Objectives To investigate the burden of recurrent sickness absence (SA), describe the time to recurrence stratified by diagnostic categories, and explore the risk factors associated with recurrent sickness absence among newly employed workers.

Methods It is a dynamic cohort composed by all civil servants of the Goiania municipality hired from 2005 to 2011, in Brazil. Administrative data of certified SA longer than three days were analysed. The frequency of recurrent SA was assessed by the recurrence density (RD) per 1000 person-years in each ICD-10 category. The time to onset of the recurrent SA was calculated by Kaplan–Meier survival analysis. Cox proportional hazard models were constructed to identify socio-demographic and occupational risk factors for recurrent SA.

Results Of the 7,017 workers with a first episode of sickness absence, 3,360 (48%) had at least one episode of recurrence totaling 323,646 work days lost. Overall, the RD was 352 per 1,000 person-years, and the median time to onset was 6 months. Mental disorders caused the majority of recurrent SA in both genders, followed by musculoskeletal diseases among women and injuries among men. The median time to recurrence of SA due to mental disorders was 3 months and due to musculoskeletal diseases was 5 months after the first episode. The number of the past sickness episodes was associated with increased risk of subsequent recurrence. Factors predicting recurrence of SA among women were older age, lower education, being a education professional and multiple jobs. Among men only professional class, blue collar and inspection, were associated with an increased risk of recurrence SA.

Conclusion The sickness absence burden could be reduced through early identification of workers with higher risk of recurrence, and follow them for at least six months after their return to work to determine whether they are able to sustain the full work status.