Key result
Bayesian external approaches reduce VTE outcome misclassification bias compared to uncorrected data.
Why the study?
To assess the validity of administrative database diagnoses of postsurgery venous thromboembolism and compare Bayesian and multiple imputation approaches in correcting outcome misclassification in logistic regression models.
Population
17 319 THR or TKR patients in Quebec, including 2136 across six Montreal hospitals
Comparison
Bayesian external vs Bayesian internal vs multiple imputation approaches to correct outcome misclassification
Design
Validation and simulation study comparing tertiary versus community hospitals
Authors
Loading...
Bayesian correction may improve VTE estimates from administrative data; extends misclassification methods for observational research but should not yet change practice.
Observational (n=17,319)
Yes
Effect estimate: OR 1.57 (95% CI 0.39-5.24)
VTE identified from administrative data has low sensitivity, but Bayesian and multiple imputation approaches can effectively reduce outcome misclassification bias in logistic regression models.
Ni et al. (2018) conducted an observational in Postsurgical venous thromboembolism following total hip and knee arthroplasty (n=17,319). Bayesian external, Bayesian internal, and multiple imputation (MI) approaches vs. Uncorrected administrative data was evaluated on Odds ratio of postsurgery VTE in tertiary versus community hospitals (OR 1.57, 95% CI 0.39-5.24). Bayesian external approaches reduced outcome misclassification bias for VTE in logistic regression, yielding an adjusted OR of 1.57 (95% CI 0.39-5.24) compared to 1.35 for uncorrected data.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: