Effectiveness of diabetes self-management education programs for US Latinos at improving emotional distress: A systematic review

Abstract

Purpose:

This systematic review examined whether diabetes self-management education (DSME) interventions for US Latino adults improve general emotional distress (e.g., depression symptoms) and/or health-specific emotional distress (e.g., diabetes distress). The topic is important given the high prevalence of type 2 diabetes (T2DM), concomitant distress, and worse health outcomes among Latinos, and considering the barriers that distress poses for effective diabetes self-management.

Methods:

Following PRISMA guidelines, a search of the online databases PsycINFO, CINAHL, PubMed, and CENTRAL was conducted from database inception through April 2018. A comprehensive search strategy identified trials testing DSME interventions for US Latinos with T2DM that reported on changes in general or health-specific emotional distress. Risk of bias was assessed using the EPHPP Quality Assessment Tool. Raw mean differences (D) and effect sizes (d) were computed where possible.

Results:

Fifteen studies were included in the review. Six of 8 studies that examined depression symptoms reported significant symptom reduction. Of 10 studies that examined health-specific emotional distress, 6 reported significant symptom reduction. Effect sizes ranged from −0.20 to −3.85. Null findings were more readily found among studies with very small sample sizes (n < 30), and among studies testing interventions without specific psychosocial content, with little cultural tailoring, with less frequent intervention sessions, and with support sessions lacking concurrent diabetes education. Most studies (11) received a weak rating of evidence quality.

Conclusions:

There is an absence of strong evidence to support that DSME programs tailored for Latino adults with T2DM are beneficial at improving emotional distress. Methodologically robust studies are needed.

Type 2 diabetes (T2DM) is the seventh most frequent cause of death¹ and affects over 20 million US adults.² US Latinos are disproportionally affected by T2DM, with a prevalence of 13% compared to 8% in non-Latino Whites.³ The landmark Hispanic Community Health Study/Study of Latinos (HCHS/SOL), a prospective cohort of 16,145 Latino adults, found a diabetes prevalence of 16.5% in Latino men and 17.1% in Latina women.⁴ Compared to non-Latino Whites, Latinos with diabetes also have poorer outcomes and higher mortality rates.^5,6 To foster adequate glycemic management and minimize complications, patients with T2DM need to follow multiple behavioral recommendations.⁷ Poor management of T2DM leads to faster disease progression and severe complications that can significantly impact quality and quantity of life.^8,9

General emotional distress (e.g., depression, anxiety) and health-specific emotional distress (e.g., diabetes distress) can impede effective diabetes self-management and adversely affect blood glucose levels.^10–12 A recent systematic review showed that 1 in 3 individuals living with diabetes reported moderate to high diabetes distress.¹³ Some studies have found depression symptoms to be highly prevalent and associated with poor glycemic management among Latinos with diabetes.^14–16 Furthermore, diabetes distress and general emotional distress appear to be more common among US Latinos with diabetes compared to their non-Latino White counterparts.^17–19

In December 2016 the American Diabetes Association (ADA) issued a “Position Statement on Psychosocial Care for People with Diabetes” calling for routine screenings and recommending diabetes self-management education (DSME) as a first level intervention for those who screen positive for emotional distress.²⁰ Further, two systematic reviews found evidence of improved emotional distress post-intervention in DSME programs.^21,22 Although research indicates that DSME programs for Latinos improve self-management behaviors and disease control,^23,24 less is known about their effects on emotional distress in this large and growing US population.

Concha et al²⁵ systematically reviewed the literature through 2008 examining multiple behavioral outcomes of DSME interventions in Latinos. The focus of this review was broad, addressing multiple behaviors, and giving limited attention to distress outcomes. Further, the review did not quantitatively evaluate the findings in respect to effect sizes of individual or aggregated studies. The review also lacked clarity on whether studies were examining DSME interventions or interventions for depression, and what specific intervention content targeted distress outcomes. Furthermore, since 2008, several new articles examining emotional distress outcomes in DSME programs among US Latinos have been published. Given the recent ADA recommendations, an updated and thorough review of the impact of DSME programs on emotional distress outcomes among US Latinos is needed.

Objectives

The primary research question this systematic review aimed to answer was: Are DSME interventions targeting diabetes self-care behaviors for adult US Latinos effective at improving general emotional distress and health-specific emotional distress? In the context of answering this research question, we also aimed to provide details regarding the components of the intervention content targeting emotional distress and the assessment methods used to measure change in emotional distress in this population. Finally, we sought to synthesize the findings to provide recommendations for future research on the effects of DSME interventions in reducing distress among Latinos.

Methods

Article Search and Selection

Procedures followed the Preferred Reporting Items for Systematic Reviews and Metal-Analyses (PRISMA) guidelines.²⁶ The following inclusion criteria were set a-priori: Studies (1) examined DSME interventions, using either a single-group pre-post design or multiple group controlled trial design, aimed at improving diabetes self-management for US Latino adults with T2DM, and (2) reported on changes in self-report measures of general emotional distress or health-specific emotional distress. For the purposes of this review we defined a DSME intervention as one that covered at least two of the target self-care behaviors recommended by the American Association of Diabetes Educators (AADE).^7,27 We did not require that the DSME interventions be accredited by the AADE because this would have reduced the already limited number of eligible studies and because most articles did not include information on accreditation. We defined an intervention aimed for US Latinos if it had a sample at least 50% Latino and if it provided some cultural tailoring (including Spanish language translation only). Intervention curriculum content specifically addressing emotional distress was not an inclusion requirement. Exclusion criteria were: (1) Studies conducted outside the US; (2) studies not published in English or Spanish; (3) participants did not have current T2DM (e.g., studies on gestational diabetes were excluded); (4) participants were younger than age 18 years; (5) studies did not report changes in general or health-specific emotional distress, (6) studies solely focused on improving depression without any DSME content, (7) studies examined co-located, integrated care models that addressed both depression and diabetes management (due to the inability to disentangle effects of psychotherapy/medication versus DSME), (8) studies published in gray literature or not peer-reviewed, (9) studies aimed at populations with serious comorbid conditions or with T2DM as only one possible eligible condition, and (10) studies of outcomes such as quality of life, since the focus was on improvement of emotional distress (rather than general mental and physical well-being).

A search of the online databases PsycINFO, CINAHL, PubMed, and the Cochrane Central Register of Controlled Trial (CENTRAL) was conducted from database inception until April 2018. The search strategy used comprehensive keywords to identify articles meeting review criteria [e.g., Hispanic (OR Latin* OR Mexican OR Puerto Rican OR Dominican OR Cuban OR Central American OR South American) AND diabetes (OR diabetic) AND intervention (OR treatment OR therapy OR trial OR randomized controlled trial OR RCT OR pilot study) AND depression (OR depressive OR anxiety OR distress OR well-being OR stress OR emotion]. Manual reference list searches were also conducted for selected articles and prior reviews. Articles were selected by 2 specially trained independent reviewers using a two-step approach: (1) the titles and abstracts of the articles were screened, and those deemed ineligible were removed; (2) the full-text of the remaining articles was reviewed. Discrepancies in article selection were discussed between the 2 reviewers, and with a third reviewer when needed, until consensus was reached.

Data Collection Process

Data collection was conducted by the 2 independent reviewers utilizing a spreadsheet that was created a-priori. Data extracted included study characteristics (e.g., sample size and demographics, study location), study methodology (e.g., sampling and recruitment methods, study design), intervention details (e.g., dosage, frequency, duration, mode of delivery, content), and outcome data (e.g., scores on self-report measures of emotional distress). Discrepancies between data extraction points were discussed between the two reviewers, and with a third reviewer when needed, until resolved. When articles included means and standard deviations, raw mean differences (D) and standardized effect sizes (d) were calculated, comparing either within-group pre- to post-intervention results (for non-randomized trials), or comparing between-group post-intervention results in the intervention vs the control group (for randomized controlled trials [RCTs]). Articles that did not report quantitative findings but collected the necessary data and summarized the findings in narrative form were included, and the narrative data were extracted. After all eligible articles were included and available data were extracted, attempts were made to contact the authors of the 4 articles with missing quantitative data; however, data could not be obtained. In such cases we reported the limited quantitative data provided in the articles and/or reported the findings in narrative form only. Given the heterogeneity in study design (e.g., intervention dosage) and in intervention content (e.g., extent of cultural tailoring), a meta-analysis of effect sizes was not conducted as per the Cochrane systematic review guidelines.²⁸

Risk of bias within each study was assessed by two independent reviewers using the Effective Public Health Practice Project Quality Assessment Tool for Quantitative Studies (EPHPP),²⁹ which was designed to assess bias in public health interventions and has been found to have good inter-rater reliability.³⁰ Each study was rated for quality of evidence on 6 possible domains: selection bias, study design, confounders, blinding, data collection methods, and withdrawals and drop-outs. Ratings range from 1 (strong), to 3 (weak). A final global rating is created by assigning the study a 1 (strong quality) if the study did not have any domains rated as a 3, assigning it a 2 (moderate quality) if the study had only one domain rated as a 3, and assigning it a 3 (weak quality) if the study had two or more domains that scored a 3. Discrepancies between quality ratings were discussed between the 2 reviewers, and with a third reviewer when needed, until consensus was reached.

Results

Article Selection and Characteristics

Figure 1 outlines the article selection process. Initial database searches yielded 1,367 articles and hand searching yielded an additional 7 articles. A total of 291 articles were removed as duplicates, leaving 1,083 articles to be screened for eligibility through title and abstract review. Out of these, 943 were excluded. The full text of 140 articles was reviewed, and 125 were excluded, for a final sample of 15 articles. Of note, during the full-text review phase, 5 studies had all but one of the eligibility criteria: they only reported on emotional distress at baseline and not post-intervention. Additionally, although Concha et al²⁵ reported 7 studies as having measured emotional distress outcomes, only 2 were deemed eligible for this review.

Figure 1.

Flow diagram of article screening and selection process.

Study characteristics are displayed in Table 1. Articles were published between 2000 and 2017, with most articles published after 2009 (n = 10). Most studies (n = 11) were RCTs,^31–41 and the rest (n = 4) were single-group pre-post trials.^42–45 Comparison groups among the RCTs included usual care control groups (n = 2),^35,39 usual care plus educational handouts (n = 2),^34,36 comparison among active treatment arms (n = 2),³⁷ wait list control group (n = 1),³² wait list group with some intervention contact (n = 1),³³ and attentional control groups (n = 4).^31,38,40,41. Sample sizes ranged from 16 to 417, with a median of 111. One study included dyads of individuals with T2DM and a family member without T2DM³³ but only the data for the individual with T2DM was included in this review. Only 1 study used random sampling,³⁶ while the rest used convenience sampling. All samples included both men and women, and most (n = 10) were at least 65% female. Most samples were 100% Latino (n = 12). The majority of studies targeted a specific Latino ancestry group, usually due to demographic characteristics in study location, while 7 studies did not report Latino ancestry.

Table 1.

Study characteristics.

Study first author (year), N Attrition %	Study location	Mean age (SD), Sex %	Latino ancestrya Latino % in sample	Study Design	Control group	Intervention dosage and duration
Castillo (2010)42 N = 70 Attrition: 33%	Chicago, IL	Age: 58.2 (13.1) Female: 76%	NR 100% Latino	One group pre-post PS	None	Two-hour weekly group sessions for 10 weeks.
Heisler (2014)31 N = 188 Attrition: 6%	Detroit, MI	Tx arm: Age: 51 (8.6) Female: 76% Control: Age: 52 (9.4) Female: 66%	NR 57% Latino	RCT: One tx arm, one control group	Comparison group had same amount of contact as tx group but used printed material	Both groups: One individual 1.5–2 hour session, plus 2 phone sessions that took place between 3–6 weeks after initial session. Tx group: Content was individually tailored and delivered in a digital format.
Lorig, (2000)43 N = 109 Attrition: 27%	Santa Clara, CA	Age: 54.6 (12.3) Female: 66%	NR 100% Latino	One group pre-post	None	Two-hour weekly group sessions for 6 weeks
Lorig (2008)32 6-month study N = 417 Attrition: 16%	San Francisco, CA	6-mo. Study Tx arm: Age: 52.9 (13.2) Female: 57%	Mexican 100% Latino	6-mo. study: RCT: One tx arm, one control group	6-mo. study: UC WL	6-month study: 2.5-hr weekly group sessions for 6 weeks (same curriculum as in the Lorig [2000]43 study).
18-month study N = 387 Attrition: 22%		6-mo. study UC: Age: 52.8 (13.4) Female: 67%		18-mo. study RCT: Two tx arms	18-mo. study: Control participants completed the 6-mo tx arm, then no other tx	18-month study: Intervention participants completed the 6-month tx arm, then received monthly automated reinforcement phone calls.
Mauldon (2006)44 N = 16 Attrition: 6%	Southern Connecticut	Age median: 50 Female: 56%	Puerto Rican and Caribbean 100% Latino	One group pre-post PS	None	Three-hour weekly group sessions for 6 weeks.
McEwen (2010)45 N = 21 Attrition: 9%	Arizona border region	Age: 53.7 (8.1) Female: 81%	Mexican 100% Latino	One group pre-post PS	None	Two-hour monthly group sessions for 6 months, plus 3 individual 60–90 minute sessions within 3 weeks of each group session.
McEwen (2017)33 N b = 157 Attrition at 3 months: 29%	Arizona border region	Age: 53.5 (9.0) Female: 65%	Mexican 100% Latino	RCT: One tx arm, one control group	WL + three weekly education classes	Two-hour weekly group educational and social support sessions for 6 weeks, followed by 2-hour weekly home visits for 3 weeks, and 20-minute weekly phone calls for 3 weeks. Intervention was delivered to family dyads (participant with T2DM plus family member).
Moncrieft (2016)34 N c = 111 Attrition at 12 months: 22%	Miami, FL	Age: 54.8 (7.4) Female: 71%	NR 85% Latino	RCT: One tx arm, one control group	UC + educational handouts	Two individual sessions, followed by 2 weekly and 4 bi-weekly group sessions, followed by 9 monthly group sessions. All sessions lasted 1.5–2 hours.
Piette (2000)35 N = 280 Attrition: 11%	NR	Tx arm: Age: 55.7 (10.2) Female: 61% UC: Age: 53.3 (10.5) Female: 57%	NR 50% Latino	RCT: One tx arm, one control group	UC	One year of bi-weekly automated assessment telephone calls lasting 5–8 minutes with the option to listen to automated health information for 1–7 minutes. “Periodic” follow-up calls were conducted with a nurse based on urgency of reported problems.
Rosal (2005)36 N = 25 Attrition: 8%	Western MA	Age: 62.6 (8.6) Female: 80%	Puerto Rican 100% Latino	PS, RCT: One tx arm, one control group	UC + educational handouts	One 1-hour individual session, then 2.5–3.5 hour weekly group sessions for 10 weeks, plus two 15-minute individual sessions prior to each group session.
Tang (2014)37 N = 116 Attrition: 41%	Detroit, MI	Age: 49.3 (11.0) Female: 58.6%	Mexican 100% Latino	RCT: Two tx arms	Comparison among two tx arms (no control group)	Both tx arms: Journey to Health DSME content delivered over 11 2-hour group sessions for 6 months, followed by two 1-hour monthly home visits, and 1 in-office visit with participant and their PCP. Peer support tx arm: Optional weekly group sessions for an additional 12 months after the 6-month DSME. CHW support tx arm: Monthly phone calls and emails on an as needed basis for an additional 12 months after the initial 6-month DSME.
Wagner (2016)38 N = 107 Attrition: 10%	Hartford, CT	Age: 60.3 (11.6) Female: 73%	NR 100% Latino	RCT: One tx arm, one control group	UC + a onetime 2.5-hr group education session	One 2.5-hour group education session, followed by eight 2-hour stress management weekly group sessions across 8–10 weeks.
Wang (2014)39 N = 252 Attrition: 7%	MA	Age majority “middle aged” Female: 76.6%	Puerto Rican 100% Latino	RCT: One tx arm, one control group	UC	One individual home visit for 1 hour, followed by weekly group sessions for 11 weeks, then 8 2.5-hour monthly group sessions.
Welch (2011)40 N = 46 Attrition: 15%	Springfield, MA	Tx arm: Age: 54.4 (10.4) Female: 68% Control: Age: 57.5 (9.5) Female: 62%	Puerto Rican 100% Latino	RCT: One tx arm, one control group	UC + seven 1-hr education visits for 1 year using ADA DSME education booklets	Seven 1-hour individual sessions over 1 year using a computerized program to guide the DSME discussion
Welch (2015)41 N = 399 Attrition: 12%	Western MA	Age: 55.0 (11.1) Female: 60%	NR 100% Latino	RCT: One tx arm, one control group	UC in the clinic’s diabetes program + individual education visits with diabetes educators; frequency of visits depended on patient need	One individual visit lasting 1 hour, followed by 30-minute visits occurring at week 2 and months 1, 3, and 6; all visits used a computerized program to guide the DSME discussions

Note. ADA = American Diabetes Association; CHW = community health worker; NR = not reported; PCP = primary care provider; Pre-post = a non-randomized prospective cohort study; PS = pilot study; RCT = randomized control trial; Tx = treatment; T2DM = Type 2 diabetes; UC = usual care control group, WL = waitlist control group

^aLatino ancestry defined as majority ancestry reported for sample or in area where sample was recruited from.

^bParticipants with T2DM without counting their non-T2DM family member dyad.

^cEligibility criteria for participants required a score of at least 11 on the Beck Depression Inventory-II.

