The Durham VAMC Human Subjects Protection committee reviewed and approved this study. We employed a cross-sectional pre–post study design with a nonequivalent control group. We had 2 distinct and nonoverlapping cross-sectional cohorts: 1) a preperiod (June 16, 2007, through June 15, 2008) cohort of veterans who were new MOVE! clinic users in the year before the copayment elimination and 2) a postperiod (June 16, 2008, through June 15, 2009) cohort of veterans who were new MOVE! clinic users in the year after the copayment elimination. No patients in the preperiod cohort were represented in the postperiod cohort. Patients were identified as new users if they had no MOVE! clinic use in the 12 months before the index visit (their first MOVE! clinic visit) in the preperiod or postperiod. The cohort included incident, rather than prevalent, MOVE! users to enable us to identify immediate effects (13) of the copayment elimination among a subgroup of MOVE! users who may have been more responsive to the policy change.

Although nonexempt veterans during the preperiod were required to pay a $15 copayment until the copayment was eliminated on June 15, 2008, for the purpose of this study we refer to veterans in the preperiod (and the postperiod) who did not meet requirements for exemption and for whom the MOVE! copayment was eliminated as the “newly exempt” cohort. We hypothesized that copayment elimination would most affect the behavior of new clinic users in the postperiod. We compared MOVE! clinic visits by newly exempt veterans to clinic visits by a subgroup of veterans in the preperiod and postperiod cohorts who were exempt from copayments for all outpatient clinic visits. Seventeen percent of the preperiod cohort were veterans who were required to pay copayments, and 16% of the postperiod cohort were newly exempt veterans who became newly exempt in the year after copayment elimination.

The sample included 126 VAMCs with high-volume MOVE! clinics, defined as having 30 or more MOVE! participants during the preperiod. This threshold was chosen to identify VAMCs that had moved beyond the start-up phase of MOVE! implementation. For the clinic-standardized rate of new MOVE! use, we aggregated data to the clinic level by month for new users in the preperiod and the postperiod. The number of MOVE! visits by new users in the 6 months after their index visit was defined for 21,193 preperiod and 26,520 postperiod new users. We excluded 23,218 preperiod new users from the analysis of attendance precopayment and postcopayment removal because their first MOVE! clinic visit occurred within 6 months of the copayment elimination (between December 16, 2007, and June 14, 2008), so they had less than 6 months of preperiod time to make a repeat visit. We excluded 37,878 postperiod new users whose first visit occurred between December 16, 2008, and June 14, 2009, for parallel cohort construction. The final sample included 44,411 preperiod and 64,398 postperiod new users.

To construct the 2 MOVE! clinic use outcomes and most patient characteristics (age, sex, race, copayment exemption status, and marital status), we used 2006–2009 VHA Outpatient Care File data, which contains information on every VHA outpatient visit. The first clinic use outcome was measured monthly at the clinic level separately for exempt and newly exempt users (126 clinics × [24 months × 2 copay groups] = 6,048) during the preperiod. This outcome was a ratio in which the numerator was the number of MOVE! first-time visits within each clinic by patients who were new users, and the clinic-specific denominator was the overall mean monthly number of MOVE! visits to that clinic. This ratio, which we refer to as the clinic-standardized rate of new MOVE! use, represents the ratio of first-time MOVE! clinic visits to mean overall clinic use by all users (prior users and new users). The denominator was constant during the 24 months and the 2 copay groups, ranging from 0.0 to 4.3, so the ratio varied only according to the monthly change in the number of veterans who were using the MOVE! clinic for the first time. The denominator reflected the overall mean number of MOVE! clinic visits, so it was possible for the number of MOVE! clinic visits by new users in a given month to be larger than the overall monthly mean preperiod use (eg, the ratio was greater than 1.0). The second use outcome was a count variable that measured at the patient level in both the preperiod and postperiod (n = 47,713), and consisted of the number of MOVE! visits by new users in the 6 months after their first clinic visit.

