2.2. Comparison with Other Published Studies

JumpstartMD weight losses were compared to those reported for other commercial programs (Weight Watchers [6, 8, 11–14, 31–42], Slimming World [12, 13, 15–22, 37, 43], Rosemary Conley [12, 13, 37, 42], Biggest Losers Club [25, 44, 45], eDiet [46, 47], Jenny Craig [23, 24, 33, 48–50], Itrim [26], Nutrisystem [9, 33, 51–53], Medifast [28, 29, 54–56], Health Management Resources [30, 57–59], Metabolic Balance Nutrition Program [27], and Diet Chef [60]) when matched for follow-up duration, BMI eligibility, age, proportion of males, and sex-specific baseline BMI effects. The studies were identified from systematic reviews, PubMed searches by program name, and references cited within each paper retrieved (Table 1). We excluded studies that used commercial weight loss products but were directed by others (e.g., Slimfast and Optifast) because they lacked standardized behavioral interventions. We also excluded very-low-calorie diets, studies conducted before the year 2000, studies not reporting LOCF or completers, and studies in cancer survivors.

Sample characteristics of studies on commercial weight loss programs.

Abbreviations: HMR: Health Management Resources; JC: Jenny Craig; NHS: National Health Service; RC: Rosemary Conley; RT: randomized trial; SW: Slimming World; and WW: Weight Watchers. ^∗Mean (standard deviation).

Results are presented in as forest plots with diamonds representing the average differences between JumpstartMD and other commercial programs (identical size for each comparison) and 95% confidence intervals for the differences. Standard errors (SE) were taken from the published reports, calculated as SD/N^0.5, or calculated from the 95 percent confidence intervals as (upper-lower bound)/(2∗1.96). When SEs were not otherwise available, the standard deviations were estimated from the relationship between mean weight change (independent variable) and the standard deviation of weight change (dependent variable) from the JumpstartMD dataset, using the coefficients of Table 2. Different coefficients were used for estimating “LOCF” and “completer” standard deviations. Median weight losses [11, 14, 32] were increased by 5.2% for weight change (∆kg), 2.1% for percent weight change (∆%), and 4.4% for BMI change (∆kg/m²) corresponding to the JumpstartMD difference between mean weight loss and median weight losses.

Estimated standard deviation for LOCF and completers as a function of the mean weight loss for use in estimating the missing standard errors (SD/N^0.5) for other commercial results.

² P ≤ 0.05.