Objectives
The objective of the study was to examine diagnosis-specific sickness
absences of different lengths as predictors of disability retirement in
different occupational classes.

Design Register-based prospective cohort study up to 8 years of follow-up.

Participants
A 70% random sample of the non-retired Finnish population aged 25–62 at
the end of 2006 was included (n=1 727 644) and linked to data on
sickness absences in 2005 and data on disability retirement in
2007–2014.

Main outcome measures
Cox proportional hazards regression was utilised to analyse the
association of sickness absence with the risk of all-cause disability
retirement during an 8-year follow-up.

Results
The risk of disability retirement increased with increasing lengths of
sickness absence in all occupational classes. A long sickness absence
was a particularly strong predictor of disability retirement in upper
non-manual employees as among those with over 180 sickness absence days
the HR was 9.19 (95% CI 7.40 to 11.40), but in manual employees the HR
was 3.51 (95% CI 3.23 to 3.81) in men. Among women, the corresponding
HRs were 7.26 (95% CI 6.16 to 8.57) and 3.94 (95% CI 3.60 to 4.30),
respectively. Adjusting for the diagnosis of sickness absence partly
attenuated the association between the length of sickness absence and
the risk of disability retirement in all employed groups.

Conclusions
A long sickness absence is a strong predictor of disability retirement
in all occupational classes. Preventing the accumulation of sickness
absence days and designing more efficient policies for different
occupational classes may be crucial to reduce the number of transitions
to early retirement due to disability.