Prevalence of use of nonpreferred bipolar medications before, during, and after a prior-authorization policya Cohorts were continuously enrolled in Medicare, Medicaid, or both. Fitted trend lines show predicted values estimated from segmented time-series regressions for the Maine study (N=5,336) and New Hampshire comparison (N=1,376) groups. The last two observation periods (September and October) for New Hampshire were omitted because they occurred after the state started its own prior-authorization policy in September 2004.