Purpose. A longstanding argument for increasing access to mental health services is that they offset the use of medical services (Fieldler & Wright, 1989). In fact, there is a substantial body of research dating back to the 1970s that has supported such an effect, although not without controversy (Mumford, et al., 1984). It has been pointed out that much of this research was undertaken prior to the widespread implementation of managed care within the mental health field, thus, its continuing relevance is uncertain (Eells, 1999). The purpose of this study, therefore, is to assess the extent that managed care is diminishing the capacity of inpatient psychiatric care to offset the utilization of medical services. It will also examine whether this effect varies based on the demographic characteristics of the patients involved, their geographic access to services, and the particular type of reimbursement available to them.

Methodology. The impact of managed care in mediating the offset effect was investigated through a secondary analysis of changes in psychiatric and medical hospital utilization over the 1995-2000 period. The data was taken from Massachusetts' statewide casemix hospital reporting system, and includes 1,896,042 unduplicated individuals. In addition to insurance type and annualized hospital days (psychiatric & medical), other variables included gender, age, race, diagnosis, severity, SES of home community, and distance from home to hospital. After extensive data preparations, including merging of annual files, unduplication, etc., the data was analyzed both through descriptive statistics as well as a latent growth curve model using LISREL (v8.71), with split half testing. This was supplemented by dynamical phase portraits.

Results. Two primary models were tested in this study with latent growth curve modeling (Hox, 2002). To assess the offset effect by itself, two independent variables -- mean and slope of psychiatric days -- were used to predict the mean and slope of medical utilization, with a three year lag. Each of these variables had three indicators or the three years of data (1995-1997 or 1998-2000). This revealed a very modest offset effect, with a negative standardized coefficient of -.15 between the slopes of psychiatric and medical utilization. However, when measures of insurance type (i.e. managed care vs. indemnity), and the various control variables were added to the above model, the offset effect becomes a substantial -.84, one that had been suppressed by the impact of both managed care provision as well as condition, both diagnosis and severity. The two models had split half cross validation (CVI) indices of .0087 and .00, respectively.

Implications. There is, thus, evidence that although psychiatric care continues to offset medical utilization (even with inpatient care considered alone); the extent of this impact is suppressed by the increasing provision of managed care, even when demographics, access, and condition variables are controlled for. This suggests that estimates of the costs of enhanced psychiatric coverage, for example that under parity laws, must take into account savings in the medical arena, as much as we might want to justify more generous benefits on humanitarian grounds alone.