Intervention Duration and Modality

The duration of the interventions varied widely, ranging from 6 weeks to 18 months, with approximately half lasting 6 months or less (n = 8). Four studies delivered the intervention in an individual format (in person and/or by phone); most used a group format (n = 3), or a mixed format of group sessions plus supplemental individual visits, home visits, and/or phone calls (n = 8). Five studies examined digital interventions, with 3 utilizing either a digital tablet or a computerized program to guide DSME discussion during individual visits.^31,40,41 Two studies utilized automated phone calls for delivering DSME and support.^32,35

DSME Intervention Content

The description of the interventions and their content is summarized in Table 2. Three studies had participants choose optional DSME content areas and thus participants likely received variable exposure to DSME content.^35,40,41 Of the remaining studies (n = 12), all provided DSME content covering both healthy eating and physical activity, most also covered problem solving (n = 8), all (n = 12) provided content that covered at least 4 of the 7 AADE target self-management behaviors, and one study covered the 7 target behaviors.³⁹ Three studies offered additional support sessions following the educational sessions: 1 study offered automated reinforcement calls after completing 6 months of DSME,³² 1 study offered 12 months of a support led by a peer or community health worker after completing 6-months of DSME,³⁷ and 1 study offered 8 weekly stress management group sessions after a one-time diabetes education session.³⁸

Table 2.

Intervention characteristics.

Study first author (year)	Study Name, DSME intervention curriculum content	Cultural tailoring of DSME content or intervention	Emotional distress intervention curriculum content	Distress outcome assessed	Assessment method
Castillo (2010)42	The Diabetes Empowerment Education Program (DEEP) Nutrition and PA Medication adherence Glucose self-monitoring Reducing risks Healthy coping	Described as “culturally competent” Family involvement Spanish language Low-literacy Community setting Led by CHW	DSME described as including content addressing “psychosocial aspects” and “self-care skills” and that it was based on “empowerment theory” Article described CHW facilitated self-efficacy, social support, and “stress relief”	Depression symptoms	PHQ-9
Heisler (2014)31	iDecide: Main focus was on medication adherence Nutrition and PA Reducing risks Problem solving Content tailored to each participant	Previous formative research using CBPR and user-centered design Spanish language Led by CHW Low-literacy and low-numeracy	No DSME content covering topic of emotional distress DSME was described as encouraging autonomy through CHW use of motivational interviewing	Diabetes distress	DDSa
Lorig, (2000)43	Study name NR Nutrition and PA Problem solving Healthy coping	Previous formative research Family involvement Spanish language Low-literacy Peer-led Community setting	DSME described as including content addressing “stress management” DSME was described as facilitating self-efficacy via skill mastery, modeling, reinterpretation of symptoms, and social persuasion	Health-specific distress	MOS health distress scale
Lorig (2008)32	Spanish Diabetes Self-Management Program ([SDSMP], same curriculum as in Lorig [2000]). 6-month study: Nutrition and PA Problem solving Healthy coping 18-month study: Upon completing 6-month DSME, received automated reinforcement calls assessing readiness and reviewing T2DM information through optional vignettes	Previous formative research Family involvement Spanish language Low-literacy Peer-led Community setting	6-month study: DSME described as including content addressing “stress management” DSME was described as facilitating self-efficacy via skill mastery, modeling, reinterpretation of symptoms, and social persuasion 18-month study: Received additional “support” from peer leader through the option of leaving messages during the automated reinforcement calls	Health-specific distress	MOS health distress scale
Mauldon (2006)44	Taking Control: Nutrition and PA Glucose self-monitoring Healthy coping	Spanish language Nutrition examples relevant to Latino culture Acknowledgment of cultural health beliefs Social emphasis Low-literacy	DSME described as including content addressing “psychosocial issues”, “stress management”, and “depression”	Diabetes distress	PAID
McEwen (2010)45	Study name NR Nutrition and PA Reducing risks Healthy coping Content tailored to each participant during home visits	Spanish language Nutrition and PA examples relevant to Mexican culture Social support linked to traditional cultural values Acknowledgment of cultural health beliefs Community setting Led by CHW	DSME described as including content addressing “distress management” Social support provided by CHWs throughout sessions	Diabetes distress	DDS
McEwen (2017)33	Study name NR Nutrition and PA Problem solving Healthy coping Content tailored to each family during home visits	Previous formative research using CBPR Described as “culturally tailored” CHW led the support sessions (but not the education sessions) Spanish language Family-based: participant plus one family member dyad	DSME described as including content addressing “stress management” CHW provided social support group sessions	Diabetes distress	DDS
Moncrieft (2016)34	Community Approach to Lifestyle Modification for Diabetes (CALM-D) Nutrition and PA Medication adherence Problem solving Healthy coping	Spanish language	CBT and social learning approaches to address depression symptoms including: Behavioral activation Active problem solving Identifying and challenging negative thoughts Stress management Coping skills training (e.g., deep breathing) Modification of behavioral, environmental and cognitive factors to promote healthy social support (e.g., assertiveness)	Depression symptoms	BDI-II
Piette (2000)35	Study name NR Automated calls included the option to listen to DSME information covering: Glucose self-monitoring Nutrition and PA Reducing risks Optional tailored feedback regarding weight and nutrition Periodic phone calls with nurse included problem solving and additional tailored feedback	Spanish language and pilot testing of translated material	No DSME content covering topic of emotional distress DSME described as facilitating self-efficacy	Depression and anxiety symptoms	CES-D and anxiety subscale of RMHI
Rosal (2005)36	Study name NR Nutrition and PA Medication adherence Glucose self-monitoring Problem solving Healthy coping	Previous formative research Spanish language Family involvement Nutrition and PA examples relevant to Latino culture Activities relevant to Latino culture Community setting Low-literacy Socializing time (i.e., coffee time before sessions)	DSME described as including content addressing “stress management”	Depression symptoms	CES-D
Tang (2014)37	Journey to Health 6-month DSME: Nutrition and PA Problem solving Healthy coping Reducing risks Content tailored to each participant during home visits 12-month PL support: Group problem solving Goal setting Review T2DM information Navigating health care system 12-month CHW support: Goal setting Information on health care resources	6-month DSME + 12-month support intervention: Previous formative research using CBPR Peer-led Spanish language Nutrition and PA examples relevant to Latino culture Family involvement Low-literacy Community setting Social support Incorporated cultural symbols and themes, cultural patterns and concepts, values, norms, and relationships	6-month DSME: Based on Social Cognitive Theory and “empowerment principles”; provided depression psychoeducation, stress-lowering activities, strategies for cognitive restructuring, improving social support and communication. Peer support tx arm Weekly emotional and behavioral peer-led support groups for addressing diabetes distress, share feelings about challenges and provide support CHW tx arm: monthly phone calls, plus as-needed emails, to provide emotional support and resources	Diabetes distress	DDS
Wagner (2016)38	Community Health Workers Assisting Latinos Manage Stress and Diabetes (CALMS-D)One-time group education session: Nutrition and PA Medication adherence Glucose self-monitoring Stress management group sessions focused on healthy coping	Led by CHW Spanish language Community setting Used culturally relevant stories and analogies for teaching stress management Relaxation exercises were “culturally tailored”	Stress management curriculum developed by clinical psychologists, previously pilot tested, and based on elements of CBT (e.g., thought restructuring), mindfulness, DBT (e.g., distress tolerance), and relaxation techniques (e.g., PMR, thermal biofeedback)	Depression and anxiety symptoms; Diabetes distress	PHQ-9 (adapted into the PHQ-8); PROMIS Anxiety Subscale 8a; PAID
Wang (2014)39	Latinos en Control: Nutrition and PA Medication adherence Glucose self-monitoring Problem solving Reducing risks Healthy coping Content and goals tailored to each participant during 10-minute individual discussion with interventionist each group session	Intervention team included CHWs Spanish language Nutrition examples relevant to Latino culture Activities and experiential exercises that were culturally familiar Family involvement Low-literacy	DSME described as including content addressing “stress management” DSME described as encouraging self-efficacy and value identification, and using elements of MI	Depression symptoms	CES-D
Welch (2011)40	The Comprehensive Diabetes Management Program (CDMP) Content of DSME visit was tailored based on: A web-based assessment tool (Diabetes Self-Care Profile) which creates diabetes behavioral alerts based on participant’s answers and laboratory data By asking the participant to identify one self-management behavior to discuss during visit Content of visit could cover the following: Nutrition and PA Medication adherence Glucose self-monitoring Reducing risks Problem solving was facilitated in each session	Spanish language Bicultural/bilingual clinical team Nutrition examples relevant to Latino culture Staff assisted with low-literacy barriers	At the start of the session the provider reviewed a summary of patient’s self-reported psychosocial challenges (i.e., diabetes distress, social distress, depression, binge eating, alcohol abuse, and low social support). NR if any DSME content covering topic of emotional distress was provided during the session.	Depression symptoms; Diabetes distress	PHQ-9 PAID
Welch (2015)41	The Comprehensive Diabetes Management Program (CDMP); same curriculum as in Welch (2015): Content of DSME visit was tailored based on: A web-based assessment tool (Diabetes Self-Care Profile) which creates diabetes behavioral alerts based on participant’s answers and laboratory data By asking the participant to identify one self-management behavior to discuss during visit Content of visit could cover the following: Nutrition and PA Medication adherence Glucose self-monitoring Reducing risks Problem solving was facilitated in each session	Spanish language Bicultural/bilingual clinical team Nutrition examples relevant to Latino culture Family involvement Literacy and numeracy “screening” Assessment of alternative medication use	At the start of the session the provider reviewed a summary of patient’s self-reported psychosocial challenges (i.e., diabetes distress, social distress, depression, binge eating, alcohol abuse, and low social support) and provided “education” tailored to patient’s psychosocial profile Participants were referred to mental health services as needed	Depression symptoms; Diabetes distress	PHQ-9 PAID-5

Note. BDI-II = Beck Depression Inventory-II⁴⁶; CBPR = community based participatory research; CBT = cognitive behavioral therapy; CES-D = Center for Epidemiologic Studies Depression Scale⁴⁷; CHW = community health worker; DBT = dialectical behavioral therapy; DDS = Diabetes Distress Scale⁴⁸; DSME = diabetes self-management education; MI = motivational interviewing; MOS = Medical Outcomes Study⁴⁹; PA = physical activity; PAID = Problem Areas in Diabetes Questionnaire⁵⁰; PAID-5 = Five item short-form version of the PAID; PHQ-9 = Patient Health Questionnaire-9⁵¹; PMR = progressive muscle relaxation; PROMIS = Patient Reported Outcomes Measurement Information System⁵²; RMHI = Rand Mental Health Inventory; SCT = social cognitive theory; T2DM = type 2 diabetes; Tx = treatment.

^aMeasure was scaled so scores would range from 0–100 and higher scores indicated more positive outcome.

Intervention Tailoring

All studies offered the intervention in participants’ preferred language (Spanish or English), and all but 2 reported additionally tailoring the intervention content for the needs of the Latino population. Extent of intervention tailoring varied widely. The following tailoring strategies were the most commonly reported: low-literacy adjustments (n = 9), delivery of the DSME sessions by a peer educator or community health worker (n = 8), activities and educational examples relevant to the culture (e.g., healthy Latino foods; n = 8), family involvement in the intervention (e.g., family permitted to attend the group sessions; n = 8), held in a community setting (e.g., churches; n = 6), and previous formative research to tailor the interventions (e.g., focus groups with target population; n = 6). Six studies included a table that outlined in detail the intervention content per session,^{34,36–39,45} and 2 also included a table detailing the tailored components of the intervention.^36,37

Emotional Distress Intervention Content

Of the 15 studies reviewed, most included some curriculum content targeting emotional distress (n = 12); however, most of these (n = 9) were vague in describing such content, generally using single-phrase descriptions such as “stress management”, “coping skills”, “social support”, “content on psychosocial issues”, or “self-efficacy”. In contrast, 3 studies provided thorough descriptions of their emotional distress content. Two of these studies^34,38 described providing content such as elements of cognitive behavioral therapy (e.g., cognitive restructuring), mindfulness exercises, dialectical behavioral therapy (e.g., distress tolerance), and stress management techniques (e.g., deep breathing, progressive muscle relaxation). The third study³⁷ reported that all participants first completed a 6-month DSME that included depression psychoeducation and strategies for cognitive restructuring, and subsequently one treatment arm received optional peer-led weekly social support groups for 12 months focusing on the challenges of living with diabetes, while the other treatment arm received optional monthly supportive phone calls with a community health worker for 12 months.

Assessment Methods

All studies reported using valid and reliable standardized self-report measures originally developed in English. Of note, although all studies reported administering the measures in Spanish, most (n = 11) only mentioned the English version of the measure without reporting the source of the Spanish version, whether the Spanish version was previously validated, or which (if any) translation procedures were used to translate the measure. For studies measuring changes in general emotional distress, approximately half assessed changes in depression symptoms (n = 8), and 2 assessed changes in anxiety symptoms. A total of 10 studies assessed changes in health-specific emotional distress. Measures used in the studies are listed in Table 2. Most studies (n = 14) administered the self-report measures in person either through a clinical interview or by having the participant completing it on their own, however 1 study administered the questionnaires by phone or by mail.³²

Intervention Outcomes

Raw mean differences and effect sizes for changes in measures of general emotional distress (depression symptoms and anxiety symptoms) are listed in Table 3. Of the 8 studies that assessed changes in depression symptoms, 6 reported statistically significant symptom reduction.^{34–36,38,39,42} Of 7 RCTs, 5 reported significant reductions in depression symptoms in the intervention arm relative to the control arm.^{34–36,38,39} The one study that was a single-group pre-post design reported statistically significant reductions in depression symptoms post-intervention compared to baseline.⁴² The two RCTs that reported null findings on depression symptom scores included few culturally tailored components, were delivered by a clinical team in an individual setting, occurred on an infrequent basis (approximately every 2 to 3 months), and did not provide any intervention content covering emotional distress. The effect sizes for reductions in depression symptom scores ranged from −0.20 to −3.24. Only two studies measured anxiety symptoms; one RCT reported a significant group by time interaction favoring the intervention group (with a small-to-medium effect size d = −0.32),³⁸ while the other RCT did not find a significant difference in anxiety symptoms between the intervention and the control group.³⁵ The study that reported null findings on anxiety symptoms included few culturally tailored components and used automated phone calls to provide optional educational content.

Table 3.

General emotional distress outcomes.

Outcome			Quantitative findingsa		Narrative findingsb	Summary of findings
			Raw mean difference (D)	Effect size (d) [95% CI]
Depression symptoms: BDI-II (0–63 range)
Between group (intervention vs control)
	Moncrieft (2016)34		−6.15*	−0.61 [−1.04, −0.18]	Intervention participants had significantly lower depression scores post-intervention.	+
Depression symptoms: CES-D (0–60 range)
Within group (pre-post)
		Wang (2014)39	−2.30*	NR	Significantly lower depression scores post-intervention compared to baseline scores.	+
Between group (intervention vs control)
		Piette (2000)35	−3.90*	−3.24 [−3.62, −2.86]	Intervention participants had significantly lower depression scores post-intervention.	+
		Rosal (2005)36	−1.70*	−0.20 [1.00, −0.60]	Intervention participants had significantly lower depression scores post-intervention.	+
		Wang (2014)39	−4.10*	−0.31 [−0.58, −0.04]	Intervention participants had significantly lower depression scores post-intervention.	+
Depression symptoms: PHQ-9 (0–27 range)
Within group (pre-post)
		Castillo (2010)42	−1.95*	NR	Significantly lower depression scores post-intervention compared to baseline scores.	+
Between group (intervention vs control)
		Wagner (2016)38	−1.50*c	−0.28 [−0.66, 0.11]	There was a significant group by time interaction favoring the intervention group.	+
		Welch (2011)40	NR	NR	No significant difference between groups in change in depression status from baseline to 12 months.	○
		Welch (2015)41	NR	NR	No significant difference between groups in proportion of patients moving from depressed to non-depressed status.	○
Anxiety symptoms: PROMIS 8a (8–40 range)
Between group (intervention vs control)
		Wagner (2016)38	−0.30*	−0.32 [−0.70, 0.07]	There was a significant group by time interaction favoring the intervention group.	+
Anxiety symptoms: Rand Mental Health Inventory Anxiety Subscale
Between group (intervention vs control)
		Piette (2000)35	+0.10	1.00 [0.73, 1.26]	No significant difference post-intervention between intervention and control groups.	○

Note. BDI-II = Beck Depression Inventory II; CES-D = Center for Epidemiologic Studies Depression Scale; NR = studies did not report sufficient information to calculate mean differences or effect sizes; PHQ-9 = Patient Health Questionnaire-9; PROMIS = Patient Reported Outcomes Measurement Information System.

+ results showed significant improvements in outcome (either within group pre-post or between intervention and control)

○ no significant change in scores (either within group pre-post, or between intervention and control).

^aIf studies reported more than one post-intervention follow up assessment, then the assessment that took place most closely to the end of the intervention was chosen for increased consistency across studies.

^bSome studies did not report quantitative information and only narrative information regarding the findings.

^cStudy used the PHQ-8 instead of the PHQ-9.

^*Change was statistically significant as reported by study authors.

Raw mean differences and effect sizes for the 10 studies that assessed changes in measures of health-specific emotional distress are listed in Table 4. Six of the 10 reported statistically significant symptom reduction post-intervention.^{31–33,37,41,43} Seven RCTs assessed between-group changes in health-specific emotional distress, with 5 reporting statistically significant reductions in health-specific distress scores in the intervention group compared to the control group.^{31–33,37,41} Across studies, the effect sizes for reductions in health-specific distress scores ranged from −0.17 to −3.85. One of the RCTs with null findings on health-specific emotional distress provided no intervention content on emotional distress, had infrequent sessions, and used few tailoring strategies.⁴⁰ The other RCTs that had null findings provided extensive support to the participants after they completed the diabetes education portion of the DSME. One of these studies reported that most of the intervention content focused on stress management (8 weekly 2-hour sessions focused on stress management skills following a one-time diabetes education session).³⁸ The second study provided 1 year of support (through either weekly support groups or monthly phone calls) after participants completed the 6-month educational portion of the DSME.³⁷ Although participants in this study did see significant reductions in health-specific distress following the 6-month DSME, the reductions in distress for those originally with high levels of distress were not sustained at the end of the one-year support portion of the DSME. The third study provided 18-months of automated reinforcement calls after participants completed the 6-month educational DSME portion.³² In this study, although participants in the intervention group had experienced significant reductions in health-specific emotional distress at the end of the 6-month DSME compared to the control group, there was no difference between groups at the end of the 18-months of reinforcement calls.

Table 4.

Health-specific emotional distress outcomes.

Outcome		Quantitative findingsa		Narrative findingsb	Summary of findings
		Raw mean difference (D)	Effect size (d) [95% CI]
Diabetes distress: PAID (0–100 range)
Within group (pre-post)
	Mauldon (2006)44	NR	NR	No significant change in diabetes distress scores post-intervention compared to baseline.	○
Between group (intervention vs control)
	Wagner (2016)38	−1.10	−0.17 [−0.56, 0.21]	No significant difference post-intervention between intervention and control groups.	○
	Welch (2011)40	−15.30	−0.57 [−1.21, 0.07]	No significant difference post-intervention between intervention and control groups.	○
	Welch (2015)41	−7.90*	−3.85 [−4.18, −3.51]	The intervention group had significantly lower diabetes distress scores post-intervention.	+
Diabetes distress: DDS (1–6 range)
Within group (pre-post)
	Heisler (2014)31	+14.10*c	NR	Significantly improved diabetes distress scores post-intervention compared to baseline scores3.	+
	McEwen (2010)45	−2.09	−0.36 [NR]	No significant difference in the total diabetes distress score post-intervention compared to baseline.	○
	Tang (2014)37	NR	NR	6-month DSME: Significant reduction in the proportion of participants who reported high distress at the end of the 6-month DSME. 12-month PL/CHW support after 6-mo. DSME: Reductions in proportion of participants with high distress achieved at the end of the 6-month DSME was not sustained at the end of the additional 12-month support in either the PL or CHW group. Significant reduction in the proportion of participants with moderate diabetes distress at the end of the additional 12-month support in both the PL and CHW groups.	+ ○ +
Between group (intervention vs control)
	Heisler (2014)31	+10.40*c	0.39 [0.09, 0.68]	The intervention group had significantly improved diabetes distress scores post-intervention3.	+
	McEwen (2017)33	−2.52*	−0.33 [−0.65, −0.02]	The intervention group had significantly lower total diabetes distress scores post-intervention.	+
Health-specific distress: MOS scale (0–5 range)
Within group (pre-post)
	Lorig (2000)43	−0.37*	NR	Significantly lower health-specific distress scores post-intervention compared to baseline.	+
Between group (intervention vs control)
	Lorig (2008)32 6-month study	−0.32*	−0.24 [−0.03, −0.45]	The intervention group had significantly lower health-specific distress scores post-intervention.	+
	Lorig (2008)32 18-month study (after completing 6-mo. DSME)	NR	NR	No significant difference in health-specific distress scores post-intervention between group that received phone reinforcement after 6-mo. DSME and group that did not receive reinforcement.	○

Note. CHW = community health worker; DDS = Diabetes Distress Scale; MOS = Medical Outcomes Study; NR = studies did not report sufficient information to calculate mean differences or effect sizes; PAID = Problem Areas in Diabetes Questionnaire; PL = Peer Leader.

+ results showed significant improvements in outcome (either within group pre-post or between intervention and control)

○ no significant change in scores (either within group pre-post, or between intervention and control).

^aIf studies reported more than one post-intervention follow up assessment, then the assessment that took place most closely to the end of the intervention was chosen for increased consistency across studies.

^bSome studies did not report quantitative information and only narrative information regarding the findings.

^cStudy scored the DDS on a scale of 0–100 where higher numbers indicated better outcomes (less distress).

^*Change was statistically significant as reported by study authors.

Risk of Bias

Ratings of risk of bias and quality of evidence using the EPHPP Quality Assessment Tool for each study are outlined in Table 5. Only one study obtained the highest global rating of 1 (strong). Three articles received a global rating of 2 (moderate) and 11 a global rating of 3 (weak). The domains most commonly rated as weak included the use of non-random convenience sampling (all but one study), low recruitment rates/high refusal rates, lack of control for confounders when groups differed at baseline (or lack of reporting on possible confounders), and lack of reporting on the use of validated/reliable Spanish measures. The domains most commonly rated as strong included study designs such as RCTs or controlled single-group pre-post studies, and low attrition rates. However, many studies received poor ratings because the articles did not provide sufficient information to accurately judge the quality of evidence and risk of bias. Areas that were commonly inadequately described included blinding of assessors to intervention allocation, randomization procedure, and reasons for withdrawals and drop-outs.

Table 5.

Quality of evidence and research bias (using EPHPP).

Study first author (year)	Selection Bias	Study Design	Confounders	Blinding	Assessment Method	Attrition	Global Rating
Castillo (2010)42	3	2	N/A	N/A	3	2	3
Heisler (2014)31	2	1	2	2	3	1	2
Lorig, (2000)43	3	2	N/A	N/A	3	2	3
Lorig (2008)32	3	1	1	2	3	2	3
Mauldon (2006)44	3	2	N/A	N/A	3	1	3
McEwen (2010)45	3	2	N/A	N/A	2	1	2
McEwen (2017)33	3	1	3	2	2	3	3
Moncrieft (2016)34	3	1	3	2	1	2	3
Piette (2000)35	2	1	1	3	1	1	2
Rosal (2005)36	3	1	3	2	1	1	3
Tang (2014)37	2	1	1	2	3	3	3
Wagner (2016)38	3	1	3	2	1	2	3
Wang (2014)39	3	1	1	2	3	1	3
Welch (2011)40	2	1	3	3	3	1	3
Welch (2015)41	2	1	1	2	2	1	1

Note. Quality of evidence and research bias was assessed using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool (Thomas et al., 2004). A rating of 1 = strong quality/low bias, 2 = moderate quality/bias, 3 = weak quality/high bias. The Global Rating score was determined across the 6 areas of quality, and determined to be a 1 = if all areas of quality scored “2” or higher, a 2 = if there was only one area that scored a “3”, and a 3 = if there were two or more areas that scored a “3”.

N/A = not applicable to one-group pre-post studies were by design there is only one group and thus no confounders between groups or blinding to treatment allocation.

Discussion

Latinos in the US are disproportionally affected by T2DM, and notably, by emotional distress in the context of this condition. General and health-specific emotional distress can act as obstacles in the pathway to appropriately managing T2DM. Interventions that improve both clinical outcomes and emotional distress have significant potential to improve health in the community of Latinos with diabetes. We evaluated the literature for studies administering DSME interventions to Latino adults with T2DM, and synthesized the evidence on the effects of these interventions on general or health-specific emotional distress. A total of 15 studies were included in the review, of which 11 were RCTs.

Of the 7 RCTs examining depression symptoms, 5 reported significant reductions compared to the control group. Results were inconclusive regarding anxiety symptoms, with one study finding significant reductions compared to the control group and another not finding significant results. Of studies assessing general emotional distress, those that did not find significant results were more likely to have fewer cultural tailoring components (e.g., delivered in individual sessions, in clinical setting with clinical staff or automated calls), had less frequent intervention contact, and/or did not provide any content covering emotional distress.

Of the 7 RCTs examining health-specific emotional distress, 5 found significant reductions in health-specific distress compared to the control group. RCTs that examined the effects of additional support (through support groups or reinforcement phone calls) after participants completed the diabetes education portion of the DSME were less likely to find significant results, even though participants had reductions in health-specific distress after completing the educational portion of the DSME. Thus, support sessions in the absence of continued delivery of educational content may be insufficient for ameliorating health-specific emotional distress. Finally, future studies should examine if improvements in health-specific emotional distress during a DSME in turn predict improved clinical outcomes.

Although one aim of this review was to describe and synthesize the intervention content targeting emotional distress, only 3 studies included detailed descriptions of such content. In contrast, most studies referred very generally to the content of focus (e.g., “stress management”, “self-care skills”), making it impossible to establish links between specific content or approaches and changes in the distress outcomes. Future studies should include precise operational definitions and example content. Studies should consider describing the intervention content with the use of a table outlining the content areas delivered in each session, either in the same article or in a separate trial protocol paper.

Importantly, when assessing risk of bias and quality of evidence, only one reviewed study received the highest rating of 1 (strong quality) and 11 received the lowest rating of 3 (weak). Upon further review, there was no apparent pattern in quality between null findings and bias ratings. However, results should still be interpreted within the context of these ratings. Based on the most common reasons studies received weak ratings, we recommend that future studies explore ways to incorporate random sampling and improve enrollment rates, report refusal reasons, and clarify if Spanish measures were used and whether translation methods were employed.

Limitations

The designs, focus, tailoring, and delivery of the reviewed studies varied widely, prohibiting meta-analysis, and limiting potential comparison across studies. Furthermore, in addition to the fact that most studies received weak quality of evidence ratings, several had small sample sizes and were pilot or feasibility studies. A review focusing only on the strongest study design (RCTs) should be carried out once there is a sufficient body of literature. Another limitation is that none of the studies reported whether participants had parallel care (e.g., medications, psychotherapy), for depression or other mental health conditions, raising the possibility that improved outcomes could be due to other treatment. In addition, analyses methods varied, and only some studies reported intent-to-treat analyses. One final limitation is that the outcomes described in this review were at the end of intervention and not at later follow-up time points, preventing discussion of maintenance of reduced distress over time in the absence of the intervention.

Conclusions and Recommendations for Research

Emotional distress can lead to worse outcomes and reduced quality of life among patients with T2DM. Based on the studies included in this review, there is an absence of strong evidence to support that DSME programs for Latino adults with T2DM are beneficial for improving general emotional distress and health-specific emotional distress. High quality research studies are needed to determine if the ADA Position Statement²⁰ suggesting diabetes self-management education as the first level intervention for emotional distress is appropriate for US Latino adults. Finally, to the extent that DSME programs do significantly improve distress for this vulnerable group, such improvements are more likely to stem from DSME programs that incorporate content targeting emotional distress and include cultural tailoring for the Latino population.

Summary of recommendations for future research studying the effects of DSME programs on emotional distress for US Latino adults with T2DM is as follows:

• Trials examining DSME programs among US Latinos should assess and report changes in general emotional distress and/or health-specific emotional distress outcomes in response to the DSME.

• Interventions that used only minimal tailoring were less likely to significantly impact emotional distress. Future studies should consider tailoring multiple aspects of interventions, beyond language, for example, by locating them in community settings, training peer leaders to deliver the interventions, utilizing group delivery to promote social support, including family members, and tailoring content or examples to ensure they resonate with the ethno-cultural group.

• If support or reinforcement sessions are offered, diabetes education should continue to be delivered in the intervention, as interventions that did not continue the educational content in the support sessions were less likely to significantly impact health-specific emotional distress.

• Studies should include precise operational definitions and examples to describe the DSME intervention content targeting emotional distress.

• Studies should consistently examine the degree to which emotional distress changes over time in responses to DSME, and whether reductions in distress relate to improvements in disease outcomes.