The use of coercive measures in psychiatry is still poorly understood. Most empirical research has been limited to compulsory admission and to risk factors on an individual patient level. This study addresses three coercive measures and the role of predictive factors at both patient and institutional levels.

Using the central psychiatric register that covers all psychiatric hospitals in Canton Zurich (1.3 million people), Switzerland, we traced all inpatients in 2007 aged 18–70 (n = 9698). We used GEE models to analyse variation in rates between psychiatric hospitals.

Overall, we found quotas of 24.8% involuntary admissions, 6.4% seclusion/restraint and 4.2% coerced medication. Results suggest that the kind and severity of mental illness are the most important risk factors for being subjected to any form of coercion. Variation across the six psychiatric hospitals was high, even after accounting for risk factors on the patient level suggesting that centre effects are an important source of variability. However, effects of the hospital characteristics ‘size of the hospital’, ‘length of inpatient stay’, and ‘work load of the nursing staff’ were only weak (‘bed occupancy rate’ was not statistically significant).

The significant variation in use of coercive measures across psychiatric hospitals needs further study.