Related Experiment Video
Updated: Mar 13, 2026
06:55
Inverse Probability of Treatment Weighting Propensity Score using the Military Health System Data Repository and National Death Index
Published on: January 8, 2020
15.4K
Variable Selection for Confounding Adjustment in High-dimensional Covariate Spaces When Analyzing Healthcare
Sebastian Schneeweiss1, Wesley Eddings, Robert J Glynn
1From the aDivision of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA; and bAetion Inc., New York, NY.
Epidemiology (Cambridge, Mass.)
|October 26, 2016
Summary
Data-adaptive methods for confounding adjustment in electronic healthcare databases showed that standard variable selection performed well. Minor gains in accuracy were observed with Bayesian regression for rare outcomes.
Area of Science:
- Health Informatics
- Epidemiology
- Biostatistics
Background:
- Data-adaptive approaches to confounding adjustment offer advantages over expert knowledge for analyzing electronic healthcare databases.
- These methods are particularly useful for rapid analysis of multiple databases.
- Reliable empirical identification and adjustment of outcome predictors are key to improving analytical performance.
Purpose of the Study:
- To evaluate data-adaptive confounding adjustment strategies in electronic healthcare databases.
- To compare the performance of a base-case high-dimensional propensity score (hdPS) algorithm with augmented variable selection methods.
- To assess the impact of different variable selection and estimation techniques on treatment effect estimates.
Main Methods:
- A base-case high-dimensional propensity score (hdPS) algorithm was implemented across five cohort studies.
- Variable selection was augmented using alternative strategies, including adjusted outcome-confounder associations (RRCD) and joint modeling (Lasso, Bayesian regression).
- Propensity scores and outcome models were directly estimated using over 1,500 variables (Lasso, Bayesian regression, Ridge).
Main Results:
- Most tested augmentations to the base-case hdPS did not significantly alter estimates due to wide confidence intervals.
- Bayesian regression and Lasso for estimating RRCD minimally improved estimates in three of five studies.
- Direct outcome estimation using Lasso yielded the poorest performance.
Conclusions:
- The standard heuristic of variable reduction in hdPS adjustment performed comparably to alternative approaches across diverse settings.
- Bayesian outcome regression may offer slight improvements in variable selection for propensity score estimation with rare outcomes.
- The findings support the utility of data-adaptive methods while highlighting the robustness of basic hdPS variable selection strategies.
Related Concept Videos
01:27Confounding in Epidemiological Studies
932
Confounding in statistical epidemiology represents a pivotal challenge, referring to the distortion in the perceived relationship between an exposure and an outcome due to the presence of a third variable, known as a confounder. This variable is associated with both the exposure and the outcome but is not a direct link in their causal chain. Its presence can lead to erroneous interpretations of the exposure's effect, either exaggerating or underestimating the true association. This...
932
01:25Strategies for Assessing and Addressing Confounding
521
Confounding is a critical issue in epidemiological studies, often leading to misleading conclusions about associations between exposures and outcomes. It occurs when the relationship between the exposure and the outcome is mixed with the effects of other factors that influence the outcome. Given that, addressing confounding is of high importance for drawing accurate inferences in research.
Confounding can be addressed at both the design phase of a study and through analytical methods after data...
Confounding can be addressed at both the design phase of a study and through analytical methods after data...
521
01:15Study Design in Statistics
10.2K
A study design is a set of techniques that allow a researcher to collect and analyze data from different variables defined for a specific research problem. Statistics is commonly for effective study design and more robust experiments,
Does aspirin reduce the risk of heart attacks? Is one brand of fertilizer more effective at growing roses than another? Is fatigue as dangerous to a driver as the influence of alcohol? Questions like these are answered using randomized experiments with proper...
Does aspirin reduce the risk of heart attacks? Is one brand of fertilizer more effective at growing roses than another? Is fatigue as dangerous to a driver as the influence of alcohol? Questions like these are answered using randomized experiments with proper...
10.2K
01:21Friedman Two-way Analysis of Variance by Ranks
531
Friedman's Two-Way Analysis of Variance by Ranks is a nonparametric test designed to identify differences across multiple test attempts when traditional assumptions of normality and equal variances do not apply. Unlike conventional ANOVA, which requires normally distributed data with equal variances, Friedman's test is ideal for ordinal or non-normally distributed data, making it particularly useful for analyzing dependent samples, such as matched subjects over time or repeated measures...
531
01:29Contingency Table
4.7K
A contingency table provides a way of portraying data that can facilitate calculating probabilities. It is a method of displaying a frequency distribution as a table with rows and columns to show how two variables may be dependent (contingent) upon each other; The table helps determine conditional probabilities quite quickly and can help systematically organize, analyze and quantify data. The table displays sample values concerning two variables that may be dependent or contingent on one...
4.7K
01:25Statistical Methods for Analyzing Epidemiological Data
1.1K
Epidemiological data primarily involves information on specific populations' occurrence, distribution, and determinants of health and diseases. This data is crucial for understanding disease patterns and impacts, aiding public health decision-making and disease prevention strategies. The analysis of epidemiological data employs various statistical methods to interpret health-related data effectively. Here are some commonly used methods:
1.1K