Performance of the high-dimensional propensity score in adjusting for unmeasured confounders

Abstract

PurposeHigh-dimensional propensity scores (hdPS) can adjust for measured confounders, but it remains unclear how well it can adjust for unmeasured confounders. Our goal was to identify if the hdPS method could adjust for confounders which were hidden to the hdPS algorithm.MethodThe hdPS algorithm was used to estimate two hdPS; the first version (hdPS-1) was estimated using data provided by 6 data dimensions and the second version (hdPS-2) was estimated using data provided from only two of the 6 data dimensions. Two matched sub-cohorts were created by matching one patient initiated on a high-dose statin to one patient initiated on a low-dose statin based on either hdPS-1 (Matched hdPS Full Info Sub-Cohort) or hdPS-2 (Matched hdPS Hidden Info Sub-Cohort). Performances of both hdPS were compared by means of the absolute standardized differences (ASDD) regarding 18 characteristics (data on seven of the 18 characteristics were hidden to the hdPS algorithm when estimating the hdPS-2).ResultsEight out of the 18 characteristics were shown to be unbalanced within the unmatched cohort. Matching on either hdPS achieved adequate balance (i.e., ASDD <0.1) on all 18 characteristics.ConclusionOur results indicate that the hdPS method was able to adjust for hidden confounders supporting the claim that the hdPS method can adjust for at least some unmeasured confounders.

Automated data-adaptive analytics for electronic healthcare data to study causal treatment effects

“…Guertin et al 70 published an empirical study but included a sensitivity analysis simulating unobserved confounding. In a study comparing low-intensity versus high-intensity lipid-lowering therapy with statins, they were concerned that sicker patients with more advanced arteriosclerosis or higher serum lipid levels were more likely to initiate high-intensity treatment.…”

The risk of sudden cardiac arrest and ventricular arrhythmia with rosiglitazone versus pioglitazone: real-world evidence on thiazolidinedione safety

“…We calculated descriptive statistics for baseline variables, crude incidence rates, and unadjusted association measures, the latter via Cox proportional hazards models. We utilized a semi-automated, data-adaptive hdPS approach-an algorithm for identifying and selecting proxies for important confounder constructs [47]-to reduce the impact of measured and unmeasured potential confounders that are correlated with measured factors [48]. First, we used the hdPS algorithm [41,47] to identify empiric candidate covariates; we identified the 200 most prevalent baseline diagnoses, procedures, and dispensed drugs for each of nine prespecified data dimensions.…”

Comparing external and internal validation methods in correcting outcome misclassification bias in logistic regression: A simulation study and application to the case of postsurgical venous thromboembolism following total hip and knee arthroplasty

“…Simulations that accounted for misclassification varied with measured and unmeasured confounders showed poorer performance in presence of unmeasured confounders (Appendix S4). Appropriate methods that adjust for unmeasured confounders such as high‐dimensional propensity score matching and instrumental variable analysis should be considered with the internal validation approaches in such cases. The structure of the Bayesian external validation approach allows the adjustment for unmeasured confounders if information exists on the effect of these cofounders on the outcome misclassification.…”