Objective Psychiatric expertises for determining a person’s work capacity are being criticised in Switzerland. Issues of concern are lack of transparency and variation in the judgment on work capacity. The aims of this study were (1) to learn about current practice among Swiss psychiatric experts in evaluating and reporting on work capacity, (2) to compare the practice to professional guidance, and (3) to identify sources for the variation and lack of transparency that may be amenable to improvement.

Methods National online survey among psychiatrists experienced in evaluation of work capacity. We identified issues of concern through the literature and discussion with opinion leaders in psychiatric expertises. We inquired current practice of disability evaluation, aspects of quality, and suggestions for improvement.

Results 129 psychiatrists participated (response rate 31%). The job description, reference for determining work capacity in the current job, was perceived as mandatory information by 90%, but was usually missing or deficient in files (74%). The psychiatrists expressed work capacity as free text plus percentage of work capacity (49%), percentage only (23%), or free text only (14%). 13% used instruments to report on work capacity. Psychiatrists reported diverse interpretations for ’percentage of work capacity’; three interpretations were reported as applicable in equal frequency around 80%. Psychiatrists usually used report forms of the insurers (77%), peer consulting (65%), and process guidelines (51%), but rarely evidence-based information. Current guidelines attach low importance to job descriptions and instruments, and recommend percentage of work capacity’ without reflecting the pros and cons.

Conclusions Our findings point to some possible causes of the variation and lack of transparency in disability evaluation. More formal and explicit approaches to a professional consensus informed by up-to-date research findings would increase the trust in the final judgments. Measures to reduce the unwanted variation need to be tested in further research.