Evaluating the comparability of osteoporosis treatments using propensity score and negative control outcome methods in UK and Denmark electronic health record databases
Authors' conclusion
Achievement of covariate balance and determination of residual bias were dependent upon several factors including the population under study, PS method, prevalence of NCO, and the threshold indicating residual confounding.
Study context
- Population
- Women aged ≥50 yr initiating denosumab or oral bisphosphonate (OBP) in 2011–2018, UK Clinical Practice Research Datalink (CPRD) and Danish National Registries (DNR)
- Participants
- 280,288 women
- Exposure
- Denosumab vs oral bisphosphonate
- Outcome
- Comparability of treatment groups: covariate balance (absolute standardized mean difference (ASMD) >0.1 = imbalance) and residual confounding by NCOs (≥100 events; HR via Cox models)
Methodological detailspropensity score weighting, matching and stratification, negative control outcome analysis, two electronic health record databases
- Propensity score weighting, matching and stratification
- Negative control outcome analysis
- Two electronic health record databases
Main finding
As reported in the abstract
Covariate balance (ASMD >0.1 = imbalance) and residual confounding (NCOs) after PS weighting, matching and stratification, denosumab vs OBP
- Imbalanced covariates before adjustment, CPRD
- 22/87
- Imbalanced covariates before adjustment, DNR
- 18/83
- Imbalance remaining after weighting, matching or stratification
- 2%-11%
- Approach 1: residual confounding present for all propensity score (PS) methods in both databases (negative control outcome (NCOs))
- ≥8%
- Approach 1: no residual confounding with stratification in DNR (NCOs)
- 3.8%
- Approach 2, CPRD: residual confounding with PS matching (NCOs)
- 5.3%
- Approach 2, CPRD: residual confounding with stratification (NCOs)
- 6.4%
- Approach 2, CPRD: no residual confounding with weighting (NCOs)
- 4.3%
Observational data: associations, not effects.
Original publication
Evaluating the comparability of osteoporosis treatments using propensity score and negative control outcome methods in UK and Denmark electronic health record databases. J Bone Miner Res. Published 15 April 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →