Jennifer Pillay, BSc; Marni J. Armstrong, PhD, RCEP; Sonia Butalia, MD, MSc; Lois E. Donovan, MD; Ronald J. Sigal, MD, MPH; Pritam Chordiya, BDS, MSc; Sanjaya Dhakal, MBBS, MPH; Ben Vandermeer, MSc; Lisa Hartling, PhD; Megan Nuspl, BSc; Robin Featherstone, MLIS; Donna M. Dryden, PhD
This article was published online first at www.annals.org on 29 September 2015.
Disclaimer: The authors of this report are responsible for its content. Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality or the U.S. Department of Health and Human Services.
Acknowledgment: The authors are grateful for the ongoing support of the AHRQ task order officer Aysegul Gozu, MD, MPH, and AHRQ associate editor, Jonathan Treadwell, PhD, who reviewed the review protocol and the full report submitted to AHRQ. They also thank the key informants and technical expert panel members (listed in the full report online at www.effectivehealthcare.ahrq.gov/reports/final.cfm) who provided input into the review; several technical expert panel members also provided peer review of the draft report, submitted to AHRQ and available online.
Grant Support: This project was funded under contract 290-2012-000131 from the Agency for Healthcare Research and Quality, U.S. Department of Health and Human Services. Dr. Armstrong was supported by doctoral awards from the Alliance for Canadian Health Outcomes for Research in Diabetes, Alberta Innovates–Health Solutions, and the University of Calgary–Eyes High program. Dr. Sigal is supported by a Health Senior Scholar award from Alberta Innovates–Health Solutions. Dr. Hartling holds a New Investigator Salary Award from the Canadian Institutes of Health Research.
Disclosures: Ms. Pillay reports other payment from the Agency for Healthcare Research and Quality during the conduct of the study. Dr. Sigal reports salary support from Alberta Innovates–Health Solutions during the conduct of the study. Dr. Chordiya reports grants from the Agency for Healthcare Research and Quality during the conduct of the study. Mr. Vandermeer reports grants from the Agency for Healthcare Research and Quality during the conduct of the study. Ms. Nuspl reports grants from the Agency for Healthcare Research and Quality during the conduct of the study. Ms. Featherstone reports grants from the Agency for Healthcare Research and Quality during the conduct of the study. Dr. Dryden reports other payment from the Agency for Healthcare Research and Quality during the conduct of the study. Authors not named here have disclosed no conflicts of interest. Disclosures can also be viewed at www.acponline.org/authors/icmje/Conflict OfInterestForms.do?msNum=M15-1399.
Editors' Disclosures: Christine Laine, MD, MPH, Editor in Chief, reports that she has no financial relationships or interests to disclose. Darren B. Taichman, MD, PhD, Executive Deputy Editor, reports that he has no financial relationships or interests to disclose. Cynthia D. Mulrow, MD, MSc, Senior Deputy Editor, reports that she has no relationships or interests to disclose. Deborah Cotton, MD, MPH, Deputy Editor, reports that she has no financial relationships or interest to disclose. Jaya K. Rao, MD, MHS, Deputy Editor, reports that she has stock holdings/options in Eli Lilly and Pfizer. Sankey V. Williams, MD, Deputy Editor, reports that he has no financial relationships or interests to disclose. Catharine B. Stack, PhD, MS, Deputy Editor for Statistics, reports that she has stock holdings in Pfizer.
Reproducible Research Statement:Study protocol: PROSPERO registration number CRD42014010515. Statistical code and data set: Available from Ms. Pillay (e-mail, firstname.lastname@example.org).
Requests for Single Reprints: Jennifer Pillay, BSc, Alberta Research Centre for Health Evidence, Department of Pediatrics, 4th Floor, Edmonton Clinic Health Academy, University of Alberta, 11405 87th Avenue, Edmonton, Alberta T6G 1C9, Canada; e-mail, email@example.com.
Current Author Addresses: Ms. Pillay; Drs. Chordiya, Dhakal, Hartling, and Dryden; Mr. Vandermeer; Ms. Nuspl; and Ms. Featherstone: Alberta Research Centre for Health Evidence, Department of Pediatrics, 4th Floor, Edmonton Clinic Health Academy, University of Alberta, 11405 87th Avenue, Edmonton, Alberta T6G 1C9, Canada.
Drs. Armstrong, Butalia, Donovan, and Sigal: Division of Endocrinology and Metabolism, Richmond Road Diagnostic and Treatment Centre, University of Calgary, 1820 Richmond Road SW, Calgary, Alberta T2T 5C7, Canada.
Author Contributions: Conception and design: J. Pillay, M.J. Armstrong, S. Butalia, L.E. Donovan, R.J. Sigal, L. Hartling, R. Featherstone, D.M. Dryden.
Analysis and interpretation of the data: J. Pillay, M.J. Armstrong, S. Butalia, L.E. Donovan, P. Chordiya, S. Dhakal, B. Vandermeer, L. Hartling, D.M. Dryden.
Drafting of the article: J. Pillay, S. Butalia, L.E. Donovan, P. Chordiya, S. Dhakal, B. Vandermeer, M. Nuspl, R. Featherstone.
Critical revision of the article for important intellectual content: J. Pillay, M.J. Armstrong, S. Butalia, L.E. Donovan, R.J. Sigal, B. Vandermeer, L. Hartling, D.M. Dryden.
Final approval of the article: J. Pillay, M.J. Armstrong, S. Butalia, L.E. Donovan, R.J. Sigal, P. Chordiya, S. Dhakal, B. Vandermeer, L. Hartling, M. Nuspl, R. Featherstone, D.M. Dryden.
Statistical expertise: B. Vandermeer.
Obtaining of funding: S. Butalia, L.E. Donovan, L. Hartling, D.M. Dryden.
Administrative, technical, or logistic support: J. Pillay, M. Nuspl, R. Featherstone, D.M. Dryden.
Collection and assembly of data: J. Pillay, P. Chordiya, S. Dhakal, L. Hartling, M. Nuspl, R. Featherstone, D.M. Dryden.
Pillay J, Armstrong MJ, Butalia S, Donovan LE, Sigal RJ, Chordiya P, et al. Behavioral Programs for Type 1 Diabetes Mellitus: A Systematic Review and Meta-analysis. Ann Intern Med. 2015;163:836-847. doi: 10.7326/M15-1399
Download citation file:
Published: Ann Intern Med. 2015;163(11):836-847.
Published at www.annals.org on 29 September 2015
Whether behavioral approaches for self-management programs benefit individuals with type 1 diabetes mellitus is unclear.
To determine the effects of behavioral programs for patients with type 1 diabetes on behavioral, clinical, and health outcomes and to investigate factors that might moderate effect.
6 electronic databases (1993 to June 2015), trial registries and conference proceedings (2011 to 2014), and reference lists.
36 prospective, controlled studies involving participants of any age group that compared behavioral programs with usual care, active controls, or other programs.
One reviewer extracted and another verified data. Two reviewers assessed quality and strength of evidence (SOE).
Moderate SOE showed reduction in glycated hemoglobin (HbA1c) at 6 months after the intervention compared with usual care (mean difference, −0.29 [95% CI, −0.45 to −0.13] percentage points) and compared with active controls (−0.44 [CI, −0.69 to −0.19] percentage points). At the end of the intervention and 12-month follow-up or longer, there were no statistically significant differences in HbA1c (low SOE) for comparisons with usual care or active control. Compared with usual care, generic quality of life at program completion did not differ (moderate SOE). Other outcomes had low or insufficient SOE. Adults appeared to benefit more for glycemic control at program completion (−0.28 [CI, −0.57 to 0.01] percentage points) than did youth (−0.12 [CI, −0.43 to 0.19] percentage points). Program intensity appeared not to influence effectiveness; some individual delivery appears beneficial.
All studies had medium or high risk of bias. There was scarce evidence for many outcomes.
Behavioral programs for type 1 diabetes offer some benefit for glycemic control, at least at short-term follow-up, but improvement for other outcomes has not been shown. (PROSPERO registration number: CRD42014010515)
Agency for Healthcare Research and Quality. (PROSPERD registration number: CRD42014010515)
Type 1 diabetes mellitus (T1DM), one of the most common chronic diseases in childhood and adolescence, is increasing in prevalence in the United States (1). The landmark DCCT (Diabetes Control and Complications Trial) and its related longitudinal study (EDIC [Epidemiology of Diabetes Interventions and Complications]) found that intensive glycemic control prevents development and progression of micro- and macrovascular complications (2, 3) and death (4). However, the intervention was initiated early (duration of T1DM <3 years for prevention group) in relatively young (mean age, 27 years), healthy patients. A meta-analysis of 12 trials of intensive control in diverse patient populations confirmed only a reduction in development of microvascular complications. Authors of that analysis stressed that benefits may apply only for interventions initiated early and should be weighed against risks for severe hypoglycemia (5). Factors other than glycemic control appear necessary to improve outcomes. For instance, intensive lowering of blood pressure has reduced major cardiovascular events by 11% (6). In addition, findings from 2 large cross-national studies support interventions to address other outcomes of importance for patients, such as diabetes-related distress (7).
All patients with diabetes are encouraged to adopt and adhere to many self-care behaviors (8, 9). This is particularly challenging for those with T1DM, who require lifelong insulin therapy and therefore should undertake rigorous self-monitoring and regulation of blood glucose levels through frequent adjustments to insulin dose, diet, and physical activity (10). Approaches for supporting patients to change several behaviors include diabetes self-management education (DSME) with or without added support (11) and lifestyle programs (12). Because knowledge acquisition alone is insufficient for behavioral changes (13, 14), the focus for DSME has shifted from traditional didactic approaches to more patient-centered methods that incorporate interaction, problem-solving, and other behavioral approaches and techniques (11, 15–17). Moreover, programs need to be tailored to the needs of the target population, such as developmental milestones in children or unique personal challenges during adolescence or adulthood (18).
Few systematic reviews on education and training in T1DM have been conducted over the past decade (19–21). Most reviews assessed only the effects on glycemic control, included highly didactic interventions, or reviewed interventions conducted outside the health care setting (such as summer camps) (19–23). All focused on children and adolescents. When calculated, effect sizes demonstrated very modest improvement at longest follow-up (21, 23). An updated evaluation—one that focuses on programs incorporating behavioral approaches and targeting several behaviors—is required to determine whether shifts in practice have translated into better outcomes for patients of all ages with T1DM. Anticipating high diversity in program content and delivery mechanisms, our evaluation also explores effect modification by program factors.
With assistance from key informants, a technical expert panel, and public commentary, we developed and followed a standard protocol. A peer- and public-reviewed technical report with additional details is available online on the Agency of Healthcare Research and Quality's (AHRQ's) Effective Healthcare Web site (24).
Our librarian searched the following bibliographic databases from 1993 to 15 January 2015: Ovid MEDLINE and Ovid MEDLINE In-Process & Other Non-Indexed Citations, Cochrane Central Register of Controlled Trials via Cochrane Library, EMBASE via Ovid, CINAHL Plus with Full Text via EBSCOhost, PsycINFO via Ovid, and PubMed (2014 only) via the National Center for Biotechnology Information Databases (MEDLINE strategy is presented in Appendix Table 1). On 3 June 2015, we updated the search in MEDLINE. We reviewed the reference lists of relevant systematic reviews and all included studies, searched ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry Platform, and searched relevant conference proceedings (2011 through 2014) and the U.S. Federal Register.
Appendix Table 1. Search Strategy for MEDLINE*
We included studies that were conducted in highly developed countries (25) and were published in English after 1993 (reflecting intensification of medical management based on the DCCT) (2). We included prospective comparative studies (that is, randomized, controlled trials [RCTs]; nonrandomized controlled trials; prospective cohort studies; controlled before–after studies) that enrolled participants of any age and compared a behavioral program with usual care (that is, medical management provided to all participants), an active control (intervention beyond usual care but not meeting our definition of a behavioral program), or another behavioral program. A behavioral program was operationally defined as a multicomponent, diabetes-specific program with repeated interactions by trained individuals, a duration of 4 weeks or longer, and DSME that entailed a behavioral approach or another program format that included at least a structured dietary or physical activity intervention with another component (Appendix Table 2).
