Does involuntary out-patient treatment work?

There has been considerable debate in the UK in recent years about the propriety of using various forms of ‘involuntary out-patient treatment’ for some individuals with serious mental illness (Eastman, 1995; Dyre, 1998; Burns, 1999; Moncrieff & Smyth, 1999; Sugarman, 1999). While most jurisdictions in the US have statutes that support involuntary out-patient treatment (Torrey & Kaplan, 1995), its use remains controversial (Slobogin, 1994; Diamond, 1995; Torrey, 1997). Involuntary out-patient treatment was originally proposed as a solution to the ‘revolving door syndrome’ (Geller, 1996). It has also been recognised, however, that it may provide a solution to the clinical and ethical dilemmas of allowing individuals who are incapable of making treatment decisions to discontinue treatment, with predictable deterioration to the point where they may harm themselves or others (Geller, 1990). This paper provides a review of controlled studies that have examined whether involuntary treatment in the community is effective.

In this article I will use the term involuntary out-patient treatment for all forms of compulsory out-patient treatment except provisions that apply to individuals who are charged with, or have been found not criminally responsible (because of a mental disorder) for, a criminal offence. It is important at the outset to note that studies of the effectiveness of involuntary out-patient treatment examined a variety of forms of involuntary out-patient treatment such as conditional leave from hospital and court-ordered community treatment. Most of the studies were undertaken in the US where decisions to commit individuals to out-patient treatment are made by the courts and not directly by mental health professionals. Therefore, the findings from these studies may not be applicable to jurisdictions where a health professional, rather than a judge, is the arbiter of the appropriateness of involuntary out-patient treatment. Moreover, the consequences of non-adherence to treatment varied markedly between the jurisdictions where these studies were conducted. In some jurisdictions there were no legal consequences if patients defaulted on treatment and in others no action was taken even though legal consequences were available.

The most common outcome measure used in involuntary out-patient treatment research, is the amount of hospitalisation an individual experiences, measured either as the total number of admissions or the cumulative days spent in hospital over a given period. The use of hospitalisation as a primary outcome measure may not be appropriate for patients who receive involuntary out-patient treatment because some of these individuals may actually benefit from extended hospital stays (Draine, 1997). The alternative to hospitalisation is often imprisonment or homelessness.

If the raison d'être for involuntary out-patient treatment is to solve the problems of the revolving door patient by maintaining these individuals in the community, then hospitalisation is the most appropriate measure. If, however, the goal is to reduce morbidity and mortality or to improve the quality of patients' lives, then other measures are more suitable. A number of the studies have looked at outcomes apart from hospitalisation, including adherence to follow-up treatment, violent behaviour, quality of life, stability of residence and ability to work.

Hospitalisation as the outcome measure

Most studies have demonstrated a reduction in hospital utilisation associated with involuntary out-patient treatment (Bursten, 1986; Zanni & deVeau, 1986; Van Putten et al, 1988; Fernandez & Nygard, 1990; Sensky et al, 1991; Munetz et al, 1996; O'Keefe et al, 1997; Geller et al, 1998; Steadman et al, 2001) although in some this reduction was not statistically significant (Steadman et al, 2001). The study by Bursten is sometimes quoted as failing to demonstrate significant reduction in hospital utilisation, but it is a complex study where the involuntary out-patient treatment patients showed significantly reduced rates of hospital utilisation compared with only one of two control groups. One study from North Carolina reported a non-significant increase in hospitalisation with involuntary out-patient treatment (Hiday & Scheid-Cook, 1989).

Some of these studies have methodological weaknesses such as unclear definition of the target population, selection bias, small sample size and lack of uniform enforcement of involuntary out-patient treatment provisions. However, the major methodological problem is that a number of these studies use a ‘pre v. post’ design, where patients act as their own controls (O'Keefe et al, 1997; Fernandez & Nygard, 1990; Sensky et al, 1991; Munetz et al, 1996; Zanni and deVeau 1986; Rohland et al, 2000). This methodology ignores the possibility of temporal changes in rates of hospitalisation. Indeed, we know that the availability of psychiatric beds in the US has been progressively decreasing (Torrey, 1997).

As most pre v. post studies were conducted over a limited time period, probably a more serious criticism of the design is the possible influence of regression towards the mean — that is, patients are placed on involuntary out-patient treatment because they have extremely high hospital utilisation and probability alone dictates that they are unlikely to maintain this extreme utilisation over time. Every study that has used patients as their own controls demonstrated a significant reduction in hospital utilisation (Zanni & deVeau, 1986; Fernandez & Nygard, 1990; Sensky et al, 1991; Munetz et al, 1996; O'Keefe et al, 1997; Rohland, 2000). On the other hand, while the five studies that have used separate groups of patients as controls (Bursten, 1986; Van Putten et al, 1988; Geller et al, 1998; Steadman et al, 2001; Swartz et al, 1999) show a reduction in the use of hospitalisation, only Van Putten et al reported an unequivocally statistical significant reduction in hospital utilisation. Moreover, Geller et al (1998) demonstrated that significant reductions in hospitalisation observed in both a pre v. post design and when the involuntary out-patient treatment group was compared with controls matched for clinical and demographic variables, disappeared when controls were matched for prior hospital utilisation.

Clearly the field needed a randomised controlled study and the results of two such studies have been recently reported. Steadman et al (2001) examined the effectiveness of a pilot study of involuntary out-patient treatment run by the Bellevue Hospital in New York City. This study identified no statistically significant difference in the number of hospital admissions or the number of days spent in hospital between a group of patients who had been directed by the court to follow a specific out-patient treatment plan when discharged, as opposed to a group of patients who had been discharged without a court ordered treatment plan. Closer scrutiny of this study reveals that patients who were on court orders spent a mean of 43 days in hospital during the 11-month follow-up period compared with a mean of 101 days for those discharged without a court order. Steadman et al (2001) did not give a P value to indicate how close their finding of reduced hospitalisation came to reaching statistical significance. The reported difference in hospital utilisation is clinically significant even though it did not reach statistical significance. It is worth noting that owing to uncertainty about the extent of legal powers of the pilot study, no patient was conveyed to hospital for evaluation and no patient was medicated over objection — that is, there were no consequences for non-compliance (Telson et al, 1999).

A second randomised control trial conducted in North Carolina also found no significant difference in number of days spent in hospital, or the number of admissions between patients committed to involuntary out-patient treatment and those released by the courts (Swartz et al, 1999). In a post hoc analysis those patients who spent more than 6 months on treatment orders had significantly fewer days in hospital and fewer admissions. Reading this study it is hard to exclude the possibility that subjects on involuntary out-patient treatment who were admitted to hospital did not have their out-patient committal orders renewed when they became due (every 3 months) and that this caused a spurious association between increased hospital utilisation and short duration on involuntary out-patient treatment. A re-analysis of the data did not find a trend for out-patient committal to lapse while patients were hospitalised: many patients actually had their out-patient committal orders renewed because they had been non-compliant with follow-up leading to hospitalisation (Dr Marvin Swartz, personal communication). Moreover, it is notable that patients on involuntary out-patient treatment for 6 months or longer were more disabled at the start of the study than those who did not continue on involuntary out-patient treatment. Swartz et al also examined the clinical characteristics of patients who did best on involuntary out-patient treatment and reported that patients with psychoses or with serious mental illness associated with substance misuse and violent behaviour showed the greatest reduction in hospital utilisation.

Other outcome measures

Patients' reliability in attending out-patient appointments (Van Putten et al, 1988; Hiday & Scheid-Cook, 1991; Sensky et al, 1991; Munetz et al, 1996; Roland, 2000) and taking prescribed medication (O'Keefe et al, 1997) is significantly improved while they are receiving involuntary out-patient treatment. Hiday and Scheid-Cook employed a control group negating the influence of regression toward the mean. Furthermore, Hiday and Scheid-Cook and Van Putten et al demonstrated that improved attendance at follow-up appointments by patients on court orders persisted even after the court orders expired.

A number of researchers have found that involuntary out-patient treatment results in a significant decrease in violent behaviour in the community when patients are placed on treatment orders (O'Keefe et al, 1997; Sensky et al, 1991). Swanson et al (2000) examined the incidence of violent behaviour in patients in the North Carolina randomised control trial and observed a similar trend as reported by Swartz for hospitalisation, that is, involuntary out-patient treatment reduced violent incidents but only in that group of patients who remained on treatment orders for longer than 6 months. Swanson et al used logistical regression analysis to examine the relationship between involuntary out-patient treatment, intensity of clinical services and violence, and concluded that “neither the court order nor services alone were effective in reducing violence [but that] with regular service use the presence of extended out-patient committal reduced the probability of violence from 48% to 24%”. In contrast the Bellevue Pilot Study found no differences in arrests for criminal behaviour between involuntary out-patient treatment patients and controls. Munetz et al (1996) and Hiday and Scheid-Cook (1989) reported a statistically non-significant decrease in criminal behaviour for patients receiving involuntary out-patient treatment.

Three studies examined the effect of involuntary out-patient treatment on patients' ability to maintain employment. O'Keefe et al (1997) found that this was significantly increased for patients receiving involuntary out-patient treatment, Hiday and Scheid-Cook (1989) reported a non-significant improvement, while Munetz et al (1996) did not find any difference. Steadman et al (2001) found no differences in measures of quality of life of patients on or off treatment orders. Munetz et al (1996) and Rohland et al (2000) reported a reduction in emergency room visits for the patients on treatment orders. O'Keefe et al reported that patients' misuse of psychoactive substances decreased when patients received involuntary out-patient treatment. Finally, O'Keefe et al also found that patients were more likely to live in a stable residence while receiving involuntary out-patient treatment.

Conclusion

The literature suggests that involuntary out-patient treatment can reduce hospital utilisation for many patients with severe mental illness, but its ability to do so likely depends on the clinical characteristics of patients for whom it is used, the consequences of non-compliance, the rigour with which consequences are applied and on the duration for which treatment orders are maintained. It is clear that involuntary out-patient treatment does not increase the use of hospitalisation by forcing large numbers of non-compliant patients into hospital, as has sometimes been suggested. Attendance for follow-up appointments and adherence to medication regimes are improved when patients receive involuntary out-patient treatment. Moreover, the limited data suggest that this enhanced engagement with services persists even after involuntary out-patient treatment is discontinued. This finding should assuage fears that the coercive nature of involuntary out-patient treatment causes long-term damage to the therapeutic relationship. The conclusions about the impact of involuntary out-patient treatment on violent behaviour in the community are less clear, with some studies demonstrating decreased violence while others have failed to show an effect. Selection bias may have influenced these findings as patients who were at high risk for violent behaviour were excluded from some studies.

Many mental health professionals who have used involuntary out-patient treatment believe that it has been very effective in bringing stability to the lives of some of their most difficult patients (Miller & Fiddleman, 1984). Individual success stories may get lost in aggregated data, for example, Munetz et al (1996) noted “striking individual differences in the effectiveness of out-patient committal in reducing hospital use”, with 13 of 20 study patients never hospitalised while receiving involuntary out-patient treatment. Of the four patients in this study who had increased levels of hospitalisation, two were known to be non-responders to medication and thus did not meet Geller's suggested clinical recommendations for the use of involuntary out-patient treatment (Geller, 1990).

Some scholars have suggested that involuntary out-patient treatment may exert its effect by committing the mental health system to provide services just as much as by committing the patient to use them (Swanson et al, 1997; Rohland et al, 2000). This view receives some empirical support from the observation that involuntary out-patient treatment must be combined with high levels of service to be effective (Swartz et al, 1999; Swanson et al, 2000) and needs further study. Future research should also try to identify the types of patients who are most likely to benefit from involuntary out-patient treatment, the optimal model of delivery of involuntary out-patient treatment and specifically test for posited negative effects such as alienation of patients from treatment providers. Ideally, quantitative research in this area would use randomised control trials. However, randomisation of the delivery services that depend on legal statute are intrinsically difficult to perform (Swartz et al, 1997) and we cannot expect a rash of such studies to assist us in our decision-making. Qualitative research methods have been relatively underutilised to date, but may enable us to address in detail the effect of involuntary out-patient treatment on the lives of individual patients, their experience of the coercive element of involuntary out-patient treatment and the views of families and professionals.