scholarly journals Selection Bias When Using Instrumental Variable Methods to Compare Two Treatments But More Than Two Treatments Are Available

2016 ◽  
Vol 12 (1) ◽  
pp. 219-232 ◽  
Author(s):  
Ashkan Ertefaie ◽  
Dylan Small ◽  
James Flory ◽  
Sean Hennessy
Keyword(s):  
Selection Bias ◽  
Treatment Effect ◽  
Instrumental Variable ◽  
Treatment Effects ◽  
Health Improvement ◽  
Effect Estimate ◽  
Diabetic Patients ◽  
Data Set ◽  
The Health Improvement Network ◽  
Theoretical Evidence

Abstract Instrumental variable (IV) methods are widely used to adjust for the bias in estimating treatment effects caused by unmeasured confounders in observational studies. It is common that a comparison between two treatments is focused on and that only subjects receiving one of these two treatments are considered in the analysis even though more than two treatments are available. In this paper, we provide empirical and theoretical evidence that the IV methods may result in biased treatment effects if applied on a data set in which subjects are preselected based on their received treatments. We frame this as a selection bias problem and propose a procedure that identifies the treatment effect of interest as a function of a vector of sensitivity parameters. We also list assumptions under which analyzing the preselected data does not lead to a biased treatment effect estimate. The performance of the proposed method is examined using simulation studies. We applied our method on The Health Improvement Network (THIN) database to estimate the comparative effect of metformin and sulfonylureas on weight gain among diabetic patients.

Bounding Treatment Effects with Contaminated and Censored Data: Assessing the Impact of Early Childbearing on Children

2005 ◽  
Vol 5 (1) ◽  
Author(s):  
Charles H Mullin
Keyword(s):  
Censored Data ◽  
Treatment Effect ◽  
Instrumental Variable ◽  
Treatment Effects ◽  
Child Outcomes ◽  
Well Being ◽  
Teenage Mothers ◽  
Age At First Birth ◽  
First Birth ◽  
The Impact

AbstractEmpirical researchers commonly invoke instrumental variable (IV) assumptions to identify treatment effects. This paper considers what can be learned under two specific violations of those assumptions: contaminated and corrupted data. Either of these violations prevents point identification, but sharp bounds of the treatment effect remain feasible. In an applied example, random miscarriages are an IV for women’s age at first birth. However, the inability to separate random miscarriages from behaviorally induced miscarriages (those caused by smoking and drinking) results in a contaminated sample. Furthermore, censored child outcomes produce a corrupted sample. Despite these limitations, the bounds demonstrate that delaying the age at first birth for the current population of non-black teenage mothers reduces their first-born child’s well-being.

Estimating Treatment Effects in the Presence of Noncompliance and Nonresponse: The Generalized Endogenous Treatment Model

2011 ◽  
Vol 19 (2) ◽  
pp. 205-226 ◽  
Author(s):  
Kevin M. Esterling ◽  
Michael A. Neblo ◽  
David M. J. Lazer
Keyword(s):  
Treatment Effect ◽  
Instrumental Variable ◽  
Treatment Effects ◽  
Mean Squared Error ◽  
Principal Stratification ◽  
Squared Error ◽  
Treatment Effect Heterogeneity ◽  
Causal Treatment ◽  
Effect Heterogeneity ◽  
And Control

If ignored, noncompliance with a treatment or nonresponse on outcome measures can bias estimates of treatment effects in a randomized experiment. To identify and estimate causal treatment effects in the case where compliance and response depend on unobservables, we propose the parametric generalized endogenous treatment (GET) model. GET incorporates behavioral responses within an experiment to measure each subject's latent compliance type and identifies causal effects via principal stratification. Using simulation methods and an application to field experimental data, we show GET has a dramatically lower mean squared error for treatment effect estimates than existing approaches to principal stratification that impute, rather than measure, compliance type. In addition, we show that GET allows one to relax and test the instrumental variable exclusion restriction assumption, to test for the presence of treatment effect heterogeneity across a range of compliance types, and to test for treatment ignorability when treatment and control samples are balanced on observable covariates.

Impact of minimal sufficient balance, minimization, and stratified permuted blocks on bias and power in the estimation of treatment effect in sequential clinical trials with a binary endpoint

2021 ◽  
pp. 096228022110558
Author(s):  
Steven D Lauzon ◽  
Wenle Zhao ◽  
Paul J Nietert ◽  
Jody D Ciolino ◽  
Michael D Hill ◽  
...  
Keyword(s):  
Clinical Trials ◽  
Logistic Regression ◽  
Selection Bias ◽  
Treatment Effect ◽  
Treatment Effects ◽  
Simulation Studies ◽  
Binary Outcome ◽  
Treatment Allocation ◽  
Analysis Phase ◽  
The Impact

Minimization is among the most common methods for controlling baseline covariate imbalance at the randomization phase of clinical trials. Previous studies have found that minimization does not preserve allocation randomness as well as other methods, such as minimal sufficient balance, making it more vulnerable to allocation predictability and selection bias. Additionally, minimization has been shown in simulation studies to inadequately control serious covariate imbalances when modest biased coin probabilities (≤0.65) are used. This current study extends the investigation of randomization methods to the analysis phase, comparing the impact of treatment allocation methods on power and bias in estimating treatment effects on a binary outcome using logistic regression. Power and bias in the estimation of treatment effect was found to be comparable across complete randomization, minimization, and minimal sufficient balance in unadjusted analyses. Further, minimal sufficient balance was found to have the most modest impact on power and the least bias in covariate-adjusted analyses. The minimal sufficient balance method is recommended for use in clinical trials as an alternative to minimization when covariate-adaptive subject randomization takes place.

Clarifying selection bias in cluster randomized trials

2021 ◽  
pp. 174077452110568
Author(s):  
Fan Li ◽  
Zizhong Tian ◽  
Jennifer Bobb ◽  
Georgia Papadogeorgou ◽  
Fan Li
Keyword(s):  
Selection Bias ◽  
Treatment Effect ◽  
Treatment Effects ◽  
Randomized Trials ◽  
Additional Data ◽  
Average Treatment Effect ◽  
Causal Effects ◽  
Cluster Randomized Trials ◽  
Average Treatment ◽  
Cluster Randomized

Background In cluster randomized trials, patients are typically recruited after clusters are randomized, and the recruiters and patients may not be blinded to the assignment. This often leads to differential recruitment and consequently systematic differences in baseline characteristics of the recruited patients between intervention and control arms, inducing post-randomization selection bias. We aim to rigorously define causal estimands in the presence of selection bias. We elucidate the conditions under which standard covariate adjustment methods can validly estimate these estimands. We further discuss the additional data and assumptions necessary for estimating causal effects when such conditions are not met. Methods Adopting the principal stratification framework in causal inference, we clarify there are two average treatment effect (ATE) estimands in cluster randomized trials: one for the overall population and one for the recruited population. We derive analytical formula of the two estimands in terms of principal-stratum-specific causal effects. Furthermore, using simulation studies, we assess the empirical performance of the multivariable regression adjustment method under different data generating processes leading to selection bias. Results When treatment effects are heterogeneous across principal strata, the average treatment effect on the overall population generally differs from the average treatment effect on the recruited population. A naïve intention-to-treat analysis of the recruited sample leads to biased estimates of both average treatment effects. In the presence of post-randomization selection and without additional data on the non-recruited subjects, the average treatment effect on the recruited population is estimable only when the treatment effects are homogeneous between principal strata, and the average treatment effect on the overall population is generally not estimable. The extent to which covariate adjustment can remove selection bias depends on the degree of effect heterogeneity across principal strata. Conclusion There is a need and opportunity to improve the analysis of cluster randomized trials that are subject to post-randomization selection bias. For studies prone to selection bias, it is important to explicitly specify the target population that the causal estimands are defined on and adopt design and estimation strategies accordingly. To draw valid inferences about treatment effects, investigators should (1) assess the possibility of heterogeneous treatment effects, and (2) consider collecting data on covariates that are predictive of the recruitment process, and on the non-recruited population from external sources such as electronic health records.

Unifying instrumental variable and inverse probability weighting approaches for inference of causal treatment effect and unmeasured confounding in observational studies

2020 ◽  
pp. 096228022097183
Author(s):  
Tao Liu ◽  
Joseph W Hogan
Keyword(s):  
Treatment Effect ◽  
Instrumental Variable ◽  
Observational Studies ◽  
Treatment Effects ◽  
Inverse Probability Weighting ◽  
Probability Weighting ◽  
Unmeasured Confounding ◽  
Inverse Probability ◽  
Causal Treatment ◽  
Weighting Methods

Confounding is a major concern when using data from observational studies to infer the causal effect of a treatment. Instrumental variables, when available, have been used to construct bound estimates on population average treatment effects when outcomes are binary and unmeasured confounding exists. With continuous outcomes, meaningful bounds are more challenging to obtain because the domain of the outcome is unrestricted. In this paper, we propose to unify the instrumental variable and inverse probability weighting methods, together with suitable assumptions in the context of an observational study, to construct meaningful bounds on causal treatment effects. The contextual assumptions are imposed in terms of the potential outcomes that are partially identified by data. The inverse probability weighting component incorporates a sensitivity parameter to encode the effect of unmeasured confounding. The instrumental variable and inverse probability weighting methods are unified using the principal stratification. By solving the resulting system of estimating equations, we are able to quantify both the causal treatment effect and the sensitivity parameter (i.e. the degree of the unmeasured confounding). We demonstrate our method by analyzing data from the HIV Epidemiology Research Study.

Increasing Precision without Altering Treatment Effects: Repeated Measures Designs in Survey Experiments

2021 ◽  
pp. 1-18
Author(s):  
SCOTT CLIFFORD ◽  
GEOFFREY SHEAGLEY ◽  
SPENCER PISTON
Keyword(s):  
Treatment Effect ◽  
Repeated Measures ◽  
Treatment Effects ◽  
Repeated Measurements ◽  
Experimental Designs ◽  
Treatment Effect Size ◽  
Standard Design ◽  
Survey Experiments ◽  
Repeated Measures Designs ◽  
Alternative Designs

The use of survey experiments has surged in political science. The most common design is the between-subjects design in which the outcome is only measured posttreatment. This design relies heavily on recruiting a large number of subjects to precisely estimate treatment effects. Alternative designs that involve repeated measurements of the dependent variable promise greater precision, but they are rarely used out of fears that these designs will yield different results than a standard design (e.g., due to consistency pressures). Across six studies, we assess this conventional wisdom by testing experimental designs against each other. Contrary to common fears, repeated measures designs tend to yield the same results as more common designs while substantially increasing precision. These designs also offer new insights into treatment effect size and heterogeneity. We conclude by encouraging researchers to adopt repeated measures designs and providing guidelines for when and how to use them.

scholarly journals Estimating and reporting treatment effects in clinical trials for weight management: using estimands to interpret effects of intercurrent events and missing data

2021 ◽  
Author(s):  
Sean Wharton ◽  
Arne Astrup ◽  
Lars Endahl ◽  
Michael E. J. Lean ◽  
Altynai Satylganova ◽  
...  
Keyword(s):  
Clinical Trials ◽  
Missing Data ◽  
Weight Management ◽  
Treatment Effect ◽  
Repeated Measures ◽  
Treatment Effects ◽  
Randomized Clinical Trials ◽  
Weight Losses ◽  
The Difference ◽  
Treatment Of Obesity

AbstractIn the approval process for new weight management therapies, regulators typically require estimates of effect size. Usually, as with other drug evaluations, the placebo-adjusted treatment effect (i.e., the difference between weight losses with pharmacotherapy and placebo, when given as an adjunct to lifestyle intervention) is provided from data in randomized clinical trials (RCTs). At first glance, this may seem appropriate and straightforward. However, weight loss is not a simple direct drug effect, but is also mediated by other factors such as changes in diet and physical activity. Interpreting observed differences between treatment arms in weight management RCTs can be challenging; intercurrent events that occur after treatment initiation may affect the interpretation of results at the end of treatment. Utilizing estimands helps to address these uncertainties and improve transparency in clinical trial reporting by better matching the treatment-effect estimates to the scientific and/or clinical questions of interest. Estimands aim to provide an indication of trial outcomes that might be expected in the same patients under different conditions. This article reviews how intercurrent events during weight management trials can influence placebo-adjusted treatment effects, depending on how they are accounted for and how missing data are handled. The most appropriate method for statistical analysis is also discussed, including assessment of the last observation carried forward approach, and more recent methods, such as multiple imputation and mixed models for repeated measures. The use of each of these approaches, and that of estimands, is discussed in the context of the SCALE phase 3a and 3b RCTs evaluating the effect of liraglutide 3.0 mg for the treatment of obesity.