Appendix Table 2. Operational Definitions of Behavioral Program and Comparators
We excluded studies in which the intervention was a disease or care management program (for example, those with active adjustment of diabetes-related medications) (26) or other quality improvement programs targeting health systems or providers (27). Studies were also excluded if they 1) focused on newly diagnosed (≤1 year) patients, 2) focused on psychological counseling or treatment without explicitly targeting several diabetes self-care behaviors, 3) had no outcomes of interest to this review (for example, reporting on only insulin sensitivity), 4) had study groups that differed only by a factor outside the review's scope (for example, low- vs. high-fat diet), and 5) included a study sample in which ≥25% of participants had type 2 diabetes (unless results were reported for T1DM).
Two reviewers independently screened all titles and abstracts. We retrieved the full text of any publications marked for inclusion by either reviewer. Two reviewers independently assessed the full texts using a priori inclusion criteria and a standard form. We resolved disagreements by consensus or by consulting another team member.
One reviewer extracted data by using a structured form created in the Systematic Review Data Repository (http://srdr.ahrq.gov/) (28); a second reviewer verified all data. Two reviewers independently assessed methodological quality. Discrepancies were resolved through discussion. We used the Cochrane Risk of Bias tool (29) for RCTs and nonrandomized controlled trials and used the Newcastle–Ottawa Scale (30) for prospective cohort studies and controlled before–after studies.
Characteristics of included studies are presented in summary tables. Our key outcomes were glycemic control (that is, glycosylated hemoglobin [HbA1c]); quality of life; development of micro- and macrovascular complications; all-cause mortality; adherence to diabetes self-management behaviors; and changes in body composition, physical activity, or dietary or nutrient intake. Secondary outcomes included episodes of severe hypo- or hyperglycemia, depression, anxiety, control of blood pressure and lipids, health care utilization, and program acceptability (via participant attrition). Harms included activity-related injury. We defined thresholds for clinical importance when the literature provided guidance; for HbA1c we used a between-group difference of 0.4–percentage point change (for example, 7.6% vs. 8.0%) (31); for patient-reported outcomes represented by continuous data, we used a one-half SD based on the mean SD from the pooled studies (32, 33). With input from our technical experts, we categorized the behavioral programs by various component and delivery factors (Appendix Table 3). Programs not classified as DSME or DSME with added support (both incorporating education or training on several diabetes self-care behaviors) were considered “lifestyle” because they generally consisted of structured dietary and physical activity interventions.
Appendix Table 3. Categorization of Program Components and Delivery Factors
When possible we used (or computed) change from baseline values. If SDs were not given, they were computed from P values, 95% CIs, z statistics, or t statistics or were estimated from upper-bound P values, ranges, interquartile ranges, or (as a last resort) imputation using the largest reported SD from the other studies in the same meta-analysis. When computing SDs for change from baseline values, we assumed a correlation of 0.5; we conducted post hoc sensitivity analyses using correlations of 0.25 and 0.75.
We pooled results for all ages and for subgroups based on age (that is, youth [aged ≤18 years] and their families, young adults [aged 19 to 30 years], adults [aged 31 to 64 years], and older adults (aged ≥65 years]) when there was more than 1 trial in each age category. We used the Hartung–Knapp–Sidik–Jonkman random-effects model (34, 35) using Stata 11.2 (Stata Corp.) and Excel 2010 (Microsoft) software. We calculated weighted mean differences (MDs) or standardized mean differences (SMDs), as appropriate, with corresponding 95% CIs. We analyzed outcomes at the end of intervention to 1-month follow-up (EOI), and at 1 to no more than 6 months (6-month), more than 6 to 12 months (12-month), more than 12 to 24 months (≥12-month), and more than 24 months (≥24-month) after the intervention. If a study included more than 1 follow-up time point in each stratum, we used the longer follow-up. We did not include observational studies in the pooled analyses.
