As in our previous study [3
], based on the same material, coercion was not related to short-term outcome of psychiatric inpatient treatment, even though a measure more likely to capture the amount of coercion experienced by the patients was used in the present study. Diagnoses and level of functioning at admission were the only significant outcome predictors. A floor effect for the less healthy patients was indicated by the fact that patients with lower GAF and higher BPRS scores at admission were more likely to be improved in GAF scores.
A follow-up time of maximum three weeks is short but still relevant. The number of psychiatric beds in Sweden has been reduced and lengths of stay have been shortened [13
]. Psychiatric hospital beds have to be used as effective as possible in order to help patients in acute phases of mental illness being able to rely on outpatient services in the community. Thus, it is important to measure mental health outcome after short periods of inpatient stay. We have used subjective as well as professionally assessed measures of coercion since many studies have shown great inconsistencies between perceived coercion and legal status at admission [7
], and between self-reported and recorded coercive measures during inpatient treatment [19
]. Since the time of our data collection, there has been a further decrease of psychiatric hospital admission rates in Sweden, while the numbers of involuntarily admitted patients have been quite stable. Consequently, the quota of involuntarily admitted patients in psychiatric inpatient care has increased. On a census day in 1997, 30% of all psychiatric inpatients were involuntarily admitted (forensic patients included), compared to 44% on a census day in 2008 [22
]. Further studies are needed to examine whether these changes might have an impact on our findings.
We were inspired to do the analyses of the present study by Iversen et al [7
]. It seems reasonable to assume that the total amount of coercion a person is subjected to is more likely to affect outcome than a single measure of coercion, only. However, while they found that accumulated coercion predicted lessened patient satisfaction, we found no association between accumulated coercion and subjective or assessed outcome. A probable explanation is that measures of coercion were different in the two studies, and that outcome, as measured in the present study, and patient satisfaction are quite different aspects. Furthermore, the legal definitions of coercive treatment may differ. According to the Swedish Compulsory Psychiatric Care Act all injections of involuntarily hospitalised patients should be recorded as coercive, but not orally given medication, while Iversen et al seem to include both oral medication and injections in the concept "involuntarily administered medication".
Previous studies, too, on coercion and outcome [23
] have shown contradictory results, probably influenced by differing legal prerequisites in different countries, differing inclusion criteria, and different measures and methodologies. Despite the serious nature of involuntary hospitalisation and treatment of persons with mental illnesses, the effects of coercive interventions in mental health services are still not known. Large-scale studies with uniform methods allowing for analyses of sub-samples and controlling for differences in patient and treatment characteristics are called for.
Even less is known about the long-term outcome of involuntary admissions to psychiatric hospitals. In a unique study with one year follow-up, Priebe et al [28
] found that patients' views of treatment within the first week are a relevant indicator for the long-term prognosis of involuntarily admitted patients. To study the impact of coercion, however, also voluntarily admitted patients need to be included in prospective studies, and as Høyer [1
] has pointed out, judged by face validity the total number of days of deprivation of liberty seems to be a more adequate measure of coercion than formal legal status of the patient at admission.
Outcome was in our study measured as subjective and assessed improvement. As to the latter, a difference of 10 could be regarded as clinically more significant in lower than in higher GAF scores. However, 80% of the patients had GAF scores below 50. We were not able to assess outcome in terms of prevention of dangerous acts against self or others. Occasional randomised studies of outpatient commitment have been performed [29
], but like in all other studies of involuntary hospitalisation, an inability to assign patients randomly to either compulsory or voluntary treatment is a limitation in the present study. RCT-studies of involuntary psychiatric treatment are, however, hardly possible to perform for ethical, legal and practical reasons. The studied sample in our study is not representative for all psychiatric inpatients, and the exclusion criteria applied may have biased the results. Furthermore, there were dropouts at inclusion and at follow-up, but considering that the patients were acutely mentally ill persons the dropout rate was acceptable.