1.9, 95% CI 0.6 to 3.1; MD = 1.2, 95% CI 0.2 to Of the 14 trials reporting BMI outcome, only one [45] 2.2, respectively, n = 61), measured by the Block Food showed an adjusted MD of 0.40 kg/m (95% CI 0.20 to frequency Questionnaire. 0.60) when enhanced care through a diabetes-specialist Eight trials studied physical activity using different nurse and link worker were compared to usual care. measures, and two reported an effect following the ex- One [56] of the two studies reporting data on weight perimental intervention. One study [59] reported results change from baseline found a significant decrease at the from the CHAMPS (Community Healthy Activities end of the nine-month intervention of − 2.47 kg (±1.87) model for Seniors) physical activity questionnaire modi- in the experimental group and + 0.88 kg (±1.84) in the fied for use among African Americans. At study end- control group (p = 0.01). point there was a statistically significant difference Seventeen trials assessed the impact of QI interven- between groups (MD = 2.517 Kcal/week; p < 0.01). tions on total cholesterol and/or HDL cholesterol, LDL Comparing usual care with two educational programs cholesterol, and triglycerides. In three studies there were provided by a different case manager (CHW or NCM), significant differences in change from baseline between Babamoto et al. [58] found a significant improvement in groups. physical activity with an increase from 28 to 63% (p <0.05) At six months follow-up, Garcia et al. [57] reported in the CHW group, and from 17 to 35% (p < 0.05) in the statistically significant differences between the control standard provider care group, without any change in the and intervention group for total cholesterol (p = 0.003) case management group. and LDL cholesterol (p = 0.014), although not for triglyc- Six studies reported data on diabetes knowledge mea- erides (p = 0.179). sured by validated instruments such as the Diabetes Know- A significant effect on total cholesterol and triglycer- ledge Questionnaire [28, 54, 58], the Spoken Knowledge in ides was found in Kim et al. [52]. The intervention Low Literacy in Diabetes Scale [57], and the Diabetes group showed significantly lower levels of total choles- Knowledge Test [46, 52]. A significant improvement in pa- terol (− 24.7 mg/dl vs 7.2 mg/dl; p = 0.03) and triglycer- tient’sskillswas observedin three studies[46, 54, 58]. ide (− 84.6 mg/dL vs − 4.2 mg/dL; p < 0.05) when In one out of three studies considering emergency and/ compared with the control group. The intervention or hospital admissions [51, 58, 60], there was a reduction group also showed a trend toward a lower HDL, but this in emergency visits from baseline to 24 months among pa- difference was not statistically significant (p = 0.059). tients receiving a culturally tailored care provided by a In Shea et al. [45], the intervention group experienced NCM and a CHW (RR = 0.77, 95% CI, 0.59–1.00) [60]. net improvement in LDL cholesterol level relative to usual One study [45] investigated the effect of telemedicine care; a significant between groups difference was reported compared with usual care on all cause mortality but no at 5 years (MD = − 3.84; 95% CI -7.77 to − 0.08). differences between groups were reported (HR 1.01, 95% Glucose monitoring was considered in four studies CI 0.82, 1.24). [19, 36, 46, 48]. The study conducted by McDermott et al. [19] showed that participants in the control group (waiting-list group) were more likely to self-monitor Studies evaluating differential intervention effects by their glucose level than the experimental group. PROGRESS factors (n =7) Nine trials reported adherence to diet but measures Seven studies conducted sub-analyses to explore a differ- and scores used varied between trials. Three studies ential intervention effects across PROGRESS-Plus factors found a difference between groups. (n = 7) and all were conducted in developed countries. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 10 of 18 They used a parallel study design with a follow up of and ethnic minority participants but heterogeneity and 12–24 months. complexity of interventions made difficult to identify the ef- Table 2 gives the details of studies and results. Females, fective components of these interventions. The evidence on age ≥ 50, African-Americans and those with low education the effect of patient level interventions on improving other showed a better improvement in glycemic control. Patient clinical and laboratory parameters, such as blood pressure, education based on low-fat dietary strategies delivered by cholesterol levels and BMI, as well as self-management be- discussion groups and supported by phone contacts, pro- haviours is scant. Few studies explored the effectiveness of duced a greater decrease in BMI, weight, and dietary behav- other patient level strategies, including incentives and re- iors among women than men [37]. At healthcare minders. The only study included in this review [34]testing organization level, diabetes self-management supported by a rewards-based incentive intervention, showed effective CHW was associated with a greater BMI reduction and an results. increase in exercise frequency among participants aged With regard to interventions at provider level, only ≥50. One study analyzed intervention differential effect by one study reported a significant between groups differ- levels of health literacy [36]. The experimental program ence in HbA1c reduction while no significant impact on aimed to supply information and promote diabetes blood pressure or BMI was observed. self-management skills by computer multi-media including Many of the studies included in this systematic review audio/video sequences. Among low literacy subjects with were designed to evaluate the effectiveness of changing, poor glycemic control, the authors found a greater decrease expanding, or integrating the roles of healthcare profes- in HbA1C in the group exposed to computer multi-media sionals combined with patient education to improve dia- education program than in the control group (− 2.1 vs. betes care and outcomes. QI interventions based on -0.3%, p = 0.036). No significant difference was found multidisciplinary teams including trained nurses or local among high-literacy subjects. Moreover, the multimedia community health workers providing culturally compe- users with low health literacy demonstrated gains in know- tent care, were associated with a significant reduction of ledge, self-efficacy, and perceived susceptibility to complica- HbA1c values. Changes in the role of health care profes- tions compared with those having higher health literacy. sionals have been shown to produce an improvement in glucose control in ethnic minority communities on eth- Discussion nic minority communities showed. Applying an equity-oriented approach, this review iden- As far other primary outcomes considered in this re- tified 58 RCTs (17.786 participants) evaluating QI strat- view, a significant improvement in cholesterol levels was egies to improve the quality of diabetes care in a reported while n differences were found for secondary primary care setting. outcome measures, except for an increase in physical ac- Forty-seven studies were from USA and evaluated in- tivity and diabetes knowledge. terventions specifically designed to reach population Seven studies reported data on the differential effect subgroups mainly defined on the basis of race or ethni- by at least one PROGRESS factor. We did not find evi- city. A narrow subset of these studies (n = 7) considered dence of a differential effect by gender and race of any other dimensions of disadvantage as defined by the intervention on HbA1c levels reduction. One study re- PROGRESS framework, such as socio-economic status ported an improvement in glucose control among a low and place of residence. literacy population subgroup, exposed to a culturally The RCTs included in this systematic review covered a competent education program delivered through wide assortment of QI strategies, varying from multi-media tools. We found some evidence of effective- patient-mediated interventions with sessions of ness of QI interventions in weight loss and BMI among self-management supported by healthcare professionals, females and weight loss among African-Americans. to provider education and other more complex pro- In general, the heterogeneity of baseline HbA1c values grams based on changes in healthcare organization. and mean age of participants can affect intervention out- Twenty-nine studies considered QI interventions con- comes due to the biomedical challenge of lowering ducted at the patient level, three at the provider level, HbA1c from a higher baseline value. Moreover, some and twenty-six at the health care organization level. studies defined a minimum A1C value as inclusion cri- Pooling of results and quantitative synthesis was pre- terion possibly considering patients which may not be cluded by marked heterogeneity (mainly clinical), be- representative of diabetic population receiving care in a cause study population, types of interventions, outcome real world clinical setting. Rather than implementing measures, outcome assessment tools, duration of minimum A1C values for participant inclusion, as many follow-up and risk of bias varied widely between studies. of the studies reviewed incorporated, it is important (it QI strategies based on patient education and self-man- may be worthwhile) to maintain the integrity of studying agement strategies improved HbA1c levels among racial quality improvement interventions in real-life clinical Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 11 of 18 � � � � � � � � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Anderson 2010 Spanish speaking only, Patient level A1c, DBP,SBP, BMI, LDL, Subroups analysis No significant differences A1C [61] education level Number of experimental conditions: diet behavior (BDA); and interaction between groups for any Spanish speakers (yes vs no) USA 2 (1 intervention, 1 control) physical activity (RAPA); analysis outcomes MD = − 0.10(− 0.53, 0.33) vs 0.35 Intervention: depression measured Retention rate (− 0.17, 0.88) telephonic disease management Patient Health 79% vs 64% Educational level: (weekly, bi-weekly, or monthly) questionnaire (PHQ-9) (high level vs low level) based on: MD = 0.14(− 0.30, 0.57) vs 0.00 1. brief clinical assessment (− 0.52, 0.52) 2. self-management: including diet, None of the interactions was exercise, stress reduction, significant smoking cessation, readiness assessment, and development of specific self-management goals 3. medication adherence 4. glucose monitoring and review of home glucose monitoring results educational materials Personnel involved: nurse Control group: Usual care at Community Health Center Anderson-Loftin Gender Patient level A1c, BMI, LDL, weight, Stratification by A1c Men vs women 2005 [37] Number of experimental conditions: dietary fat behaviors gender No significant differences A1c USA 2 (1 intervention, 1 control) assessed by FHQ, physical Mean weight No significant differences Intervention: activity, psychological Significant effect Mean weight Education in low fat dietary status I: - 4 lb. Significant effect strategies (4 weekly classes) C: + 4.2 lb. + 5.4 lb. vs − 1.5 lb.; 1-h peer-professional discussion BMI MD = 6.9 lb. groups (5 monthly) I: − 0.81 kg/mm2 BMI Additional educational support by C: + 0.57 kg/mm2 + 2 kg/mm2 vs 0.16 kg/mm2 phone (weekly) MD = 1.38 kg/mm2 p = 0.02 Incentives for attendance p = 0.009 Dietary behaviors Personnel involved: nurse case Dietary behaviors (FHQ score) Significant effect manager I: 2.5 ± 0.4 (FHQ score) Control group: C: 2.6 ± 0.4 − 0.24 vs − 0.17 Usual care including a referral MD = 0.2 to a local 8-h traditional diabetes class p = 0.005 (information on nature and complications of diabetes) Incentives for attendance Babamoto 2009 Age Healthcare level BMI, A1C, medication Logistic regression Mean A1c Patients aged≥50 were less [58] Number of experimental conditions: adherence, diet, physical models Within group likely to have reduced BMI at USA 3 (2 intervention, 1 control) activity, emergency CHW = 8.6 to 7.2%; p < 0.05 follow-up Intervention: department admission CM =8.5 to 7.4%; p < 0.05 OR = 0.4 (95% CI = 0.2–0.8) Group A, CHW program, Amigos en (ED) Standard care = 9.5 to Exercise frequency 3 times Salud (Friends in Health): education 7.4%;p < 0.05 or more per week vs 2 times through individual session and No significant differences or fewer per week monitoring services; individual were found between groups OR = 2.2 (95% CI = 1.1–4.1) Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 12 of 18 � � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis sessions with participants and BMI Significantly greater family member; telephone calls to decrease for the CHW group participants to monitor self- compared with the standard management, to help participants care group OR = 2.9 improve their diabetes self- (95% CI 1.1–6.6) management skills ED Group B, case management: Change from baseline CHW: education from two linguistically total visit decrease 11% competent and culturally sensitive. Case management: total visit Patients case management were increase 40% Standard care: usually seen on a monthly basis + increase 15% follow-up calls. between groups at 6-month Personnel involved: bilingual, trained follow-up p < 0.05 community health workers, nurse Diet case manager CHW group were more likely Setting: Community, home, clinic (OR = 2.43; 95% CI =1. Control group 13–5.23) to report having Standard Provider Care: standardized two or more servings of fresh clinical care by physicians and nurse fruit per day than standard practitioners, without case care management or CHW services Physical activity CHW group was more likely (OR = 2.87, 95% CI = 1.34– 6.17) than standard care to report exercising three or more times per week Brown, 2011 Gender Patient level A1c, FBG, lipids, BP, BMI, Interaction terms Over time, both the FBG, BMI: [63] Number of experimental conditions: diabetes-related in hierarchical experimental and control No significant differences between USA 2 (1 intervention, 1 control) knowledge, health linear and groups showed gender Intervention: behaviors (physical nonlinear models improvements in FBG levels The rate of change in A1c over time Diabetes self-management activity, dietary intake, to test for at three and did not differ significantly by gender education (DSME) including 8 glucose monitoring) differential impact At six months (coefficient^ = − 0.06, t ratio = 0.25, consecutive weeks of education of treatment by For A1c the control group p = 0.806) followed by a support group gender had greater clinical session at 3 and 6 months improvements at both Experienced NCM providing: intervals culturally tailored diabetes self- Self-reported physical activity management education; and fat intake individualized health guidance Improvement for both and assistance with overcoming experimental and control cultural and environmental barriers groups to improving health; guidance on locating, accessing, and navigating healthcare services; enhanced coordination of health care and communication with physicians and other healthcare providers Random observations visits Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 13 of 18 � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Personnel involved: bilingual NCM, nurses, dietitians, and CHWs Control Group: DSME intervention only Forjuoh 2014 Race/ethnicity Patient level A1C, physical activity, Interaction terms BMI and BP: Modest A1c [64] Number of experimental conditions: BMI, BP, diet in multilevel reductions from baseline to Modest reductions occurred in A1c USA 4 (3 intervention, 1 control) models to test for 12 months of follow-up for from baseline to 12 months of Intervention: differential impact all four groups. follow-up for all/ethnic groups. Group A. self-management through of treatment by No significant difference for There was no significant difference personal digit assistant (PDA). Dia race/ethnicity other outcomes. in A1c change over time by betes Pilot Chronic Disease Self Self care activities: race/ethnicity. Management Program (CDSMP): Hispanic washing feet 6 week group education program significantly more than other to increase self efficacy racial/ethnic groups (P = 0.02) Group B. self-management through Retention rate: PDA CDSMP: 85%; PDA 64%, Group C. combination of A + B CDSMP + PDA 64%; Control