Method: Systematic searches were conducted using PubMed, Cochrane Library, CINAHL, and PsycINFO to retrieve articles published in the past two decades. A total of 29 OI trials were selected using inclusion criteria (i.e., RCTs, middle-aged adults, and an OI for weight loss) and evaluated methodological quality using 10 criteria (e.g., more than 6-month follow-up and at least 70% retention at follow-ups), and intervention features (e.g., opportunities for interactions with health professionals or other participants). In accordance with MOOSE guidelines, we used narrative summary tables to review the selected 29 RCTs. Using RevMan 5.2 software, a set of meta-analyses were performed to synthesize the data and compare the effectiveness of OIs versus non-OIs–e.g., self-directed management and personal care. CMA 3.0 was used to test the relative effects of demographics, methodological quality and intervention features.
Results: Results of systematic review showed the mean methodological quality score of the trials was 12.1 (out of 16), and the mean intervention features score was 6.6 (out of 8). Use of blinding and multisite trials were the main difference between higher and lower scores of quality. Opportunity for interactions with professionals and other participants were the differences between higher and lower scores of features. RCTs with higher scores of methodological quality and intervention features were more effective in weight loss than those with lower scores. Results of meta-analyses showed that OIs were significantly more effective than non-OIs (18 trials, n=4,326; standardized mean difference [SMD], –1.74, 95% CI, –2.32 to –1.17; I2=99%). Compared to OIs only, OIs with interactions with other participants (6 trials, n=1,070; SMD, 0.60, 95% CI, 0.03–1.17; I2=93%) and professionals (6 trials, n=1,137; SMD, 0.34, 95% CI, 0.08–0.59; I2=76%) were more effective. However, there was no difference in weight loss between OIs only and in-person interventions. Results of meta-regression analyses showed that methodological quality was the only significant predictor of OIs versus non-OIs (B = -0.856, SE = 0.279, 95% CI [ -1.4042, -0.3088], P < .01).
Conclusions & Implications: Both methodological quality and intervention features were important to increase the effectiveness of OIs for weight loss. OIs should be based on rigorous methodological quality. OIs with interactions with others and professionals were more beneficial for middle-aged overweight and obese adults than OIs only. We also found that OIs were effective as much as in-person interventions for weight loss. These results should be incorporated into designing an OI to enhance its effectiveness.