To control for comorbidity, we obtained Diagnostic Cost Group (DCG) risk-adjustment scores for each patient in the preperiod and postperiod cohorts. DCG scores have been as predictive of veterans’ 1-year expenditures and mortality as other comorbidity scores (14,15). Patient characteristics were aggregated to the clinic level for assessing new users because the outcomes and unit of analysis were clinic-year. Finally, we obtained details on MOVE! clinic characteristics from VHA MOVE! annual reports. This report, required by VHA, completed by each facility’s MOVE! coordinator, and reviewed by facility leadership before submission, measures program implementation and MOVE! staffing levels at each VAMC. We selected clinic-level characteristics that may affect a clinic’s ability to meet new demand that resulted from elimination of the MOVE! copayment, including having a local MOVE! coordinator, the coordinating committee’s make-up and linkage to primary care, and the availability of facility space for meetings and exercise sessions.

The analysis for the first outcome was a linear mixed-effects model based on the framework of a split-plot design (16). In this design, the clinic was the “whole plot,” and the 2 factors (treatments) of interest were exempt versus nonexempt and preperiod versus postperiod (ie, copayment elimination phase), where exempt versus nonexempt is the “whole plot factor” and preperiod versus postperiod is the “subplot factor.” Therefore, the linear mixed-effects model included fixed effects for copayment exemption status, copayment elimination phase (ie, preperiod or postperiod), and their interaction; and random effects for clinic and the clinic by copayment exemption status interaction. We included a third random effect for month, nested in the copayment exemption status by copayment elimination phase interaction. This model also controlled for clinic average age, average DCG score, proportion female, proportion white, and proportion married as additional fixed effects in the model.

We used a negative binomial mixed-effects model to estimate differences in the number of MOVE! clinic visits. The fixed effects in the model included the main factors of interest: an indicator for copayment eligibility, an indicator for the preperiod or postperiod cohort, and an interaction of the copay eligibility and pre–post indicators. This interaction term indicates whether use of the MOVE! clinic changed significantly in exempt vs nonexempt veterans between the preperiod and postperiod. Additional fixed effects included veterans’ age, DCG score, sex, race, and marital status. A clinic-level random effect was included in the model to account for the intraclass correlation of patients from the same clinic. All analyses were conducted using SAS version 9.2 (SAS Institute, Inc, Cary, North Carolina) and Stata version 10 (StataCorp LP, College Station, Texas).

During the year before the copayment elimination, the mean clinic-standardized rate of new MOVE! clinic use was higher for new users who were copayment exempt (18.0%) than for those required to pay copayments (3.6%), which suggests that for every 2 nonexempt veterans newly using MOVE! in the preperiod, 10 exempt veterans newly used MOVE! The mean clinic-standardized rate by veterans exempt from copayments and new to the MOVE! program increased from 18% to 30% of all clinic visits, while the mean clinic-standardized rate by veterans formerly required to pay copayments and new to the MOVE! program increased from 3.6% to 5.8%. Compared with nonexempt new users, the relative increase in the clinic-standardized rate new MOVE! clinic visits by exempt veterans was 12% in the year before the copayment elimination and 2.2% after. In the adjusted analysis (Table 1), we found a significant interaction between copayment elimination phase and copayment exemption status (−0.10, 95% confidence interval [CI], −0.15 to −0.04).

In the preperiod and postperiod, compared with nonexempt patients, exempt patients were younger (mean age, 55 y vs 63 y), more likely to be female (12% vs 7%), less likely to be white (50% vs 54%), and less likely to be married (52% vs 70%) and had higher mean DCG scores (0.75 vs 0.61) (all P < .001, Table 2). The unadjusted mean number of MOVE! clinic visits in the 6 months after the first visit was 3.64 (standard deviation [SD] = 4.55) by exempt veterans and 3.63 (SD = 4.22) by nonexempt veterans in the preperiod and 3.59 (SD = 4.82) and 3.60 (SD = 4.97) in the postperiod, which corresponds to $54 per person ($15/visit × 3.6 visits) in waived copayments.

On the basis of the interaction between newly exempt and postperiod indicators in the adjusted analysis, we found no change from the preperiod to postperiod in the number of visits by new users who were exempt versus those who were required to pay copayments (Table 3). Veterans in the postperiod had a lower visit rate (0.98; 95% CI, 0.96-0.99) to MOVE! clinics than veterans in the preperiod. Visit rates were higher among female veterans than male veterans (1.12; 95% CI, 1.10–1.15), among married veterans than unmarried veterans (1.04; 95% CI, 1.02–1.05), and among white veterans than nonwhite veterans (1.13; 95% CI, 1.11–1.15).