Effects of a behavioral health carve-out on inpatient-related quality indicators for major depression treatment

OBJECTIVES. TO analyze the effects of the 1993 Massachusetts behavioral hea lth carve-out for state employees on readmissions and follow-up treatment a fter hospitalization for major depressive disorder (MDD). METHODS. The sample consisted of 218 continuous enrollees in preferred prov ider organization and/or indemnity plans who had any MDD admissions during fiscal years 1992 to 1995. These users accounted for 310 MDD admissions. El igibility files and behavioral health claims were used to analyze readmissi ons and follow-up treatment after discharge. Kaplan-Meier survival function s were obtained for pre/post (pre-carveout vs. post-carveout) comparisons o f the two indicators. Cox regression models were used to estimate carve-out effects on readmission and follow-up treatment while controlling for patie nt variables. Postdischarge contact categories were also compared. RESULTS. The risk of readmission did not change significantly after the car ve-out, in either the Kaplan-Meier or Cox regression analyses. Follow-up tr eatment was significantly more likely after the carve-out, including in the early postdischarge period. There was a significant decrease in the propor tion of discharged patients followed by readmission only, and a significant increase in patients receiving follow-up treatment prior to a readmission. CONCLUSIONS. Under this behavioral health carve-out, follow-up treatment wa s more likely, and estimated risk of readmission did not change significant ly for a seriously ill subgroup of enrollees. This was true even when contr olling for patient variables and using data for extended time "at risk" for each indicator. Future research on carve-outs should move toward direct cl inical quality measurement.