A Pay-for-Performance Initiative to Reduce Pediatric Psychiatric Inpatient Length of Stay

All hospitalized youths at the eight P4P hospitals were identified by using Beacon authorization data matched with DSS claims data to obtain all community-based behavioral health care service information. In the year prior to the P4P program, there were 2,237 inpatient admissions for 1,506 members. For P4P periods 1, 2, and 3, the number of inpatient stays were 882 (715 members), 1,975 (1,408 members), and 951 (782 members), respectively.

The average LOS decreased significantly from baseline, by 25%, by the final year of the P4P program (Figure 1). Moreover, there were no significant changes in 7- or 30-day readmission rates over the course of the P4P program.

With respect to timely postdischarge care, the first service for most members was outpatient care from either a behavioral health provider or medical practitioner. For those who were readmitted within 30 days, 69% (N=555) saw an outpatient provider within an average of four days of leaving the hospital. Although 75% (N=3,934) of those who were not readmitted within 30 days also received timely outpatient aftercare, the average number of days from discharge was notably longer (19 days for behavioral health appointments, 20 days for medical visits). Six percent (N=48) of youths who were readmitted within 30 days went directly into residential treatment compared with 13% (N=682) of those who were not readmitted. Of those who were readmitted to the hospital within 30 days, 67% (N=537) had DCF involvement compared with 52% (N=2,728) of members who remained out of the hospital.

Whereas case management and home care services were not used by any members as a first service after leaving the hospital, case management services were received by the 30-day readmitted cohort and the non-readmitted cohort at similar rates during the first 180 days postdischarge (4.0% and 3.6%, respectively). Disproportionately more youths who were readmitted within 30 days received services from outpatient medical providers (5.6% versus 0.6%).

Notable differences were observed in the estimated cost per member for 180-day postdischarge services (i.e., inpatient, ED, case management, outpatient medical and behavioral, and residential care). For members who were rehospitalized within 30 days, the average cost was $98,787 compared with $25,136 for youths who were not readmitted within 30 days, with 83% of the greater cost due to repeated hospitalization. Six-month postdischarge services costs for DCF-involved youths ($76,174) were higher than those for non-DCF youths ($10,768) due to higher use of all community-based services.

With respect to service use during the first 180 days postdischarge, the cohort who was readmitted within 30 days received a higher average number of services compared with non-readmitted youths (29.5 versus 26.5). DCF-involved youths received a higher average number of paid units for case management or home care services compared with non-DCF-involved youths (16.3 versus 2.23) and outpatient behavioral health services (136.3 versus 14.8).

The results of this P4P initiative to reduce psychiatric inpatient hospitalizations for children and adolescents at eight hospitals in Connecticut demonstrate that reductions in the average LOS are achievable without apparent deleterious outcomes. From 2008 to 2010, when hospitals received financial incentives for performance improvements, the average duration that youths spent on psychiatric inpatient units was reduced from 18.1 to 13.6 days. During the same period, the 7- and 30-day readmission rates remained stable with nonsignificant decreases.

For youths who were readmitted within 30 days, there are no indications that barriers to accessing timely aftercare were a factor. Compared with youths who were not rehospitalized within 30 days, the readmitted cohort received outpatient follow-up care significantly sooner after discharge (3.3 days for readmitted versus 20.5 days for non-readmitted) and received more services overall in the 180 days after leaving the hospital.

Although these findings suggest that the P4P was effective in reducing average LOS, caution is warranted in concluding that these changes resulted solely from the P4P initiative. Although consistent with the scope of other P4P evaluations (9), this uncontrolled, pre–post design did not account for other initiatives that were concurrently implemented and may have affected outcomes, including expanded emergency mobile psychiatric services for youths and increased rates for child-focused behavioral health outpatient clinics that agreed to provide access to services within specific shorter timeframes. The descriptive nature and lack of information about other clinically pertinent information (e.g., diagnoses, medications, and parental involvement) prevented more advanced statistical modeling of service use patterns and outcomes.

The process used to develop and the timing of the initiative may also have affected the results. The P4P methodology and goals were developed collaboratively with the participating hospitals and DCF during a time when youths were frequently delayed for several days in the ED waiting for a pediatric behavioral health inpatient bed. The ED delays were a burden to the hospitals, because they are not equipped to care for psychiatric patients for prolonged periods in an ED setting. Inpatient stays were also often protracted, particularly for youths with DCF involvement, because DCF delayed making necessary decisions about placement after discharge. Those delays, beyond the immediate control of the hospital, extended the LOS of youths ready to leave the hospital. Enhanced assistance in discharge planning to hospital staff by Beacon intensive case managers, including on-site participation in clinical rounds, coupled with quarterly regional and statewide meetings between the hospitals and DCF, was also included in the initiative to address barriers to timely discharge.

This column appears to be the first published evaluation of a P4P initiative focused on pediatric mental health services. Moreover, most of the extant literature on P4P in behavioral health is focused on adult substance use (9). Additional research is needed to explore patient, hospital, and regional service environment characteristics that may influence health care processes and outcomes to identify metrics of quality for pediatric mental health treatment across the continuum of care.

These results suggest that P4P initiatives can facilitate necessary change in systems of care providing pediatric mental health services. In addition to changes in care practices at each of the hospitals, equally important were changes between the hospitals and with DCF, including improved collaboration and communication between all organizations. Although complete service use data are not available, the CT BHP continues to monitor the average LOS at the hospitals that participated in the P4P program. The P4P initiative was discontinued due to economic constraints associated with recovering from the Great Recession of 2008; however, data reveal that shortened LOS has persisted with linear annual reductions through 2016 (10). We believe much of the success is attributable to continuation of the activities that evolved during the initiative. For the example, the CT BHP continues to meet quarterly with each hospital to review data on LOS, readmission, and postdischarge care. Additionally, the quarterly and semiannual meetings have continued, during which hospitals and DCF share continuous quality improvement efforts and discuss solutions to barriers. Dedicated DCF staff have also been assigned to each hospital. In brief, activities that promoted a sense of unity between the involved organizations during the P4P have persisted. These outcomes suggest that a period of incentivized quality improvement efforts can spur positive and enduring changes in the provision of acute behavioral health services for youths that affect the entire system of care.

Steven S. Sharfstein, M.D., Haiden A. Huskamp, Ph.D., and Alison Evans Cuellar, Ph.D., are editors of this column.