CRD summary

This review concluded that lifestyle interventions modelled on the US Diabetes Prevention Program produced a clinically significant weight loss in adults at high risk of diabetes. Using lay personnel could reduce costs without sacrificing effectiveness. The review process was effective, but the studies were of limited quality and the authors' conclusions should be treated with caution.

Authors' objectives

To evaluate the effectiveness of lifestyle interventions, modelled on the US Diabetes Prevention Program, in real-world settings, in reducing weight in adults at high risk of diabetes.

Searching

MEDLINE, EMBASE, The Cochrane Library and ClinicalTrials.gov were searched for items from 1st January 2003 to 30th April 2011; search terms were reported. The bibliographies of relevant reviews and publications from the scientific sessions of the American Diabetes Association in 2010 were handsearched.

Study selection

Controlled and uncontrolled USA-based studies were eligible for inclusion if they translated the US Diabetes Prevention Program lifestyle intervention into real-world settings, and included people 18 years old or older who were at a high risk of developing diabetes. The US Diabetes Prevention Program findings were published in February 2002, and studies published from 1st January 2003 were included. High risk participants were defined as: those with biochemically confirmed pre-diabetes (impaired fasting glucose, impaired glucose tolerance, or both); or overweight patients with a body mass index (BMI) of 25kg per m² or greater and with one or more metabolic risk, such as a history of family or gestational diabetes. Studies had to report the starting weight and weight loss achieved. They were included if less than half the participants had pre-existing diabetes. They were excluded if they applied other weight loss principles or were commercial programmes differing from the study programme. The essential criteria of the US Diabetes Prevention Program intervention were: calorie restriction; 150 minutes per week of physical activity; and a target of 7% weight loss. The primary outcome was the participant's percentage change from starting weight.

Most studies were set in urban areas; 43% were in the community and 39% were in health care facilities; 14% used electronic media to engage with patients. The number of core intervention sessions ranged from one to over 24; there were most commonly 12 or 16 group sessions. The number of maintenance sessions varied widely from none to monthly sessions (most common). Full details of the interventions were provided including retention incentives. The average age of participants was 55.1 years; their average BMI was 34.0kg per m²; the percentage of females was 70; and 71% of them were non-Hispanic and White.

Two independent reviewers selected the studies, with discrepancies resolved by consensus.

Assessment of study quality

The quality assessment framework by Juni, et al. 2001 was modified to assess four criteria. Three of these were: participants defined as being at high risk of diabetes; attrition addressed, for example by using an intention-to-treat (ITT) analysis; and data limitations reported. The fourth criterion was that the study reported the practicality of the interventions by describing four or more of the following: programme design process; enrolment process; session attendance; cost or resource input; training and qualifications of personnel; and qualitative feedback from participants or providers.

The authors did not report how many reviewers assessed quality.

Data extraction

Two independent reviewers extracted the data, with discrepancies resolved by consensus. The authors were contacted for missing data. The mean percentage weight loss was extracted for each intervention group.

Methods of synthesis

Pooled mean socio-demographic characteristics were sample weighted. The percentage weight changes were pooled using a random-effects model, because significant heterogeneity was expected, to produce the mean and 95% confidence interval. Between-study heterogeneity was assessed using Ι². The results were reported by the type of personnel delivering the intervention.

Spearman's Rank Correlation Test was used to assess whether the number of core sessions (those given in the first three to six months) was related to session attendance. For studies with complete data, meta-regression was used to assess whether summary weight loss was related to programme factors, such as core sessions attended, personnel delivering the intervention, presence of a maintenance phase, and adjustment for patients' characteristics. A sensitivity analysis was performed to assess the effect of study duration. Publication bias was assessed using the Begg and Egger tests.

Results of the review

Twenty-eight studies were identified, with 3,797 participants (range eight to 1,003). There were four randomised controlled trials (RCTs; n=286 participants, range 39 to 151), two cluster-RCTs (n=31 and n=46), two non-randomised controlled studies (n=121 and n=820), and 20 single group pre-post studies (n=2,493, range eight to 1,003). Study attrition ranged from zero to 49 percent; 17 studies used either an ITT or last-observation-carried-forward analysis. Median study duration was 12 months (range three to 12 months). Three studies met all four quality criteria; 12 studies met three criteria; 11 studies met two criteria; and two studies met one criterion.

There was a significant pre-post weight change (-3.99%, 95% CI -5.16 to -2.83; Ι²=52.4%; 28 studies) after 12 months follow-up. There were similar weight changes in studies with intervention delivery by medical and allied health professionals (-4.27%, 95% CI -5.85 to -2.70; 19 studies), those by lay community educators (-3.15%, 95% CI -5.46 to -0.83; five studies) and those using electronic media-assisted interventions (-4.20%, 95% CI -7.62 to -0.77; four studies). The results did not change when only studies with over nine months of follow-up were analysed (17 studies). Point estimates suggested a better weight loss with delivery by lay community educators (from -1.84% to -0.36%) than by medical and allied health professionals and a poorer weight loss (a gain) using electronic media (from 1.11% to 1.93%). There was no significant evidence using point estimates that the presence of a maintenance phase was beneficial for weight loss.

In the meta-regression, there was a strong correlation between the number of core sessions attended and the number offered (r=0.90; p<0.01). Every additional core session attended was associated with an additional weight change of -0.26% (95% CI -0.54 to 0.01; not significant), after adjustment for gender and ethnicity.

The Begg test showed no evidence of publication bias. The Egger test showed a tendency for smaller studies to have greater effects (p=0.02).

The RCTs showed that more intensive programmes had significantly greater weight loss. Focus group results were reported in three studies (including one RCT) of factors influencing patient attendance and intervention uptake.

Cost information

Intervention costs were reported in six studies. The total material costs were similar for six-session ($934) and 16-session programmes (£1,075). Intervention staff costs varied by level of formal training, from $10 to $15 per participant per session for lay personnel to £25 per participant per session for clinically trained staff. Costs were reduced by the use of capillary glucose measurement and the use of databases to identify suitable participants.

Authors' conclusions

Clinically significant weight reduction could be achieved in people at high risk of diabetes, using structured lifestyle interventions from the US Diabetes Prevention Program adapted to the real world. The costs of diabetes prevention could be reduced, without sacrificing effectiveness, by using non-medical personnel and encouraging attendance at programme sessions.

CRD commentary

The review addressed a well-defined question for study design, participants, interventions and relevant outcomes. The search was sufficient and some effort was made to identify unpublished studies. There was no evidence of publication bias. Study quality was assessed and the results were reported. Quality was low for nearly half the studies and relatively few controlled studies were identified. Efforts were made to reduce error and bias in study selection and data extraction, but they were not reported for quality assessment. The relevant study details were reported. The authors noted that some studies lacked description, and some minor misclassifications of some intervention features were inevitable. The methods of synthesis were appropriate and the sources of heterogeneity were investigated by meta-regression.

The review process was effective, but the studies were of limited quality and the authors' conclusions should be treated with caution.

Implications of the review for practice and research

Practice: The finding that non-medical personnel were as effective as health professionals at motivating weight loss was important for the size and cost of diabetes interventions. Their training was neither costly nor time consuming and the authors proposed that during training basic knowledge, organisational skills, and empathy should be emphasised. Electronic-media assisted programmes might be useful for some people; often attrition was low for those using DVD or online formats.

Research: The authors recommended that a large trial should rigorously evaluate costs to produce reliable information for health service payers. Large well-designed comparative trials, randomising patients to different programme lengths and different types of delivery were needed.

Funding

Not stated.

Bibliographic details

Ali MK, Echouffo-Tcheugui JB, Williamson DF. How effective were lifestyle interventions in real-world settings that were modeled on the Diabetes Prevention Program? Health Affairs 2012; 31(1): 67-75. [PubMed: 22232096]

Original Paper URL

http://content.healthaffairs.org/content/31/1/67.abstract

Indexing Status

Subject indexing assigned by NLM

MeSH

Diabetes Mellitus /prevention & control; Female; Humans; Male; Middle Aged; Program Evaluation; Risk Reduction Behavior; United States; Weight Reduction Programs

AccessionNumber

12012003021

Database entry date

06/10/2012

Record Status

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.