Dataset Information

Propensity score analysis for time-dependent exposure.

ABSTRACT: Propensity score analysis (PSA) is widely used in medical literature to account for confounders. Conventionally, the propensity score (PS) is calculated by a binary logistic regression model using time-fixed covariates. In the presence of time-varying treatment or exposure, the conventional method may cause bias because subjects with early and late exposure are treated as the same. In effect, subjects who are treated latter can be different from those who are treated early. Thus, the conventional PSA must be modified to address this bias. In this paper, we illustrate how to perform analysis in the presence of time-dependent exposure. We conduct a simulation study with a known treatment effect. In the simulation study, we find the PSA method that directly adjust PS estimated by either a binary logistic regression model or a Cox regression model using time-fixed covariates still introduce significant bias. On the other hand, the time-dependent PS matching can help to achieve a result approaching the true effect. After time-dependent PS matching, the matched cohort can be analyzed with conventional Cox regression model or conditional logistic regression (CLR) model with time strata. The performance is comparable to the correctly specified Cox regression model with time-varying covariates (i.e., adjusting the exposure in a multivariable model as a time-varying covariate). We further develop a function called TDPSM() for time-dependent PS matching and it is applied to a real world dataset.

SUBMITTER: Zhang Z

PROVIDER: S-EPMC7154493 | biostudies-literature | 2020 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Propensity score analysis for time-dependent exposure.

Zhang Zhongheng Z Li Xiuyang X Wu Xiao X Qiu Huixian H Shi Hongying H

Annals of translational medicine 20200301 5

Propensity score analysis (PSA) is widely used in medical literature to account for confounders. Conventionally, the propensity score (PS) is calculated by a binary logistic regression model using time-fixed covariates. In the presence of time-varying treatment or exposure, the conventional method may cause bias because subjects with early and late exposure are treated as the same. In effect, subjects who are treated latter can be different from those who are treated early. Thus, the conventiona ...[more]

PMID: 32309393

Similar Datasets

Project description:Propensity score methodology has become increasingly popular in recent years as a tool for estimating causal effects in observational studies. Much of the related research has been directed at settings with binary or discrete exposure variables with more recent work involving continuous exposure variables. In environmental epidemiology, a substantial proportion of individuals is often completely unexposed while others may experience heavy exposure leading to an exposure distribution with a point mass at zero and a heavy right tail. We suggest a new approach to handle this type of exposure data by constructing a propensity score based on a two-part model and show how this model can be used to more reliably adjust for covariates of a semi-continuous exposure variable. We also consider the case when a misspecified propensity score is used in a regression adjustment and derive an explicit form of the bias. We show that the potential bias gets smaller as the estimated propensity score gets closer to the true expectation of the exposure variable given a set of observed covariates. While this result pertains to a more general setting, we use it to evaluate the potential bias in settings in which the true exposure has a semi-continuous structure. We also evaluate and compare the performance of our proposed method through simulation studies relative to a simpler linear regression-based propensity score for a continuous exposure variable as well as through direct covariate adjustment. Overall, we find that using a propensity score constructed via a two-part model significantly improves the regression estimate when the exposure variable is semi-continuous in nature. Specifically when the proportion of non-exposed subjects is high and the effects of covariates on exposure and outcome are strong, the proposed two-part propensity score method outperforms the more standard competing methods. We illustrate our method using data from the Detroit Longitudinal Cohort Study in which the exposure variable reflects gestational alcohol exposure featuring zero values and a long tail.

