Propensity Score Estimation Research Articles

The lasting impact of historical residential security maps on experienced segregation

AbstractWe study the impact of 1930s Residential Security maps on experienced segregation as measured by geolocation data. We compare adjacent neighborhoods with different creditworthiness grades. Using a sample of neighborhood boundaries, which, based on estimated propensity scores, were likely drawn for idiosyncratic reasons, residents of neighborhoods on the lower graded side of the boundary are more likely to visit other historically lower graded destination neighborhoods and neighborhoods that currently tend to have lower household income and lower educational attainment. Disparities in visits are not driven by work commutes, very local visits, or differences in income.

Machine Learning Algorithms to Estimate Propensity Scores in Health Policy Evaluation: A Scoping Review

Machine Learning Algorithms to Estimate Propensity Scores in Health Policy Evaluation: A Scoping Review

Favorable Impact of Azithromycin on Patients in the Intensive Care Unit with Coronavirus Disease 2019: Insights from the First Wave Using a Japanese Database

BackgroundAzithromycin has favorable effects on critical respiratory diseases owing to its antimicrobial and anti-inflammatory properties. During the early stages of the coronavirus disease 2019 (COVID-19) pandemic, azithromycin was frequently administered before specific treatments were developed. However, the efficacy of this treatment has not been verified. We retrospectively investigated the effects of its intravenous (IV) administration in patients with severe/critical COVID-19 using the National Administrative Database of Japan during the first wave (February–April 2020). MethodsPatients were categorized based on whether they received IV azithromycin within three days of hospitalization. An overlap weighting method with estimated propensity scores was used to reduce bias. ResultsAmong the 830 patients with severe/critical COVID-19, 148 (17.8%) received azithromycin, and 682 (82.2%) did not. After adjustment, the use of azithromycin was associated with a shorter duration of intensive care unit (ICU) management (-3.48 days, 95% confidence interval [CI]: -4.59 to -2.38). However, other endpoints, including mortality rate, duration of mechanical ventilation, and duration of hospital stay, did not suggest any associations. Furthermore, of the 115 ICU patients, 27 (23.5%) were treated with IV azithromycin and 88 (76.5%) were not. After adjustment, azithromycin was associated with favorable outcomes, including reduced in-hospital mortality (odds ratio [OR], 0.45, 95% CI: 0.22 to 0.92), 30-day mortality (OR, 0.46, 95% CI: 0.22 to 0.94), and a shorter duration of ICU management (-2.94 days, 95% CI: -5.15 to -0.73). ConclusionWe verified that IV azithromycin was associated with favorable impact in patients with COVID-19 requiring ICU management.

Causal Inference with Complex Surveys: A Unified Perspective on Sample Selection and Exposure Selection

Probability surveys are a major source of population representative data for policy research and program evaluation. However, the data come with the added complications of being observational and selected with unequal probabilities. Propensity score adjustments have become increasingly popular for inferring causal relationships in non-randomized studies, but when using survey data, estimates of the population level causal effect may be biased if the sampling design is not adequately adjusted for. The current practice of using propensity score estimators with complex surveys is somewhat ad-hoc. We propose a potential-outcome super-population framework to streamline the causal analysis. We also develop propensity-score-and-survey weighted estimators and corresponding variance estimators, as well as their asymptotic properties. Our framework clarifies the confusion regarding the use of survey weighted propensity score in practice. The choice actually depends on the available sampling weights. Various estimators are compared in a simulation study, which shows that the proposed estimators perform better than the competing methods in terms of bias and confidence interval coverage when treatment effects are heterogeneous. To address an important public health issue, we evaluate the impact of e-cigarette use on future tobacco use intention in teens, using a large nationally representative survey in the United States.

Machine learning methods for propensity and disease risk score estimation in high-dimensional data: a plasmode simulation and real-world data cohort analysis

IntroductionMachine learning (ML) methods are promising and scalable alternatives for propensity score (PS) estimation, but their comparative performance in disease risk score (DRS) estimation remains unexplored.MethodsWe used real-world data comparing antihypertensive users to non-users with 69 negative control outcomes, and plasmode simulations to study the performance of ML methods in PS and DRS estimation. We conducted a cohort study using UK primary care records. Further, we conducted a plasmode simulation with synthetic treatment and outcome mimicking empirical data distributions. We compared four PS and DRS estimation methods: 1. Reference: Logistic regression including clinically chosen confounders. 2. Logistic regression with L1 regularisation (LASSO). 3. Multi-layer perceptron (MLP). 4. Extreme Gradient Boosting (XgBoost). Covariate balance, coverage of the null effect of negative control outcomes (real-world data) and bias based on the absolute difference between observed and true effects (for plasmode) were estimated. 632,201 antihypertensive users and nonusers were included.ResultsML methods outperformed the reference method for PS estimation in some scenarios, both in terms of covariate balance and coverage/bias. Specifically, XgBoost achieved the best performance. DRS-based methods performed worse than PS in all tested scenarios.DiscussionWe found that ML methods could be reliable alternatives for PS estimation. ML-based DRS methods performed worse than PS ones, likely given the rarity of outcomes.

Association between warfarin and direct oral anticoagulants users and risk of mortality: cohort studies in England and Hong Kong

Abstract Background The evidence of the association between warfarin and risk of mortality compared with direct oral anticoagulants (DOACs) has been mixed in current literature. Purpose To investigate the association between all-cause mortality and warfarin, compared with DOACs. Methods Population-based cohort studies were conducted using the data from the Hong Kong Clinical Data Analysis and Reporting System (CDARS) and United Kingdom Clinical Practice Research Datalink (CPRD) Aurum. People with atrial fibrillation aged ≥18 years who initiated warfarin or a DOAC during the study period (1st Jan 2014 - 31st Dec 2019) were identified. Follow-up started from the first date of oral anticoagulant prescription and ended at earliest of the date of death, the day before drug switching (between warfarin and DOACs), transferring out of the practice (CPRD Aurum only), last data collection date for the practice (CPRD Aurum only) or the end of study period. We estimated propensity score (PS) based on the list of potential confounders including demographics, co-morbidities, co-medications and polypharmacy. Cox proportional hazards regression model with PS-weighting was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Results Among 227,190 new oral anticoagulant users, 54,367 (23.93%) died. The median follow-up was 2.96 and 2.45 years in CPRD and CDARS, respectively. Compared with DOAC use, a lower hazard of all-cause mortality was found in warfarin use (PS-weighted hazard ratio [HR]=0.94, 95% confidence interval [CI]=0.89-0.99) in CPRD; while a higher hazard was observed in warfarin users (PS-weighted HR=1.31, 95% CI=1.24-1.39) in CDARS, versus DOAC users. Consistent results were seen in both databases when stratified warfarin users by Time in Therapeutic Range (TTR). We observed a lower hazard of all-cause mortality in warfarin users with TTR≥65% (PS-weighted HR=0.90, 95% CI=0.85-0.94 in CPRD; 0.87, 0.79-0.96 in CDARS) and increased hazard in warfarin users with TTR<65% (PS-weighted HR=1.37, 95% CI=1.25-1.49 in CPRD; 1.57, 1.48-1.66 in CDARS), versus DOAC users. Conclusions The hazard of all-cause mortality associated with warfarin compared with DOACs were found to be conflicting across databases. However, both databases consistently showed that an increased hazard of all-cause mortality was only found in warfarin users with TTR<65% but not in those with TTR≥65%, suggesting that the differences in hazard of all-cause mortality associated with warfarin compared with DOAC, in part may depend on anticoagulation control in warfarin users. We recommend that warfarin users with poor INR control warrant closer clinical attention and could be considered for a switch to DOACs.

Mismatch Repair Status and Surgical Outcomes in Localized Colorectal Cancer: A Nationwide Cohort Study

Objective: This study examined the association between deficient mismatch repair (dMMR) versus proficient MMR (pMMR) status and overall survival and disease-free survival in patients with localized colorectal cancer. Background: Several distinctions exist between patients with dMMR and pMMR colorectal cancer. However, the impact on prognosis is yet to be investigated in large-scale cohort studies. Methods: In this cohort study, we included patients who underwent curative-intent surgery for localized colorectal cancer between 2009 and 2020. Patients were identified in the Danish Colorectal Cancer Group database and patient-level data were extracted from 6 registry databases. After inclusion, patients with dMMR status were matched 1:1 to patients with pMMR status using an estimated propensity score. Results: After matching, 5994 patients were included. The patients had a median age of 74 years and a median follow-up of 4.1 years. There was no significant association between mismatch repair (MMR) status and overall survival (hazard ratio, 0.91; 95% confidence interval [CI], 0.81–1.03) or disease-free survival (hazard ratio, 0.89; 95% CI, 0.78–1.01). However, the restricted 5-year mean disease-free survival time, calculated due to violation of the proportional hazards assumption, showed a significant absolute difference of 0.13 years (95% CI, 0.03–0.23; P = 0.01) in favor of the dMMR group. Conclusions: No significant association with overall survival was found according to MMR status. dMMR status was, however, found to be associated with marginally improved disease-free survival compared to pMMR status in patients with localized colorectal cancer undergoing curative-intent surgery.

Association between an Early Revascularization Strategy and Major Cardiac Events Based on Left Ventricular Dyssynchrony in Patients at Intermediate Risk of Major Cardiac Events Using the J-ACCESS Risk Model.

Objective This retrospective study aimed to investigate the association between therapeutic strategies and the development of major cardiac events (MCEs) in intermediate-risk patients by using the the Japanese Assessment of Cardiac Events and Survival Study by Quantitative Gated SPECT (JACCESS) risk model in combination with the stress phase bandwidth (SPBW), an index of left ventricular dyssynchrony. Methods Patients were followed-up for three years to confirm their prognosis. Based on the estimated propensity scores, the patients who underwent revascularization within the first 60 days after SPECT and those who did not were matched 1:1 (n=367 per group). The composite endpoint was the occurrence of MCEs, consisting of cardiac death, non-fatal myocardial infarction, and severe heart failure. SPBW was calculated by a phase analysis using the Heart Risk View-F software program, and the MCE rate was compared between the two groups by applying the normal value of SPBW (38°). Patients The study included 2,053 patients with either known or suspected coronary artery disease who underwent electrocardiogram-gated single-photon emission computed tomography myocardial perfusion imaging and were at intermediate risk of MCE according to the J-ACCESS risk model. Results During follow-up, 54 of the 734 patients (7.4%) experienced MCEs. The overall incidence of MCE in intermediate-risk patients was not significantly different between the two groups. However, the incidence of MCE in patients with an abnormal SPBW was significantly lower in those who underwent early revascularization (4.8% vs. 11.9%, p=0.0407). Conclusion The combination of the J-ACCESS risk model and the SPBW is thus considered to be an optimal treatment strategy for patients at intermediate risk of MCE, and early revascularization may lead to an improved prognosis in intermediate-risk patients with an abnormal SPBW.

Comparative analysis of income level in matched small farms with and without eucalyptus plantations in Senan, Ethiopia

The cultivation of eucalyptus trees is a widely practiced economic activity in the highlands of the Amhara region, Ethiopia. This study investigates the causal impact of eucalyptus plantation ownership on household income within the Senan District of Ethiopia. Using cross-sectional data from 332 rural households, comprising 166 households with eucalyptus plantations and 166 without, the study employed a two-sample t-test comparison to compare the income level of households in the two groups. To identify the key factors influencing household income, the study conducted a Principal Component Analysis (PCA) and Propensity Score Matching (PSM) to isolate the causal impact of eucalyptus plantations on household income. This involved estimating propensity scores using a binary logistic regression model. The Average Treatment Effect on the Treated (ATT) analysis revealed that households that plant eucalyptus trees had 40.2 % higher total household income than those that did not participate in such activities. Therefore, considering potential demand and market fluctuations, eucalyptus tree growers need to diversify their livelihood strategies by utilizing the income generated from their plantations.

Prognostic score-based model averaging approach for propensity score estimation

BackgroundPropensity scores (PS) are typically evaluated using balance metrics that focus on covariate balance, often without considering their predictive power for the outcome. This approach may not always result in optimal bias reduction in the treatment effect estimate. To address this issue, evaluating covariate balance through prognostic scores, which account for the relationship between covariates and the outcome, has been proposed. Similarly, using a typical model averaging approach for PS estimation that minimizes prediction error for treatment status and covariate imbalance does not necessarily optimize PS-based confounding adjustment. As an alternative approach, using the averaged PS model that minimizes inter-group differences in the prognostic score may further reduce bias in the treatment effect estimate. Moreover, since the prognostic score is also an estimated quantity, model averaging in the prognostic scores can help identify a better prognostic score model. Utilizing the model-averaged prognostic scores as the balance metric for constructing the averaged PS model can contribute to further decreasing bias in treatment effect estimates. This paper demonstrates the effectiveness of the PS model averaging approach based on prognostic score balance and proposes a method that uses the model-averaged prognostic score as a balance metric, evaluating its performance through simulations and empirical analysis.MethodsWe conduct a series of simulations alongside an analysis of empirical observational data to compare the performances of weighted treatment effect estimates using the proposed and existing approaches. In our examination, we separately provid four candidate estimates for the PS and prognostic score models using traditional regression and machine learning methods. The model averaging of PS based on these candidate estimators is performed to either maximize the prediction accuracy of the treatment or to minimize intergroup differences in covariate distributions or prognostic scores. We also utilize not only the prognostic scores from each candidate model but also an averaged score that best predicted the outcome, for the balance assessment.ResultsThe simulation and empirical data analysis reveal that our proposed model-averaging approaches for PS estimation consistently yield lower bias and less variability in treatment effect estimates across various scenarios compared to existing methods. Specifically, using the optimally averaged prognostic scores as a balance metric significantly improves the robustness of the weighted treatment effect estimates.DiscussionThe prognostic score-based model averaging approach for estimating PS can outperform existing model averaging methods. In particular, the estimator using the model averaging prognostic score as a balance metric can produce more robust estimates. Since our results are obtained under relatively simple conditions, applying them to real data analysis requires adjustments to obtain accurate estimates according to the complexity and dimensionality of the data.ConclusionsUsing the prognostic score as the balance metric for the PS model averaging enhances the performance of the treatment effect estimator, which can be recommended for a wide variety of situations. When applying the proposed method to real-world data, it is important to use it in conjunction with techniques that mitigate issues arising from the complexity and high dimensionality of the data.

Comparison of Open and Endovascular Repair of Complex Abdominal Aortic Aneurysms

To compare perioperative and 5-year outcomes following endovascular (FEVAR) and open repair (OAR) of complex abdominal aortic aneurysms (cAAA) in males and females separately, given the known sex related differences in perioperative outcomes. We studied all elective cAAA repairs between 2014-2019 in the Vascular Implant Surveillance and Interventional Outcomes Network (VISION) registry. We stratified patients based on sex. We calculated propensity scores for assignment to either OAR or FEVAR. Covariates including age, race, diameter, baseline comorbidities, proximal extent of repair, annual center volumes, and annual surgeon volumes were introduced into the model for estimating propensity scores. Within matched cohorts, perioperative outcomes and 5-year outcomes (mortality, reinterventions, and ruptures) were evaluated using multivariable logistic and Cox regression models. We identified 2,825 patients of whom 29% were female. Within both the sexes, OAR was more commonly performed (OAR vs FEVAR: males: 53% vs 47%; females: 63% vs 37%). After matching, among males (n=1326), FEVAR was associated with lower perioperative mortality (FEVAR vs OAR: 2.3% vs 5.1%; p<.001). However, FEVAR was associated with comparable 5-year mortality (38% vs 28%; hazard ratio (HR) 1.2 [0.92-1.4]; p=.22) and a higher hazard of 5-year reintervention (19% vs 3.7%; aHR: 4.5 [2.6-7.6], p<.001). Among females (n=456), FEVAR and OAR showed similar perioperative mortality (8.3% vs 7.0%; p=.73). At 5 years, FEVAR was associated with higher hazards of mortality (43% vs 32%; aHR: 1.5 [1.03-2.2], p=.034) and reintervention (20% vs 3.0%; aHR: 4.8 [2.1-11], p<.001) compared with OAR. Among males, FEVAR was associated with favorable perioperative outcomes compared with OAR, though these advantages attenuate over time. However, among females, FEVAR was associated with similar perioperative outcomes, eventually leading to higher reinterventions and possibly higher mortality within 5 years. Future efforts should focus on determining the factors associated with these sex disparities to improve outcomes following FEVAR in females. Based on current evidence, females undergoing elective cAAA repair should be selected with due caution, especially for endovascular repair.

A Retrospective, Longitudinal Registry Study on the Long-Term Durability of Ivacaftor Treatment in People with Cystic Fibrosis.

Ivacaftor (IVA) has been shown to change the trajectory of cystic fibrosis (CF) disease progression by slowing the rate of lung function decline in clinical studies. Long-term real-world data help to confirm the durability of this response. This non-interventional, longitudinal study used data from the US CF Foundation Patient Registry to describe the annualized rate of change in lung function in people with CF receiving IVA. The IVA-treated cohort included people with CF aged ≥ 6years who had ≥ 1 CF transmembrane conductance regulator (CFTR)-gating mutation and initiated IVA between 31 January 2012 and 31 December 2018. An age-matched comparator cohort included people with CF heterozygous for the F508del-CFTR mutation and a minimal function mutation (R117H excluded) and had not received CFTR modulator therapy. Baseline characteristics were balanced using standardized mortality ratio (SMR) weights computed from estimated propensity scores. The annualized rate of change in percent predicted forced expiratory volume in 1s (ppFEV1) was estimated over 5years and used to calculate the relative annualized rate of change in lung function in the IVA-treated versus comparator cohorts. In the 5-year follow-up period, 548 people were in the IVA-treated and 541 in the comparator cohorts after SMR weighting. The annualized rate of change in ppFEV1 over 5years was -1.23 (95% CI -1.45, -1.03) and -2.03 (-2.16, -1.90) percentage points in the IVA-treated and comparator cohorts, respectively. There was a 39% reduction (95% CI: 28, 50) in the rate of lung function decline in the IVA-treated versus comparator cohort over 5years. Findings were generally consistent with those of shorter follow-up periods. IVA showed a durable clinical benefit by slowing the rate of lung function decline over 5years. Results support a sustained and consistent impact of IVA on lung function trajectory in people with CF. Word count: 300 (limit: 300 words).

Communication‐Efficient Distributed Estimation of Causal Effects With High‐Dimensional Data

ABSTRACTWe propose a communication‐efficient algorithm to estimate the average treatment effect (ATE), when the data are distributed across multiple sites and the number of covariates is possibly much larger than the sample size in each site. Our main idea is to calibrate the estimates of the propensity score and outcome models using some proper surrogate loss functions to approximately attain the desired covariate balancing property. We show that under possible model misspecification, our distributed covariate balancing propensity score estimator (disthdCBPS) can approximate the global estimator, obtained by pooling together the data from multiple sites, at a fast rate. Thus, our estimator remains consistent and asymptotically normal. In addition, when both the propensity score and the outcome models are correctly specified, the proposed estimator attains the semi‐parametric efficiency bound. We illustrate the empirical performance of the proposed method in both simulation and empirical studies.

Can supervised deep learning architecture outperform autoencoders in building propensity score models for matching?

PurposePropensity score matching is vital in epidemiological studies using observational data, yet its estimates relies on correct model-specification. This study assesses supervised deep learning models and unsupervised autoencoders for propensity score estimation, comparing them with traditional methods for bias and variance accuracy in treatment effect estimations.MethodsUtilizing a plasmode simulation based on the Right Heart Catheterization dataset, under a variety of settings, we evaluated (1) a supervised deep learning architecture and (2) an unsupervised autoencoder, alongside two traditional methods: logistic regression and a spline-based method in estimating propensity scores for matching. Performance metrics included bias, standard errors, and coverage probability. The analysis was also extended to real-world data, with estimates compared to those obtained via a double robust approach.ResultsThe analysis revealed that supervised deep learning models outperformed unsupervised autoencoders in variance estimation while maintaining comparable levels of bias. These results were supported by analyses of real-world data, where the supervised model’s estimates closely matched those derived from conventional methods. Additionally, deep learning models performed well compared to traditional methods in settings where exposure was rare.ConclusionSupervised deep learning models hold promise in refining propensity score estimations in epidemiological research, offering nuanced confounder adjustment, especially in complex datasets. We endorse integrating supervised deep learning into epidemiological research and share reproducible codes for widespread use and methodological transparency.

Minimizing confounding in comparative observational studies with time-to-event outcomes: An extensive comparison of covariate balancing methods using Monte Carlo simulation.

Observational studies are frequently used in clinical research to estimate the effects of treatments or exposures on outcomes. To reduce the effects of confounding when estimating treatment effects, covariate balancing methods are frequently implemented. This study evaluated, using extensive Monte Carlo simulation, several methods of covariate balancing, and two methods for propensity score estimation, for estimating the average treatment effect on the treated using a hazard ratio from a Cox proportional hazards model. With respect to minimizing bias and maximizing accuracy (as measured by the mean square error) of the treatment effect, the average treatment effect on the treated weighting, fine stratification, and optimal full matching with a conventional logistic regression model for the propensity score performed best across all simulated conditions. Other methods performed well in specific circumstances, such as pair matching when sample sizes were large (n = 5000) and the proportion treated was < 0.25. Statistical power was generally higher for weighting methods than matching methods, and Type I error rates were at or below the nominal level for balancing methods with unbiased treatment effect estimates. There was also a decreasing effective sample size with an increasing number of strata, therefore for stratification-based weighting methods, it may be important to consider fewer strata. Generally, we recommend methods that performed well in our simulations, although the identification of methods that performed well is necessarily limited by the specific features of our simulation. The methods are illustrated using a real-world example comparing beta blockers and angiotensin-converting enzyme inhibitors among hypertensive patients at risk for incident stroke.

The impact of incomplete revascularization on survival in minimal invasive off-pump coronary artery surgery: a propensity score analysis of 1,149 cases.

Minimally invasive concepts are increasingly influential in modern cardiac surgery. This study aimed to evaluate the effect of completeness of revascularization on clinical outcomes and overall survival in minimally invasive, thoracoscopic coronary artery bypass grafting (CABG) surgery. We retrospectively evaluated a consecutive series of 1,149 patients who underwent minimally invasive off-pump CABG with single, double, or triple-vessel revascularization between 2007 and 2018. Of these patients, 185 (16.1%) had incomplete revascularization (IR) (group I), and 964 (83.9%) had complete revascularization (CR) (group C). We used gradient boosted propensity score estimation to account for possible confounding variables. Median age was 69 years, interquartile range (IQR) 60-76 years, and median EuroSCORE II was 4, IQR 2-7. Of the 1,149 patients, 495 patients suffered from two vessel disease (VD) and 353 presented with three VD. Long-term median follow-up 5.58 (3.27-8.48) years was available for 1,089 patients (94.8%). The incidence of recurrent or persisting angina, myocardial infarction, redo-bypass surgery, and stroke during follow-up did not differ significantly between groups. During follow-up, there were 47 deaths in group I and 172 deaths in group C. The 1-, 3-, 5-, 8-, and 10-year unadjusted survival rates were 94%, 84%, 75%, 62%, and 51% for group I, and 97%, 94%, 88%, 77%, and 72% for group C, respectively (long-rank test P<0.001), favouring CR. Following risk adjustment the long-rank test P value for survival was 0.23. In minimally invasive coronary surgery, IR resulted in decreased long-term survival, but did not achieve statistical significance after risk adjustment. However, IR should only be used in carefully selected cases.

Consanguineous marriage and associated diseases among their children and grandchildren in India: evidence from large-scale data.

