Dataset Information

Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance.

ABSTRACT:

Background

Double-adjustment can be used to remove confounding if imbalance exists after propensity score (PS) matching. However, it is not always possible to include all covariates in adjustment. We aimed to find the optimal imbalance threshold for entering covariates into regression.

Methods

We conducted a series of Monte Carlo simulations on virtual populations of 5,000 subjects. We performed PS 1:1 nearest-neighbor matching on each sample. We calculated standardized mean differences across groups to detect any remaining imbalance in the matched samples. We examined 25 thresholds (from 0.01 to 0.25, stepwise 0.01) for considering residual imbalance. The treatment effect was estimated using logistic regression that contained only those covariates considered to be unbalanced by these thresholds.

Results

We showed that regression adjustment could dramatically remove residual confounding bias when it included all of the covariates with a standardized difference greater than 0.10. The additional benefit was negligible when we also adjusted for covariates with less imbalance. We found that the mean squared error of the estimates was minimized under the same conditions.

Conclusion

If covariate balance is not achieved, we recommend reiterating PS modeling until standardized differences below 0.10 are achieved on most covariates. In case of remaining imbalance, a double adjustment might be worth considering.

SUBMITTER: Nguyen TL

PROVIDER: S-EPMC5408373 | biostudies-literature | 2017 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance.

Nguyen Tri-Long TL Collins Gary S GS Spence Jessica J Daurès Jean-Pierre JP Devereaux P J PJ Landais Paul P Le Manach Yannick Y

BMC medical research methodology 20170428 1

<h4>Background</h4>Double-adjustment can be used to remove confounding if imbalance exists after propensity score (PS) matching. However, it is not always possible to include all covariates in adjustment. We aimed to find the optimal imbalance threshold for entering covariates into regression.<h4>Methods</h4>We conducted a series of Monte Carlo simulations on virtual populations of 5,000 subjects. We performed PS 1:1 nearest-neighbor matching on each sample. We calculated standardized mean diffe ...[more]

PMID: 28454568

Similar Datasets

Project description:The association between the initial cardiac rhythm and short-term survival in patients with in-hospital cardiac arrest (IHCA) has not been extensively studied despite the fact that it is thought to be a prognostic factor in patients with out-of-hospital cardiac arrest. This study aimed to look at the relationship between initial shockable rhythm and survival to hospital discharge in individuals with IHCA. 1516 adults with IHCA who received chest compressions lasting at least two minutes at the National Taiwan University Hospital between 2006 and 2014 made up the study population. Propensity scores were estimated using a fitted multivariate logistic regression model. Various statistical methodologies were employed to investigate the association between shockable rhythm and the probability of survival to discharge in patients experiencing IHCA, including multivariate adjustment, propensity score adjustment, propensity score matching, and logistic regression based on propensity score weighting. In the original cohort, the multivariate-adjusted odds ratio (OR) was 2.312 (95% confidence interval [CI]: 1.515-3.531, P < 0.001). In additional propensity score adjustment, the OR between shockable rhythm and the probability of survival to hospital discharge in IHCA patients was 2.282 (95% CI: 1.486, 3.504, P < 0.001). The multivariate-adjusted logistic regression model analysis revealed that patients with shockable rhythm had a 1.761-fold higher likelihood of surviving to hospital release in the propensity score-matched cohort (OR = 2.761, 95% CI: 1.084-7.028, P = 0.033). The multivariate-adjusted OR of the inverse probability for the treatment-weighted cohort was 1.901 (95% CI: 1.507-2.397, P < 0.001), and the standardized mortality ratio-weighted cohort was 2.692 (95% CI: 1.511-4.795, P < 0.001). In patients with in-hospital cardiac arrest, Initial cardiac rhythm is an independent predictor of survival to hospital discharge. Depending on various statistical methods, patients with IHCA who have a shockable rhythm have a one to two fold higher probability of survival to discharge than those who have a non-shockable rhythm. This provides a reference for optimizing resuscitation decisions for IHCA patients and facilitating clinical communication.

Project description:BackgroundAlthough congestive heart failure (CHF) is considered a risk factor for postoperative mortality, reliable quantification of the relationship between CHF and postoperative mortality risk is limited. We aimed to investigate the association between CHF and 1-year mortality after surgery in a large cohort of the Singaporean population.MethodsIn this retrospective cohort study, the study population included 69,032 adult patients who underwent surgery at Singapore General Hospital between 1 January 2012 and 31 October 2016. The target independent and dependent variables were CHF and 1-year mortality after surgery, respectively. Propensity score was estimated using a non-parsimonious multivariable logistic regression model. Multivariable adjustment, propensity score matching, propensity score adjustment, and propensity score-based weighting Cox proportional-hazards regression were performed to investigate the association between CHF and 1-year mortality after surgery.ResultsThe multivariate-adjusted hazard ratio (HR) in the original cohort was 1.39 (95% confidence interval (CI): 1.20-1.61, P < 0.001). In additional propensity score adjustment, the HR between CHF and 1-year mortality after surgery was 1.34 (95% CI: 1.15-1.56, P < 0.001). In the propensity score-matched cohort, the multivariate-adjusted Cox proportional hazard regression model analysis showed participants with CHF had a 54% increased risk of 1-year mortality after surgery (HR 1.54, 95% CI: 1.19-1.98, P < 0.001). The multivariate-adjusted HR of the inverse probability of treatment-weighted and standardised mortality ratio-weighted cohorts was 1.34 (95% CI: 1.10-1.62, P = 0.004) and 1.24 (95% CI: 1.17-1.32, P < 0.001), respectively.ConclusionCHF is an independent risk factor for 1-year mortality after surgery in patients undergoing surgery. Depending on the statistical method, patients with CHF had a 24-54% increased risk of 1-year all-cause mortality after surgery. This provides a reference for optimising clinical decision-making, improving preoperative consultation, and promoting clinical communication.

Project description:BackgroundThe diagnostic process is a key element of medicine but it is complex and prone to errors. Infectious diseases are one of the three categories of diseases in which diagnostic errors can be most harmful to patients. In this study we aimed to estimate the effect of initial misdiagnosis of the source of infection in patients with bacteraemia on 14 day mortality using propensity score methods to adjust for confounding.MethodsData from a previously described longitudinal cohort of patients diagnosed with monobacterial bloodstream infection (BSI) at the Leiden University Medical Centre (LUMC) between 2013 and 2015 were used. Propensity score matching and inversed probability of treatment weighting (IPTW) were applied to correct for confounding. The average treatment effect on the treated (ATT), which in this study was the average effect of initial misdiagnosis on the misdiagnosed (AEMM), was estimated. Methodological issues that were encountered when applying propensity score methods were addressed by performing additional sensitivity analyses. Sensitivity analyses consisted of varying caliper in propensity score matching and using different truncated weights in inversed probability of treatment weighting.ResultsData of 887 patients were included in the study. Propensity scores ranged between 0.015 and 0.999 and 80 patients (9.9%) had a propensity score > 0.95. In the matched analyses, 35 of the 171 misdiagnosed patients died within 14 days (20.5%), versus 10 of the 171 correctly diagnosed patients (5.8%), yielding a difference of 14.6% (7.6%; 21.6%). In the total group of patients, the observed percentage of patients with an incorrect initial diagnosis that died within 14 days was 19.8% while propensity score reweighting estimated that their probability of dying would have been 6.5%, if they had been correctly diagnosed (difference 13.3% (95% CI 6.9%;19.6%)). After adjustment for all variables that showed disbalance in the propensity score a difference of 13.7% (7.4%; 19.9%) was estimated. Sensitivity analyses yielded similar results. However, performing weighted analyses without truncation yielded unstable results.ConclusionThus, we observed a substantial increase of 14 day mortality in initially misdiagnosed patients. Furthermore, several patients received propensity scores extremely close to one and were almost sure to be initially misdiagnosed.

Dataset Information

Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance.

Background

Methods

Results

Conclusion

Publications

Double-adjustment in propensity score matching analysis: choosing a threshold for considering residual imbalance.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets