Spondyloarthritis & psoriatic arthritis
Challenge of missing data in observational studies: investigating cross-sectional imputation methods for assessing disease activity in axial spondyloarthritis
Authors' conclusion
Hot deck imputation using predictive mean matching exhibited the highest robustness
Study context
- Population
- Patients with axial spondyloarthritis (axSpA) in longitudinally collected observational data from ten European registries
- Participants
- 8,583 patients10 European registries
- Exposure
- Cross-sectional imputation methods for missing disease activity
- Outcome
- Performance of imputation (relative bias and coverage of 95% CIs) for mean Axial Spondyloarthritis Disease Activity Score (ASDAS) and proportion in LDA (ASDAS<2.1)
- Follow-up
- Baseline, 6 and 12 months
Methodological detailsregistry data from ten European registries, nine single and five multiple imputation methods
- Registry data from ten European registries
- Nine single and five multiple imputation methods
Main finding
As reported in the abstract
Performance of imputation methods (relative bias, coverage)
- Hot deck methods outperformed mean and regression methods, particularly for low disease activity (LDA)
- Multiple imputation gave better coverage than corresponding single imputation
- No method gave unbiased estimates with adequate coverage across all time points; worse for missing baseline than follow-up data
- Predictive mean and weighted predictive mean hot deck consistently gave low bias
Observational data: associations, not effects.
Original publication
Challenge of missing data in observational studies: investigating cross-sectional imputation methods for assessing disease activity in axial spondyloarthritis. RMD Open. Published 20 February 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →