Use of Longer Periods of Temporary Detention to Reduce Mental Health Civil Commitments

Data from the CMS database on commitment hearings for three quarters—from July 1, 2008, through March 31, 2009—were combined with Virginia's Medicaid claims files for individuals with a mental health diagnosis. The study was reviewed by the University of Virginia's institutional review board and received exempt status because all data were blind-coded and deidentified before being provided to the researchers. Data were analyzed using Stata, version 10, statistical software. The CMS database contained an entry for each hearing, regardless of whether it was at the beginning or end of the TDO. Anindividual's hearings were matched on the basis of the second hearing's having a later date than the TDO and the person's first name and birth date. This method was less likely to yield errors than matching by last name and birth date. It resulted in 9,005 matches. The CMS data were then matched to the Medicaid claims file with MySQL. Initially there were 5,497 observations. After exclusion of repeat entries for the same visit, care related to pharmacy, personal care, physician visits, and hospital outpatient care, 508 matches resulted. Because of data entry errors, we do not know the precise number of Medicaid matches. For example, we excluded 402 entries without an identification number and 180 without a CSB number, some of which may have been a match.

The Medicaid claims file provided information on the length of the hospitalization, if any, that immediately followed the commitment hearing, as well as information on the age, race, sex, and primary diagnosis of the Medicaid recipients. We excluded individuals with more than 21 days of hospitalization or two standard deviations above the mean (N=8). Many of these extended stays were by individuals who were hospitalized indefinitely, and the commitment hearings were held in order to renew their stay. However, excluding them did not significantly change the results. The final data set included a sample of Medicaid recipients (N=500) who had a mental health diagnosis and at least one TDO during the study period.

A logistic model was used to formally test whether longer TDOs were correlated with an increased probability of dismissal, including mandatory outpatient treatment, rather than hospitalization; and an increased probability of voluntary rather than involuntary hospitalization. A multivariate regression was then used to investigate the relationship between TDO length and the length of subsequent hospitalization. Covariates included the CSB serving the individual with the TDO, as well as the person's sex, age, and race. A dummy variable indicated whether the primary diagnosis from the Medicaid claims file involved a serious mental illness, which we defined as an ICD-9 diagnosis of schizophrenic disorder (ICD-9 code 295) (N=155) or affective psychosis (ICD-9 code 296) (N=232). TDO was treated as a continuous variable in the logistic regression and a categorical variable in the multivariate regression analysis.

Table 1 presents data on the characteristics of the 500 persons in the sample. Race was categorized as white (71%), black (26%), or other (3%). The mean age was 39.7 years, and women made up more than half of the sample (62%). Thirty-one percent of the sample had a diagnosis of schizophrenic disorder, and 46% had a diagnosis of affective psychosis.

Involuntary commitment can result in both stigma and trauma to individuals with mental illness. Therefore, an important policy goal is to reduce involuntary hospital commitments. The findings are consistent with extending TDO length as a means to this goal. The study found that extending TDOs beyond the 48 hours required under current law in Virginia could reduce the level of coercion needed to restrain and treat people experiencing mental health crises. In our sample of Medicaid recipients, longer TDO periods were correlated with increased rates of dismissal. When hospitalization occurred, the stay was more likely to be voluntary rather than involuntary. Furthermore, longer TDO periods were correlated with shorter subsequent hospital stays. These results are promising and point to the need for more research to establish a causal link between TDO length and health outcomes.

Although improving health outcomes is in itself an important goal, policy makers must also consider the costs of proposed changes. Results from this study allowed us to make a preliminary estimation of the effect of changing Virginia's law to extend the maximum period of temporary detention before a commitment hearing from 48 to 72 hours. The results suggest a net increase of 1,782 treatment days; however, that estimate might be reduced if a minimum TDO of 24 hours was also implemented.

Ultimately, the impact on the state's budget of increasing the maximum TDO period would depend on the differences in costs and funding sources between TDO detention days and post-TDO hospitalization days, as well as on indirect costs related to public health and safety. This assessment highlights the complexity of the interaction between TDO length, hospitalization, and costs. Although the estimates provide an indication of the expected effect, more comprehensive data are needed to more accurately predict the budgetary implications of changing the TDO law.

This study has several limitations. The data available allowed us to exploit the natural variation that occurs as a result of provisions in the law and the institutional structure of the CSBs and courts. This quasi-experiment is valid only if we can rule out selection bias. The most likely source of bias is the influence of an individual's health on the length of both the TDO period and the hospitalization. For example, clinicians may fast-track certain individuals into involuntary commitment because they are familiar with the cases and know what the final disposition will be. These individuals would also have longer hospitalizations because of their health. However, evidence suggests that clinicians generally have little discretion over TDO length. For example, an informal survey conducted by Virginia's Commission on Mental Health Law Reform found that many jurisdictions hold TDO hearings every other day (10). When a person is detained, the TDO hearing is scheduled for the next day that hearings are held. Other jurisdictions have different policies, but TDO length is still determined at the institutional rather than the clinician level. For example, officials at the Prince William County CSB reported that one aspect of their civil commitment process that helped make mandatory outpatient treatment more feasible was that they usually allow a full 48 hours for the TDO period (10). This suggests that an individual's health does not determine TDO length and reduces the possibility that an individual's health is a third variable influencing both the length of the TDO period and the subsequent hospitalization.

There are additional sources of potential bias that may have influenced the results. The sample was drawn entirely from Medicaid recipients. However, persons with private insurance represent the largest proportion of inpatient mental health care users in Virginia (11). Thus, to the extent that Medicaid recipients differ from other mental health patients, the results are not applicable to the broader population of the state. For example, we might expect women to be more likely to qualify for Medicaid, and, in fact, women made up more than half the sample. However, even if the results apply only to Medicaid recipients, they have important implications for public policy. In addition, certain types of individuals may be more likely to be committed over a long weekend, such as those with a history of violence or comorbid substance abuse, which represents another source of bias. We have no evidence of such a bias, but future research should explore this possibility.

The study found that the location of the TDO had a significant effect on hearing outcome and hospitalization length. This result is consistent with the Commission on Mental Health Law Reform hearing study, which found significant variation in TDO practice across the CSBs (4). It is unlikely that the sample from the CSBs was random, and their record keeping accuracy also likely varied in systematic ways. Our data set contained a disproportionately large number of observations from the Blue Ridge Behavioral Healthcare CSB, which accounted for 37% of the total. In the jurisdiction served by this CSB, periods of TDOs were significantly shorter—a mean±SD of 1.36±.83 days, compared with 1.75±1.14 days for the rest of the sample. However, hospitalization days were not significantly different in this CSB location compared with the rest of the sample. Because each CSB structures its TDO period independently, it is not surprising that TDO periods differed by CSB. However, including the CSB as a control variable accounted for most of the variation. Some individual characteristics that influenced hospitalization but were not fully captured by the CSB variable may have been more prevalent in some CSB locations. Socioeconomic status might be an example, but it was not relevant in this study of Medicaid recipients.

This study is the first to clearly identify the relationship between TDO length and the result of the commitment hearing and subsequent hospitalization. Although more research is needed to determine costs, the findings are consistent with use of longer TDO periods as a means to improve health outcomes of individuals experiencing mental health crises.

This study was funded by Virginia's Commission on Mental Health Law Reform. The authors acknowledge the data assistance of Rhonda Newsome, M.P.P., and Aaron Bloomfield, Ph.D.

The authors report no competing interests.