Diagnoses have the greatest impact on variation in sick-leave certification rate among primary-care patients in Sweden: A multilevel analysis including patient, physician and primary health-care centre levels.

Aims: The aims of this study were to determine and evaluate simultaneously the importance of factors known to influence sick-leave certification such as the sick leave-related diagnoses, the patients' socio-economic status, and characteristics of the physicians. Methods: Computerised medical records from 24 public primary health-care centres (PHCC) were used in a multilevel logistic regression analysis at three levels: patients (n=64,354; sex, age, socio-economic status, workplace factors and diagnoses), physicians (n=574; sex and level of experience) and PHCC (n=24). The variation of sick-leave certification at each level was the outcome. Results: Most of the variation was attributed to the patient level and only 3.5% to the physician and 1.2% to the PHCC levels. Among the patient characteristics, psychiatric diagnoses (mostly acute stress) had the highest odds ratio (OR) for sick leave (OR=16.0; 95% confidence interval [CI] 15-17.2), followed by musculoskeletal diagnoses (OR=6.1; 95% CI 5.8-6.5). Other factors with increased OR were low education (OR=1.7; 95% CI 1.6-1.8), use of social allowance (OR=1.4; 95% CI 1.2-1.7) and certain workplaces (manufacture and health and social care). Being older was not associated with increased certified sick leave. Conclusions: The greatest variation in sick-leave certification rate was seen at the patient level, specifically psychiatric diagnoses. Socio-economic factors increasing the risk for sick-leave certification were education, social allowance and occupations in manufacture and caregiving. Understanding the impact of the different factors that influence certified sick leave is important both for targeted interventions in order to facilitate patients' return to work.