Impact of State Psychiatric Hospital Waitlists on Monthly Admissions

Nationally, deinstitutionalization led to the closure of many SPHs and the conversion of remaining hospitals to long-term care for civil and forensic populations. In North Carolina (NC), this rapid SPH deinstitutionalization did not occur; no SPHs were closed, and only gradual reductions in use were observed (7).

The situation began to change following the U.S. Supreme Court’s Olmstead v. L.C. decision in 1999 and NC’s passage of Mental Health Reform House Bill 381 in 2001, which aimed, in part, to reduce reliance on SPHs and shift care to community settings (8). Between 2001 and 2006, NC’s four SPHs decreased bed size from 1,755 to 1,180 (9). However, reliance on SPHs for acute care persisted because the state’s funding and policy positions did not fully support development of local inpatient alternatives (10). By fiscal year 2007, NC had the highest absolute number of annual SPH admissions nationally (11).

On February 6, 2007, NC’s Division of Mental Health, Developmental Disabilities, and Substance Abuse Services announced a SPH waitlist policy to address urgent issues precipitated by SPHs operating over capacity, including patient and staff safety concerns (7; Moseley M, memorandum to community hospital CEOs and LME directors, Feb 6, 2007). The policy instructed SPH administrators to begin delaying patient admissions when the hospital’s admission unit capacity exceeded 110%. Dates of waitlist implementation varied across the state’s four SPHs as a function of each hospital’s actual level of overcrowding (February 2007, western/south-central regions; March 2007, north-central region; and April 2008, eastern region).

At each SPH, waitlists generally operated as first-in, first-out queues, in which the patient waiting longest was admitted when a bed became available. However, patients with extremely acute illness were prioritized for admission. Not all patients placed on waitlists were ultimately admitted to SPHs. This study estimates the effects of these waitlist practices on the volume and case mix of monthly admissions to SPHs in NC.

Administrative data from NC’s Division of State-Operated Healthcare Facilities were used to determine the monthly number and proportion of admissions with patient-level characteristics at each SPH. State hospital waitlist data were used to determine when each hospital began operating on a waitlist.

Other data sources were used to control for time-varying regional factors potentially affecting SPH admissions, including demographic composition, unemployment rates, and alternative treatment availability in each region [see online supplement].

The sample included all nonforensic short-term and long-term treatment unit admissions to NC’s four SPHs between January 2004 and November 2010 for patients ages 18–64 years at admission. This analysis excluded forensic admissions, which were not subject to the waitlist policy. Visit-level admissions data were aggregated to the SPH-month level to reflect the monthly number and case mix of admissions at each SPH.

Outcome variables included the number of monthly SPH admissions overall, as well as the number and proportion of monthly admissions by sex, age, race, insurance status, criminal involvement, involuntary commitment status, and diagnosis.

The key independent variable was a binary indicator for whether the hospital-month observation was pre- or postwaitlist interacted with a linear time trend to examine the main effects of the waitlist policy, allowing both the regression intercept and slope to change postwaitlist. In all analyses, the postwaitlist period was defined by taking into account the differential implementation of waitlists (6).

Other independent variables in regression analyses controlled for potentially confounding factors affecting SPH admissions, including indicators for calendar months to control for seasonality and time-varying SPH region characteristics. Regional characteristics included each SPH catchment area’s full population demographic composition (by sex, age, and race), unemployment rate, and available mental health services (numbers of licensed psychiatrists and adult beds in general hospital psychiatric units or private psychiatric hospitals per 100,000 population).

Descriptive analyses compared monthly admissions pre- and postwaitlist by using two-sample t tests with unequal variances, with subgroup analyses of the patient characteristics described above. Hospital-level fixed-effects ordinary least-squares regression models with robust standard errors were estimated to further examine the effect of the waitlist policy on the overall number of admissions, as well as the number and proportion of admissions with specific diagnoses. Fixed effects controlled for time-invariant regional and SPH-level characteristics that may have affected hospital use.

Analyses were conducted using Stata, version 12.1, with an alpha of .05 for statistical significance. The study was approved by the institutional review board at the University of North Carolina at Chapel Hill.

Between January 1, 2004, and November 30, 2010, there were a total of 72,035 nonforensic adult admissions to NC’s four SPHs. The unadjusted mean±SD absolute number of monthly SPH admissions decreased by 46.4% following waitlist policy implementation (from 281.0±42.2 to 150.6±65.8 admissions; p<.001). The scale of reductions varied by hospital from a mean of 39.2% to 60.2% fewer monthly admissions postwaitlist (p<.001). Reductions occurred for all subgroups analyzed, although changes in the unadjusted relative percentage distribution of demographic characteristics prewaitlist and postwaitlist were not clinically meaningful (Table 1).

Regression results indicate that the waitlist policy was associated with decreases in the absolute number of monthly admissions both overall and for most admissions with specific diagnoses [see online supplement]. Waitlist implementation was associated with an average of 53.1 fewer total monthly admissions per hospital across all months postwaitlist (p<.001) and an incremental 1.6 fewer admissions in each additional month postwaitlist (p=.01). Waitlists were associated with a decrease of 31.8 in the absolute number of monthly admissions with severe mental illness diagnoses across all months postwaitlist (p<.001) and an incremental 1.4 fewer monthly admissions with severe mental illness in each additional month postwaitlist (p<.001). Furthermore, there was a decrease of 43.7 in the absolute number of monthly admissions with substance abuse diagnoses across all months postwaitlist (p<.001).

Estimates of the changes in proportions of monthly admissions further indicate that the waitlist policy was associated with a 4.2% reduction in the percentage of monthly admissions with substance abuse diagnoses agnostic to severe mental illness status across all months postwaitlist (p=.003), with an offsetting increase in the percentage of non–substance abuse admissions. Among the five mutually exclusive diagnosis categories, this effect was concentrated among individuals with severe mental illness, with a decrease in the percentage of admissions for severe mental illness/substance abuse (−4.1%; p=.005) and an increase in the percentage of admissions for severe mental illness/nonsubstance abuse (3.7%; p=.003).

Results quantified the extent to which waitlists helped to limit the number of SPH admissions. Although the hospitals processed fewer patients postwaitlist, the demographic characteristics of patients remained largely the same before and after the waitlist from a clinically meaningful standpoint, according to descriptive analyses. However, regression results indicated that the diagnostic composition of postwaitlist admissions changed during the study period, with the relative percentage of patients with substance abuse diagnoses markedly lower compared with the prewaitlist percentage. In this respect, the proportion of patients with co-occurring diagnoses decreased, and the proportion of patients with single diagnoses increased, potentially transforming the case mix toward less complexity and reducing overcrowding on hospital treatment units, both long sought-after goals for SPHs (12). However, additional research is needed to determine whether patients with co-occurring disorders were admitted to state-operated substance abuse treatment facilities to a greater extent postwaitlist and, if so, whether these facilities also operated on waitlists or experienced compensatory changes in their admissions case mix.

Because the waitlist policy was implemented statewide, this study lacked a control group of NC SPHs. It is possible that other changes to the mental health system affecting SPH utilization may have occurred around the same time as the waitlist policy. In an effort to rule out these other possibilities, the postwaitlist period was defined on the basis of SPH-specific policy implementation rather than statewide policy announcement.

Analyses also controlled for time-varying regional characteristics and used fixed effects to control for time-invariant factors. Analyses were unable to control for certain measures, such as regional capacity of psychiatric crisis facilities and assertive community treatment teams, because of a lack of time-series data on these measures. However, analyses controlled for the regional number of psychiatrists, which may fluctuate on the basis of the capacities of these specialized psychiatric services.

This study was unable to evaluate the type of care received and outcomes for waitlisted patients who were not admitted to an SPH. Further research is required to understand whether increased admissions control leads to improved SPH outcomes, how psychiatric crises are managed at the community level, and the circumstances that lead to SPH referral.

Waitlists and patient delays will likely persist, given the substantial cuts to state mental health budgets and decreased availability of psychiatric beds over the past decades (13). These admission pressures will intensify until efforts are made either to increase psychiatric beds (at state, private, and local general hospitals) or to expand community capacity to deal with psychiatric crises in other ways.

Waitlist policies were associated with reduced monthly admissions and a transformed monthly admissions case mix at NC SPHs. Whether these intrahospital effects also had external consequences for management of psychiatric crises in alternative community settings is an important agenda item for future research.

North Carolina’s Division of State-Operated Healthcare Facilities is acknowledged for their assistance providing data for the study.