Project description:BackgroundEarly use of antimicrobials is a critical intervention in the treatment of patients with sepsis. The exact time of initiation is controversial and its early administration may be a difficult task in crowded emergency departments (ED). The aim of this study was to estimate, using a matched propensity score, the effect on hospital mortality of administration of antimicrobials within 1 or 3 hours, in patients admitted to the ED with sepsis.MethodsThis was a secondary analysis of a multicenter prospective cohort. Patients included in the study were older than 18 years, hospitalized between 2014 and 2016 with suspected sepsis, and admitted to ED of three tertiary care university hospitals in Medellín, Colombia. A propensity score analysis for administration of antimicrobials, both within 1 and 3 h of admission by the ED, was fitted with 28 variables related with clinical attention and physiological changes. As a sensitivity analysis, a logistic regression model was fitted for antimicrobial use adjusted both by propensity score and confounding variables.ResultsThe study cohort was composed of 2454 patients with a median age of 62 years (IQR = 46-74). Among them, 32% (n = 781) received antibiotics within 3 h and 14% (n = 340) within the first hour. The main diagnoses were urinary tract infection (28%, n = 682) and pneumonia (27%, n = 671). Blood cultures were obtained in 87% (n = 2140) and yielded positive in 29% (n = 629), mainly with Escherichia coli (37%, n = 230), Staphylococcus aureus (21%, n = 132), and Klebsiella pneumoniae (10.2%, n = 64). The hospital mortality rate was 11.5% (n = 283). There were no significant differences in mortality, after adjustment, using antimicrobials either in the first hour (OR 1.03; 95% CI = 0.63; 1.70) or 3 h (OR 0.85; 95% CI = 0.61; 1.20). There were no changes with different models for sensitivity analysis.ConclusionsDespite the obvious constraints given for sample size and residual confounding, our results suggest that we need a more comprehensive approach to sepsis and its treatment, considering early detection, multiple interventions, and goals beyond the simple time-to-antimicrobials.

Project description:ObjectivesThe aim of this study is to compare the postoperative complications, perioperative course, and survival among patients from the multicentric Spanish Video-assisted Thoracic Surgery Group database who received video-assisted thoracic surgery lobectomy or video-assisted thoracic surgery anatomic segmentectomy.MethodsFrom December 2016 to March 2018, a total of 2250 patients were collected from 33 centers. Overall analysis (video-assisted thoracic surgery lobectomy = 2070; video-assisted thoracic surgery anatomic segmentectomy = 180) and propensity score-matched adjusted analysis (video-assisted thoracic surgery lobectomy = 97; video-assisted thoracic surgery anatomic segmentectomy = 97) were performed to compare postoperative results. Kaplan-Meier and competing risks method were used to compare survival.ResultsIn the overall analysis, video-assisted thoracic surgery anatomic segmentectomy showed a lower incidence of respiratory complications (relative risk, 0.56; confidence interval, 0.37-0.83; P = .002), lower postoperative prolonged air leak (relative risk, 0.42; 95% confidence interval, 0.23-0.78; P = .003), and shorter median postoperative stay (4.8 vs 6.2 days; P = .004) than video-assisted thoracic surgery lobectomy. After propensity score-matched analysis, prolonged air leak remained significantly lower in video-assisted thoracic surgery anatomic segmentectomy (relative risk, 0.33; 95% confidence interval, 0.12-0.89; P = .02). Kaplan-Meier and competing risk curves showed no differences during the 3-year follow-up (median follow-up in months: 24.4; interquartile range, 20.8-28.3) in terms of overall survival (hazard ratio, 0.73; 95% confidence interval, 0.45-1.7; P = .2), tumor progression-related mortality (subdistribution hazard ratio, 0.41; 95% confidence interval, 0.11-1.57; P = .2), and disease-free survival (subdistribution hazard ratio, 0.73; 95% confidence interval, 0.35-1.51; P = .4) between groups.ConclusionsVideo-assisted thoracic surgery segmentectomy showed results similar to lobectomy in terms of postoperative outcomes and midterm survival. In addition, a lower incidence of prolonged air leak was found in patients who underwent video-assisted thoracic surgery anatomic segmentectomy.

Dataset Information

Propensity score analysis for time-dependent exposure.

Publications

Propensity score analysis for time-dependent exposure.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets