Schizophrenia is the most common diagnosis among patients who are involuntarily admitted to psychiatric hospitals and treated against their will (1). In most Western countries, the legal requirements for involuntary treatment include severe mental illness and dangerousness to self or others (2,3). These characteristics are frequently found among persons with schizophrenia (4) and often lead to involuntary hospitalization, especially of persons who lack insight into their illness (5). However, coercive treatment can lead to the development of posttraumatic stress disorder among patients with schizophrenia (6,7).
In previous studies, researchers found that involuntary treatment was successful for patients with anorexia nervosa (8) and substance abuse (9). However, very little research on involuntary treatment has been done with patients with schizophrenia, although use of involuntary treatment for these patients is widespread and rarely questioned. In one of the few studies in this area, Kjellin and associates (10) examined involuntarily admitted patients' reports of the ethical costs and benefits of treatment. The patient group consisted predominantly of patients with schizophrenia. Ethical benefits were defined as fulfillment of the ethical principles of beneficence and autonomy, and ethical costs were defined as any violation of those principles. The authors found that the great majority of patients reported improvement associated with psychiatric care and that a third reported experiencing ethical benefits only, without ethical costs. Høyer (11) pointed out that almost no scientifically sound studies have addressed the outcome of coercive treatment. The aim of the study reported here was to examine associations between the voluntary or coercive nature of patients' participation in inpatient treatment and the short-term outcome of inpatient treatment in a group of patients with schizophrenia who were receiving routine hospital care.
All patients between the ages of 18 and 65 years with schizophrenia or delusional disorder who were consecutively admitted to a 320-bed psychiatric hospital in Baden-Wurttemberg in southern Germany over a three-month period in 2002 were included in the study. The hospital serves a catchment area of 440,000 inhabitants that includes several small towns but no large cities. The patients were treated on seven wards, each of which serves a particular sector of the catchment area. Neither treatment concepts nor legal and clinical practices differed among the wards during the study period, as far as we were able to ascertain. All seven wards operate with a common set of guidelines for treatment and administrative procedures. The study site was the only psychiatric hospital within the catchment area. The patients included in the study met the criteria for an F2 disorder (schizophrenia or delusional disorder) according to the tenth revision of the International Statistical Classification of Diseases and Related Health Problems (12).
The physician who was responsible for the patient's treatment completed the Positive and Negative Syndrome Scale (PANSS) (13) and the German Version of the Global Assessment of Functioning (GAF) (14) at admission and discharge. Possible scores on the PANSS range from 30 to 210, with higher scores representing more severe symptoms. GAF scores can range from 0 to 100, with higher scores indicating better functioning. The psychopathological data were recorded as part of the competence network schizophrenia study, a large, multicenter study funded by the German Ministry of Education and Research to examine quality management in the treatment of schizophrenia. The study was approved by the ethics committees of the medical schools associated with the hospital. German legislation allows data recording and processing without individual informed consent as part of quality management in hospitals, and hospitals are obliged by federal legislation to perform quality management to achieve best clinical practices. Therefore, complete clinical data were available for all patients in the study, and even patients who were treated involuntarily had no missing data. As part of this project, all physicians on the wards participated repeatedly in rater training for the PANSS and the GAF.
Outcome was defined as the change in the PANSS total score between admission and discharge, the change in the GAF score between admission and discharge, the change in PANSS total score divided by number of inpatient days, and the change in GAF score divided by number of inpatient days. The latter measures were calculated to standardize the effect of different lengths of stay.
A more complex approach seemed necessary to define the voluntary or coercive nature of patients' participation in treatment. Voluntariness of participation was considered for five aspects of inpatient treatment—admission, hospital stay, taking medication as prescribed, and agreement with plans for discharge and continuation of treatment after discharge. The laws of the state of Baden-Wurttemberg—like those of other German states—formally define the legal status of admission and hospital stay as voluntary or involuntary. The law also stipulates that involuntary patients must accept the required treatment. Thus medication can be administered by coercion to patients with involuntary legal status during the hospital stay, and patients are so informed by the judge during the commitment hearing. Patients with voluntary legal status can be forced to take medication only in case of emergency. In most cases of involuntary commitment, the physician encourages the patient to continue the hospital stay on a voluntary basis, if necessary, after the period of involuntary commitment is over. The time of discharge is determined by agreement between the patient and the physician, and the patient's voluntary participation is needed for discharge to occur as planned. Because involuntary outpatient commitment does not exist in Germany, continuation of treatment is always a matter of the patient's voluntary participation.
We did not distinguish between coercion in treatment related to the legal status of involuntary patients and patients' subjective perception of coercion, which could be assessed through interviews with patients. In this study setting, the classification of a patient's admission, hospital stay, and medication as voluntary or involuntary reflects the patient's formal, legal status and may involve coercion for involuntary patients. Whether a patient's discharge and intention to continue treatment are classified as voluntary or involuntary reflects the personal choice of the patient and is not subject to formal coercion.
The voluntary or involuntary nature of patients' participation in treatment exists on a continuum from complete coercion through strong persuasion to complete voluntary participation. To avoid reliability problems in this study, we used a dichotomous definition of voluntariness for the five aspects of treatment we examined. Voluntariness of admission and hospital stay was determined according to the patient's legal status as a voluntary or involuntary patient. Medication was classified as involuntary if the patient was touched or held in any way in order to administer medication. Instances in which staff told the patient that he or she would be forced to receive medication by physical coercion but in which the patient ultimately was not touched were thus not classified as involuntary medication. For discharge, the patient was classified as involuntary if the treatment episode was terminated by eloping or leaving the hospital against medical advice. For intention to continue treatment after discharge, the patient was classified as involuntary if he or she refused to continue treatment as suggested by the physician at the time of discharge, if the hospitalization was terminated by the patient's eloping, or if no reasonable agreement could be made with the patient about continued treatment. Patients who agreed to continue treatment, including taking medication as prescribed and making follow-up appointments with physicians and social service agencies, were classified as voluntary.
In addition, we recorded data on clinical indicators that are frequently associated with involuntary treatment, including aggressive behavior and use of seclusion or restraint. Staff members are required to use special forms in the patient's chart to record instances of seclusion and restraint. Data on aggressive behavior was obtained from the charts and was recorded by using the Modified Overt Aggression Scale (MOAS) (15,16). The MOAS is used to classify aggressive behavior in four categories, each with four degrees of severity, and provides a weighted total score with a possible score of 1 to 40 points. Higher scores indicate more severe aggressive behavior. For this study, we defined aggressive behavior as a MOAS score ≥4. A score of 4 indicates destruction of minor objects, petty assault against another person, or severe verbal threats.
Calculations were performed by using the Statistica program (17). Correlations of voluntariness of participation in the various aspects of treatment were determined by using Spearman's rank-order correlation. Differences between admission and discharge PANSS and GAF scores and differences in scores weighted for length of stay were compared by using t tests. The level of significance was set at p<.05. Categorical data were analyzed by using chi square tests. Because multiple tests were necessary to analyze the five aspects of treatment, the level of significance was adjusted to p<.01 (.05/5).
The sample (N=88) consisted of 44 men and 44 women. Their mean±SD age was 40.1±12.1 years. Fifty-two patients had a diagnosis of paranoid schizophrenia; eight had schizoaffective disorder, manic; seven had schizoaffective disorder, depressive; five had residual schizophrenia; nine had transient psychotic disorders; and two each had the disorganized, catatonic, and undifferentiated types of schizophrenia. Twenty patients had no previous admissions, nine had been admitted once before, and the remainder had two or more previous admissions. The mean length of stay was 36.2±33.6 days, with a range from 1 to 247 days.
The mean total PANSS total score for all patients was 94.1±24.6, with a range from 33 to 163, at admission and 55±21.4, with a range from 30 to 115, at discharge. The mean GAF score for all patients was 34.3±13.4, with a range from 5 to 80, at admission and 59.2±17.9, with a range from 20 to 97, at discharge. Thirty-six patients lived alone, 28 lived with a partner, four lived with the family of origin, five had social services assistance in their living situation, and 13 lived in sheltered homes; the living situation of two patients was not specified.
t1 presents the mean PANSS and GAF scores at admission for patients with voluntary and involuntary participation in the five aspects of treatment examined in this study. No significant differences were found between male and female patients. For all aspects of treatment, the mean PANSS score at admission was higher for involuntary patients than for patients who were voluntary participants. However, because of high standard deviations, the differences between groups in mean PANSS scores did not reach statistical significance, except for the difference for hospital stay. The patients who were discharged against their physicians' advice had significantly lower GAF scores at admission, compared with the patients who were discharged in agreement with their physicians. The significance of this association persisted after the significance level was adjusted for multiple testing (t=−3.715, df=68, p<.001).
Voluntary participation in admission, hospital stay, and medication was highly intercorrelated, but no significant correlations were found between voluntary participation in these aspects of treatment and voluntary participation at the time of discharge. However, a significant correlation was found between voluntary participation at discharge and agreement to continue treatment after discharge (t2).
For all aspects of treatment except hospital stay, no significant difference between the voluntary and involuntary patients was found in the mean change in the PANSS total score from admission to discharge or the mean change per inpatient day. For hospital stay, the mean reduction in the PANSS total score from admission to discharge was 36±24.2 points for the voluntary group and 53.8±27.2 points for the involuntary group (t=2.538, df=86, p<.05). However, because of the need to adjust the significance level to account for multiple tests, this finding cannot be defined as statistically significant. The involuntary patients generally had improvement that was similar to or even greater than that of the voluntary patients.
As for gender differences, men who had an involuntary hospital stay had greater improvement in the PANSS total score from admission to discharge than men who had a voluntary hospital stay (t=2.061, df=42, p<.05); no differences in improvement were found between women who had an involuntary stay and women who had a voluntary stay. The reduction in the PANSS total score per inpatient day did not differ between voluntary patients and involuntary patients in any of the aspects of treatment: admission, hospital stay, medication, discharge, and intention to continue treatment after discharge. No significant differences between the voluntary and involuntary groups were found for the increase in GAF score from admission to discharge or the increase in GAF score per inpatient day. The mean increase in GAF score was 41.2±23.5 for involuntarily admitted men, compared with 22.5±19 for voluntarily admitted men (t=2.137, df=36, p<.04). With adjustment for multiple tests, this result did not reach statistical significance.
The charts of 13 patients included records of aggressive behavior that met the criteria for a MOAS rating of 4 or more. Compared with patients without aggressive behavior, the patients with aggressive behavior were more likely to have been administered medication involuntarily (Spearman's r=.390, df=75, p<.001) and to have had an involuntary hospital stay (Spearman's r=-.304, df=75, p<.01), but their likelihood of involuntary admission, discharge against medical advice, and refusal of continued treatment was not significantly different. Thirteen patients, including eight patients in the group with aggressive behavior, experienced seclusion or restraint while in the hospital. Compared with the patients who were not secluded or restrained, the patients who experienced those interventions were more likely to have been admitted involuntarily (χ2=10.858, df=3, p<.05), to have been administered medication involuntarily (χ2=214.277, df=3, p<.01), and to have an involuntary hospital stay (χ2=20.181, df=3, p<.001), but their likelihood of being discharged against medical advice and of refusing continued treatment did not differ. For each of the five aspects of treatment, the voluntary patients did not differ from the involuntary patients in length of stay, history of violent behavior, occupation, or living situation.
The principal result of this study is that patients with schizophrenia improve with inpatient treatment, irrespective of whether they receive treatment voluntarily or involuntarily. The reduction in PANSS score, the increase in GAF score, and changes in the respective measures per inpatient day did not differ significantly between voluntary and involuntary patients in five aspects of treatment, including admission, hospital stay, medication, discharge, and intention to continue treatment. The involuntary patients tended to show greater improvement than the voluntary patients. Thus the fact that improvement was not found among voluntary patients, compared with the involuntary patients, is probably not due to insufficient sample sizes. Compared with the patients who had a voluntary hospital stay, the patients who had an involuntary hospital stay had significantly higher PANSS scores at admission, suggesting that refusal of treatment is characteristic of severe psychopathology. The patients with an involuntary admission and those who received medication involuntarily also had higher PANSS scores at admission, but these differences did not reach statistical significance.
Furthermore, the results provide some evidence that voluntariness is not an unvarying characteristic of patients but one that undergoes profound changes during inpatient treatment. The patients who terminated their hospital stay against medical advice and those who refused to continue treatment as proposed by their doctors generally were not those who had been involuntarily admitted or who had received medication involuntarily. This pattern was indicated by the small and insignificant correlations between the respective aspects of treatment (t2).
It is noteworthy that no significant correlation was found between the voluntariness of participation in the hospital stay and willingness to follow through with outpatient treatment after discharge. This finding is supported by the work of Rain and colleagues (18), who found that perceived coercion was not associated with treatment adherence at ten-week follow-up. We found that although voluntariness of participation in treatment at the beginning of the inpatient stay seems to be associated with psychopathological features, insight into the necessity of continuing treatment was associated with more global aspects of functioning. This relationship was indicated by a strong statistical association between discharge against medical advice and lower GAF scores at admission.
We did not use the term "compliance" to describe participation in treatment in this study, although compliance is a factor in many of the issues we addressed. However, voluntariness, as examined in this study, is different from compliance in some respects. For example, a patient could be noncompliant with medication treatment if he or she tried to deceive staff by hiding medication or by not taking medication during weekends at home. If this patient did not receive any medication by physical coercion, the patient would be classified as voluntary in this study. In addition, a patient who received medication by coercion once and later demonstrated good compliance would have been classified as involuntary. The voluntariness of patients' participation in treatment is primarily a matter of ethics and law, while compliance is a clinical issue. Previous research has provided strong evidence to suggest that treatment compliance and insight into illness affect the long-term course of schizophrenia (19,20). In contrast, our focus was the short-term course of acute episodes, a period of a few weeks.
We could find no other study findings on voluntary and coerced treatment that could be compared with our results. Obviously, our results cannot be used to draw conclusions about long-term outcome or on the effect of involuntary treatment on patients' attitudes, patients' treatment adherence, and the emergence of adverse effects such as posttraumatic stress disorder (6). The finding that the outcome of involuntary inpatient treatment is similar to that of voluntary treatment should not lead to underestimation of the subjective suffering of patients who experience involuntary treatment, especially patients with serious disturbances who are repeatedly treated involuntarily. Involuntary admissions strongly predict future involuntary admissions (21). Many patients revise their beliefs about the necessity of involuntary hospitalization after discharge, but their negative attitudes toward the coercion they experienced does not change (22). We found no evidence to suggest that clinicians should decrease their efforts to develop a positive therapeutic alliance and to respect patients' autonomy whenever possible.
The sample included consecutive admissions of patients between the ages of 18 and 65 years from all wards of the hospital where the study was conducted. This hospital is required to treat all patients from its catchment area. Thus the sample can be considered representative of inpatients with schizophrenia in this catchment area. However, the results cannot necessarily be generalized beyond the structural characteristics of the hospital and its catchment area. The catchment area includes no large cities, the unemployment rate in the catchment area at the time of the study was rather low for Germany at 5.6 percent, and cooperative arrangements between the hospital and several outpatient service agencies were well established and had been in existence for many years. Only one patient was homeless at admission, and none were homeless at discharge. Furthermore, in Baden-Wurtemberg, the state in which this study was conducted, a judge must approve an involuntary hospital stay within 72 hours after involuntary admission, instead of within 24 hours, as in other parts of Germany. This difference results in a lower rate of involuntary hospital stays in this state than in other parts of Germany (23).
The results of this study provide some empirical evidence to ethically justify involuntary commitment and treatment of patients with schizophrenia from both a legal and a medical perspective. Filling the gap of scientifically sound literature on this issue (11) is the first step toward an evidence-based ethics in this aspect of psychiatry. Involuntary commitment laws and practices vary considerably between countries (3,24). Further research should examine the influences of involuntary treatment on various outcome measures in medium- and long-term follow-up studies.
Dr. Steinert is chair of and Mr. Schmid is a medical statistician in the psychiatric care research department of the Weissenau Psychiatric Center at the University of Ulm, Ravensburg, Germany. Address correspondence to Dr. Steinert at Zentrum für Psychiatrie, Weissenau D88214, Ravensburg-Weissenau, Germany (email, email@example.com).
Positive and Negative Syndrome Scale (PANSS) and Global Assessment of Functioning (GAF) scores at admission of 88 inpatients with schizophrenia and delusional disorders, by patients' involuntary or voluntary participation in five aspects of treatment
Spearman's rank-order correlations of voluntariness of participation in five aspects of treatment among 88 inpatients with schizophrenia and delusional disorders