Propensity Score Weighting and Trimming Strategies for Reducing Variance and Bias of Treatment Effect Estimates: A Simulation Study

2021/02/16 by Til Stürmer‎, Til Stürmer, Michael Webster‐Clark +11 · 118 citations
Economics, Econometrics and Finance · Mathematics · Medicine · #Advanced Causal Inference Techniques #Computer science #Econometrics #Economics #Health Systems, Economic Evaluations, Quality of Life #Mathematics #Medicine #Propensity score matching #Repeated measures design #Selection bias #Statistical Methods in Clinical Trials #Statistics #Trimming #Variance (accounting) #Weighting

paper · open access · doi:10.1093/aje/kwab041

published in American Journal of Epidemiology 190(8), 1659-1670 (Oxford University Press)

openalex publication_date 2021/02/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Abstract

To extend previous simulations on the performance of propensity score (PS) weighting and trimming methods to settings without and with unmeasured confounding, Poisson outcomes, and various strengths of treatment prediction (PS c statistic), we simulated studies with a binary intended treatment T as a function of 4 measured covariates. We mimicked treatment withheld and last-resort treatment by adding 2 "unmeasured" dichotomous factors that directed treatment to change for some patients in both tails of the PS distribution. The number of outcomes Y was simulated as a Poisson function of T and confounders. We estimated the PS as a function of measured covariates and trimmed the tails of the PS distribution using 3 strategies ("Crump," "Stürmer," and "Walker"). After trimming and reestimation, we used alternative PS weights to estimate the treatment effect (rate ratio): inverse probability of treatment weighting, standardized mortality ratio (SMR)-treated, SMR-untreated, the average treatment effect in the overlap population (ATO), matching, and entropy. With no unmeasured confounding, the ATO (123%) and "Crump" trimming (112%) improved relative efficiency compared with untrimmed inverse probability of treatment weighting. With unmeasured confounding, untrimmed estimates were biased irrespective of weighting method, and only Stürmer and Walker trimming consistently reduced bias. In settings where unmeasured confounding (e.g., frailty) may lead physicians to withhold treatment, Stürmer and Walker trimming should be considered before primary analysis.

Citations

• Dealing with limited overlap in estimation of average treatment effects
• Treatment Effects in the Presence of Unmeasured Confounding: Dealing With Observations in the Tails of the Propensity Score Distribution--A Simulation Study
• The central role of the propensity score in observational studies for causal effects

Cited by

• Matching Methods for Confounder Adjustment: An Addition to the Epidemiologist’s Toolbox
• Nonoperative Versus Neurosurgical Treatment of Brain Abscess: An Emulated Trial Nested Within a Nationwide, Population-Based Cohort
• Real-world effectiveness and causal mediation study of BNT162b2 on long COVID risks in children and adolescents
• A generalized bootstrap procedure of the standard error and confidence interval estimation for inverse probability of treatment weighting
• Synthetic Negative Controls: Using Simulation to Screen Large-scale Propensity Score Analyses. [europepmc]
• Association of Inappropriate Outpatient Pediatric Antibiotic Prescriptions With Adverse Drug Events and Health Care Expenditures. [europepmc]
• Association of vancomycin plus piperacillin-tazobactam with early changes in creatinine versus cystatin C in critically ill adults: a prospective cohort study. [europepmc]
• Rheumatoid Arthritis Disease Activity and Hospitalized Infection in a Large US Registry. [europepmc]
• Cardiovascular Outcomes of α-Blockers vs 5-α Reductase Inhibitors for Benign Prostatic Hyperplasia. [europepmc]
• Effectiveness of early treatment with plasma exchange in patients with Stevens-Johnson syndrome and toxic epidermal necrolysis. [europepmc]
• Treatment switches of disease-modifying therapies in people with multiple sclerosis: long-term experience from the German MS Registry. [europepmc]
• Comparative effectiveness of second line oral antidiabetic treatments among people with type 2 diabetes mellitus: emulation of a target trial using routinely collected health data. [europepmc]
• Pneumocystis pneumonia in intensive care: clinical spectrum, prophylaxis patterns, antibiotic treatment delay impact, and role of corticosteroids. A French multicentre prospective cohort study. [europepmc]
• Use of sensitivity analyses to assess uncontrolled confounding from unmeasured variables in observational, active comparator pharmacoepidemiologic studies: a systematic review. [europepmc]
• GLP-1 Receptor Agonist Use and Risk of Suicide Death. [europepmc]
• High-dimensional Iterative Causal Forest (hdiCF) for Subgroup Identification Using Health Care Claims Data. [europepmc]
• Antidiabetic Medication and Asthma Attacks. [europepmc]
• Single-microglia transcriptomic transition network-based prediction and real-world patient data validation identifies ketorolac as a repurposable drug for Alzheimer's disease. [europepmc]
• Prenatal exposure to selective serotonin reuptake inhibitors and risk of disorders of gut-brain interaction in children. [europepmc]
• Real-world effectiveness and causal mediation study of BNT162b2 on long COVID risks in children and adolescents. [europepmc]
• Tranexamic acid is associated with post-injury mortality in a resource-limited trauma system: Findings from the epidemiology and outcomes of prolonged trauma care cohort study. [europepmc]
• Missing data approaches for longitudinal neuroimaging research: Examples from the Adolescent Brain and Cognitive Development (ABCD) Study®. [europepmc]
• Human milk oligosaccharides and Campylobacter jejuni and Campylobacter coli gastroenteritis risk in a Nicaraguan Birth Cohort. [europepmc]
• Who Are We Missing? : A Principled Approach to Characterizing the Underrepresented Population. [europepmc]
• Instrumental variable approaches for estimating time-varying treatment effects in comparative effectiveness research. [europepmc]
• First-Trimester mRNA COVID-19 Vaccination and Risk of Major Congenital Anomalies. [europepmc]
• Comparative Gastrointestinal Safety of Dulaglutide, Semaglutide, and Tirzepatide in Adults With Type 2 Diabetes. [europepmc]
• The effect of time to surgery on clinical outcomes and hospitalization costs in older adults with femoral shaft fractures: A nationwide retrospective cohort study in Japan. [europepmc]
• Comparative effectiveness of Paxlovid versus sotrovimab and molnupiravir for preventing severe COVID-19 outcomes in non-hospitalised patients: observational cohort study using the OpenSAFELY platform [europepmc]

Related

• Treatment Effects in the Presence of Unmeasured Confounding: Dealing With Observations in the Tails of the Propensity Score Distribution--A Simulation Study
• Comparing Propensity Score-Based Methods in Estimating the Treatment Effects: A Simulation Study
• Average treatment effect on the treated, under lack of positivity
• Efficient Estimation of Average Treatment Effects Using the Estimated Propensity Score
• On variance estimation of the inverse probability‐of‐treatment weighting estimator: A tutorial for different types of propensity score weights