scholarly journals Improving the causal treatment effect estimation with propensity scores by the bootstrap

2021 ◽  
Author(s):  
Maeregu W. Arisido ◽  
Fulvia Mecatti ◽  
Paola Rebora
Keyword(s):  
Propensity Score ◽  
Selection Bias ◽  
Propensity Scores ◽  
Treatment Selection ◽  
Treatment Effect Estimation ◽  
Causal Treatment ◽  
Effect Estimation ◽  
Propensity Score Weight ◽  
Stage Renal Disease ◽  
End Stage

AbstractWhen observational studies are used to establish the causal effects of treatments, the estimated effect is affected by treatment selection bias. The inverse propensity score weight (IPSW) is often used to deal with such bias. However, IPSW requires strong assumptions whose misspecifications and strategies to correct the misspecifications were rarely studied. We present a bootstrap bias correction of IPSW (BC-IPSW) to improve the performance of propensity score in dealing with treatment selection bias in the presence of failure to the ignorability and overlap assumptions. The approach was motivated by a real observational study to explore the potential of anticoagulant treatment for reducing mortality in patients with end-stage renal disease. The benefit of the treatment to enhance survival was demonstrated; the suggested BC-IPSW method indicated a statistically significant reduction in mortality for patients receiving the treatment. Using extensive simulations, we show that BC-IPSW substantially reduced the bias due to the misspecification of the ignorability and overlap assumptions. Further, we showed that IPSW is still useful to account for the lack of treatment randomization, but its advantages are stringently linked to the satisfaction of ignorability, indicating that the existence of relevant though unmeasured or unused covariates can worsen the selection bias.

scholarly journals Comparison of Methods for Handling Covariate Missingness in Propensity Score Estimation with a Binary Exposure

2019 ◽  
Author(s):  
Donna Coffman ◽  
Jiangxiu Zhou ◽  
Xizhen Cai
Keyword(s):  
Propensity Score ◽  
Multiple Imputation ◽  
Missing Values ◽  
Propensity Scores ◽  
Causal Effect ◽  
Nonparametric Approach ◽  
Split Method ◽  
Mean Imputation ◽  
Substantial Bias ◽  
Effect Estimation

Abstract Background Causal effect estimation with observational data is subject to bias due to confounding, which is often controlled for using propensity scores. One unresolved issue in propensity score estimation is how to handle missing values in covariates.Method Several approaches have been proposed for handling covariate missingness, including multiple imputation (MI), multiple imputation with missingness pattern (MIMP), and treatment mean imputation. However, there are other potentially useful approaches that have not been evaluated, including single imputation (SI) + prediction error (PE), SI+PE + parameter uncertainty (PU), and Generalized Boosted Modeling (GBM), which is a nonparametric approach for estimating propensity scores in which missing values are automatically handled in the estimation using a surrogate split method. To evaluate the performance of these approaches, a simulation study was conducted.Results Results suggested that SI+PE, SI+PE+PU, MI, and MIMP perform almost equally well and better than treatment mean imputation and GBM in terms of bias; however, MI and MIMP account for the additional uncertainty of imputing the missingness.Conclusions Applying GBM to the incomplete data and relying on the surrogate split approach resulted in substantial bias. Imputation prior to implementing GBM is recommended.

scholarly journals Causal Subclassification Tree Algorithm and Robust Causal Effect Estimation via Subclassification

2020 ◽  
Vol 10 (1) ◽  
pp. 40
Author(s):  
Tomoshige Nakamura ◽  
Mihoko Minami
Keyword(s):  
Propensity Score ◽  
Propensity Scores ◽  
Causal Effect ◽  
Model Misspecification ◽  
Robust Estimator ◽  
Propensity Score Model ◽  
Confounding Variables ◽  
Score Model ◽  
Effect Estimation ◽  
Weighting Methods

In observational studies, the existence of confounding variables should be attended to, and propensity score weighting methods are often used to eliminate their e ects. Although many causal estimators have been proposed based on propensity scores, these estimators generally assume that the propensity scores are properly estimated. However, researchers have found that even a slight misspecification of the propensity score model can result in a bias of estimated treatment effects. Model misspecification problems may occur in practice, and hence, using a robust estimator for causal effect is recommended. One such estimator is a subclassification estimator. Wang, Zhang, Richardson, & Zhou (2020) presented the conditions necessary for subclassification estimators to have $\sqrt{N}$-consistency and to be asymptotically well-defined and suggested an idea how to construct subclasses.

Abstract 17474: End-Stage Renal Disease is Associated With Increased Complications From Watchman Implantation: A Propensity-Score Matched Analysis From National Readmission Database

Circulation ◽  
2020 ◽  
Vol 142 (Suppl_3) ◽  
Author(s):  
Moghniuddin Mohammed ◽  
Amit Noheria ◽  
Seth Sheldon ◽  
Madhu Reddy
Keyword(s):  
Propensity Score ◽  
Renal Disease ◽  
Propensity Score Matching ◽  
End Stage Renal Disease ◽  
Early Mortality ◽  
Index Hospitalization ◽  
Index Hospital ◽  
Stage Renal Disease ◽  
End Stage ◽  
National Readmission Database

Introduction: End-stage renal disease (ESRD) is associated with increased complications due to oral anticoagulation (OAC) use for stroke prevention in atrial fibrillation (AF). Left atrial appendage occlusion (LAAO) is indicated for patients who cannot tolerate or prefer not to use OAC but the outcomes of LAAO in ESRD has not been well studied. Methods: Using National Readmission Database January 2016-December 2017, we identified all adult AF patients who had LAAO performed in the months of January to November with no missing length of stay and/or mortality information. We excluded patients who had ablation, device implantation/revision, other form of LAAO and/or coronary artery bypass graft surgery performed during index hospitalization. 1:1 propensity score matching was performed for patients with and without ESRD based on variables shown in Table 1. The main outcome of interest was early mortality defined as mortality of index hospitalization or 30-day readmissions and index hospital complications. Results: A total of 13,790 procedures were included and of these 370 patients had history of ESRD. The baseline characteristics before and after matching are shown in table 1. After propensity score matching, ESRD group was associated with significantly higher early mortality, 30-day readmissions, systemic embolism and pericardial complications (Table 2). In the matched cohort, none of the patients had postprocedural cerebrovascular accident, transient ischemic attack, device thrombosis and device embolization. After propensity matching 2 (0.5%) developed acute kidney injury requiring hemodialysis. Conclusions: ESRD is associated with higher LOS, index hospital complications and early mortality from LAAO compared to patients without ESRD. Further studies comparing outcomes between OAC use and LAAO are warranted.

P5731Optimal blood pressure in diabetic hypertensive patients with overt proteinuria

2019 ◽  
Vol 40 (Supplement_1) ◽  
Author(s):  
C J Lee ◽  
J Hwang ◽  
C Y Kang ◽  
H Kim ◽  
J Ha ◽  
...  
Keyword(s):  
Blood Pressure ◽  
Diabetic Nephropathy ◽  
Lower Risk ◽  
Propensity Scores ◽  
Renal Outcome ◽  
Health Examination ◽  
Hypertensive Patients ◽  
Overt Proteinuria ◽  
Stage Renal Disease ◽  
End Stage

Abstract Background Evidence for the benefit of intensive blood pressure lowering in diabetic nephropathy is not clear at this time. The objective of this study was to demonstrate whether lower mean blood pressure (BP) in treated hypertensive patients with diabetic nephropathy is associated with better prognosis. Methods From the National Health Insurance Service (NHIS) Health Examination Database, diabetic hypertensive subjects with proteinuria between 2009 and 2010 were selected and followed-up until 2015 (N=8,663). Mean of the recorded systolic and diastolic BP during follow-up health examinations were stratified into five categories (SBP: <120, 120 to <130, 130 to <140, 140 to <150, and ≥150 mmHg; DBP: <70, 70 to <80, 80 to <90, 90 to <100, and ≥100 mmHg). All-cause death, myocardial infarction (MI), stroke, and renal outcome (progression to end stage renal disease or doubling of serum creatinine) were examined by Cox proportional hazard models with the propensity scores adjusted method. Results Compared to SBP of 130 to <140 mmHg, SBP of 120 to <130 mmHg was associated with lower risk of all-cause death (HR=0.78; 95% CI, 0.64–0.95), stroke (HR: 0.65; 95% CI, 0.45–0.94), and renal outcome (HR: 0.81; 95% CI, 0.68–0.97). SBP of <120 mmHg was associated with benefit for renal outcomes (HR: 0.69; 95% CI 0.55–0.88) but not with elevated risk of other outcomes. Compared to DBP of 80 to <90 mmHg, DBP of 70 to <80 mmHg were associated with lower risk of all-cause death (HR: 0.75; 95% CI, 0.64–0.88) but with higher risk of MI (HR: 1.52; 95% CI, 1.05–2.21). DBP of <70 mmHg was associated with reduced risk of all-cause death (HR: 0.79; 95% CI, 0.64–0.98). Conclusion In diabetic hypertensive subjects with overt proteinuria, deterioration of renal function decreased with decreasing SBP and the lowest risk of all-cause death and stroke were observed in SBP <130 mmHg. Low DBP was associated with low risk of all-cause death but there was a J curve phenomenon for MI in DBP of 70 to <80 mmHg.

Using Propensity Scores to Reduce Selection Bias in Mathematics Education Research

2010 ◽  
Vol 41 (2) ◽  
pp. 147-168 ◽  
Author(s):  
Suzanne E. Graham
Keyword(s):  
Education Research ◽  
Mathematics Education ◽  
Propensity Score ◽  
Selection Bias ◽  
Propensity Scores ◽  
Mathematics Education Research ◽  
Causal Inferences ◽  
Propensity Score Methods ◽  
Mathematics Classrooms ◽  
The Impact

Selection bias is a problem for mathematics education researchers interested in using observational rather than experimental data to make causal inferences about the effects of different instructional methods in mathematics on student outcomes. Propensity score methods represent 1 approach to dealing with such selection bias. This article describes general principles underlying propensity score methods and illustrates their application to mathematics education research using 2 examples investigating the impact of problem-solving emphasis in mathematics classrooms on students' subsequent mathematics achievement and course taking. Limitations of the method are discussed.

Risk of and Fatality from Acute Pancreatitis in Long-Term Hemodialysis and Peritoneal Dialysis Patients

2018 ◽  
Vol 38 (1) ◽  
pp. 30-36 ◽  
Author(s):  
I-Kuan Wang ◽  
Shih-Wei Lai ◽  
Hsueh-Chou Lai ◽  
Cheng-Li Lin ◽  
Tzung-Hai Yen ◽  
...  
Keyword(s):  
Acute Pancreatitis ◽  
Peritoneal Dialysis ◽  
Propensity Score ◽  
Dialysis Patients ◽  
Hazard Ratios ◽  
Significant Difference ◽  
Taiwan National Health Insurance ◽  
Stage Renal Disease ◽  
End Stage

Background This study was conducted to evaluate the risk of developing acute pancreatitis (AP) and the fatality from AP in hemodialysis (HD) and peritoneal dialysis (PD) patients, using the claims data of Taiwan National Health Insurance. Methods From patients with newly diagnosed end-stage renal disease (ESRD) in 2000–2010, we identified a PD cohort ( N = 9,766), a HD cohort ( N = 18,841), and a control cohort ( N = 114,386) matched by sex, age, and the diagnosis year of the PD cohort. We also established another 2 cohorts with 9,744 PD patients and 9,744 propensity score-matched HD patients. The incident AP and fatality from AP were evaluated for all cohorts by the end of 2011. Results The adjusted hazard ratios (HRs) of acute pancreatitis were 5.68 (95% confidence interval [CI] = 5.05 – 6.39), 4.91 (95% CI = 4.32 – 5.59), and 7.47 (95% CI = 6.48 – 8.62) in the all dialysis, HD, and PD patients, compared with the controls, respectively. Peritoneal dialysis patients had an adjusted HR of 1.41 (95% CI = 1.21 – 1.65) for AP, compared with propensity score-matched HD patients. Peritoneal dialysis patients under icodextrin treatment had a lower incidence of AP than those without the treatment, with an adjusted HR of 0.59 (95% CI = 0.47 – 0.73). There was no significant difference in the 30-day mortality from AP between HD and PD patients. Conclusions Peritoneal dialysis patients were at a higher risk of developing AP than HD patients. Icodextrin solution could reduce the risk of developing AP in PD patients.

scholarly journals A propensity-score matched observational study of remdesivir in patients with COVID-19 and severe kidney disease

Kidney360 ◽  
2021 ◽  
pp. 10.34067/KID.0006152021
Author(s):  
Rituvanthikaa Seethapathy ◽  
Sophia Zhao ◽  
Joshua D. Long ◽  
Ian A. Strohbehn ◽  
Meghan E. Sise
Keyword(s):  
Kidney Disease ◽  
Propensity Score ◽  
Propensity Scores ◽  
Altered Mental Status ◽  
Control Group ◽  
Increased Risk ◽  
Kidney Replacement Therapy ◽  
Clinically Significant ◽  
End Stage ◽  
Matched Study

Background: Remdesivir is not currently approved for patients with estimated glomerular filtration rate (eGFR) < 30mL/min/1.73m2. We aimed to determine the safety of remdesivir in patients with kidney failure. Methods: Retrospective cohort study of patients with COVID-19 hospitalized between May 2020 to January 2021 with eGFR <30 mL/min/1.73m2 who received remdesivir and historical controls with COVID-19 hospitalized between March 1, 2020 - April 30, 2020 prior to Emergency Use Authorization of remdesivir within a large healthcare system. Patients were 1:1 matched by propensity scores accounting for factors associated with treatment assignment. Adverse events and hospital outcomes were recorded by manual chart review. Results: The overall cohort included 34 hospitalized patients who initiated remdesivir within 72 hours of hospital admission with eGFR <30 mL/min/1.73m2 and 217 COVID-19 controls with eGFR <30 mL/min/1.73m2. The propensity score-matched cohort included 31 remdesivir treated cases and 31 non-remdesivir-treated controls. The mean age was 74.0 (SD: 13.8), 56.6% female, 67.7% white. A total of 25.5% had end-stage kidney disease. Among patients who were not on dialysis prior to initiating remdesivir, one developed worsening kidney function (defined ≥ 50% increase in creatinine or initiation of kidney replacement therapy) compared to three in the historical control group. There was no increased risk of cardiac arrythmia, cardiac arrest, altered mental status, or clinically significant anemia or liver function test abnormalities. There was a significantly increased risk of hyperglycemia, which may be partly explained by the increased use of dexamethasone in the remdesivir-treated population. Conclusion: In this propensity-score matched study, remdesivir was well tolerated in patients with eGFR < 30 mL/min/1.73m2.

scholarly journals Management of Aortic Stenosis in Patients With End-Stage Renal Disease on Hemodialysis

2020 ◽  
Vol 13 (8) ◽  
Author(s):  
Amgad Mentias ◽  
Milind Y. Desai ◽  
Marwan Saad ◽  
Phillip A. Horwitz ◽  
James D. Rossen ◽  
...  
Keyword(s):  
Heart Failure ◽  
Propensity Score ◽  
Aortic Stenosis ◽  
Renal Disease ◽  
Conservative Management ◽  
End Stage Renal Disease ◽  
Rate Ratio ◽  
Incidence Rate Ratio ◽  
Stage Renal Disease ◽  
End Stage

Background: Patients with end-stage renal disease on hemodialysis (ESRD-HD) and aortic stenosis have poor prognosis. The role of transcatheter aortic valve replacement (TAVR) in this high-risk population is debated. Methods: We compared the outcomes among ESRD-HD Medicare beneficiaries who were managed with TAVR, surgical AVR (SAVR), or conservative management for aortic stenosis between 2015 and 2017, using overlap propensity score weighting analysis to control for differences in treatment assignment. The primary outcome was all-cause mortality and was compared between treatment groups as well as to age-sex matched mortality for ESRD-HD in the US population. Secondary outcomes included trend of heart failure hospitalizations. Results: A total of 8107 ESRD-HD patients with aortic stenosis were included, 4130 (50%) underwent TAVR, 2565 (31.6%) underwent SAVR, and 1412 (17.4%) were managed conservatively. TAVR patients had more comorbidities and higher frailty compared with the other 2 groups. Thirty-day mortality was lower with TAVR compared with SAVR (4.6% versus 12.8%, P <0.01). After a median follow-up of 465 days (interquartile range, 261–759), on overlap propensity score weighting analysis, there was no difference in mortality between TAVR and SAVR (adjusted hazard ratio, 1.02 [95% CI, 0.91–1.15], P =0.7), and mortality was lower with TAVR compared with conservative management (adjusted hazard ratio, 0.53 [95% CI, 0.47–0.60], P <0.001). Standardized mortality ratios with TAVR, SAVR, and conservative management compared with age-sex matched ESRD-HD US population were 1.24, 1.27, and 1.83, respectively. The rate of heart failure admissions declined after TAVR (incidence rate ratio, 0.55 [95% CI, 0.48–0.62], P <0.001) and SAVR (incidence rate ratio, 0.76 [95% CI, 0.65–0.88], P <0.001). Conclusions: In ESRD-HD patients with aortic stenosis, mortality was lower in the short-term with TAVR compared with SAVR but comparable in the mid-term. AVR is associated with an improvement in survival and reduction in heart failure hospitalizations compared with conservative management.

scholarly journals Propensity Score-Matched Analysis of the Survival Benefit from Kidney Transplantation in Patients with End-Stage Renal Disease

2018 ◽  
Vol 7 (11) ◽  
pp. 388
Author(s):  
Ming-Ju Wu ◽  
Tung-Min Yu ◽  
Cheng-Li Lin ◽  
Chia-Hung Kao
Keyword(s):  
Kidney Transplantation ◽  
Propensity Score ◽  
Renal Disease ◽  
End Stage Renal Disease ◽  
Survival Outcome ◽  
Research Database ◽  
Newly Diagnosed ◽  
Dialysis Patients ◽  
Stage Renal Disease ◽  
End Stage

Background: Several comparison studies have suggested that kidney transplantation (KT) could reduce mortality in patients with end-stage renal disease (ESRD). Selection criteria bias is common in the selection of dialysis patients for control groups. In this study, we compared the survival outcome between KT recipients and comparable propensity score-matched dialysis patients. Methods: We used Taiwan’s National Health Insurance Research Database to identify patients newly diagnosed with ESRD between 2000 and 2010. We separated them into two groups: a KT group and non-KT dialysis-only group. To evaluate the survival outcome, we compared each patient with KT to a patient on dialysis without KT using propensity score matching. Results: In total, 1276 KT recipients and 1276 propensity score-matched dialysis patients were identified. Compared with the propensity score-matched dialysis patients, the patients who underwent KT exhibited significantly higher 5-year and 10-year survival rates (88% vs. 92% and 74% vs. 87%, both p < 0.05). The crude and adjusted hazard ratios for mortality were 0.55 and 0.52 in patients with KT (both p < 0.001). Mortality was insignificantly higher for patients who were on dialysis for longer than 1 year prior to KT compared with those on dialysis for less than 1 year. Conclusion: This study used a propensity score-matched cohort to confirm that KT is associated with lower risk of mortality than dialysis alone in patients newly diagnosed with ESRD.

Propensity score analysis for correlated subgroup effects

2019 ◽  
Vol 29 (4) ◽  
pp. 1067-1080
Author(s):  
Shan-Yu Liu ◽  
Chunyan Liu ◽  
Eddie Nehus ◽  
Maurizio Macaluso ◽  
Bo Lu ◽  
...  
Keyword(s):  
Propensity Score ◽  
Variable Selection ◽  
Propensity Scores ◽  
Propensity Score Analysis ◽  
Real Data ◽  
Treatment Selection ◽  
Propensity Score Model ◽  
Interaction Terms ◽  
Selection Processes ◽  
Subgroup Effects

As individuals may respond differently to treatment, estimating subgroup effects is important to understand the characteristics of individuals who may benefit. Factors that define subgroups may be correlated, complicating evaluation of subgroup effects, especially in observational studies requiring control of confounding variables. We address this problem when propensity score methods are used for confounding control. A common practice is to evaluate candidate subgroup identifiers one at a time without adjusting for other candidate identifiers. We show that this practice can be misleading if the treatment effect modification attributed to a candidate identifier is in truth due to the effect of other correlated true effect modifiers. Whereas jointly analyzing multiple identifiers provides estimates of the desired subgroup effects adjusted for the effects of the other identifiers, it requires the propensity scores to adequately reflect the underlying treatment selection processes and balance the covariates within each subgroup of interest. Satisfying the requirement in practice is hard since the number of strata may increase quickly, while the per stratum sample size may decrease dramatically. A practically helpful approach is utilizing the whole cohort for the propensity score estimation with modeling of interaction terms to reflect the potentially different treatment selection processes across strata. We empirically examine the performance of the whole cohort approach by itself and with subjecting the interaction terms to variable selection. Our results using both simulations and real data analysis suggest that the whole cohort approach should explore inclusion of high-order interactions in the propensity score model to ensure adequate covariate balance across strata, and that variable selection is of limited utility.

Sign in / Sign up

Export Citation Format

Share Document