Personnel involved: trained facilitator, 78% project coordinators Setting: outpatient clinic, community Control group: usual clinical diabetes care, along with patient education materials Gerber 2005 Health literacy Patient level A1c, BMI, BP, eye exam, Stratification by No significant differences for Lower literacy group [36] Number of experimental conditions: diabetes knowledge, self- level of health all outcomes but perceived % change A1c USA 2 (1 intervention, 1 control) efficacy, self-reported literacy susceptibility to diabetes − 0.21 ± 2.0 vs − 0.1% ± 1.3 Intervention: medical care, and per- complications MD = − 0.10 [− 0.67, 0.47] Education by computer multi-media ceived susceptibility to People with A1c > 9% including audio/video sequences complications − 2.1 vs − 0.3 (p = 0.036) (“Living Well with Diabetes”)to Perceived susceptibility to communicate information, provide complications psychosocial support and promote % change score= self-management. Subject received 1.48 ± 2.7 vs 0.19 ± 2.5 (p = 0.016) compensation based on computer Self-efficacy usage. Lessons in English and trend toward greater improvement Spanish. Navigation provided in self-efficacy through a simplified interface, 1.51 ± 1.5 vs. 0.99 ± 1.4 including forward/backward buttons (p = 0.113) for user control. Advanced features Higher literacy included “pop-up” supplementary % change A1c text information or additional + 0.3% ± 1.6 vs. -0.5 ± 1.5 testimonials related to the MD = 0.80 [0.22, 1.38] concurrent screen concept Perceived susceptibility to Personnel involved: bilingual complications research assistant 0.76 ± 2.5 vs. 0.29 ± 2.4 (p = 0.267) Setting: urban outpatient clinics Medical care Improvement over Control group: simple multiple- time (p < 0.012 for time interaction) choice quizzes on diabetes-related but no effect for either lower- or concepts higher-literacy groups Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 14 of 18 � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Sixta 2008 [28] age Healthcare level A1C, knowledge, beliefs Stratified analysis A1C, knowledge, beliefs A1C USA Number of experimental conditions: by age No difference between groups 2 (1 intervention, 1 control) DKQ, HBQ. Intervention: No difference between groups Diabetes culturally self-management DKQ, HBQ, and HbA1c results were education with group sessions significantly affected by age; Personnel involved: promotores in Slightly negative effect on DKQ consultation with a care team scores per year of age. Control group:Usual care delivered Slightly negative effect on HBQ by provider at the clinic or to a scores and HbA1c levels per year self-care of age management West 2007 [70] Race/ethnicity Patient level A1C, glucose monitoring The weight Weight Weight at 6 months regardless USA Number of experimental conditions: patterns over time At 6 months treatment: 2 (1 intervention, 1 control) by race were Means: − 4.7 ± 5.4 kg vs − African-American vs White Intervention: examined using 3.1 ± 3.9 kg (p = 0.03) -3 kg ± 3.9 vs. -4.5 ± 5.1 kg 42 group session of behavioral a two-factor Over 18 months: (p = 0.03) weight control program focusing repeated Means: − 3.5 ± 6.8 Kg vs − Weight at 12 months regardless on attainable and sustainable measures ANOVA 1.7 ± 5.7Kg (p = 0.04) treatment:: changes in dietary and physical stratified by A1C − 2.3 kg ± 4.4 vs − 4.6 ± 6.8 kg activity habits treatment Decrease in both groups (p = 0.09) Motivational interviewing: 5 (p < 0.0001) at 6 months Weight at 18 months regardless individual sessions lasted 45 min but not sustained at treatment: Personnel involved: Behaviorist, 18 months − 1.4 kg ± 4.7 vs − 3.3 ± 7.1 kg nutritionist, diabetes educator, Greater decrease in the (p = 0.09) trained clinical psychologist intervention than in the For African-American experimental Setting: outpatient clinic control group (p = 0.002) intervention produced greater Control group: health education weight loss than control group at 3 sessions with focus on women’s and 6 months. The benefit was not health topics sustained after 12 months A1c African American had high A1c values regardless of treatment assignment. No interaction by race Attendance between groups was comparable. Data are means ± SD; I intervention group, C control group, OR odds ratio, A1c, Glycated hemoglobin; BMI Body Mass Index, LDL low density cholesterol, BP blood pressure, SBP systolic blood pressure, DBP diastolic blood pressure, MD mean difference, FHQ food habit questionnaire, PHQ-9 Patient Health Questionnaire, DSME Diabetes self-management education, DKQ diabetes knowledge questionnaire, HBQ Health Beliefs Questionnaire a b multivariate analysis adjusted for study group, gender, dietary, exercise activity; univariate analysis (did not persist after the other covariates were controlled for); ^b = regression coefficient Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 15 of 18 settings and therefore address differences in baseline publication. The issue of small sample size extends be- A1C values across studies in ways other than restricting yond the quality of those studies included in this review. patient participant inclusion. There were a number of studies, both pilot and not, that Another relevant issue in the evaluation of QI strat- were excluded from this review because they had a sam- egies is that the control groups received a wide range of ple size smaller than 50. Furthermore, since most studies interventions, from basic education materials, usual care, were carried out in USA, their degree of external validity to individualized coaching from community health is uncertain. Results from these studies may be less workers. Furthermore, in many of these studies, the con- transferrable to other countries and settings due to their trol group intervention was not described in detail. This being tested in a market-based health care system. It is is important as the usual or routine care in different set- likely that the patients’ population covered by universal- tings varies by a multitude of variables including pay- istic care is more heterogeneous with regard to ment system, geographic location, country, and more socio-demographic and clinical characteristics. For ex- generally, the resources and quality of services routinely ample, those countries with universal health care sys- provided to patients. In addition, type and quality of tems may have more heterogeneous patient populations usual care at a health center can impact baseline values, in a single community. It is therefore necessary to plan especially HbA1c. Moreover, biases may exist depending trials in other countries. By the same token, interven- on previous improvement activities implemented and tions addressing health disparities in other countries are general commitment of medical staff and organizational likely to involve groups of varying social advantage or leadership to reducing disparities and improving care. disadvantage being served under the same health center The conclusions of this systematic review are largely or system. The approach to addressing inequity becomes in accord with those in a previous review on this topic more about reducing health disparities on a more granu- among socially disadvantaged population living in indus- lar level requiring tools such as health equity audit. trialized countries published in 2006 [74]. The review Although the PROGRESS framework provides a vast identified 17 studies, seven trials were with low SES pop- array of disadvantage categories, there was limited het- ulations, and ten focused on etno-racial groups. The erogeneity in the dimensions of disadvantage considered small number of studies in Glazier’s review provided in RCTs. The most common PROGRESS factor were age limited and inconclusive evidence on intervention attri- and race/ethnicity, this underlines the needs of further butes that improved diabetes quality of care and health research with a focus on other characteristics such as so- outcomes, underlining the potential effect of some fea- cioeconomic status, social capital, place of residence, oc- tures in reducing health disparities. cupation, education, and religion. Researchers studying Our review provides an update and a more complete populations at social disadvantage must also describe the overview of the available evidence considering three spe- study population and the nature of their disadvantage cific aspects: use of PROGRESS framework to capture more specifically. This is of further importance because different socio-economic dimensions; assessment of the a lack of description or definition of a socially disadvan- risk of bias of included studies; and the inclusion of taged group was a common reason for study exclusion studies evaluating QI strategies defined according to in this review and others. international classification. There is also a clear need for more RCTs at the provider Using an equity oriented approach, we identified a level, especially those evaluating interventions based on large number of randomized studies showing that con- computerized provider reminder systems. With the wide- siderable strides have been made to test interventions to spread uptake of recognition and certification programs in address health inequities in diabetes care and outcomes. primary care (e.g. medical home, diabetes recognition pro- Despite the increase of the number of trials, the meth- grams,), it is likely that audit and feedback strategies using odological quality resulted to be low. This finding is con- benchmarking are common among primary care practices, sistent with a previous review [75] reporting that the but are less frequently reported for effectiveness among increase in the number of RCTs on QI strategies runs disadvantaged patient populations. parallel to the proportion of trials having at least one do- This research reveals an overall lack of focus on inter- main with high risk of bias. Most included trials did not ventions that address outcomes related to adherence to report the method of randomization and description of guidelines where disparities are stark according to the the allocation process. The area of the greatest potential literature. The paucity of studies measuring process of risk of bias was the inadequate blinding of participants care may be a reflection of the few number of QI inter- and outcome assessors, and poor follow up. In some of ventions at the provider level who, in conjunction with included trials the general lack of reporting of methods other members of the primary care team, are responsible made it difficult to assess methodological quality and for performing or referring to these services. Clinical thereby judge risk of bias, independently of year of outcomes should derived from electronic health record Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 16 of 18 systems, but may not be as recurrently funded as bio- Additional file 3: Table S2. Quality improvement strategies: level and chemical diabetes outcomes. Process outcomes or adher- description. (DOCX 13 kb) ence to guidelines is crucial to measure and address due Additional file 4: Table S3. Characteristics of eligible studies assessing the efficacy of QI interventions in participants with type 2 diabetes. to the evidence of disparities that exist on the level of (DOCX 71 kb) clinical quality and care. It is also important to note that several studies measured diabetes “self-care” or “self-- Authors’ contributions management” activities but did not report results on dis- NT, AB, and NA made substantial contributions to the conception and tinct components such as medication adherence or design of this systematic review. ZM completed the literature search. NT and AB screened studies against eligibility criteria, extracted data, and analysed glucose monitoring. As these clinical outcome measures and interpreted data. All authors contributed to writing and revising the final are crucial in measuring effectiveness of diabetes inter- manuscript. All authors read and approved the final manuscript LA and MD vention, it is important to report on these components contributed to the critical revision. as distinctive measures. Ethics approval and consent to participate We see many studies that aim to evaluate interven- Not applicable tions to improve care and/or outcomes among a disad- vantaged group, but seldom do we find studies Competing interests The authors declare that they have no competing interests. investigating the effect of QI interventions disentangled by different levels of indicators of socio-economic pos- Publisher’sNote ition or relevant socio-demographic factors. This may Springer Nature remains neutral with regard to jurisdictional claims in because practices are not disaggregating data to identify published maps and institutional affiliations. disparities within patient populations and are therefore Author details not initiating action to address them. It should be neces- 1 2 Trenton Health Team, Trenton, New Jersey, USA. Department of sary to promote and sustain a different approach includ- Epidemiology, Lazio Region- ASL Rome1, Rome, Italy. ing audit activities to identify inequities in care and Received: 28 February 2018 Accepted: 16 May 2018 outcomes, and then work to address these disparities. Moreover, an “equity lens” approach should be adopted by the scientific community when identifying research References priorities aimed at contrasting socioeconomic differen- 1. World Health Organization. Global Report on Diabetes 2016. http://apps.who. int/iris/bitstream/handle/10665/204871/9789241565257_eng.pdf;jsessionid= tials. This equity-oriented approach is necessary to iden- 45F529CEFB7FB49CB7EE39B0F63BA11F?sequence=1. Accessed 20 Feb 2016. tify and describe the appropriate target population, to 2. Agardh E, Allebeck P, Hallqvist J, Moradi T, Sidorchuk A. Type 2 diabetes define inequalities indicators, and select process and incidence and socio-economic position: a systematic review and meta- analysis. Int J Epidemiol. 2011;40:804–18. outcome indicators useful for assessing the differential 3. Redle EE, Atkins D. The applicability of quality improvement research for effect of an intervention. comparative effectiveness. Implement Sci. 2013;8(Suppl 1):S6. 4. Espelt A, Arriola L, Borrell C, Larrañaga I, Sandín M, Escolar-Pujolar A. Socioeconomic position and type 2 diabetes mellitus in Europe 1999-2009: Conclusions a panorama of inequalities. Curr Diabetes Rev. 2011;7(3):148–58. 5. NICE. Type 2 diabetes in adults: management 2015. https://www.nice.org. Because of the methodological differences and weak- uk/guidance/ng28/evidence/full-guideline-pdf-78671532569. Accessed 20 nesses that precluded meta-analytic synthesis, we can Feb 2016. draw no strong conclusions concerning the potential 6. Grintsova O, Maier W, Mielck A. Inequalities in health care among patients with type 2 diabetes by individual socio-economic status (SES) and regional benefits or harms of QI strategies to reduce inequalities deprivation: a systematic literature review. Int J Equity Health. 2014;13:43. in access to care for patients with diabetes in primary https://doi.org/10.1186/1475-9276-13-43. care. Moreover, the included studies did not allow for an 7. Heidemann DL, Joseph NA, Kuchipudi A, Perkins DW, Drake S. Racial and economic disparities in Diabetes in a Large Primary Care Patient Population. analysis of the differential effects of interventions across Ethn Dis. 2016;26(1):85–90. https://doi.org/10.18865/ed.26.1.85. population sub-groups. 8. Calvert M, Shankar A, McManus RJ, Lester H, Freemantle N. Effect of the This review highlights some QI strategies for consider- quality and outcomes framework on diabetes care in the United Kingdom: retrospective cohort study. BMJ. 2009;338:b1870. https://doi.org/10.1136/ ation and in need of further study. Health care profes- bmj.b1870. Erratum in: BMJ. 2009;339:b2768 sionals and policy makers need the best available evidence 9. Rossi MC, Candido R, Ceriello A, Cimino A, Di Bartolo P, Giorda C, et al. to administer and support those interventions most likely Trends over 8 years in quality of diabetes care: results of the AMD annals continuous quality improvement initiative. Acta Diabetol. 2015;52(3):557–71. to be effective to reduce disparities in diabetes care. https://doi.org/10.1007/s00592-014-0688-6. 10. Tricco AC, Ivers NM, Grimshaw JM, Moher D, Turner L, Galipeau J, et al. Effectiveness of quality improvement strategies on the management of Additional files diabetes: a systematic review and meta-analysis. Lancet. 2012;379(9833): 2252–61. https://doi.org/10.1016/S0140-6736(12)60480-2. 11. O'Neill J, Tabish H, Welch V, Petticrew M, Pottie K, Clarke M, et al. Applying Additional file 1: Search strategy for PubMed. (DOCX 17 kb) an equity lens to interventions: using PROGRESS ensures consideration of Additional file 2: Table S1. Inclusion and exclusion criteria (PICOS). socially stratifying factors to illuminate inequities in health. J Clin Epidemiol. (DOCX 14 kb) 2014;67(1):56–64. https://doi.org/10.1016/j.jclinepi.2013.08.005. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 17 of 18 12. Welch V, Petticrew M, Petkovic J, Moher D, Waters E, White H, et al. PRISMA- randomized controlled trial of a community health worker intervention for equity Bellagio group. Extending the PRISMA statement to equity-focused Mexican Americans with type 2 diabetes mellitus. Am J Public Health. 2014; systematic reviews (PRISMA-E 2012): explanation and elaboration. J Clin 104(8):1540–8. Epidemiol. 2016;70:68–89. https://doi.org/10.1016/j.jclinepi.2015.09.001. 31. Palmas W, Findley SE, Mejia M, Batista M, Teresi J, Kong J, et al. Results of 13. McPheeters ML, Kripalani S, Peterson NB, Idowu RT, Jerome RN, Potter SA, the northern Manhattan diabetes community outreach project: a et al. Quality improvement interventions to address health disparities. In: randomized trial studying a community health worker intervention to Closing the quality gap: revisiting the state of the science. Evidence report improve diabetes care in hispanic adults. Diabetes Care. 2014;37(4):963–9. no. 208. (prepared by the Vanderbilt University evidence-based practice 32. Walker EA, Shmukler C, Ullman R, Blanco E, Scollan-Koliopoulus M, Cohen center under contract no. 290–2007-10065.) AHRQ publication no. 12-E009- HW. Results of a successful telephonic intervention to improve diabetes EF. Rockville: Agency for Healthcare Research and Quality; 2012. https:// control in urban adults: a randomized trial. Diabetes Care. 2011;34(1):2–7. effectivehealthcare.ahrq.gov/sites/default/files/pdf/disparities-quality- 33. Davis RM, Hitch AD, Salaam MM, Herman WH, Zimmer-Galler IE, Mayer-Davis improvement_research.pdf. EJ. TeleHealth improves diabetes self-management in an underserved 14. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. community: diabetes TeleCare. Diabetes Care. 2010;33(8):1712–7. Cochrane Bias methods group; Cochrane Statistical Methods Group The 34. Philis-Tsimikas A, Fortmann A, Lleva-Ocana L, Walker C, Gallo LC. Peer-led Cochrane Collaboration's tool for assessing risk of bias in randomised trials. diabetes education programs in high-risk Mexican Americans improve BMJ. 2011;343:d5928. https://doi.org/10.1136/bmj.d5928. glycemic control compared with standard approaches: a project Dulce 15. RevMan 2014. The Nordic Cochrane Centre, The Cochrane Collaboration. promotora randomized trial. Diabetes Care. 2011;34(9):1926–31. Review Manager (RevMan). Version 5.3. Copenhagen: The Nordic Cochrane 35. Brown SA, Blozis SA, Kouzekanani K, Garcia AA, Winchell M, Hanis CL. Centre, The Cochrane Collaboration; 2014. Dosage effects of diabetes self-management education for Mexican 16. Bellary S, O'Hare JP, Raymond NT, Gumber A, Mughal S, Szczepura A, et al. Americans: the Starr County border health initiative. Diabetes Care. 2005; Enhanced diabetes care to patients of south Asian ethnic origin (the United 28(3):527–32. Kingdom Asian Diabetes Study): a cluster randomised controlled trial. 36. Gerber BS, Brodsky IG, Lawless KA, Smolin LI, Arozullah AM, Smith EV, et al. Lancet. 2008;371(9626):1769–76. Implementation and evaluation of a low-literacy diabetes education 17. Berry DC, Williams W, Hall EG, Heroux R, Bennett-Lewis T. Imbedding computer multimedia application. Diabetes Care. 2005;28(7):1574–80. interdisciplinary diabetes group visits into a community-based medical 37. Anderson-Loftin W, Barnett S, Bunn P, Sullivan P, Hussey J, Tavakoli A. Soul setting. Diabetes Educ. 2016;42(1):96–107. food light: culturally competent diabetes education. Diabetes Educ. 2005; 18. DePue JD, Rosen RK, Seiden A, Bereolos N, Chima ML, Goldstein MG, et al. 31(4):555–63. Implementation of a culturally tailored diabetes intervention with 38. Tang TS, Funnell MM, Sinco B, Spencer MS, Heisler M. Peer-led, community health workers in American Samoa. Diabetes Educat. 2013;39(6): empowerment-based approach to self-management efforts in diabetes 761–71. (PLEASED): a randomized controlled trial in an African American community. 19. McDermott RA, Schmidt B, Preece C, Owens V, Taylor S, Li M, et al. Ann Fam Med. 2015;13(Suppl 1):S27–35. Community health workers improve diabetes care in remote Australian 39. Toobert DJ, Strycker LA, King DK, Barrera M Jr, Osuna D, Glasgow RE. Long- indigenous communities: results of a pragmatic cluster randomized term outcomes from a multiple-risk-factor diabetes trial for Latinas: -íViva controlled trial. BMC Health Serv Res. 2015;15:68. Bien! Transl Behav Med. 2011;1(3):416–26. 20. Gucciardi E, DeMelo M, Lee RN, Grace SL. Assessment of two culturally 40. Tang TS, Funnell M, Sinco B, Piatt G, Palmisano G, Spencer MS, Kieffer EC, competent diabetes education methods: individual versus individual plus Heisler M. Comparative effectiveness of peer leaders and community health group education in Canadian Portuguese adults with type 2 diabetes. Ethn workers in diabetes selfmanagement support: results of a randomized Health. 2007;12(2):163–87. controlled trial. Diabetes Care. 2014;37(6):1525–34. 21. Guo H, Tian X, Li R, Lin J, Jin N, Wu Z, Yu D. Reward-based, task-setting 41. Negarandeh R, Mahmoodi H, Noktehdan H, Heshmat R, Shakibazadeh E. education strategy on glycemic control and self-management for low-income Teach back and pictorial image educational strategies on knowledge about outpatients with type-á2 diabetes. J Diabetes Investig. 2014;5(4):410–7. diabetes and medication/dietary adherence among low health literate 22. Hotu C, Collins J, Harwood L, Whalley G, Doughty R, Gamble G, Braatvedt G. patients with type 2 diabetes. Prim Care Diabetes. 2013;7(2):111–8. A community-based model of care improves blood pressure control and 42. Weinstein E, Galindo RJ, Fried M, Rucker L, Davis NJ. Impact of a focused delays progression of proteinuria, left ventricular hypertrophy and diastolic nutrition educational intervention coupled with improved access to fresh dysfunction in Maori and Pacific patients with type 2 diabetes and chronic produce on purchasing behavior and consumption of fruits and vegetables kidney disease: a randomized controlled trial. Nephrol Dial Transplant. 2010; in overweight patients with diabetes mellitus. Diabetes Educat. 2014;40(1): 25(10):3260–6. 100–6. 23. Partapsingh VA, Maharaj RG, Rawlins JM. Applying the Stages of Change 43. Phillips LS, Ziemer DC, Doyle JP, Barnes CS, Kolm P, Branch WT, et al. An model to Type 2 diabetes care in Trinidad: A randomised trial. J Negat endocrinologist-supported intervention aimed at providers improves Results Biomed. 2011;10(1):13. diabetes management in a primary care site: improving primary care of 24. Shahid M, Mahar SA, Shaikh S, Shaikh Z-U-D. Mobile phone intervention to African Americans with diabetes (IPCAAD) 7. Diabetes Care. 2005;28(10): improve diabetes care in rural areas of Pakistan: a randomized controlled 2352–60. trial. J Coll Physicians Surg Pak. 2015;25(3):166–71. 44. Welch G, Zagarins SE, Santiago-Kelly P, Rodriguez Z, Bursell SE, Rosal MC, 25. Frosch DL, Uy V, Ochoa S, Mangione CM. Evaluation of a behavior support et al. An internet-based diabetes management platform improves team intervention for patients with poorly controlled diabetes. Arch Intern Med. care and outcomes in an urban Latino population. Diabetes Care. 2015; 2011;171(22):2011–7. 38(4):561–7. 26. Khan MA, Shah S, Grudzien A, Onyejekwe N, Banskota P, Karim S, et al. A 45. Shea S, Weinstock RS, Teresi JA, Palmas W, Starren J, Cimino JJ, et al. diabetes education multimedia program in the waiting room setting. IDEATel consortium. A randomized trial comparing telemedicine case Diabetes Ther. 2011;2(3):178–88. https://doi.org/10.1007/s13300-011-0007-y. management with usual care in older, ethnically diverse, medically Epub 2011 Aug 22 underserved patients with diabetes mellitus: 5 year results of the IDEATel 27. Rosal MC, Ockene IS, Restrepo A, White MJ, Borg A, Olendzki B, et al. study. J Am Med Inform Assoc. 2009;16(4):446–56. Randomized trial of a literacy-sensitive, culturally tailored diabetes self- 46. Kim MT, Kim KB, Huh B, Nguyen T, Han H-R, Bone LR, et al. The effect of a management intervention for low-income latinos: latinos en control. community-based self-help intervention: Korean Americans with type 2 Diabetes Care. 2011;34(4):838–44. diabetes. Am J Prev Med. 2015 Nov;49(5):726–37. 28. Sixta CS, Ostwald S. Texas-Mexico border intervention by promotores for 47. Perez-Escamilla R, Damio G, Chhabra J, Fernandez ML, Segura-Perez S, Vega- patients with type 2 diabetes. Diabetes Educ. 2008;34(2):299–309. Lopez S, et al. Impact of a community health workers-led structured 29. Hill-Briggs F, Lazo M, Peyrot M, Doswell A, Chang YT, Hill MN, et al. Effect of program on blood glucose control among Latinos with type 2 diabetes: problem-solving-based diabetes self-management training on diabetes The DIALBEST Trial. Diabetes Care. 2015;38(2):197–205. // () *National Center control in a low income patient sample. J Gen Intern Med. 2011;26(9):972–8. for Advancing Translational Sciences* 30. Rothschild SK, Martin MA, Swider SM, Tumialán Lynas CM, Janssen I, Avery 48. Spencer MS, Rosland AM, Kieffer EC, Sinco BR, Valerio M, Palmisano G, et al. EF, et al. Mexican American trial of community health workers: a Effectiveness of a community health worker intervention among African Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 18 of 18 American and Latino adults with type 2 diabetes: a randomized controlled 68. Skelly AH, Carlson JR, Leeman J, Holditch-Davis D, Soward AC. Symptom- trial. Am J Public Health. 2011;101(12):2253–60. focused management for African American women with type 2 diabetes: a 49. Prezio EA, Cheng D, Balasubramanian BA, Shuval K, Kendzor DE, Culica D. pilot study. J Nurs Scholarsh. 2005;18(4):213–20. Community diabetes education (CoDE) for uninsured Mexican Americans: a 69. Wayne N, Perez DF, Kaplan DM, Ritvo P. Health coaching reduces HbA1c in randomized controlled trial of a culturally tailored diabetes education and type 2 diabetic patients from a lower-socioeconomic status community: a management program led by a community health worker. Diabetes Res randomized controlled trial. J Med Internet Res. 2015;17(10):e224. Clin Pract. 2013;100(1):19–28. https://doi.org/10.1016/j.diabres.2013.01.027. 70. West DS, DiLillo V, Bursac Z, Gore SA, Greene PG. Motivational interviewing improves weight loss in women with type 2 diabetes. Diabetes Care. 2007; 50. Thom DH, Ghorob A, Hessler D, Vore D, Chen E, Bodenheimer TA. Impact of 30(5):1081–7. peer health coaching on glycemic control in low-income patients with 71. Seligman HK, Wang FF, Palacios JL, Wilson CC, Daher C, Piette JD, et al. diabetes: a randomized controlled trial. Ann Fam Med. 2013;11(2):137–44. Physician notification of their diabetes patients' limited health literacy. A 51. Gary TL, Batts-Turner M, Yeh HC, Hill-Briggs F, Bone LR, Wang NY, et al. The randomized, controlled trial. J Gen Intern Med. 2005;20(11):1001–7. effects of a nurse case manager and a community health worker team on 72. Ruggiero L, Moadsiri A, Butler P, Oros SM, Berbaum ML, Whitman S, et al. diabetic control, emergency department visits, and hospitalizations among Supporting diabetes self-care in underserved populations: a randomized urban African Americans with type 2 diabetes mellitus: a randomized pilot study using medical assistant coaches. Diabetes Educ. 2010;36(1):127–31. controlled trial. Arch Intern Med. 2009;169(19):1788–94. 73. Willard-Grace R, Chen EH, Hessler D, DeVore D, Prado C, Bodenheimer T, 52. Kim MT, Han HR, Song HJ, Lee JE, Kim J, Ryu JP, et al. A community-based, et al. Health coaching by medical assistants to improve control of diabetes, culturally tailored behavioral intervention for Korean Americans with type 2 hypertension, and hyperlipidemia in low-income patients: a randomized diabetes. Diabetes Educat. 2009;35(6):986–94. controlled trial. Ann Fam Med. 2015;13(2):130–8. 53. Liou J-K, Soon M-S, Chen C-H, Huang T-F, Chen Y-P, Yeh Y-P, et al. Shared 74. Glazier RH, Bajcar J, Kennie NR, Willson K. A systematic review of care combined with telecare improves glycemic control of diabetes interventions to improve diabetes care in socially disadvantaged patients in a rural, underserved community. J Diabetes Sci Technol. 2013; populations. Diabetes Care. 2006;29(7):1675–88. 7(1):A82. 75. Ivers NM, Tricco AC, Taljaard M, Halperin I, Turner L, Moher D, et al. Quality 54. Lujan J, Ostwald SK, Ortiz M. Promotora diabetes intervention for Mexican improvement needed in quality improvement randomised trials: systematic Americans. Diabetes Educ. 2007;33(4):660–70. review of interventions to improve care in diabetes. BMJ Open. 2013;3(4): 55. Levy NK, Moynihan V, Nilo A, Singer K, Etiebet M-A, Bernik L, et al. The e002727. https://doi.org/10.1136/bmjopen-2013-002727. mobile insulin titration intervention (MITI) study: innovative chronic disease management of diabetes. J Gen Intern Med. 2015;30:S547–S8. 56. Cramer JS, Sibley RF, Bartlett DP, Kahn LS, Loffredo L. An adaptation of the diabetes prevention program for use with high-risk, minority patients with type 2 diabetes. Diabetes Educat. 2007;33(3):503–8. 57. Garcia AA, Brown SA, Horner SD, Zuniga J, Arheart KL. Home-based diabetes symptom self-management education for Mexican Americans with type 2 diabetes. Health Educ Res. 2015;30(3):484–96. 58. Babamoto KS, Sey KA, Camilleri AJ, Karlan VJ, Catalasan J, Morisky DE. Improving diabetes care and health measures among hispanics using community health workers: results from a randomized controlled trial. Health Educ Behav. 2009;36(1):113–26. 59. Lynch EB, Liebman R, Ventrelle J, Avery EF, Richardson D. A self- management intervention for African Americans with comorbid diabetes and hypertension: a pilot randomized controlled trial. Prev Chronic Dis. 2014;11:E90. 60. Tobe SW, Pylypchuk G, Wentworth J, Kiss A, Szalai JP, Perkins N, et al. Effect of nurse-directed hypertension treatment among first nations people with existing hypertension and diabetes mellitus: the diabetes risk evaluation and microalbuminuria (DREAM 3) randomized controlled trial. CMAJ. 2006; 174(9):1267–71. 61. Anderson DR, Christison-Lagay J, Villagra V, Liu H, Dziura J. Managing the space between visits: a randomized trial of disease management for diabetes in a community health center. J Gen Intern Med. 2010;25(10):1116–22. 62. Baradaran HR, Knill-Jones RP, Wallia S, Rodgers A. A controlled trial of the effectiveness of a diabetes education programme in a multi-ethnic community in Glasgow [ISRCTN28317455]. BMC Public Health. 2006;6:134. 63. Brown SA, García AA, Winter M, Silva L, Brown A, Hanis CL. Integrating education, group support, and case management for diabetic Hispanics. Ethn Dis. 2011;21(1):20–6. 64. Forjuoh SN, Bolin JN, Huber Jr JC, Vuong AM, Adepoju OE, Helduser JW, et al. Behavioral and technological interventions targeting glycemic control in a racially/ethnically diverse population: a randomized controlled trial. BMC Public Health. 2014;14:71. 65. Gregg JA, Callaghan GM, Hayes SC, Glenn-Lawson JL. Improving diabetes self-management through acceptance, mindfulness, and values: a randomized controlled trial. J Consult Clin Psychol. 2007;75(2):336–43. 66. Heisler M, Choi H, Palmisano G, Mase R, Richardson C, Fagerlin A, et al. Comparison of community health worker-led diabetes medication decision- making support for low-income Latino and African American adults with diabetes using e-health tools versus print materials: a randomized, controlled trial. Ann Intern Med. 2014;161(10 Suppl):S13–22. 67. Schillinger D, Hammer H, Wang F, Palacios J, McLean I, Tang A, et al. Seeing in 3-D: examining the reach of diabetes self-management support strategies in a public health care system. Health Educ Behav. 2008;35(5): 664–82. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Endocrine Disorders Springer Journals

Quality improvement strategies at primary care level to reduce inequalities in diabetes care: an equity-oriented systematic review

Free
18 pages

Loading next page...
 
/lp/springer_journal/quality-improvement-strategies-at-primary-care-level-to-reduce-jLhonla1NP
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s).
Subject
Medicine & Public Health; Endocrinology; Metabolic Diseases; Diabetes; Andrology
eISSN
1472-6823
D.O.I.
10.1186/s12902-018-0260-4
Publisher site
See Article on Publisher Site

Abstract

Background: There is evidence that disparities exist in diabetes prevalence, access to diabetes care, diabetes- related complications, and the quality of diabetes care. A wide range of interventions has been implemented and evaluated to improve diabetes care. We aimed to review trials of quality improvement (QI) interventions aimed to reduce health inequities among people with diabetes in primary care and to explore the extent to which experimental studies addressed and reported equity issues. Methods: Pubmed, EMBASE, CINAHL, and the Cochrane Library were searched to identify randomized controlled studies published between January 2005 and May 2016. We adopted the PROGRESS Plus framework, as a tool to explore differential effects of QI interventions across sociodemographic and economic factors. Results: From 1903 references fifty-eight randomized trials met the inclusion criteria (with 17.786 participants), mostly carried out in USA. The methodological quality was good for all studies. Almost all studies reported the age, gender/sex and race distribution of study participants. The majority of trials additionally used at least one further PROGRESS-Plus factor at baseline, with education being the most commonly used, followed by income (55%). Large variation was observed between these studies for type of interventions, target populations, and outcomes evaluated. Few studies examined differential intervention effects by PROGRESS-plus factors. Existing evidence suggests that some QI intervention delivered in primary care can improve diabetes-related health outcomes in social disadvantaged population subgroups such as ethnic minorities. However, we found very few studies comparing health outcomes between population subgroups and reporting differential effect estimates of QI interventions. Conclusions: This review provides evidence that QI interventions for people with diabetes is feasible to implement and highly acceptable. However, more research is needed to understand their effective components as well as the adoption of an equity-oriented approach in conducting primary studies. Moreover, a wider variety of socio-economic characteristics such as social capital, place of residence, occupation, education, and religion should be addressed. Keywords: Type 2 diabetes, Quality improvement strategies, Equity, Systematic review * Correspondence: s.vecchi@deplazio.it Department of Epidemiology, Lazio Region- ASL Rome1, Rome, Italy Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 2 of 18 Background Methods Diabetes is a complex, chronic disease recognized as an For the purpose of the review, a “socially disadvantaged important cause of premature death and disability [1] group” is defined by differences that place the group at dis- and disproportionately affects socially and economically tinct levels in a social hierarchy. To explicitly consider disadvantaged populations [2–4]. According the Na- health equity and to capture characteristics possibly indicat- tional Institute for Health and Care Excellence guide- ing disadvantaged status, we adopted the PROGRESS-Plus lines [5], patients with type 2 diabetes should receive a framework recommended by the Campbell and Cochrane clear gamut of care to be provided by primary care pro- Equity Methods Group and the Cochrane Public Health viders. Annual routine monitoring of health indicators Group to identify studies with a focus on reducing health such as urinary albumin, BMI, cholesterol, blood creatin- inequalities [11]. PROGRESS-Plus stands for place of resi- ine, HbA1c and BP measured, eyes and feet examined dence, race/ethnicity/culture/language, occupation, gender/ and a smoking review, forms a major part of patient dia- sex, religion, socioeconomic status and social capital. This betes care. In addition patients should expect to receive systematic review was conducted in accordance with an evidenced-based education and access to specialist PRISMA-E 2012 (Preferred Reporting Items for Systematic healthcare professionals including ophthalmologists, po- Reviews and Meta-Analyses, Equity 2012 Extension), a vali- diatrists and dieticians. dated tool to improve both the reporting and conducting of Quality of care among diabetic patient can be influenced equity focused systematic reviews, were upheld in this re- by a range of factors that has been already described. Pre- view [12]. vious systematic reviews showed that low individual socio-economic status and residential area deprivation are Data sources and searches often associated with both worse process indicators and We searched all relevant biomedical databases such as worse intermediate outcomes among patients with type 2 Pubmed, EMBASE, CINAHL, and the Cochrane Li- diabetes [6]. These differences are present even in coun- brary for relevant published RCTs and cluster-RCTs tries with a significant level of economic development that published in English. We limited the search from 1 have a universal health care system. Moreover, disparities January 2005 to 31 May 2016. A combination of MeSH in diabetes care exist among racial or ethnic minority terms and keywords were chosen to reflect selection cri- groups, independent of economic status [7]. teria tailored to each database. Details of the full search To improve diabetes care, it might be important to focus strategy for PubMed are included in supplemental mater- on quality management (QM), especially because the com- ial (Additional file 1). In addition, we scanned the refer- plexity of healthcare system and patients complexities has ence lists of relevant reviews to track relevant RCTs. dramatically increased. QM comprises procedures to moni- tor, assess, and enhance the quality of care. In the last years many countries have developed quality improvement inter- Study selection ventions (QI) to improve both patient outcomes and the Two authors (NT, AMB) independently screened all title quality of diabetes care [8, 9]. A meta-analysis of studies in- and abstracts of all studies obtained from electronic vestigating QI strategies [10] found that interventions target- searches. For studies meeting the inclusion criteria, we ing the entire system of disease management (team changes, retrieved full texts and the same authors independently case management, promotion of self-management) along evaluated them for inclusion. Any disagreements were with patient-mediated QI activities were important compo- resolved through consensus or in discussion with the ex- nents of strategies to improve diabetes care. However, the tended authorial group. studies included in this review were targeted to the general We used the “population, intervention, comparison, population, irrespective of socio-demographic characteristics outcome, setting” (PICOS) logic to guide the systematic or socio-economic status. review (Additional file 2). We included randomized con- Acknowledging the existence of such disparities, our trolled trials (RCTs) and cluster-randomized trials, aims are to: a) describe the extent to which effects on so- evaluating all QI interventions designed to improve cial inequalities are considered in randomized controlled health outcomes in social disadvantaged people with trials (RCTs) evaluating the effects of QI interventions to type 2 diabetes and designed to reduce inequalities in improve quality of diabetes care and b) synthesize evi- diabetes care. We considered studies that reported quan- dence on the effectiveness of QI strategies to reduce titative estimates of total effect of treatment and differ- health inequities in diabetes care in the primary care set- ential effects for the PROGRESS-Plus factors. ting. We conducted an equity-oriented systematic review We used the Agency for Healthcare Research and including RCTs only, using an international taxonomy of Quality [13] taxonomy to identify QI strategies QI interventions, and assessing the quality of included (Additional file 3). QI strategies can be delivered to studies with a methodological rating tool. specific levels of influence: Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 3 of 18 Patient level (e.g. patient education, patient and subjective outcome measures. We defined clinical reminders, or promotion of self-management); and laboratory measures, process indicators, diabetes Health care provider level (e.g. electronic medical complications, hospital admissions, emergency admis- record reminders, audit & feedback, cultural sions and all-cause mortality as objective outcome mea- competency training); sures. We defined measures of self-management/ Health care system level (e.g. change in the health adherence to recommendations as subjective outcome system structure or delivery, adjusting roles of care measures. With respect to missing data, we judged indi- team members, nurse care management model). vidual trials at high risk of bias if data from more than 10% of participants were not available. We used the Data extraction and quality assessment quality criteria for descriptive purposes only to highlight Two authors independently extracted data (NT, SV), and differences between studies. We used RevMan 2014 soft- disagreements were resolved by discussion. Data from ware [15] to generate figures related to risk of bias. multiple publications of the same study was considered as a single study. A data extraction form was designed to Data synthesis document the following study details: trials characteris- We synthesized findings from the included studies by tics; participants (total number at baseline, age range, intervention level (patients, health care provider, and gender, clinical features); type of intervention and com- health care system). The wide variety of interventions parator; clinical and no clinical outcomes; timing; risk of (in terms of mode of delivery, frequency and duration of bias; study results. For continuous outcomes, we ex- follow up assessment) and population groups considered tracted the mean change from baseline (with the stand- in the included studies did not allow for a meaningful ard deviation) and the mean difference, if available, with meta-analysis to be conducted. We summarized results the corresponding 95% confidence interval (CIs). Relative using narrative methods. We described in more detail risk (RR), and absolute risk differences, with the corre- studies reporting differences in QI interventions effects sponding 95% CI, was extracted for binary primary out- across subgroups. comes. If studies reported data for more than one time point, we extracted data for the longest-term outcomes. Results Baseline population characteristics relevant for ad- The search strategy generated 1903 citations after remov- dressing potential issues in health equity were extracted ing duplicates. Upon reviewing titles and abstracts, we re- using the PROGRESS-Plus framework. We extracted trieved full text articles for 247 studies that were screened data on outcome assessed, according to whether by two authors independently (NT, AMB). We excluded PROGRESS-Plus factors were considered as control vari- 189 trials. Most common reasons for exclusion were not ables (e.g., by adjusting in regression analyses) and the addressing a socially disadvantaged group, an evaluation methods utilized to investigate differential effects (strati- of primary prevention intervention, and being conducted fied analysis or modification/interaction analysis). We in a setting other than primary care. Fifty-eight RCTs met also extracted details on the duration of intervention, eligibility criteria. PRISMA Flow Diagram Fig. 1 shows the duration of follow up, health professional group in- details of study selection process. volved, details of the strategy being implemented (i.e. modality, delivery format). Overview of the included studies Two authors independently assessed risk of bias of in- A substantial synthesis of the characteristics of all 58 cluded studies using the Cochrane ‘Risk of bias’ tool for studies included in this review is reported in Table 1. RCTs [14]. We considering the following domains: se- Overall the majority of studies (n = 54) used a parallel quence generation, allocation concealment, blinding of RCT design while four trials were cluster RCTs [16–19]. participants and personnel, blinding of outcome assess- Follow-up periods varied in duration from less than 1 ment, incomplete data, selective reporting, and other month to 5 years, with the majority lasting 6 to biases. For each domain, risk of bias was classified as 12 months. Most of trials were conducted in the USA “high,”“low,” or “unclear”. Since we included (n = 47); the remaining studies were carried out in cluster-randomized controlled trials, additional items Canada [20], Asia [21], the United Kingdom [16], New were considered: (1) recruitment bias: did recruitment of Zealand [22], Australia [19], Trinidad and American diabetes patients take place before or after Samoa [18, 23]. randomization of the clusters?, (2) did the intervention Almost all studies reported the age, gender/sex and and control group differ in baseline characteristics?, (3) race distribution of study participants. The majority of did any of the clusters drop out during follow-up, (4) studies additionally used at least one further was clustering accounted for in the statistical analyses? PROGRESS-Plus factor for the description of partici- We investigated detection bias separately for objective pants’ baseline characteristics. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 4 of 18 Records identified through database Additional records identified through searching other sources (n =2931) (n = 4 ) Records after duplicates removed (n =1903) Records excluded on title Records screened and abstracts (n = 1903) (n = 1656) Full-text articles assessed for Full-text articles excluded eligibility (n = 189) (n = 247) -167 not addressing social disadvantaged people -1 no outcome defined in the inclusion criteria Studies included in -21 no RCT qualitative synthesis (n =58 ) Studies evaluating differential intervention effects by PROGRESS-Plus factors (n =7 ) Fig. 1 PRISMA 2009 Flow Diagram. Study selection process Among these, education was the most commonly re- delayed intervention or no intervention. Health profes- ported factor (n = 45), followed by income (n = 32). sionals who participated in studies included physicians, Twenty-six studies considered at least one PROGRESS-Plus specialist nurses, social workers, dietitians, diabetes educa- factor as control variable when measuring intervention ef- tors, community health workers, general practitioners, fects (e.g., by adjusting in multivariate analyses). Again, age practice nurses and home care nurses. (n = 23) and gender/sex (n = 20) were the factors most com- The majority of trials (96%) provided data on change monly controlled for, followed by education (n = 9). Seven in HbA1c. Thirty-seven trials (63%) reported BMI out- (12%) trials used at least one PROGRESS-Plus factors for come; blood pressure and cholesterol data in 38 and 30 examining differential intervention effects, and gender, age, trials, respectively. Process measures including diabetic race and education were those most often considered. foot exam, dilated eye exam and attendance at office ap- Detailed descriptions of the QI interventions were not pointments were seldom reported. always clearly provided in the trials. In order of frequency, For secondary outcomes, data were available for were twenty-nine studies (50%) focused on interventions patient-reported measures including diet and physical activ- delivered at the patient level [17, 20, 21, 24–27, 29, 32–39, ity (n = 28) using a considerable variety of instruments. 41, 42, 55, 61–70], and twenty-six at the health care Medication adherence and home glucose monitoring were organization level (45%) [16, 18, 19, 22, 23, 28, 30, 31, 40, measured less consistently (in 17 and 15 studies, respectively) 45–54, 56–60, 72, 73]. The remaining three studies (5%) as were diabetes complications and hospital admissions. [43, 44, 71] described interventions at the provider level. A detailed description of trials characteristics and In the majority of studies comparators were “usual” or intervention components by intervention level is pre- “standard” care (69%), five studies reported waiting list, sented in Additional file 4. Eligibility Screening Identification Included Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 5 of 18 Table 1 Synthesis of the characteristics of the included studies by level of intervention and PROGRESS factors Level of intervention Patient level Provider level Health care systems level Total QI strategies Total of studies 29 3 26 58 N % N% N % N% Sample characteristics Age 55.13 - 55.37 - 53.82 - 55.06 - Sex, female (%) 64.05 58.69 57.84 60.20 Baseline HgA1c (%; mmol/mol) 8.88; 74 7.0–11.8; 53–105 9.53; 81 8.1–12.05; 31–109 8.51; 70 7.6–10.5; 60–91 8.88; 74 7.0–12.05; 53–109 Progress factors reported at baseline Place of residence 29 50 3 5.2 26 44.8 58 - Race/ethnicity 26 49.1 3 5.7 24 45.3 53 - Occupation 12 54.5 – - 10 45.5 22 - Gender/sex 24 46.2 3 5.8 25 48.1 52 - Religion – - – - – - – - Education 26 57.7 1 2.2 18 40.1 45 - Socioeconomic status (SES) – - – - – - – - Income 20 62.5 – - 12 37.5 32 - Social capital 10 62.5 - - 6 37.5 16 - Age 28 50.0 3 5.4 25 44.7 56 - Disability – - – - – - – - Sexual orientation – - – - – - – - Study characteristics Year of publication 2005–2010 11 19 2 3.5 11 19 24 41.4 2011–2016 18 31 1 1.7 15 25.9 34 58.6 Study location North America 25 86.2 3 100 22 85 47 UK 1 3.4 – - 1 3.8 2 3.4 Australia –– – - 2 - 2 3.4 Asia 3 10.4 – - 1 7.7 4 6.9 Duration of study (months) 10 3–26 4.5 0.25–36 12 6–60 8.9 0.25–60 Average sample size (range) 190 (56–526) 1573 (182–4138) 290 (65–1665) 684 (50–4138) Risk of bias in included studies studies blinded providers [20, 21, 25–30]. For studies A summary of ‘risk of bias’ for each study and compara- reporting objective outcomes with standardized collec- tive data across the studies is reported in Figs. 2 and 3 . tion methods (e.g. automated blood test), we assigned a All studies were described as individual RCT (n = 54) or low risk of detection bias (79%), as knowledge of treat- cluster-RCTs (n = 4). None of the randomized studies ment assignment was considered unlikely to affect the had uniformly low risk of bias. The allocation sequence outcome. Twenty-eight studies reporting subjective out- was adequately reported in 48% of the studies (28/58), comes, those that used self-reported measures (i.e. ques- with random number tables or a computer-generated tionnaire on dietary habits or physical activities) were at randomized list as the most commonly used methods. high risk of bias due to the lack of blinding of outcome One study was categorized as high risk due to the use of assessment (24 studies). In the remaining 30 studies, in- a gender-based randomization procedure [24]. Most dependent research personnel who were not involved in RCTs (40/58) did not describe or described in sufficient the intervention performed outcome assessments, which detail the allocation concealment to allow a judgment we evaluated as low risk of detection bias. and were evaluated to be at unclear risk of bias. Thirty studies were at low risk of incomplete outcome In the majority of the trials, all participants were aware data due to a low attrition rate (< 10%) or an of the treatment they were receiving, and only eight intention-to-treat (ITT) analysis for primary outcomes. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 6 of 18 Fig. 2 Risk of bias graph Thirteen studies were at high risk of bias because a high with problem-solving training sessions [29], was effective proportion of participants were lost to follow-up or were in improving glycemic control (MD = − 0.72, 95% CI − missing outcome measurements. Selective reporting bias 1.42 to − 0.01, p = 0.02, n = 56). was difficult to detect in most studies because published Two studies (n = 265) showed an improvements in gly- protocols were often unavailable. Most trials reported all cemic control as measured by HbA1c (8.2% ± 0.4 vs 8.6% outcomes. One study [30] collected a large quantity of ±0.3, p = 0.004 and 7.6 ± 1.8 vs 8.2 ± 2.5; p = 0.006, respect- baseline data but did not adequately describe follow-up ively), comparing behavioral education programs via tele- data. One paper [31] did not report some subjective health [33] or using a computerized self-management measures listed in the published protocol. Risk of con- program [26]vs standard care. tamination was high in most of the studies because pa- Berry et al. [17] reported a greater improvement in tients receiving interventions and those receiving usual HbA1c levels in low-income participants receiving ses- care or other interventions were seen within the same sions led by a multidisciplinary team than in the control health center. Among cluster RCTs, three accounted for group (7.6% vs 9.3%; p = 0.001, n = 80). the effects of clustering in their results analysis. One study [21] found that an education program with incentives and self-monitoring devices produced a sig- Study evaluating the effect of QI strategies by nificant reduction in HbA1c (7.29% ±0.58 vs 7.73% intervention level (n = 51) ±0.57; p < 0.05, n = 132). Patient level Philis-Tsimikas et al. [34] did not report difference be- More than half (n = 17) of the studies showed significant tween groups but a significant decrease of HbA1c from effect in at least one of the outcomes considered in this baseline to follow-up (− 1.5%, p < 0.01) was observed in review; most (n = 11) of these interventions include the experimental group. group education sessions or visits and principles of Finally, two trials [35, 36] did not find a significant de- self-management. crease in HbA1c in the study population, but reported a Twenty-seven out of 29 trials reported data on gly- positive association for a subgroup of participants. cemic control measured as HbA1c level. Ten studies re- Brown et al. [35](n = 460) found that for those who ported an improvement in HbA1c levels in the attended ≥50% of the self-management patient education experimental group compared to the control group. sessions, the reduction of HbA1c was − 0.6% for the An education program based on telephone calls [32] “compressed” group and − 1.7% for the “extended” was found to be associated with a decrease in HbA1c group. In Gerber et al. [36](n = 244), the intervention both in the unadjusted (− 0.23 ± 0.11% vs 0.13 ± 0.13%, resulted in significant improvement in HbA1c among p <0.04, n = 526) and adjusted analysis (MD = 0.40, low–health literacy subjects with poor glycemic control. 95% CI 0.10–0.70; p = 0.009). Eighteen trials reported data on change in BMI, three Rosal et al. [27] evaluated a nutritionist or health found a significant improvement in the experimental educator-led self-management education program sup- group. ported by counseling and a self-monitoring device. The Anderson-Loftin et al. [37] reported that the group ex- study showed a difference between groups in HbA1c level posed to the dietary self-management intervention had a at 4 months (MD = − 0.53, 95% CI-0.92 to − 0.14; decrease in BMI while the control group showed an in- p > 0.008, n = 252) but not sustained at 12 months. crease in BMI control group (− 0.81 kg/m2 vs + 0,57 Kg/ An intensive training group intervention addressing m ; p =0.009, n = 97). Tang et al. [38] reported a decrease both diabetes and cardiovascular diseases, combined in BMI in the intervention group receiving behavioral Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 7 of 18 support delivered by a peer leader compared with the con- trol group; the benefit was observed at different follow-up times and maintained at the longest one (15 months) (MD = − 0.8 Kg/m 95CI%-1.6 to − 0.1; p =0.032, n =106). Toobert et al. [39] showed a significant difference in BMI (MD of − 0.40 Kg/m ; p <0.05, n =280) in an underserved and high-risk Latino population treated with a long-term multiple-behavior-change program. Fifteen of the 26 studies examining healthcare inter- ventions in diabetes care considered blood pressure among the outcomes. Two studies showed differences favoring the experimental intervention. In the study con- ducted by Hill-Briggs et al. [29], participants receiving a self-management training adapted for low literacy expe- rienced an individual improvement in DBP and SBP (median reduction = − 7.17 mmHg, n = 8, median reduc- tion of − 14.67 mmHg, n = 9, respectively). Tang et al. [40] also reported a greater reduction in the group that re- ceived a combination of self-management and peer support interventions than the control group, both in SBP (MD = − 10.0 mmHg (95% CI -17.6 to − 2.4, p = 0.01) and DBP (MD = − 8.3 mmHg (95% CI -13.2 to − 3.4, p =0 .001). A significant improvement (p < 0.001) in hypertension in both groups was found by Shahid et al. [24](n = 440) but between-group differences were not reported. Eighteen studies reported data on diet adherence. Seven studies [22, 25, 31, 35, 39, 44, 51] observed be- tween group differences although using different instru- ments and scales. Anderson-Loftin et al. [37] used the Food Habits Questionnaire (FHQ) adapted for southern African Americans to measure dietary pattern. The intervention was a patient education program delivered by nurse case manager with nutrition focus combined with support groups, and weekly telephone follow-up. The authors re- ported a significant improvement in the experimental group with a decrease in high-fat diet while the control group continued previous high-fat dietary behaviors (MD =0.2 points, p = 0.005). One trial [20] used the Summary of Diabetes Self-care Activities Questionnaire (SDCA) to assess the nutrition adherence in Canadian Portuguese-speaking adults. There was an improvement in self-reported nutrition ad- herence at 3 months in favor of the experimental inter- vention (MD = 0.42 ± 0.14, p < 0.05, n = 87). Negarandeh [41] evaluated patient education program based on different format (Pictorial or teach back strategy) compared to usual care. Adherence to dietary pattern was measured through a self-structured nine-item scale. The score improved in all study participants (n =130) in follow up measurements but the improvement was more pro- Fig. 3 Risk of bias summary nounced for the intervention groups than the control group (p < 0.05). The mean difference between groups was − 2.24 (95% CI- 2.67 to-1.81) for the Pictorial format Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 8 of 18 group, and − 2.52 (95% CI:-2.95 to − 2.09) for the Teach strategies including interventions such as individualized back format group. case management activities [23], and culturally tailored A culturally tailored self-management intervention counseling delivered by a CHW [46, 47, 49] and/or adapted for a low income Latino group [27], improved NCM [18, 45], and promotoras [50]. Three RCTs in- the quality of diet as measured by the Alternative cluded additional activities, in particular home visits to Healthy Eating Index. Significant between group differ- support patient’s progress [30, 47, 48]. ences were found at 12 months (MD = 2.83 95% CI 0.58 Seven studies found a significantly greater reduction in to 5.08, p = 0.014, n = 252). HbA1c levels in the experimental group between base- A similar intervention was evaluated by Shahid et al. line and follow up. One study [51] evaluating individual [24] among people residing in rural areas in Pakistan. In culturally tailored care provided by NCM and CHW the intervention group there was a significant increase compared to minimal care, showed a significant decrease in the proportion of participants compliant to the diet in HBA1c levels. The effect was significant only in the plan (17.3% at baseline to 43.6% at follow up, p < 0.01) group of participants receiving a higher number of home while in the control group there was no significant in- visits (− 0.68% vs 0.43%, p = 0.03, n = 522). Another study crease (13.6% at baseline to 15.9% follow up, p = 0.522). conducted with Korean Americans immigrants [52] Weinstein’s trial [42] assessed fruit and vegetable con- found that a culturally tailored program including sumption self-reported daily following brief educational psycho-behavioral education, home glucose monitoring intervention. At 12 weeks, the percentage of participants with tele-transmission, and bilingual nurse telephone who reported ever purchasing from a produce market counseling, was associated with a greater improvement increased significantly in the intervention group (81% vs in HbA1c values (− 1.3% vs − 0.4%; p = 0.01, n = 79). 48%; p = 0.003, n = 79). Moreover, there was an overall A study conducted in a rural setting [53], showed an im- decrease of the percentage of participants reporting diffi- provement in HbA1c levels among patients exposed to culty affording fresh fruits and vegetables (55% vs 74% at diabetes education with interactive online sessions, deliv- baseline, p = 0.008). This decrease was not significantly ered by a multidisciplinary team (0.7 ± 1.3% vs 0.1 ± 1.0%; different between arms. p < 0.03 after adjustment for baseline HbA1c, n =95). Toobert et al. [39] reported the percent of calories A significant decrease of HbA1c was observed follow- from saturated fat measured using a food frequency ing a case management program delivered by a CHW questionnaire following a culturally adapted Mediterra- with the support of a clinical outreach team that in- nean lifestyle intervention. He found an improvement of cluded home visits [19](− 1.0% vs − 0.2%, p = 0.02, n = 0.33 points at the 24-month follow-up. 233). Lujan et al. [54] tested the effectiveness of a multi-component education program led by promotoras Provider level showing a mean change of HbA1c in the intervention Two studies evaluating reminder and reminder+feed- group significantly greater than that of the control group back interventions [43, 44] showed an improvement in at 6 months (p < 0.001, n = 149). glycemic control (HbA1c) compared to the usual care or A multicenter study [55] considered a composite out- no intervention group (0.6% vs 0.2%, p < 0.02, n = 399; come measure based on the achievement of target values MD = − 0.80 p < 0.001, n = 2046, respectively). Both of for HbA1c, SBP, and LDL. Participants assigned to the these interventions utilized computerized systems to intervention arm (health coaching group) showed higher produce physician reminders. One study [43] found an proportions of people reaching all clinical goals (46.4% improvement for LDL cholesterol for all intervention vs 34.3%, p = 0.02, n = 389) compared to usual care. arms, with the greater change observed in the reminders A study evaluated an education program [56] supervised +feedback group (− 18 mg/dl). No studies reported dif- by a nurse specifically trained for case management (DPP ferences between intervention and control arms for Lifestyle Program) where participants in the experimental blood pressure and BMI. group also received an evidence-based medication algo- rithm. The authors observed a significant improvement in Health care system level HbA1c levels in the experimental group compared to the The majority of studies that evaluated interventions tar- control (− 1.87% ± 0.81 vs − 0.54% ± 0.55; p = 0.011). How- geting the health care system (n = 20), showed significant ever, no information on sample size and participant char- effect in at least one of the outcomes considered in this acteristics were reported. review. Significant differences in blood pressure were found As far HbA1c, nine studies reported a significant re- between groups in three studies [16, 22, 45]. A difference duction of HbA1c values [18, 23, 30, 45–50] with a in means of change from baseline in diastolic blood mean difference ranging from − 0.29% to − 0.8%. The pressure significantly favored the intervention in a multi- studies considered a range of health care system-based center study [16] where participants received intensive Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 9 of 18 disease management led by practice nurse supported by Babamoto et al. [58] found that the proportion of patients link workers and a diabetes specialist (adjusted MD = − consuming two or more servings of fruits and vegetables 1.91 mmHg; p < 0.001, n = 1486). In the study of Hotu et daily increased significantly in the CHW and case manage- al. [22], Maori and Pacific patients with diabetes and ment groups but not in the standard provider care group. chronic kidney diseases who received twelve months of Patients’ self-reported intake of fatty foods decreased sig- home visits by a nurse, achieved a significant lower sys- nificantly from 29 to 16% (p < 0.05) in the CHW group but tolic blood pressure compared to usual care group remained unchanged in the other groups. (149 mmHg vs 140 mmHg; p < 0.05, n = 55). In a Cramer et al. [56] used the Dietary Questionnaire to long-term follow-up study [45] (60 months, n = 1665), a measure eating habits and observed a significant im- significant reduction in SBP (MD = − 4.32 mmHg, 95% provement in the experimental group compared with CI -6.72 to − 1.92] and DPB (MD = − 2.63 mmHg, 95% the usual care group (p < 0.001). Lynch et al. [59] also CI -3.74 to − 1.52] was detected among ethnically di- observed a significant increase in the number of days verse, medically underserved patients receiving a following a general and specific diet among participants self-management intervention with the support of home receiving a culturally-oriented self-management program telemedicine and a nurse case manager. (MD = 1.9, 95% CI 0.6 to 3.1; MD = 1.2, 95% CI 0.2 to Of the 14 trials reporting BMI outcome, only one [45] 2.2, respectively, n = 61), measured by the Block Food showed an adjusted MD of 0.40 kg/m (95% CI 0.20 to frequency Questionnaire. 0.60) when enhanced care through a diabetes-specialist Eight trials studied physical activity using different nurse and link worker were compared to usual care. measures, and two reported an effect following the ex- One [56] of the two studies reporting data on weight perimental intervention. One study [59] reported results change from baseline found a significant decrease at the from the CHAMPS (Community Healthy Activities end of the nine-month intervention of − 2.47 kg (±1.87) model for Seniors) physical activity questionnaire modi- in the experimental group and + 0.88 kg (±1.84) in the fied for use among African Americans. At study end- control group (p = 0.01). point there was a statistically significant difference Seventeen trials assessed the impact of QI interven- between groups (MD = 2.517 Kcal/week; p < 0.01). tions on total cholesterol and/or HDL cholesterol, LDL Comparing usual care with two educational programs cholesterol, and triglycerides. In three studies there were provided by a different case manager (CHW or NCM), significant differences in change from baseline between Babamoto et al. [58] found a significant improvement in groups. physical activity with an increase from 28 to 63% (p <0.05) At six months follow-up, Garcia et al. [57] reported in the CHW group, and from 17 to 35% (p < 0.05) in the statistically significant differences between the control standard provider care group, without any change in the and intervention group for total cholesterol (p = 0.003) case management group. and LDL cholesterol (p = 0.014), although not for triglyc- Six studies reported data on diabetes knowledge mea- erides (p = 0.179). sured by validated instruments such as the Diabetes Know- A significant effect on total cholesterol and triglycer- ledge Questionnaire [28, 54, 58], the Spoken Knowledge in ides was found in Kim et al. [52]. The intervention Low Literacy in Diabetes Scale [57], and the Diabetes group showed significantly lower levels of total choles- Knowledge Test [46, 52]. A significant improvement in pa- terol (− 24.7 mg/dl vs 7.2 mg/dl; p = 0.03) and triglycer- tient’sskillswas observedin three studies[46, 54, 58]. ide (− 84.6 mg/dL vs − 4.2 mg/dL; p < 0.05) when In one out of three studies considering emergency and/ compared with the control group. The intervention or hospital admissions [51, 58, 60], there was a reduction group also showed a trend toward a lower HDL, but this in emergency visits from baseline to 24 months among pa- difference was not statistically significant (p = 0.059). tients receiving a culturally tailored care provided by a In Shea et al. [45], the intervention group experienced NCM and a CHW (RR = 0.77, 95% CI, 0.59–1.00) [60]. net improvement in LDL cholesterol level relative to usual One study [45] investigated the effect of telemedicine care; a significant between groups difference was reported compared with usual care on all cause mortality but no at 5 years (MD = − 3.84; 95% CI -7.77 to − 0.08). differences between groups were reported (HR 1.01, 95% Glucose monitoring was considered in four studies CI 0.82, 1.24). [19, 36, 46, 48]. The study conducted by McDermott et al. [19] showed that participants in the control group (waiting-list group) were more likely to self-monitor Studies evaluating differential intervention effects by their glucose level than the experimental group. PROGRESS factors (n =7) Nine trials reported adherence to diet but measures Seven studies conducted sub-analyses to explore a differ- and scores used varied between trials. Three studies ential intervention effects across PROGRESS-Plus factors found a difference between groups. (n = 7) and all were conducted in developed countries. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 10 of 18 They used a parallel study design with a follow up of and ethnic minority participants but heterogeneity and 12–24 months. complexity of interventions made difficult to identify the ef- Table 2 gives the details of studies and results. Females, fective components of these interventions. The evidence on age ≥ 50, African-Americans and those with low education the effect of patient level interventions on improving other showed a better improvement in glycemic control. Patient clinical and laboratory parameters, such as blood pressure, education based on low-fat dietary strategies delivered by cholesterol levels and BMI, as well as self-management be- discussion groups and supported by phone contacts, pro- haviours is scant. Few studies explored the effectiveness of duced a greater decrease in BMI, weight, and dietary behav- other patient level strategies, including incentives and re- iors among women than men [37]. At healthcare minders. The only study included in this review [34]testing organization level, diabetes self-management supported by a rewards-based incentive intervention, showed effective CHW was associated with a greater BMI reduction and an results. increase in exercise frequency among participants aged With regard to interventions at provider level, only ≥50. One study analyzed intervention differential effect by one study reported a significant between groups differ- levels of health literacy [36]. The experimental program ence in HbA1c reduction while no significant impact on aimed to supply information and promote diabetes blood pressure or BMI was observed. self-management skills by computer multi-media including Many of the studies included in this systematic review audio/video sequences. Among low literacy subjects with were designed to evaluate the effectiveness of changing, poor glycemic control, the authors found a greater decrease expanding, or integrating the roles of healthcare profes- in HbA1C in the group exposed to computer multi-media sionals combined with patient education to improve dia- education program than in the control group (− 2.1 vs. betes care and outcomes. QI interventions based on -0.3%, p = 0.036). No significant difference was found multidisciplinary teams including trained nurses or local among high-literacy subjects. Moreover, the multimedia community health workers providing culturally compe- users with low health literacy demonstrated gains in know- tent care, were associated with a significant reduction of ledge, self-efficacy, and perceived susceptibility to complica- HbA1c values. Changes in the role of health care profes- tions compared with those having higher health literacy. sionals have been shown to produce an improvement in glucose control in ethnic minority communities on eth- Discussion nic minority communities showed. Applying an equity-oriented approach, this review iden- As far other primary outcomes considered in this re- tified 58 RCTs (17.786 participants) evaluating QI strat- view, a significant improvement in cholesterol levels was egies to improve the quality of diabetes care in a reported while n differences were found for secondary primary care setting. outcome measures, except for an increase in physical ac- Forty-seven studies were from USA and evaluated in- tivity and diabetes knowledge. terventions specifically designed to reach population Seven studies reported data on the differential effect subgroups mainly defined on the basis of race or ethni- by at least one PROGRESS factor. We did not find evi- city. A narrow subset of these studies (n = 7) considered dence of a differential effect by gender and race of any other dimensions of disadvantage as defined by the intervention on HbA1c levels reduction. One study re- PROGRESS framework, such as socio-economic status ported an improvement in glucose control among a low and place of residence. literacy population subgroup, exposed to a culturally The RCTs included in this systematic review covered a competent education program delivered through wide assortment of QI strategies, varying from multi-media tools. We found some evidence of effective- patient-mediated interventions with sessions of ness of QI interventions in weight loss and BMI among self-management supported by healthcare professionals, females and weight loss among African-Americans. to provider education and other more complex pro- In general, the heterogeneity of baseline HbA1c values grams based on changes in healthcare organization. and mean age of participants can affect intervention out- Twenty-nine studies considered QI interventions con- comes due to the biomedical challenge of lowering ducted at the patient level, three at the provider level, HbA1c from a higher baseline value. Moreover, some and twenty-six at the health care organization level. studies defined a minimum A1C value as inclusion cri- Pooling of results and quantitative synthesis was pre- terion possibly considering patients which may not be cluded by marked heterogeneity (mainly clinical), be- representative of diabetic population receiving care in a cause study population, types of interventions, outcome real world clinical setting. Rather than implementing measures, outcome assessment tools, duration of minimum A1C values for participant inclusion, as many follow-up and risk of bias varied widely between studies. of the studies reviewed incorporated, it is important (it QI strategies based on patient education and self-man- may be worthwhile) to maintain the integrity of studying agement strategies improved HbA1c levels among racial quality improvement interventions in real-life clinical Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 11 of 18 � � � � � � � � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Anderson 2010 Spanish speaking only, Patient level A1c, DBP,SBP, BMI, LDL, Subroups analysis No significant differences A1C [61] education level Number of experimental conditions: diet behavior (BDA); and interaction between groups for any Spanish speakers (yes vs no) USA 2 (1 intervention, 1 control) physical activity (RAPA); analysis outcomes MD = − 0.10(− 0.53, 0.33) vs 0.35 Intervention: depression measured Retention rate (− 0.17, 0.88) telephonic disease management Patient Health 79% vs 64% Educational level: (weekly, bi-weekly, or monthly) questionnaire (PHQ-9) (high level vs low level) based on: MD = 0.14(− 0.30, 0.57) vs 0.00 1. brief clinical assessment (− 0.52, 0.52) 2. self-management: including diet, None of the interactions was exercise, stress reduction, significant smoking cessation, readiness assessment, and development of specific self-management goals 3. medication adherence 4. glucose monitoring and review of home glucose monitoring results educational materials Personnel involved: nurse Control group: Usual care at Community Health Center Anderson-Loftin Gender Patient level A1c, BMI, LDL, weight, Stratification by A1c Men vs women 2005 [37] Number of experimental conditions: dietary fat behaviors gender No significant differences A1c USA 2 (1 intervention, 1 control) assessed by FHQ, physical Mean weight No significant differences Intervention: activity, psychological Significant effect Mean weight Education in low fat dietary status I: - 4 lb. Significant effect strategies (4 weekly classes) C: + 4.2 lb. + 5.4 lb. vs − 1.5 lb.; 1-h peer-professional discussion BMI MD = 6.9 lb. groups (5 monthly) I: − 0.81 kg/mm2 BMI Additional educational support by C: + 0.57 kg/mm2 + 2 kg/mm2 vs 0.16 kg/mm2 phone (weekly) MD = 1.38 kg/mm2 p = 0.02 Incentives for attendance p = 0.009 Dietary behaviors Personnel involved: nurse case Dietary behaviors (FHQ score) Significant effect manager I: 2.5 ± 0.4 (FHQ score) Control group: C: 2.6 ± 0.4 − 0.24 vs − 0.17 Usual care including a referral MD = 0.2 to a local 8-h traditional diabetes class p = 0.005 (information on nature and complications of diabetes) Incentives for attendance Babamoto 2009 Age Healthcare level BMI, A1C, medication Logistic regression Mean A1c Patients aged≥50 were less [58] Number of experimental conditions: adherence, diet, physical models Within group likely to have reduced BMI at USA 3 (2 intervention, 1 control) activity, emergency CHW = 8.6 to 7.2%; p < 0.05 follow-up Intervention: department admission CM =8.5 to 7.4%; p < 0.05 OR = 0.4 (95% CI = 0.2–0.8) Group A, CHW program, Amigos en (ED) Standard care = 9.5 to Exercise frequency 3 times Salud (Friends in Health): education 7.4%;p < 0.05 or more per week vs 2 times through individual session and No significant differences or fewer per week monitoring services; individual were found between groups OR = 2.2 (95% CI = 1.1–4.1) Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 12 of 18 � � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis sessions with participants and BMI Significantly greater family member; telephone calls to decrease for the CHW group participants to monitor self- compared with the standard management, to help participants care group OR = 2.9 improve their diabetes self- (95% CI 1.1–6.6) management skills ED Group B, case management: Change from baseline CHW: education from two linguistically total visit decrease 11% competent and culturally sensitive. Case management: total visit Patients case management were increase 40% Standard care: usually seen on a monthly basis + increase 15% follow-up calls. between groups at 6-month Personnel involved: bilingual, trained follow-up p < 0.05 community health workers, nurse Diet case manager CHW group were more likely Setting: Community, home, clinic (OR = 2.43; 95% CI =1. Control group 13–5.23) to report having Standard Provider Care: standardized two or more servings of fresh clinical care by physicians and nurse fruit per day than standard practitioners, without case care management or CHW services Physical activity CHW group was more likely (OR = 2.87, 95% CI = 1.34– 6.17) than standard care to report exercising three or more times per week Brown, 2011 Gender Patient level A1c, FBG, lipids, BP, BMI, Interaction terms Over time, both the FBG, BMI: [63] Number of experimental conditions: diabetes-related in hierarchical experimental and control No significant differences between USA 2 (1 intervention, 1 control) knowledge, health linear and groups showed gender Intervention: behaviors (physical nonlinear models improvements in FBG levels The rate of change in A1c over time Diabetes self-management activity, dietary intake, to test for at three and did not differ significantly by gender education (DSME) including 8 glucose monitoring) differential impact At six months (coefficient^ = − 0.06, t ratio = 0.25, consecutive weeks of education of treatment by For A1c the control group p = 0.806) followed by a support group gender had greater clinical session at 3 and 6 months improvements at both Experienced NCM providing: intervals culturally tailored diabetes self- Self-reported physical activity management education; and fat intake individualized health guidance Improvement for both and assistance with overcoming experimental and control cultural and environmental barriers groups to improving health; guidance on locating, accessing, and navigating healthcare services; enhanced coordination of health care and communication with physicians and other healthcare providers Random observations visits Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 13 of 18 � � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Personnel involved: bilingual NCM, nurses, dietitians, and CHWs Control Group: DSME intervention only Forjuoh 2014 Race/ethnicity Patient level A1C, physical activity, Interaction terms BMI and BP: Modest A1c [64] Number of experimental conditions: BMI, BP, diet in multilevel reductions from baseline to Modest reductions occurred in A1c USA 4 (3 intervention, 1 control) models to test for 12 months of follow-up for from baseline to 12 months of Intervention: differential impact all four groups. follow-up for all/ethnic groups. Group A. self-management through of treatment by No significant difference for There was no significant difference personal digit assistant (PDA). Dia race/ethnicity other outcomes. in A1c change over time by betes Pilot Chronic Disease Self Self care activities: race/ethnicity. Management Program (CDSMP): Hispanic washing feet 6 week group education program significantly more than other to increase self efficacy racial/ethnic groups (P = 0.02) Group B. self-management through Retention rate: PDA CDSMP: 85%; PDA 64%, Group C. combination of A + B CDSMP + PDA 64%; Control Personnel involved: trained facilitator, 78% project coordinators Setting: outpatient clinic, community Control group: usual clinical diabetes care, along with patient education materials Gerber 2005 Health literacy Patient level A1c, BMI, BP, eye exam, Stratification by No significant differences for Lower literacy group [36] Number of experimental conditions: diabetes knowledge, self- level of health all outcomes but perceived % change A1c USA 2 (1 intervention, 1 control) efficacy, self-reported literacy susceptibility to diabetes − 0.21 ± 2.0 vs − 0.1% ± 1.3 Intervention: medical care, and per- complications MD = − 0.10 [− 0.67, 0.47] Education by computer multi-media ceived susceptibility to People with A1c > 9% including audio/video sequences complications − 2.1 vs − 0.3 (p = 0.036) (“Living Well with Diabetes”)to Perceived susceptibility to communicate information, provide complications psychosocial support and promote % change score= self-management. Subject received 1.48 ± 2.7 vs 0.19 ± 2.5 (p = 0.016) compensation based on computer Self-efficacy usage. Lessons in English and trend toward greater improvement Spanish. Navigation provided in self-efficacy through a simplified interface, 1.51 ± 1.5 vs. 0.99 ± 1.4 including forward/backward buttons (p = 0.113) for user control. Advanced features Higher literacy included “pop-up” supplementary % change A1c text information or additional + 0.3% ± 1.6 vs. -0.5 ± 1.5 testimonials related to the MD = 0.80 [0.22, 1.38] concurrent screen concept Perceived susceptibility to Personnel involved: bilingual complications research assistant 0.76 ± 2.5 vs. 0.29 ± 2.4 (p = 0.267) Setting: urban outpatient clinics Medical care Improvement over Control group: simple multiple- time (p < 0.012 for time interaction) choice quizzes on diabetes-related but no effect for either lower- or concepts higher-literacy groups Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 14 of 18 � � Table 2 Evidence synthesis on differential effect analyses by PROGRESS-Plus factors (Continued) Study, country PROGRESS-factor Intervention type Outcome Method of Overall intervention effect Differential effect analysis Sixta 2008 [28] age Healthcare level A1C, knowledge, beliefs Stratified analysis A1C, knowledge, beliefs A1C USA Number of experimental conditions: by age No difference between groups 2 (1 intervention, 1 control) DKQ, HBQ. Intervention: No difference between groups Diabetes culturally self-management DKQ, HBQ, and HbA1c results were education with group sessions significantly affected by age; Personnel involved: promotores in Slightly negative effect on DKQ consultation with a care team scores per year of age. Control group:Usual care delivered Slightly negative effect on HBQ by provider at the clinic or to a scores and HbA1c levels per year self-care of age management West 2007 [70] Race/ethnicity Patient level A1C, glucose monitoring The weight Weight Weight at 6 months regardless USA Number of experimental conditions: patterns over time At 6 months treatment: 2 (1 intervention, 1 control) by race were Means: − 4.7 ± 5.4 kg vs − African-American vs White Intervention: examined using 3.1 ± 3.9 kg (p = 0.03) -3 kg ± 3.9 vs. -4.5 ± 5.1 kg 42 group session of behavioral a two-factor Over 18 months: (p = 0.03) weight control program focusing repeated Means: − 3.5 ± 6.8 Kg vs − Weight at 12 months regardless on attainable and sustainable measures ANOVA 1.7 ± 5.7Kg (p = 0.04) treatment:: changes in dietary and physical stratified by A1C − 2.3 kg ± 4.4 vs − 4.6 ± 6.8 kg activity habits treatment Decrease in both groups (p = 0.09) Motivational interviewing: 5 (p < 0.0001) at 6 months Weight at 18 months regardless individual sessions lasted 45 min but not sustained at treatment: Personnel involved: Behaviorist, 18 months − 1.4 kg ± 4.7 vs − 3.3 ± 7.1 kg nutritionist, diabetes educator, Greater decrease in the (p = 0.09) trained clinical psychologist intervention than in the For African-American experimental Setting: outpatient clinic control group (p = 0.002) intervention produced greater Control group: health education weight loss than control group at 3 sessions with focus on women’s and 6 months. The benefit was not health topics sustained after 12 months A1c African American had high A1c values regardless of treatment assignment. No interaction by race Attendance between groups was comparable. Data are means ± SD; I intervention group, C control group, OR odds ratio, A1c, Glycated hemoglobin; BMI Body Mass Index, LDL low density cholesterol, BP blood pressure, SBP systolic blood pressure, DBP diastolic blood pressure, MD mean difference, FHQ food habit questionnaire, PHQ-9 Patient Health Questionnaire, DSME Diabetes self-management education, DKQ diabetes knowledge questionnaire, HBQ Health Beliefs Questionnaire a b multivariate analysis adjusted for study group, gender, dietary, exercise activity; univariate analysis (did not persist after the other covariates were controlled for); ^b = regression coefficient Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 15 of 18 settings and therefore address differences in baseline publication. The issue of small sample size extends be- A1C values across studies in ways other than restricting yond the quality of those studies included in this review. patient participant inclusion. There were a number of studies, both pilot and not, that Another relevant issue in the evaluation of QI strat- were excluded from this review because they had a sam- egies is that the control groups received a wide range of ple size smaller than 50. Furthermore, since most studies interventions, from basic education materials, usual care, were carried out in USA, their degree of external validity to individualized coaching from community health is uncertain. Results from these studies may be less workers. Furthermore, in many of these studies, the con- transferrable to other countries and settings due to their trol group intervention was not described in detail. This being tested in a market-based health care system. It is is important as the usual or routine care in different set- likely that the patients’ population covered by universal- tings varies by a multitude of variables including pay- istic care is more heterogeneous with regard to ment system, geographic location, country, and more socio-demographic and clinical characteristics. For ex- generally, the resources and quality of services routinely ample, those countries with universal health care sys- provided to patients. In addition, type and quality of tems may have more heterogeneous patient populations usual care at a health center can impact baseline values, in a single community. It is therefore necessary to plan especially HbA1c. Moreover, biases may exist depending trials in other countries. By the same token, interven- on previous improvement activities implemented and tions addressing health disparities in other countries are general commitment of medical staff and organizational likely to involve groups of varying social advantage or leadership to reducing disparities and improving care. disadvantage being served under the same health center The conclusions of this systematic review are largely or system. The approach to addressing inequity becomes in accord with those in a previous review on this topic more about reducing health disparities on a more granu- among socially disadvantaged population living in indus- lar level requiring tools such as health equity audit. trialized countries published in 2006 [74]. The review Although the PROGRESS framework provides a vast identified 17 studies, seven trials were with low SES pop- array of disadvantage categories, there was limited het- ulations, and ten focused on etno-racial groups. The erogeneity in the dimensions of disadvantage considered small number of studies in Glazier’s review provided in RCTs. The most common PROGRESS factor were age limited and inconclusive evidence on intervention attri- and race/ethnicity, this underlines the needs of further butes that improved diabetes quality of care and health research with a focus on other characteristics such as so- outcomes, underlining the potential effect of some fea- cioeconomic status, social capital, place of residence, oc- tures in reducing health disparities. cupation, education, and religion. Researchers studying Our review provides an update and a more complete populations at social disadvantage must also describe the overview of the available evidence considering three spe- study population and the nature of their disadvantage cific aspects: use of PROGRESS framework to capture more specifically. This is of further importance because different socio-economic dimensions; assessment of the a lack of description or definition of a socially disadvan- risk of bias of included studies; and the inclusion of taged group was a common reason for study exclusion studies evaluating QI strategies defined according to in this review and others. international classification. There is also a clear need for more RCTs at the provider Using an equity oriented approach, we identified a level, especially those evaluating interventions based on large number of randomized studies showing that con- computerized provider reminder systems. With the wide- siderable strides have been made to test interventions to spread uptake of recognition and certification programs in address health inequities in diabetes care and outcomes. primary care (e.g. medical home, diabetes recognition pro- Despite the increase of the number of trials, the meth- grams,), it is likely that audit and feedback strategies using odological quality resulted to be low. This finding is con- benchmarking are common among primary care practices, sistent with a previous review [75] reporting that the but are less frequently reported for effectiveness among increase in the number of RCTs on QI strategies runs disadvantaged patient populations. parallel to the proportion of trials having at least one do- This research reveals an overall lack of focus on inter- main with high risk of bias. Most included trials did not ventions that address outcomes related to adherence to report the method of randomization and description of guidelines where disparities are stark according to the the allocation process. The area of the greatest potential literature. The paucity of studies measuring process of risk of bias was the inadequate blinding of participants care may be a reflection of the few number of QI inter- and outcome assessors, and poor follow up. In some of ventions at the provider level who, in conjunction with included trials the general lack of reporting of methods other members of the primary care team, are responsible made it difficult to assess methodological quality and for performing or referring to these services. Clinical thereby judge risk of bias, independently of year of outcomes should derived from electronic health record Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 16 of 18 systems, but may not be as recurrently funded as bio- Additional file 3: Table S2. Quality improvement strategies: level and chemical diabetes outcomes. Process outcomes or adher- description. (DOCX 13 kb) ence to guidelines is crucial to measure and address due Additional file 4: Table S3. Characteristics of eligible studies assessing the efficacy of QI interventions in participants with type 2 diabetes. to the evidence of disparities that exist on the level of (DOCX 71 kb) clinical quality and care. It is also important to note that several studies measured diabetes “self-care” or “self-- Authors’ contributions management” activities but did not report results on dis- NT, AB, and NA made substantial contributions to the conception and tinct components such as medication adherence or design of this systematic review. ZM completed the literature search. NT and AB screened studies against eligibility criteria, extracted data, and analysed glucose monitoring. As these clinical outcome measures and interpreted data. All authors contributed to writing and revising the final are crucial in measuring effectiveness of diabetes inter- manuscript. All authors read and approved the final manuscript LA and MD vention, it is important to report on these components contributed to the critical revision. as distinctive measures. Ethics approval and consent to participate We see many studies that aim to evaluate interven- Not applicable tions to improve care and/or outcomes among a disad- vantaged group, but seldom do we find studies Competing interests The authors declare that they have no competing interests. investigating the effect of QI interventions disentangled by different levels of indicators of socio-economic pos- Publisher’sNote ition or relevant socio-demographic factors. This may Springer Nature remains neutral with regard to jurisdictional claims in because practices are not disaggregating data to identify published maps and institutional affiliations. disparities within patient populations and are therefore Author details not initiating action to address them. It should be neces- 1 2 Trenton Health Team, Trenton, New Jersey, USA. Department of sary to promote and sustain a different approach includ- Epidemiology, Lazio Region- ASL Rome1, Rome, Italy. ing audit activities to identify inequities in care and Received: 28 February 2018 Accepted: 16 May 2018 outcomes, and then work to address these disparities. Moreover, an “equity lens” approach should be adopted by the scientific community when identifying research References priorities aimed at contrasting socioeconomic differen- 1. World Health Organization. Global Report on Diabetes 2016. http://apps.who. int/iris/bitstream/handle/10665/204871/9789241565257_eng.pdf;jsessionid= tials. This equity-oriented approach is necessary to iden- 45F529CEFB7FB49CB7EE39B0F63BA11F?sequence=1. Accessed 20 Feb 2016. tify and describe the appropriate target population, to 2. Agardh E, Allebeck P, Hallqvist J, Moradi T, Sidorchuk A. Type 2 diabetes define inequalities indicators, and select process and incidence and socio-economic position: a systematic review and meta- analysis. Int J Epidemiol. 2011;40:804–18. outcome indicators useful for assessing the differential 3. Redle EE, Atkins D. The applicability of quality improvement research for effect of an intervention. comparative effectiveness. Implement Sci. 2013;8(Suppl 1):S6. 4. Espelt A, Arriola L, Borrell C, Larrañaga I, Sandín M, Escolar-Pujolar A. Socioeconomic position and type 2 diabetes mellitus in Europe 1999-2009: Conclusions a panorama of inequalities. Curr Diabetes Rev. 2011;7(3):148–58. 5. NICE. Type 2 diabetes in adults: management 2015. https://www.nice.org. Because of the methodological differences and weak- uk/guidance/ng28/evidence/full-guideline-pdf-78671532569. Accessed 20 nesses that precluded meta-analytic synthesis, we can Feb 2016. draw no strong conclusions concerning the potential 6. Grintsova O, Maier W, Mielck A. Inequalities in health care among patients with type 2 diabetes by individual socio-economic status (SES) and regional benefits or harms of QI strategies to reduce inequalities deprivation: a systematic literature review. Int J Equity Health. 2014;13:43. in access to care for patients with diabetes in primary https://doi.org/10.1186/1475-9276-13-43. care. Moreover, the included studies did not allow for an 7. Heidemann DL, Joseph NA, Kuchipudi A, Perkins DW, Drake S. Racial and economic disparities in Diabetes in a Large Primary Care Patient Population. analysis of the differential effects of interventions across Ethn Dis. 2016;26(1):85–90. https://doi.org/10.18865/ed.26.1.85. population sub-groups. 8. Calvert M, Shankar A, McManus RJ, Lester H, Freemantle N. Effect of the This review highlights some QI strategies for consider- quality and outcomes framework on diabetes care in the United Kingdom: retrospective cohort study. BMJ. 2009;338:b1870. https://doi.org/10.1136/ ation and in need of further study. Health care profes- bmj.b1870. Erratum in: BMJ. 2009;339:b2768 sionals and policy makers need the best available evidence 9. Rossi MC, Candido R, Ceriello A, Cimino A, Di Bartolo P, Giorda C, et al. to administer and support those interventions most likely Trends over 8 years in quality of diabetes care: results of the AMD annals continuous quality improvement initiative. Acta Diabetol. 