MENTAL-HEALTH-CARE UTILIZATION IN PREPAID AND FEE-FOR-SERVICE PLANS AMONG DEPRESSED-PATIENTS IN THE MEDICAL OUTCOMES STUDY

Objective. We compare mental health utilization in prepaid and fee-for -service plans and analyze selection biases. Data Source. Primary data were collected every six months over a two-year interval for a panel of depressed patients participating in the Medical Outcomes Study, an observational study of adults in competing systems of care in three ur ban areas (Boston, Chicago, and Los Angeles). Study Design. Patients v isiting a participating clinician at baseline were screened for depres sion, followed by a telephone interview, which included the depression section of the NIMH Diagnostic Interview Schedule. Patients with curr ent or past lifetime depressive disorder and those with depressed mood and three other lifetime symptoms were eligible for this analysis. We analyze mental health utilization based on periodic patient self-repo rt. Analytic Methods. We use two-part models because of the presence o f both nonuse and skewness of use. Standard errors are corrected nonpa rametrically for correlations across observations due to clustered sam pling within participating physicians and repeated observations on the same individual. Principal Findings. The average number of mental hea lth visits was 35-40 percent lower in the prepaid system, adjusted and unadjusted for observed differences in patient characteristics, inclu ding health status. Utilization differences were concentrated among pa tients of psychiatrists, with only minor differences among patients of general medical providers. Analyzing the effect of switches that pati ents make between payment systems over time, we found some evidence of adverse selection into fee-for-service plans based on baseline utiliz ation, but not based on utilization at the end of the study. In partic ular, after adjusting for observed patient characteristics and health status, patients switching out of prepaid plans had higher baseline us e than predicted, whereas patients switching out of fee-for-service ha d lower use than predicted. Switching itself appears to be related to an immediate decline in utilization and was not followed by an increas e or ''catch-up'' effect. Conclusions. The absence of the commonly fou nd ''catch-up'' effect following switching and the significant decreas e in utilization during the switching period suggests an interruption in care that does not occur for patients staying within a payment syst em. This finding emphasizes the need for integrating new patients quic kly into a system, an issue that should not be neglected in the curren t policy discussion.