High‐dimensional propensity score algorithm in comparative effectiveness research with time‐varying interventions

Abstract

The high-dimensional propensity score (hdPS) algorithm was proposed for automation of confounding adjustment in problems involving large healthcare databases. It has been evaluated in comparative effectiveness research (CER) with point treatments to handle baseline confounding through matching or covariance adjustment on the hdPS. In observational studies with time-varying interventions, such hdPS approaches are often inadequate to handle time-dependent confounding and selection bias. Inverse probability weighting (IPW) estimation to fit marginal structural models can adequately handle these biases under the fundamental assumption of no unmeasured confounders. Upholding of this assumption relies on the selection of an adequate set of covariates for bias adjustment. We describe the application and performance of the hdPS algorithm to improve covariate selection in CER with time-varying interventions based on IPW estimation and explore stabilization of the resulting estimates using Super Learning. The evaluation is based on both the analysis of electronic health records data in a real-world CER study of adults with type 2 diabetes and a simulation study. This report (i) establishes the feasibility of IPW estimation with the hdPS algorithm based on large electronic health records databases, (ii) demonstrates little impact on inferences when supplementing the set of expert-selected covariates using the hdPS algorithm in a setting with extensive background knowledge, (iii) supports the application of the hdPS algorithm in discovery settings with little background knowledge or limited data availability, and (iv) motivates the application of Super Learning to stabilize effect estimates based on the hdPS algorithm.

Strategies to Prioritize Clinical Options in Primary Care

“…19 The availability of large databases that include detailed clinical data on millions of patients and novel analytic approaches, such as marginal structural models and machine learning, will likely lead to improved risk prediction and prioritization methods in the near future. [20][21][22] It may be difficult, however, to explain these complex statistical approaches to clinicians and patients, who thus may be skeptical of their results.…”

Collaborative-controlled LASSO for constructing propensity score-based estimators in high-dimensional data

“…44,45 Schneeweiss et al 46 evaluated a range of algorithms to improve covariate ranking based on the empirical covariate outcome relationship without any meaningful improvement over the ranking using the Bross formula. Ju et al 20 evaluated various choices for the parameters k 1 and k 2 within the hdPS algorithm, and found that the performance of the hdPS was not sensitive to choices for k 1 and k 2 as long as the hyperparameter pair were within a reasonable range.…”

simcausal R Package: Conducting Transparent and Reproducible Simulation Studies of Causal Effect Estimation with Complex Longitudinal Data

“…In this example we replicate results from the longitudinal data simulation protocol used in two published manuscripts Neugebauer et al (2014, 2015). We first describe the structural equation model that implies the data generating distribution of the observed data, with time-to-event outcome, as reported in Section 5.1 of Neugebauer et al (2015).…”

“…We first describe the structural equation model that implies the data generating distribution of the observed data, with time-to-event outcome, as reported in Section 5.1 of Neugebauer et al (2015). We then show how to specify this model using the simcausal R interface, simulate observed data, define static and dynamic intervention, simulate counterfactual data, and calculate various causal parameters based on these interventions.…”

“…In Section 4, we describe the use of the package for a more realistic and complex simulation study example based on survival data with repeated measures and dynamic interventions at multiple time points. In Section 4, we also apply the simcausal package to replicate some of the results of a previously published simulation study by Neugebauer et al (2014, 2015). In Section 5, we apply the simcausal package to replicate results of another published simulation study by Lefebvre et al (2008).…”