Worldwide, more than 130 million infants are born each year and a considerable number of 13.5 million of these children have inbred parents. The present study aimed to investigate the association between parents' consanguinity and chronic illness among their children and grandchildren in India. The nationally representative data, Longitudinal Aging Study in India, 2017-2018, Wave 1 was used for the present study. Bivariate analysis, a probit model, and propensity score estimation were employed to conduct the study. The study observed the highest prevalence of consanguinity marriage in the state of Andhra Pradesh (28%) and the lowest in Kerala (5%) among the south Indian States. People who lived in rural areas, belonged to the richer wealth quintile and Hindu religion were the significant predictors of consanguinity marriage in India. For individuals who were in consanguineous marriages, there was 0.85%, 0.84%, 1.57% 0.43%, 0.34%, and 0.14% chances of their children and grandchildren developing psychotic disorders, heart disease, hypertension stroke, cancer, and diabetes, respectively. Moreover, around 4.55% of the individuals have a history of birth defects or congenital disorders. To address the risk of complicated illnesses due to the consanguinity of marriage, medical, genetic, and social counselling services are required.

Propensity score matching for estimating a marginal hazard ratio.

Propensity score matching is commonly used to draw causal inference from observational survival data. However, its asymptotic properties have yet to be established, and variance estimation is still open to debate. We derive the statistical properties of the propensity score matching estimator of the marginal causal hazard ratio based on matching with replacement and a fixed number of matches. We also propose a double-resampling technique for variance estimation that takes into account the uncertainty due to propensity score estimation prior to matching.

Characterizing the Association Between Traumatic Brain Injury and Discontinuation of Medications for Opioid Use Disorder in a Commercially Insured Adult Population.

Extending prior research that has found that people with traumatic brain injury (TBI) experience worse substance use treatment outcomes, we examined whether history of TBI was associated with discontinuation of medication to treat opioid use disorder (MOUD), an indicator of receiving evidence-based treatment. We used MarketScan claims data to capture inpatient, outpatient, and retail pharmacy utilization from large employers in all 50 states from 2016 to 2019. We identified adults aged 18 to 64 initiating non-methadone MOUD (ie, buprenorphine, injectable naltrexone, and oral naltrexone) in 2016-2019. The exposure was whether an individual had a TBI diagnosis in the 2years before initiating MOUD. During this period, there were 709 individuals with TBI who were then matched with 709 individuals without TBI. We created a retrospective cohort of matched individuals with and without TBI and used quasi-experimental methods to identify the association between TBI status and MOUD use. We estimated propensity scores by TBI status and created a 1:1 matched cohort of people with and without TBI who initiated MOUD. We used a Cox proportional hazards model to identify the association between TBI and MOUD discontinuation. The outcome was discontinuation of MOUD (ie, a gap of 14days or more of MOUD). Among those initiating MOUD, the majority were under 26years of age, male, and living in an urban setting. Nearly 60% of individuals discontinued medication by 6months. Adults with TBI had an elevated risk of MOUD discontinuation (hazard ratio [HR] 1.13; 95% confidence interval [CI], 1.01-1.27) compared to those without TBI. Additionally, initiating oral naltrexone was associated with a higher risk of discontinuation (HR 1.63; 95% CI, 1.40-1.90). We found evidence of reduced MOUD retention among people with TBI. Differences in MOUD retention may reflect health care inequities, as there are no medical contraindications to using MOUD for people with TBI or other disabilities.

A comparison of random forest-based missing imputation methods for covariates in propensity score analysis.

Propensity score analysis (PSA) is a prominent method to alleviate selection bias in observational studies, but missing data in covariates is prevalent and must be dealt with during propensity score estimation. Through Monte Carlo simulations, this study evaluates the use of imputation methods based on multiple random forests algorithms to handle missing data in covariates: multivariate imputation by chained equations-random forest (Caliber), proximity imputation (PI), and missForest. The results indicated that PI and missForest outperformed other methods with respect to bias of average treatment effect regardless of sample size and missing mechanisms. A demonstration of these five methods with PSA to evaluate the effect of participation in center-based care on children's reading ability is provided using data from the Early Childhood Longitudinal Study, Kindergarten Class of 2010-2011. (PsycInfo Database Record (c) 2024 APA, all rights reserved).