This nationwide study included all veterans receiving care in VHA who participated in at least one MOVE! session. Due to the episodic nature of weight management care, each veteran’s MOVE! participation was conceptualized into episodes that were separated by the absence of a MOVE! visit for ≥6 months, as applicable. In this study, we assessed veterans’ most recent MOVE! episodes that were initiated during the 10-year period from October 1, 2004, to September 30, 2014 (n = 588,839, Fig. 1).
A veteran’s most recent episode of MOVE! care was defined by the initiation date, i.e., the date of the first MOVE! visit or contact as identified in the electronic health record (EHR). We defined the 12 months prior to initiation as “pre-initiation” and the 12 months following initiation as “post-initiation.” A contact was defined as any clinical encounter between the veteran and a clinical professional in the EHR coded as “MOVE!” that was conducted in person, by telephone, or by video (Online Appendix A).
Veterans whose EHR indicated bariatric surgery (Current Procedural Terminology codes 43644, 43645, 43770, 43775, 43842, 43843, 43845, 43846, or 43847) anytime prior to the end of post-initiation were excluded from this study (n = 1830).
Predictor, covariate, and outcome data from the EHR were accessed through the VHA Corporate Data Warehouse, a repository of all EHRs for VHA facilities across the nation. Data on participation were available from every MOVE! program, as well as demographic information, weight measurements, medication use, and procedure coding.
The primary predictor of interest was the number of MOVE! contacts post-initiation. These were categorized as follows: 1, 2–5, 6–9, 10–13, 14–17, and 18 or more, for a total of six categories.
For purposes of this study, episodes of MOVE! that continued longer than 12 months were truncated, and veterans with one contact during their most recent episode of care were used as the reference group.
Weight and height measurements taken during routine health care, including MOVE! participation, were obtained from the VHA’s EHR. Using a multi-step process, weight and height values that were identified as outliers were excluded (Online Appendix B).
The primary outcome was the proportion of veterans with clinically relevant weight loss (≥5% body weight loss) at 12 months following initiation.6
21 We assigned each veteran’s initiation weight as the weight taken closest to initiation up to 30 days prior to or 7 days after initiation. Similarly, we assigned each veteran’s post-initiation weight as the weight taken closest to 365 days after initiation, up to 60 days prior to and 60 days following the target date. Veterans who did not have weight values meeting the aforementioned time windows were excluded from the analysis (n = 200,181, Online Appendix C). The percentage body weight change was calculated as the difference between initiation weight and post-initiation weight divided by initiation weight.
The secondary outcome of interest was the mean change in weight trajectory from pre-initiation to post-initiation. Veterans with fewer than three weight measurements documented in the medical record during pre-initiation and three during post-initiation were also excluded from the analysis (n = 149,251), resulting in a sample of 237,577 veterans. The linear trends (i.e., slopes) of all weights available in the EHR separately during pre- and post-initiation were calculated in the percentage change in body weight from initiation per year. Change in weight trajectory was defined as the difference between pre-initiation and post-initiation weight trends, where a negative change in weight trajectory indicated that an individual reduced their slope from pre- to post-initiation (not necessarily that the individual lost weight). For example, a veteran who was gaining weight pre-initiation at a rate of +5% per year, and had since attenuated that gain to maintain a stable weight (which is a post-initiation weight trajectory of 0% per year), would experience a −5% change in weight trajectory per year (i.e. change in slope).
A variety of demographic variables were available from the EHR, including age, gender, self-identified race, and marital status. Clinical variables were used to characterize the cohort, including body mass index (BMI; categorized into overweight: 25 to <30, obese class I: 30 to <35, obese class II: 35 to <40, and obese class III: ≥40), participation in TeleMOVE!22 (an 82-day disease management protocol for weight management delivered through home telehealth technologies, i.e., in-home messaging devices) during pre- or post-initiation, history of obesogenic23 medication use (at least a 30-day fill of amitriptyline, mirtazapine, olanzapine, quetiapine, risperidone, gabapentin, tolbutamide, pioglitazone, glimepiride, glyburide, glipizide, sitagliptin, or nateglinide) anytime before the end of post-initiation, and history of leptogenic medication use (at least a 30-day fill of sibutramine, orlistat, phentermine, phentermine/topiramate, or lorcaserin) anytime before the end of post-initiation. Naltrexone/bupropion and liraglutide, approved for the treatment of obesity near the end of our study period, were not included as leptogenic medications because no veterans in the cohort were prescribed them before the end of post-initiation.
Analyses were conducted using SAS version 9.4 software (SAS Institute Inc., Cary, NC) with an alpha of 0.05. Primary analyses used logistic regression to estimate the odds of achieving clinically relevant weight loss at 1 year post-initiation, with unadjusted and adjusted analyses (with age, gender, marital status, participation in TeleMOVE!, history of obesogenic medication use, and history of leptogenic medication use). Secondary analyses involved estimating linear trends in pre-initiation and post-initiation weight trajectories, then assessing the difference from pre- to post-initiation as the change in weight trajectory per veteran.