Background

Evaluation of outcomes in commercial weight management programs is challenging. Participants access a service for as long as they wish and missing weight data across longitudinal time points is typical. Analysis using imputation from baseline (BOCF) or last observation (LOCF) are recognised methods for assessing outcomes. This analysis explores and compares outcomes using different imputation and reporting methods in the context of such a program.

Methods

Weight data from adults (n=1,094,676) joining Slimming World’s weight management program during 2016 were analysed at weeks 12 and 52 using BOCF and LOCF methods. Weight outcomes in members with ≥75% (high) attendance over the time period was explored.

Results

Mean attendance by 12 weeks was 8.0 ± 4.4 sessions. 58% attended at the 12-week time point and mean weight change was -6.6%. Using LOCF and BOCF methods to include those who did not attend at 12 weeks, mean weight change was -5.0% and -3.8%, respectively. Those attending ≥75% of sessions but not the 12th session had a mean weight change of -5.7%. When using the LOCF method for all high attenders it was -7.7%. Mean attendance by 52 weeks was 15.8 ± 14.7 sessions. 16.4% attended at the 52-week time point and mean weight change was -12.5%. Using LOCF and BOCF methods mean weight change was -6.0% and -2.1%, respectively. High attenders not present at week 52 had a mean weight change of -11.4% and -14.1% using the LOCF method.

Conclusions

Weight outcomes varied by analysis method. The BOCF method assumes weight has remained as at baseline if an individual does not have a measure at a specified time point. This can underestimate weight outcomes particularly in individuals who are otherwise high attenders. Methods that account for the free-living nature of participants, fluctuating patterns of attendance and weight change over the whole time period in question as opposed to at one single, arbitrary time point may be more suitable for evaluating weight outcomes in these types of intervention.