scholarly journals Propensity Score Weighting And Trimming Strategies To Reduce Variance And Bias Of Treatment Effect Estimates: A Simulation Study

2021 ◽  
Author(s):  
Til Stürmer ◽  
Michael Webster-Clark ◽  
Jennifer L Lund ◽  
Richard Wyss ◽  
Alan R Ellis ◽  
...  
Keyword(s):  
Propensity Score ◽  
Treatment Effect ◽  
Relative Efficiency ◽  
Rate Ratio ◽  
Primary Analysis ◽  
Weighting Method ◽  
Unmeasured Confounding ◽  
C Statistic ◽  
Treatment Prediction ◽  
Poisson Function

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 two “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 based on measured covariates and trimmed the tails of the PS distribution using three strategies (“Crump”, “Stürmer”, and “Walker”). After trimming and re-estimation, we used alternative PS weights to estimate the treatment effect (rate ratio): IPTW, SMR-treated, SMR-untreated, overlap (ATO), matching, and entropy. With no unmeasured confounding, ATO (123%) and “Crump” trimming (112%) improved relative efficiency compared with untrimmed IPTW. 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.

What is the treatment effect of surgery compared with nonoperative treatment in patients with lumbar spinal stenosis at 1-year follow-up?

2019 ◽  
Vol 31 (2) ◽  
pp. 185-193
Author(s):  
Ulrike Held ◽  
Johann Steurer ◽  
Giuseppe Pichierri ◽  
Maria M. Wertli ◽  
Mazda Farshad ◽  
...  
Keyword(s):  
Quality Of Life ◽  
Propensity Score ◽  
Propensity Score Matching ◽  
Spinal Stenosis ◽  
Treatment Effect ◽  
Nonoperative Treatment ◽  
Unmeasured Confounding ◽  
Lumbar Spinal

OBJECTIVEThe aim of this study was to obtain an unbiased causal treatment estimate of the between-group difference of surgery versus nonoperative treatment with respect to outcomes on quality of life, pain, and disability in patients with degenerative lumbar spinal stenosis (DLSS) 12 months after baseline.METHODSThe authors included DLSS patients from a large prospective multicenter observational cohort study. Propensity score matching was used, including 15 demographic, clinical, and MRI variables. Linear and logistic mixed-effects regression models were applied to quantify the between-group treatment effect. Unmeasured confounding was addressed in a sensitivity analysis, assessing the robustness of the results.RESULTSA total of 408 patients were included in this study, 222 patients after matching, with 111 in each treatment group. Patients with nonoperative treatment had lower quality of life at the 12-month follow-up (−6.21 points, 95% CI −9.93 to −2.49) as well as lower chances of reaching a minimal clinically important difference in Spinal Stenosis Measure (SSM) symptoms (OR 0.26, 95% CI 0.13 to 0.53) and SSM function (OR 0.26, 95% CI 0.14 to 0.49), than patients undergoing surgery. These results were very robust in case of unmeasured confounding. The surgical complication rate was low; 5 (4.5%) patients experienced a durotomy during intervention, and 5 (4.5%) patients underwent re-decompression.CONCLUSIONSThe authors used propensity score matching to assess the difference in treatment efficacy of surgery compared with nonoperative treatment in elderly patients with DLSS. This study delivers strong evidence that surgical treatment is superior to nonoperative treatment. It helps in clinical decision-making, especially when patients suffer for a long time, sometimes over many years, hoping for a spontaneous improvement of their symptoms. In light of these new results, the number of years with disability can hopefully be reduced by providing adequate treatment at the right time for this ever-growing elderly and frail population.

scholarly journals A novel weighting method to remove bias from within-subject exposure dependency in case-crossover studies

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Kiyoshi Kubota ◽  
Thu-Lan Kelly ◽  
Tsugumichi Sato ◽  
Nicole Pratt ◽  
Elizabeth Roughead ◽  
...  
Keyword(s):  
Logistic Regression ◽  
Crossover Study ◽  
Odds Ratio ◽  
Time Trends ◽  
Rate Ratio ◽  
Conditional Logistic Regression ◽  
Time Varying ◽  
Weighting Method ◽  
Case Crossover ◽  
Crossover Studies

Abstract Background Case-crossover studies have been widely used in various fields including pharmacoepidemiology. Vines and Farrington indicated in 2001 that when within-subject exposure dependency exists, conditional logistic regression can be biased. However, this bias has not been well studied. Methods We have extended findings by Vines and Farrington to develop a weighting method for the case-crossover study which removes bias from within-subject exposure dependency. Our method calculates the exposure probability at the case period in the case-crossover study which is used to weight the likelihood formulae presented by Greenland in 1999. We simulated data for the population with a disease where most patients receive a cyclic treatment pattern with within-subject exposure dependency but no time trends while some patients stop and start treatment. Finally, the method was applied to real-world data from Japan to study the association between celecoxib and peripheral edema and to study the association between selective serotonin reuptake inhibitor (SSRI) and hip fracture in Australia. Results When the simulated rate ratio of the outcome was 4.0 in a case-crossover study with no time-varying confounder, the proposed weighting method and the Mantel-Haenszel odds ratio reproduced the true rate ratio. When a time-varying confounder existed, the Mantel-Haenszel method was biased but the weighting method was not. When more than one control period was used, standard conditional logistic regression was biased either with or without time-varying confounding and the bias increased (up to 8.7) when the study period was extended. In real-world analysis with a binary exposure variable in Japan and Australia, the point estimate of the odds ratio (around 2.5 for the association between celecoxib and peripheral edema and around 1.6 between SSRI and hip fracture) by our weighting method was equal to the Mantel-Haenszel odds ratio and stable compared with standard conditional logistic regression. Conclusion Case-crossover studies may be biased from within-subject exposure dependency, even without exposure time trends. This bias can be identified by comparing the odds ratio by the Mantel-Haenszel method and that by standard conditional logistic regression. We recommend using our proposed method which removes bias from within-subject exposure dependency and can account for time-varying confounders.

Abstract TP77: Race/Ethnic Disparity Trends in tPA Administration in the United States

Stroke ◽  
2016 ◽  
Vol 47 (suppl_1) ◽  
Author(s):  
Michelle P Lin ◽  
Steven Cen ◽  
Amytis Towfighi ◽  
May Kim-Tenser ◽  
William Mack ◽  
...  
Keyword(s):  
United States ◽  
Ischemic Stroke ◽  
Acute Ischemic Stroke ◽  
Rate Ratio ◽  
The United States ◽  
Racial Groups ◽  
Primary Analysis ◽  
Targeted Interventions ◽  
Yearly Change ◽  
Administration Rate

Introduction: Prior studies have shown racial disparities in tPA use for acute ischemic stroke. With the implementation of nationwide quality improvement measures, we sought to describe the temporal change in racial disparity in tPA administration. Hypothesis: Disparity in tPA administration improved across all racial groups in the past decade Methods: Data were obtained from all US states that contributed to the Nationwide Inpatient Sample. All patients (N=5,932,175) admitted to hospitals between 2000 and 2010 with a discharge diagnosis of ischemic stroke (ICD9 codes) were included. Primary analysis was the proportion of patients who received tPA administration stratified by race (white, black, Hispanic, Asian) temporally. Survey-weighted Poisson regression was used to estimate the rate ratio and compare the trend for yearly change between race categories. Results: Of the patients with ischemic stroke, 55.4% were white, black 11.89%, Hispanic 5.32%, Asian 1.89%, others 1.77%, missing race 23.31%. tPA administration rate increased from 2000 to 2010 regardless of race. In 2000, tPA administration rate was 0.96%, 0.40%, 0.73%, 0.59% in white, black, Hispanic, Asian, respectively. In 2010, tPA administration rate was 4.0%, 2.14%, 2.09%, 2.13% respectively. The relative change was the greatest in black with rate ratio of 6.7 (5.95-7.54), compared to other racial groups, Asian 5.36 (4.23-6.78), Hispanic 3.93 (3.42-4.51), and white 3.88 (3.74-4.03). Conclusions: Over the last decade, the rate of tPA administration for acute ischemic stroke in the United States have increased for every racial group. There is a lasting but improved disparity in tPA administration in non-white race. Targeted interventions designed to increase treatment and close disparity gap focusing on culturally tailored education and communications to address barriers need to be further explored.

Analysis of the effect of crossover from placebo (PBO) to darolutamide (DARO) on overall survival (OS) benefit in the ARAMIS Trial.

2021 ◽  
Vol 39 (6_suppl) ◽  
pp. 240-240
Author(s):  
Neal D. Shore ◽  
Karim Fizazi ◽  
Teuvo Tammela ◽  
Murilo Luz ◽  
Manuel Philco Salas ◽  
...  
Keyword(s):  
Treatment Effect ◽  
Safety Profile ◽  
Final Analysis ◽  
Phase Iii ◽  
Sensitivity Analyses ◽  
Primary Analysis ◽  
Open Label ◽  
Double Blind ◽  
Secondary Endpoints ◽  
The Impact

240 Background: DARO is a structurally distinct androgen receptor inhibitor approved for the treatment of non-metastatic castration-resistant prostate cancer (nmCRPC) based on significantly prolonged metastasis-free survival compared with PBO (median 40.4 vs 18.4 months; hazard ratio [HR] 0.41; 95% confidence interval [CI] 0.34–0.50; P < 0.0001) and a favorable safety profile in the phase III ARAMIS trial. Following unblinding at the primary analysis, crossover from PBO to DARO was permitted for the subsequent open-label treatment phase. Sensitivity analyses were performed to assess the effect of PBO–DARO crossover on OS benefit. Methods: Patients (pts) with nmCRPC receiving androgen deprivation therapy were randomized 2:1 to DARO (n = 955) or PBO (n = 554). In addition to OS, secondary endpoints included times to pain progression, first cytotoxic chemotherapy, first symptomatic skeletal event, and safety. The OS analysis was planned to occur after approximately 240 deaths, and secondary endpoints were evaluated in a hierarchical order. Iterative parameter estimation (IPE) and rank-preserving structural failure time (RPSFT) analyses were performed as pre-planned sensitivity analyses to adjust for the treatment effect of PBO–DARO crossover. The IPE method used a parametric model for the survival times and iteratively determined the model parameter describing the magnitude of the treatment effect, whereas a grid search and non-parametric log-rank test were used for the RPSFT analysis. The IPE and RPSFT analyses both generated a Kaplan–Meier curve for the PBO arm that predicts what would have been observed in the absence of PBO–DARO crossover. Results: After unblinding, 170 pts (30.7% of those randomized to PBO) crossed over from PBO to DARO; median treatment duration from unblinding to the final data cut-off was 11 months. Final analysis of the combined double-blind and open label periods was conducted after 254 deaths (15.5% of DARO and 19.1% of PBO pts) and showed a statistically significant OS benefit for DARO vs PBO (HR 0.69; 95% CI 0.53–0.88; P = 0.003). Results from the IPE (HR 0.66; 95% CI 0.51–0.84; P < 0.001) and RPSFT (HR 0.68; 95% CI 0.51–0.90; P = 0.007) analyses were similar to those from the intention-to-treat population, showing that the impact of PBO–DARO crossover was small. Additional analyses accounting for the effect of PBO–DARO crossover will be presented. The safety profile of DARO continued to be favorable at the final analysis, and discontinuation rates at the end of the double-blind period remained unchanged from the primary analysis (8.9% with DARO and 8.7% with PBO). Conclusions: Early treatment with DARO in men with nmCRPC is associated with significant improvement in OS regardless of pts crossing over from PBO to DARO. The safety profile of DARO remained favorable at the final analysis. Clinical trial information: NCT02200614.

Propensity score specification for optimal estimation of average treatment effect with binary response

2020 ◽  
Vol 29 (12) ◽  
pp. 3623-3640
Author(s):  
John A Craycroft ◽  
Jiapeng Huang ◽  
Maiying Kong
Keyword(s):  
Propensity Score ◽  
Treatment Effect ◽  
Propensity Scores ◽  
Optimal Estimation ◽  
Theoretical Explanation ◽  
Average Treatment Effect ◽  
Adaptive Lasso ◽  
Confounding Bias ◽  
Average Treatment ◽  
The Impact

Propensity score methods are commonly used in statistical analyses of observational data to reduce the impact of confounding bias in estimations of average treatment effect. While the propensity score is defined as the conditional probability of a subject being in the treatment group given that subject’s covariates, the most precise estimation of average treatment effect results from specifying the propensity score as a function of true confounders and predictors only. This property has been demonstrated via simulation in multiple prior research articles. However, we have seen no theoretical explanation as to why this should be so. This paper provides that theoretical proof. Furthermore, this paper presents a method for performing the necessary variable selection by means of elastic net regression, and then estimating the propensity scores so as to obtain optimal estimates of average treatment effect. The proposed method is compared against two other recently introduced methods, outcome-adaptive lasso and covariate balancing propensity score. Extensive simulation analyses are employed to determine the circumstances under which each method appears most effective. We applied the proposed methods to examine the effect of pre-cardiac surgery coagulation indicator on mortality based on a linked dataset from a retrospective review of 1390 patient medical records at Jewish Hospital (Louisville, KY) with the Society of Thoracic Surgeons database.

Impact of mobile money access on internal remittances, consumption expenditure and household welfare in Ghana

2020 ◽  
Vol ahead-of-print (ahead-of-print) ◽  
Author(s):  
Priscilla Twumasi Baffour ◽  
Wassiuw Abdul Rahaman ◽  
Ibrahim Mohammed
Keyword(s):  
Propensity Score ◽  
Propensity Score Matching ◽  
Treatment Effect ◽  
Average Treatment Effect ◽  
Consumption Expenditure ◽  
Content Type ◽  
Mobile Money ◽  
Money Transfer ◽  
Per Capita ◽  
Transfer Services

PurposeThe purpose of this study is to examine the impact of mobile money access on internal remittances received, per capita consumption expenditure and welfare of household in Ghana.Design/methodology/approachThe study used data from the latest round of the Ghana Living Standards Survey (GLSS 7) and employed the propensity score matching technique to estimate average treatment effect between users and non-users of mobile money transfer services.FindingsThe study finds that using mobile money is welfare enhancing, particularly for poor households and the channel by which it impacts on welfare is through higher internal remittances received and per capita expenditure. The results from the average treatment effect indicate that mobile money users receive significantly higher remittances and consequently spend averagely higher on consumption than non-users.Research limitations/implicationsAlthough the data employed in this study is limited to one country, the findings support the financial inclusion role and developmental impact of mobile money transfer services. Hence, mobile money transfer services should be promoted and facilitated by the telecommunication and financial sector regulators.Originality/valueIn addition to making original contribution to the literature on the welfare impact of mobile money, the study's use of the propensity score matching is unique.

scholarly journals Estimating Treatment Effects of Unit-Based Pricing of Household Solid Waste Disposal

2018 ◽  
Vol 48 (1) ◽  
pp. 21-43
Author(s):  
Christopher Wright ◽  
John M. Halstead ◽  
Ju-Chin Huang
Keyword(s):  
Propensity Score ◽  
Municipal Solid Waste ◽  
Solid Waste ◽  
Propensity Score Matching ◽  
Waste Disposal ◽  
Treatment Effect ◽  
Treatment Effects ◽  
Average Treatment Effect ◽  
Solid Waste Disposal ◽  
Annual Reduction

Propensity score matching is used to estimate treatment effects when data are observational. Results presented in this study demonstrate the use of propensity score matching to evaluate the average treatment effect of unit-based pricing of household trash for reducing municipal solid waste disposal. Average treatment effect of the treated for 34 New Hampshire communities range from an annual reduction of 631 pounds per household to 823 pounds per household. This represents an annual reduction of 42 percent to 54 percent from an average of 1530 pounds per household if a town did not adopt municipal solid waste user fees.

Sign in / Sign up

Export Citation Format

Share Document