Do Low Participation Rates Bias Health Risk Estimates in Studies of Psychosocial Factors?
Author Information
Author(s): Kaerlev Linda, Kolstad Henrik A, Hansen Åse Marie, Thomsen Jane Frølund, Kærgaard Anette, Rugulies Reiner, Mikkelsen Sigurd, Andersen Johan Hviid, Mors Ole, Grynderup Matias B, Bonde Jens Peter
Primary Institution: Danish Ramazzini Centre, Department of Occupational Medicine, Aarhus University Hospital
Hypothesis
Does low participation in follow-up studies of psychosocial factors lead to biased health risk estimates?
Conclusion
The study found no evidence that low participation at baseline distorts the estimates of associations between psychosocial work environment and mental health outcomes.
Supporting Evidence
- The participation rate was 45%, with 4,489 out of 10,036 employees participating.
- Respondents differed from non-respondents by gender, age, and social class.
- Risk estimates for sick leave and antidepressant prescriptions differed marginally between respondents and the entire population.
Takeaway
If fewer people join a study about work stress and health, it doesn't necessarily mean the results are wrong. This study showed that the health risks were similar whether you looked at everyone or just those who participated.
Methodology
The study analyzed data from a cohort of public service workers in Denmark, comparing health outcomes between respondents and the entire source population.
Potential Biases
There is a risk of bias if non-respondents perceive the work environment differently than respondents.
Limitations
The findings may not apply to other exposures or outcomes, and the study relied on self-reported measures which can introduce bias.
Participant Demographics
Participants were more often women, older than 45, and of higher social class compared to non-respondents.
Statistical Information
Confidence Interval
95% CI
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website