Heterogeneity was considered substantial when the I2 statistic was greater than 50% (36). In cases of substantial heterogeneity, sensitivity analysis related to outlying effect sizes and incomplete (<70%) outcome data was performed. Publication bias was assessed visually and quantitatively by using the Egger test for comparisons with at least 10 studies (37).
We searched for subgroup analyses reported by individual studies that focused on whether a particular program was more or less effective for reducing HbA1c on the basis of age, baseline glycemic control (HbA1c <7% vs. ≥7%), duration of diabetes (≤1 vs. >1 year), race/ethnicity, and socioeconomic status. We also compared results between subgroups of studies based on study-level data (for example, mean age). We assessed whether program effectiveness was moderated by program factors (Appendix Table 3) by performing univariate meta regressions for comparisons between behavioral programs and usual care at longest follow-up. For harms (activity-related injury), we planned to descriptively summarize results.
We followed guidance by AHRQ (38) to evaluate the strength of evidence (SOE) for key outcomes across all studies using the domains most relevant to RCTs: study limitations (risk of bias for RCTs), consistency, directness, precision, and reporting bias. The overall SOE was graded by one reviewer and reviewed by another, with disagreements resolved through discussion. High, moderate, and low SOE reflect the confidence we have in the effect estimate and the likelihood that the estimate will not change with further research. Insufficient SOE implies that we cannot estimate an effect because of no or very little evidence or substantial discrepancies in the results (for example, when 95% CIs of the estimated effects include clinically important values both for and against behavioral programs).
Staff of AHRQ participated in development of the scope of the work and reviewed drafts of the manuscript. Approval by AHRQ was required before the manuscript could be submitted for publication, but the authors are solely responsible for its content and the decision to submit for publication. AHRQ staff did not participate in the conduct of the review, data selection or collection, data analysis, interpretation of the data, or preparation of the manuscript.
Our main searches identified 47 857 citations (Figure 1); by searching reference lists we identified 11 more studies. We included 36 studies (39–74) described in 47 publications (75–85), including 31 RCTs, 2 nonrandomized controlled trials (44, 71), and 3 controlled before–after studies (72–74). One study included adults with T1DM (49%) and adults with type 2 diabetes (51%) and reported HbA1c results for each patient group (68). See Appendix Tables 4 and 5 for study characteristics.
Summary of evidence search and selection.
T1DM = type 1 diabetes mellitus; T2DM = type 2 diabetes mellitus.
* One study was included for both T1DM and T2DM.
† T2DM not reported.
Appendix Table 4. Description of Studies and Interventions for Type 1 Diabetes in Youth
Appendix Table 5. Description of Studies and Interventions for Type 1 Diabetes in Adults
Most studies were 2-group trials comparing DSME with usual care. Twenty-seven studies were conducted in youths and 9 in adults (no study focused on young or older adults). One RCT compared the same DSME program delivered in person versus a program that used Internet-based videoconferencing (Skype, Microsoft) (48). Both studies focusing on lifestyle programs compared them with usual care; 1 was a 2-group RCT (65) and the other an observational study (72). Total duration of the behavioral programs ranged from 1.5 to 25 months (median, 5.6 months). Number of contact hours ranged from 1 to 48 hours (median, 9.5 hours). Settings, content, delivery mechanisms and personnel, and degree of program integration within standard medical care varied widely.
Mean age of youth participants ranged from 9.7 to 15.4 years (median, 13.5 years) and mean baseline HbA1c value ranged from 7.4% to 15.7% (median, 9.6%). In adult trials, mean age of participants ranged from 30 to 49 years and mean HbA1c value ranged from 7.7% to 9.6%. Mean body mass index ranged from 24.8 to 27.6 kg/m2.
All trials were assessed as having medium or high risk of bias. For objective outcomes, 60% of trials had medium (unclear) risk of bias and 40% had high risk; risk of bias was largely driven by incomplete outcome data (that is, loss to follow-up). For trials reporting on subjective outcomes of interest to this review (n = 22), all but 1 had high risk of bias (95%), primarily because of lack of blinding of participants, study personnel, and outcome assessors.
Tables 1 and 2 summarize the findings and SOE assessments for our key outcomes in comparisons between behavioral programs and usual care and between behavioral programs and active controls, respectively. These results represent studies from all ages combined and focus on trial data. Funnel plots and statistical tests for HbA1c compared with usual care did not indicate presence of publication bias.
Table 1. Findings and Strength of Evidence Assessments for Key Outcomes of Behavioral Programs Compared With Usual Care in Type 1 Diabetes (All Ages)
Table 2. Findings and Strength of Evidence Assessments for Key Outcomes of Behavioral Programs Compared With Active Controls in Type 1 Diabetes (All Ages)
There was no statistically significant difference in comparisons with usual care or active controls for HbA1c at EOI or 12-month follow-up (Figure 2; Appendix Figures 1, 2, and 3); the SOE was low because of risk of bias and inconsistent (vs. usual care at EOI) or imprecise (other comparisons) effect estimates. Moderate SOE showed greater reduction in HbA1c at 6-month follow-up with changes in HbA1c of −0.29 (CI, −0.45 to −0.13) percentage points and −0.44 (CI, −0.69 to −0.19) percentage points for behavioral programs compared with usual care (Figure 3) and active controls, respectively (Appendix Figure 4). For comparisons at 12-month or greater follow-up, the CIs included our threshold for clinical importance (≥0.4% HbA1c) such that we cannot rule out benefit for behavioral programs based on available evidence. Results from observational studies were inconsistent (72–74); the only study assessed as low risk of bias found no difference (72). The results from the sensitivity analysis using correlations of 0.25 and 0.75 instead of 0.5 when SDs were calculated for change from baseline values are presented in Appendix Table 6.
Effect of behavioral programs versus usual care on hemoglobin A1c at end of intervention.
Effect of behavioral programs versus usual care on hemoglobin A1c at 6 months after the intervention.
Effect of behavioral programs versus usual care on hemoglobin A1cat 12-month follow-up.
Effect of behavioral programs versus an active control on hemoglobin A1cat end of intervention.
Effect of behavioral programs versus an active control on hemoglobin A1cat 12-month follow-up.
Effect of behavioral programs versus an active control on hemoglobin A1cat 6-month follow-up.
Appendix Table 6. Changes to 95% CIs Using Results of Sensitivity Analysis for Calculations of SDs for Change From Baseline Values
Low SOE showing no difference, or insufficient SOE, was found for all comparisons with usual care or active controls for adherence to diabetes self-management (that is, frequency of self-monitoring blood glucose or overall self-management behaviors). In the RCT (71 youths) comparing in-person with Skype delivery of a DSME program, the authors used the Diabetes Self-Management Profile (in which higher scores show beneficial effect) to assess adherence and found no difference between groups at EOI or 6-month follow-up (MD, 2.80 [CI, −1.93 to 7.53] and 2.94 [CI, −1.67 to 7.55], respectively) (48).
For generic health-related quality of life (for example, World Health Organization Well-Being Index, Pediatric Quality of Life, Wellbeing Questionnaire), there was moderate SOE showing no difference at EOI. Three RCTs in youth reported generic health-related quality of life for longer follow-up periods, with no difference between groups at any time point. There was no difference for diabetes distress at EOI or at 6-month follow-up; the SOE was low because of high risk of bias and imprecision. The CIs for diabetes distress included our threshold for clinical importance (SMD ≥0.5) such that we cannot rule out a favorable effect.
There was insufficient SOE for other key outcomes that we graded for SOE (that is, changes in body composition, changes to dietary intake or physical activity, diabetes-specific quality of life). No studies reported on diabetes-related complications, all-cause mortality, or associated harms (activity-related injury).
Appendix Table 7 summarizes results for other clinical (hypoglycemic or hyperglycemic episodes, serum lipids, blood pressure) and health care utilization (hospital admissions and emergency department visits) outcomes, most of which were reported in single trials.
Appendix Table 7. Results of Secondary Outcomes
On the basis of between-study results for comparisons with usual care based on age, adults appeared to benefit more than youths at EOI (MD, −0.28 [CI, −0.57 to 0.01] and −0.12 [CI, −0.43 to 0.19] percentage points, respectively); the effect for adults approached statistical significance and the CI included a clinically important value (Figure 2). At 6 months, results were generally consistent with those from combining youth and adult studies (MD, −0.26 [CI, −0.47 to −0.05] and −0.38 [CI, −0.82 to 0.06] percentage points), except that results for adults did not reach statistical significance (Figure 3). The small number of adult studies (particularly at 6-month follow-up) limits the precision of these results. None of the point estimates exceeded a 0.4–percentage point difference in HbA1c.
The effectiveness of behavioral programs compared with active controls appeared higher for youth (MD, −0.52 [CI, −1.04 to 0.00] percentage points) than for adults (MD, −0.14 [CI, −1.28 to 1.00] percentage points) at 12-month follow-up (Appendix Figure 3); the effectiveness for youths was clinically important. The small number and size of studies in comparisons with active controls limited the ability to draw conclusions.
One RCT (n = 101) conducted a subgroup analysis of 54 youths with suboptimal baseline glycemic control (HbA1c ≥8%) (54). At EOI, it found greater odds of maintaining or improving HbA1c with behavioral programs compared with usual care (odds ratio, 3.4 [CI, 1.0 to 11.9]). This compares favorably to the overall study results of no difference (MD, 0.30 [CI, −0.22 to 0.82]). We did not conduct subgroup analysis at the study level because the mean baseline HbA1c was greater than 7% for all studies—our a priori cutoff for glycemic control. No trials reported on HbA1c by race or ethnicity, socioeconomic status, or time since diagnosis.
Univariate meta regressions were conducted with 27 studies (40–48, 50, 51, 54–64, 66, 67, 69, 72, 73). Variables of duration, contact hours, and contact frequency seem not to influence program effectiveness; coefficients were essentially 0 (for example, an additional month of program duration would not reduce HbA1c to any greater extent) and CIs were very precise without any indication of potentially producing a clinically important effect considering our threshold of 0.4% (Appendix Table 8). Incorporating some delivery to individuals seems to be beneficial compared with sole delivery in a group format (that is, positive coefficient indicating switching to group delivery increased HbA1c); the result approached statistical significance and the CI included a value meeting our threshold for clinical importance. Evidence was insufficient for other program factors; lack of reporting for community engagement precluded any interpretation of the results.
Appendix Table 8. Results From Univariate Metaregressions Analyzing the Association Between Different Program Delivery Factors and the Effectiveness of Behavioral Programs in Improving Hemoglobin A1c
Overall, behavioral programs seem to have some benefit for reducing HbA1c when follow-up extends beyond the immediate postintervention period. The delay in benefit may in part reflect the time required for this marker of glycemic control, which indicates control during the past 2 to 3 months, to demonstrate change. Notable, however, is the large diversity in program durations (range, 1.5 to 25 months). The beneficial findings for HbA1c are tempered by the apparent finding of no difference at longer follow-up time points, although we could not confidently rule out any benefit at long-term follow-up.
Our findings may underestimate the effect of programs should they be implemented in routine practice. The usual care group in several studies received some form of attention from the investigators (such as periodic telephone calls to maintain contact and encourage study participation); this may have resulted in improved glycemic control for the comparator group and reduced the relative effects of the behavioral program. Similarly, differences in usual care between studies may have played a role, such that a similar intervention may differ in its relative effect between low- and high-efficacy usual care conditions. Although variations in standard care in studies of behavioral interventions for youths (aged 8 to 21 years) with T1DM did not significantly affect results (86), better reporting of what constitutes usual care would allow readers, and systematic reviewers, to better apply the studies to their target populations or account for differences between studies. Underestimation, particularly for the outcome of glycemic control, may stem in part from lack of blinding of participants, and their health care providers, and the potential for co-intervention; adjustments of insulin may have been performed to a greater extent in the comparison groups than in the intervention groups.
Many studies were directed at adolescents. Self-management of T1DM during adolescence is complex, often characterized by personal challenges and uncertainty in schedules and frequent changes in location due to education and employment, transitions to adult care, and diminished parental involvement; consequently, glycemic control often deteriorates during childhood and adolescence (87–90). Accordingly, many studies aimed to prevent deterioration of glycemic control rather than improve control. The large reductions in HbA1c for some individuals in several studies (lower boundary of CI well past our threshold for clinical significance) may be considered quite promising. Likewise, practicing more demanding self-management behaviors could negatively affect social and emotional functioning, such that our findings of no difference in generic health-related quality of life at EOI may be viewed positively.
Most studies were undertaken in populations with a baseline HbA1c value of 8.5% or greater. Although this level of glycemic control could affect the applicability of the findings, many clinicians deal with patient populations—particularly when patients are in their pubertal years—that are struggling to achieve optimal control. The results are most applicable to older children, adolescents, and adults through middle age. Mean duration of diabetes ranged from 2.7 to 7.3 years (studies in youths) and 7.5 to 23 years (studies in adults); results may not apply to individuals with recently diagnosed T1DM. We did not find evidence to determine whether behavioral programs are more or less efficacious for other subgroups, including sex or racial/ethnic minorities. The findings are highly relevant to the United States, where there is an increasing trend to incorporate psychological techniques and theories within educational programs; they may not apply as well to countries where this integration does not commonly occur.
We did not identify enough studies of lifestyle programs, or of DSME programs adding a support component, to draw conclusions about difference in benefit in terms of these factors. Many individuals with T1DM under good glycemic control may have other risk factors (such as overweight, hyperlipidemia, and hypertension) for which lifestyle programs may be warranted and beneficial. With our focus on programs incorporating interaction with program personnel, we cannot comment on the effects of programs delivered entirely by way of technology, which may provide sophisticated mechanisms to interact with and motivate participants. Further research in these areas would be informative. We also cannot comment on the effectiveness of programs with durations less than 4 weeks that had similar content, contact hours, and delivery mechanisms.
Some potential limitations are inherent in systematic reviews, such as those related to selection and reporting biases. We were able to locate several trial registries and protocols to compare planned and published outcome reporting; most included studies were judged as having low bias in this respect. Our prespecified tests for publication bias indicated no important bias, although these tests were limited to the outcome of HbA1c in comparisons with usual care. We included only English-language studies because we felt that these would be most applicable to the end-users of this review who create recommendations or implement programs within the United States. Moreover, effect sizes in language-restricted reviews do not differ significantly from those without restrictions (91).
The evidence base was limited in terms of allowing us to make conclusions for many outcomes, such as behavioral outcomes related to changes in dietary intake or physical activity, and for clinical and health outcomes apart from HbA1c and generic health-related quality of life. Considering that behavioral changes are key mediators to achieving clinical and health outcomes, analysis based on valid outcomes of changes to physical activity or diet would be ideal; greater use of these outcomes, especially via objective means, would be beneficial for future research. No studies contributed data related to harms. Most data for examination of subgroup effects relied on between-study rather than within-study comparisons, such that the effect of randomization is removed and the results are considered observational and possibly biased through confounding by other study-level characteristics. Many trials had high risk of bias, especially for subjective outcomes. Blinding of participants and personnel, or outcome assessors, was rarely reported or sufficient. Lack of intention-to-treat analysis and high participant attrition were prevalent.
In conclusion, current evidence does not support encouraging patients with T1DM to participate in behavioral programs to improve outcomes apart from HbA1c. Program evaluation is an important component to build into the implementation of any behavioral program for diabetes to ensure that it is the correct fit for the target population. At this time, clinicians still must monitor patients after participating in these programs, should additional means be necessary to control their disease more adequately to prevent devastating complications.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Cardiology, Endocrine and Metabolism, Diabetes, Coronary Risk Factors.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only