2015;52(3):557–71. to be effective to reduce disparities in diabetes care. https://doi.org/10.1007/s00592-014-0688-6. 10. Tricco AC, Ivers NM, Grimshaw JM, Moher D, Turner L, Galipeau J, et al. Effectiveness of quality improvement strategies on the management of Additional files diabetes: a systematic review and meta-analysis. Lancet. 2012;379(9833): 2252–61. https://doi.org/10.1016/S0140-6736(12)60480-2. 11. O'Neill J, Tabish H, Welch V, Petticrew M, Pottie K, Clarke M, et al. Applying Additional file 1: Search strategy for PubMed. (DOCX 17 kb) an equity lens to interventions: using PROGRESS ensures consideration of Additional file 2: Table S1. Inclusion and exclusion criteria (PICOS). socially stratifying factors to illuminate inequities in health. J Clin Epidemiol. (DOCX 14 kb) 2014;67(1):56–64. https://doi.org/10.1016/j.jclinepi.2013.08.005. Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 17 of 18 12. Welch V, Petticrew M, Petkovic J, Moher D, Waters E, White H, et al. PRISMA- randomized controlled trial of a community health worker intervention for equity Bellagio group. Extending the PRISMA statement to equity-focused Mexican Americans with type 2 diabetes mellitus. Am J Public Health. 2014; systematic reviews (PRISMA-E 2012): explanation and elaboration. J Clin 104(8):1540–8. Epidemiol. 2016;70:68–89. https://doi.org/10.1016/j.jclinepi.2015.09.001. 31. Palmas W, Findley SE, Mejia M, Batista M, Teresi J, Kong J, et al. Results of 13. McPheeters ML, Kripalani S, Peterson NB, Idowu RT, Jerome RN, Potter SA, the northern Manhattan diabetes community outreach project: a et al. Quality improvement interventions to address health disparities. In: randomized trial studying a community health worker intervention to Closing the quality gap: revisiting the state of the science. Evidence report improve diabetes care in hispanic adults. Diabetes Care. 2014;37(4):963–9. no. 208. (prepared by the Vanderbilt University evidence-based practice 32. Walker EA, Shmukler C, Ullman R, Blanco E, Scollan-Koliopoulus M, Cohen center under contract no. 290–2007-10065.) AHRQ publication no. 12-E009- HW. Results of a successful telephonic intervention to improve diabetes EF. Rockville: Agency for Healthcare Research and Quality; 2012. https:// control in urban adults: a randomized trial. Diabetes Care. 2011;34(1):2–7. effectivehealthcare.ahrq.gov/sites/default/files/pdf/disparities-quality- 33. Davis RM, Hitch AD, Salaam MM, Herman WH, Zimmer-Galler IE, Mayer-Davis improvement_research.pdf. EJ. TeleHealth improves diabetes self-management in an underserved 14. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. community: diabetes TeleCare. Diabetes Care. 2010;33(8):1712–7. Cochrane Bias methods group; Cochrane Statistical Methods Group The 34. Philis-Tsimikas A, Fortmann A, Lleva-Ocana L, Walker C, Gallo LC. Peer-led Cochrane Collaboration's tool for assessing risk of bias in randomised trials. diabetes education programs in high-risk Mexican Americans improve BMJ. 2011;343:d5928. https://doi.org/10.1136/bmj.d5928. glycemic control compared with standard approaches: a project Dulce 15. RevMan 2014. The Nordic Cochrane Centre, The Cochrane Collaboration. promotora randomized trial. Diabetes Care. 2011;34(9):1926–31. Review Manager (RevMan). Version 5.3. Copenhagen: The Nordic Cochrane 35. Brown SA, Blozis SA, Kouzekanani K, Garcia AA, Winchell M, Hanis CL. Centre, The Cochrane Collaboration; 2014. Dosage effects of diabetes self-management education for Mexican 16. Bellary S, O'Hare JP, Raymond NT, Gumber A, Mughal S, Szczepura A, et al. Americans: the Starr County border health initiative. Diabetes Care. 2005; Enhanced diabetes care to patients of south Asian ethnic origin (the United 28(3):527–32. Kingdom Asian Diabetes Study): a cluster randomised controlled trial. 36. Gerber BS, Brodsky IG, Lawless KA, Smolin LI, Arozullah AM, Smith EV, et al. Lancet. 2008;371(9626):1769–76. Implementation and evaluation of a low-literacy diabetes education 17. Berry DC, Williams W, Hall EG, Heroux R, Bennett-Lewis T. Imbedding computer multimedia application. Diabetes Care. 2005;28(7):1574–80. interdisciplinary diabetes group visits into a community-based medical 37. Anderson-Loftin W, Barnett S, Bunn P, Sullivan P, Hussey J, Tavakoli A. Soul setting. Diabetes Educ. 2016;42(1):96–107. food light: culturally competent diabetes education. Diabetes Educ. 2005; 18. DePue JD, Rosen RK, Seiden A, Bereolos N, Chima ML, Goldstein MG, et al. 31(4):555–63. Implementation of a culturally tailored diabetes intervention with 38. Tang TS, Funnell MM, Sinco B, Spencer MS, Heisler M. Peer-led, community health workers in American Samoa. Diabetes Educat. 2013;39(6): empowerment-based approach to self-management efforts in diabetes 761–71. (PLEASED): a randomized controlled trial in an African American community. 19. McDermott RA, Schmidt B, Preece C, Owens V, Taylor S, Li M, et al. Ann Fam Med. 2015;13(Suppl 1):S27–35. Community health workers improve diabetes care in remote Australian 39. Toobert DJ, Strycker LA, King DK, Barrera M Jr, Osuna D, Glasgow RE. Long- indigenous communities: results of a pragmatic cluster randomized term outcomes from a multiple-risk-factor diabetes trial for Latinas: -íViva controlled trial. BMC Health Serv Res. 2015;15:68. Bien! Transl Behav Med. 2011;1(3):416–26. 20. Gucciardi E, DeMelo M, Lee RN, Grace SL. Assessment of two culturally 40. Tang TS, Funnell M, Sinco B, Piatt G, Palmisano G, Spencer MS, Kieffer EC, competent diabetes education methods: individual versus individual plus Heisler M. Comparative effectiveness of peer leaders and community health group education in Canadian Portuguese adults with type 2 diabetes. Ethn workers in diabetes selfmanagement support: results of a randomized Health. 2007;12(2):163–87. controlled trial. Diabetes Care. 2014;37(6):1525–34. 21. Guo H, Tian X, Li R, Lin J, Jin N, Wu Z, Yu D. Reward-based, task-setting 41. Negarandeh R, Mahmoodi H, Noktehdan H, Heshmat R, Shakibazadeh E. education strategy on glycemic control and self-management for low-income Teach back and pictorial image educational strategies on knowledge about outpatients with type-á2 diabetes. J Diabetes Investig. 2014;5(4):410–7. diabetes and medication/dietary adherence among low health literate 22. Hotu C, Collins J, Harwood L, Whalley G, Doughty R, Gamble G, Braatvedt G. patients with type 2 diabetes. Prim Care Diabetes. 2013;7(2):111–8. A community-based model of care improves blood pressure control and 42. Weinstein E, Galindo RJ, Fried M, Rucker L, Davis NJ. Impact of a focused delays progression of proteinuria, left ventricular hypertrophy and diastolic nutrition educational intervention coupled with improved access to fresh dysfunction in Maori and Pacific patients with type 2 diabetes and chronic produce on purchasing behavior and consumption of fruits and vegetables kidney disease: a randomized controlled trial. Nephrol Dial Transplant. 2010; in overweight patients with diabetes mellitus. Diabetes Educat. 2014;40(1): 25(10):3260–6. 100–6. 23. Partapsingh VA, Maharaj RG, Rawlins JM. Applying the Stages of Change 43. Phillips LS, Ziemer DC, Doyle JP, Barnes CS, Kolm P, Branch WT, et al. An model to Type 2 diabetes care in Trinidad: A randomised trial. J Negat endocrinologist-supported intervention aimed at providers improves Results Biomed. 2011;10(1):13. diabetes management in a primary care site: improving primary care of 24. Shahid M, Mahar SA, Shaikh S, Shaikh Z-U-D. Mobile phone intervention to African Americans with diabetes (IPCAAD) 7. Diabetes Care. 2005;28(10): improve diabetes care in rural areas of Pakistan: a randomized controlled 2352–60. trial. J Coll Physicians Surg Pak. 2015;25(3):166–71. 44. Welch G, Zagarins SE, Santiago-Kelly P, Rodriguez Z, Bursell SE, Rosal MC, 25. Frosch DL, Uy V, Ochoa S, Mangione CM. Evaluation of a behavior support et al. An internet-based diabetes management platform improves team intervention for patients with poorly controlled diabetes. Arch Intern Med. care and outcomes in an urban Latino population. Diabetes Care. 2015; 2011;171(22):2011–7. 38(4):561–7. 26. Khan MA, Shah S, Grudzien A, Onyejekwe N, Banskota P, Karim S, et al. A 45. Shea S, Weinstock RS, Teresi JA, Palmas W, Starren J, Cimino JJ, et al. diabetes education multimedia program in the waiting room setting. IDEATel consortium. A randomized trial comparing telemedicine case Diabetes Ther. 2011;2(3):178–88. https://doi.org/10.1007/s13300-011-0007-y. management with usual care in older, ethnically diverse, medically Epub 2011 Aug 22 underserved patients with diabetes mellitus: 5 year results of the IDEATel 27. Rosal MC, Ockene IS, Restrepo A, White MJ, Borg A, Olendzki B, et al. study. J Am Med Inform Assoc. 2009;16(4):446–56. Randomized trial of a literacy-sensitive, culturally tailored diabetes self- 46. Kim MT, Kim KB, Huh B, Nguyen T, Han H-R, Bone LR, et al. The effect of a management intervention for low-income latinos: latinos en control. community-based self-help intervention: Korean Americans with type 2 Diabetes Care. 2011;34(4):838–44. diabetes. Am J Prev Med. 2015 Nov;49(5):726–37. 28. Sixta CS, Ostwald S. Texas-Mexico border intervention by promotores for 47. Perez-Escamilla R, Damio G, Chhabra J, Fernandez ML, Segura-Perez S, Vega- patients with type 2 diabetes. Diabetes Educ. 2008;34(2):299–309. Lopez S, et al. Impact of a community health workers-led structured 29. Hill-Briggs F, Lazo M, Peyrot M, Doswell A, Chang YT, Hill MN, et al. Effect of program on blood glucose control among Latinos with type 2 diabetes: problem-solving-based diabetes self-management training on diabetes The DIALBEST Trial. Diabetes Care. 2015;38(2):197–205. // () *National Center control in a low income patient sample. J Gen Intern Med. 2011;26(9):972–8. for Advancing Translational Sciences* 30. Rothschild SK, Martin MA, Swider SM, Tumialán Lynas CM, Janssen I, Avery 48. Spencer MS, Rosland AM, Kieffer EC, Sinco BR, Valerio M, Palmisano G, et al. EF, et al. Mexican American trial of community health workers: a Effectiveness of a community health worker intervention among African Terens et al. BMC Endocrine Disorders (2018) 18:31 Page 18 of 18 American and Latino adults with type 2 diabetes: a randomized controlled 68. Skelly AH, Carlson JR, Leeman J, Holditch-Davis D, Soward AC. Symptom- trial. Am J Public Health. 2011;101(12):2253–60. focused management for African American women with type 2 diabetes: a 49. Prezio EA, Cheng D, Balasubramanian BA, Shuval K, Kendzor DE, Culica D. pilot study. J Nurs Scholarsh. 2005;18(4):213–20. Community diabetes education (CoDE) for uninsured Mexican Americans: a 69. Wayne N, Perez DF, Kaplan DM, Ritvo P. Health coaching reduces HbA1c in randomized controlled trial of a culturally tailored diabetes education and type 2 diabetic patients from a lower-socioeconomic status community: a management program led by a community health worker. Diabetes Res randomized controlled trial. J Med Internet Res. 2015;17(10):e224. Clin Pract. 2013;100(1):19–28. https://doi.org/10.1016/j.diabres.2013.01.027. 70. West DS, DiLillo V, Bursac Z, Gore SA, Greene PG. Motivational interviewing improves weight loss in women with type 2 diabetes. Diabetes Care. 2007; 50. Thom DH, Ghorob A, Hessler D, Vore D, Chen E, Bodenheimer TA. Impact of 30(5):1081–7. peer health coaching on glycemic control in low-income patients with 71. Seligman HK, Wang FF, Palacios JL, Wilson CC, Daher C, Piette JD, et al. diabetes: a randomized controlled trial. Ann Fam Med. 2013;11(2):137–44. Physician notification of their diabetes patients' limited health literacy. A 51. Gary TL, Batts-Turner M, Yeh HC, Hill-Briggs F, Bone LR, Wang NY, et al. The randomized, controlled trial. J Gen Intern Med. 2005;20(11):1001–7. effects of a nurse case manager and a community health worker team on 72. Ruggiero L, Moadsiri A, Butler P, Oros SM, Berbaum ML, Whitman S, et al. diabetic control, emergency department visits, and hospitalizations among Supporting diabetes self-care in underserved populations: a randomized urban African Americans with type 2 diabetes mellitus: a randomized pilot study using medical assistant coaches. Diabetes Educ. 2010;36(1):127–31. controlled trial. Arch Intern Med. 2009;169(19):1788–94. 73. Willard-Grace R, Chen EH, Hessler D, DeVore D, Prado C, Bodenheimer T, 52. Kim MT, Han HR, Song HJ, Lee JE, Kim J, Ryu JP, et al. A community-based, et al. Health coaching by medical assistants to improve control of diabetes, culturally tailored behavioral intervention for Korean Americans with type 2 hypertension, and hyperlipidemia in low-income patients: a randomized diabetes. Diabetes Educat. 2009;35(6):986–94. controlled trial. Ann Fam Med. 2015;13(2):130–8. 53. Liou J-K, Soon M-S, Chen C-H, Huang T-F, Chen Y-P, Yeh Y-P, et al. Shared 74. Glazier RH, Bajcar J, Kennie NR, Willson K. A systematic review of care combined with telecare improves glycemic control of diabetes interventions to improve diabetes care in socially disadvantaged patients in a rural, underserved community. J Diabetes Sci Technol. 2013; populations. Diabetes Care. 2006;29(7):1675–88. 7(1):A82. 75. Ivers NM, Tricco AC, Taljaard M, Halperin I, Turner L, Moher D, et al. Quality 54. Lujan J, Ostwald SK, Ortiz M. Promotora diabetes intervention for Mexican improvement needed in quality improvement randomised trials: systematic Americans. Diabetes Educ. 2007;33(4):660–70. review of interventions to improve care in diabetes. BMJ Open. 2013;3(4): 55. Levy NK, Moynihan V, Nilo A, Singer K, Etiebet M-A, Bernik L, et al. The e002727. https://doi.org/10.1136/bmjopen-2013-002727. mobile insulin titration intervention (MITI) study: innovative chronic disease management of diabetes. J Gen Intern Med. 2015;30:S547–S8. 56. Cramer JS, Sibley RF, Bartlett DP, Kahn LS, Loffredo L. An adaptation of the diabetes prevention program for use with high-risk, minority patients with type 2 diabetes. Diabetes Educat. 2007;33(3):503–8. 57. Garcia AA, Brown SA, Horner SD, Zuniga J, Arheart KL. Home-based diabetes symptom self-management education for Mexican Americans with type 2 diabetes. Health Educ Res. 2015;30(3):484–96. 58. Babamoto KS, Sey KA, Camilleri AJ, Karlan VJ, Catalasan J, Morisky DE. Improving diabetes care and health measures among hispanics using community health workers: results from a randomized controlled trial. Health Educ Behav. 2009;36(1):113–26. 59. Lynch EB, Liebman R, Ventrelle J, Avery EF, Richardson D. A self- management intervention for African Americans with comorbid diabetes and hypertension: a pilot randomized controlled trial. Prev Chronic Dis. 2014;11:E90. 60. Tobe SW, Pylypchuk G, Wentworth J, Kiss A, Szalai JP, Perkins N, et al. Effect of nurse-directed hypertension treatment among first nations people with existing hypertension and diabetes mellitus: the diabetes risk evaluation and microalbuminuria (DREAM 3) randomized controlled trial. CMAJ. 2006; 174(9):1267–71. 61. Anderson DR, Christison-Lagay J, Villagra V, Liu H, Dziura J. Managing the space between visits: a randomized trial of disease management for diabetes in a community health center. J Gen Intern Med. 2010;25(10):1116–22. 62. Baradaran HR, Knill-Jones RP, Wallia S, Rodgers A. A controlled trial of the effectiveness of a diabetes education programme in a multi-ethnic community in Glasgow [ISRCTN28317455]. BMC Public Health. 2006;6:134. 63. Brown SA, García AA, Winter M, Silva L, Brown A, Hanis CL. Integrating education, group support, and case management for diabetic Hispanics. Ethn Dis. 2011;21(1):20–6. 64. Forjuoh SN, Bolin JN, Huber Jr JC, Vuong AM, Adepoju OE, Helduser JW, et al. Behavioral and technological interventions targeting glycemic control in a racially/ethnically diverse population: a randomized controlled trial. BMC Public Health. 2014;14:71. 65. Gregg JA, Callaghan GM, Hayes SC, Glenn-Lawson JL. Improving diabetes self-management through acceptance, mindfulness, and values: a randomized controlled trial. J Consult Clin Psychol. 2007;75(2):336–43. 66. Heisler M, Choi H, Palmisano G, Mase R, Richardson C, Fagerlin A, et al. Comparison of community health worker-led diabetes medication decision- making support for low-income Latino and African American adults with diabetes using e-health tools versus print materials: a randomized, controlled trial. Ann Intern Med. 2014;161(10 Suppl):S13–22. 67. Schillinger D, Hammer H, Wang F, Palacios J, McLean I, Tang A, et al. Seeing in 3-D: examining the reach of diabetes self-management support strategies in a public health care system. Health Educ Behav. 2008;35(5): 664–82.

Journal

BMC Endocrine DisordersSpringer Journals

Published: May 29, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off