scholarly journals Do statins reduce mortality in older people? Findings from a longitudinal study using primary care records

2021 ◽  
Vol 9 (2) ◽  
pp. e000780
Author(s):  
Lisanne Andra Gitsels ◽  
Ilyas Bakbergenuly ◽  
Nicholas Steel ◽  
Elena Kulinskaya
Keyword(s):  
Primary Care ◽  
Older People ◽  
Clinical Information ◽  
Cardiac Risk ◽  
Health Improvement ◽  
Network Database ◽  
Statin Prescription ◽  
The Health Improvement Network ◽  
Care Practices ◽  
Primary Care Practices

ObjectiveAssess whether statins reduce mortality in the general population aged 60 years and above.DesignRetrospective cohort study.SettingPrimary care practices contributing to The Health Improvement Network database, England and Wales, 1990–2017.ParticipantsCohort who turned age 60 between 1990 and 2000 with no previous cardiovascular disease or statin prescription and followed up until 2017.ResultsCurrent statin prescription was associated with a significant reduction in all-cause mortality from age 65 years onward, with greater reductions seen at older ages. The adjusted HRs of mortality associated with statin prescription at ages 65, 70, 75, 80 and 85 years were 0.76 (95% CI 0.71 to 0.81), 0.71 (95% CI 0.68 to 0.75), 0.68 (95% CI 0.65 to 0.72), 0.63 (95% CI 0.53 to 0.73) and 0.54 (95% CI 0.33 to 0.92), respectively. The adjusted HRs did not vary by sex or cardiac risk.ConclusionsUsing regularly updated clinical information on sequential treatment decisions in older people, mortality predictions were updated every 6 months until age 85 years in a combined primary and secondary prevention population. The consistent mortality reduction of statins from age 65 years onward supports their use where clinically indicated at age 75 and older, where there has been particular uncertainty of the benefits.

scholarly journals Association between Androgen Deprivation Therapy and Risk of Dementia in Men with Prostate Cancer

Cancers ◽  
2021 ◽  
Vol 13 (15) ◽  
pp. 3861
Author(s):  
Jui-Ming Liu ◽  
Chin-Yao Shen ◽  
Wallis C. Y. Lau ◽  
Shih-Chieh Shao ◽  
Kenneth K. C. Man ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Androgen Deprivation Therapy ◽  
Incidence Rates ◽  
Androgen Deprivation ◽  
Health Improvement ◽  
The United Kingdom ◽  
Incident Dementia ◽  
The Health Improvement Network ◽  
The Uk ◽  
Incidence Of Dementia

The risk of dementia after androgen deprivation therapy (ADT) in patients with advanced prostate cancer (PCa) remains controversial. This study aimed to evaluate the association between ADT and the incidence of dementia in patients with PCa. We identified patients newly diagnosed with PCa in the National Health Insurance Database of Taiwan from 1 January 2002 to 30 June 2016 and in The Health Improvement Network of the United Kingdom (UK) from 1 January 1998 to 31 March 2018. We classified patients with PCa into ADT and ADT-naïve groups. Propensity score (PS) methods were used to minimize the differences in characteristics between the groups. We performed a Cox proportional hazard model to obtain the adjusted hazard ratio (HR) to compare the incidence of dementia between the groups. Our ADT group comprised 8743 and 73,816 patients in Taiwan and the UK, respectively, which were matched 1:1 to ADT-naïve patients by PS. The incidence rates of dementia in the ADT group were 2.74 versus 3.03 per 1000 person-years in the ADT naïve groups in Taiwan, and 2.81 versus 2.79 per 1000 person-years in the UK. There was no statistical difference between ADT and ADT-naïve groups (adjusted HR: 1.12; 95% confidence interval (CI): 0.87–1.43 in Taiwan and adjusted HR: 1.02; 95% CI: 0.85–1.23 in the UK). We found no association between the incidence of dementia and ADT in patients with advanced PCa in either database. Further studies are warranted to evaluate other possible triggers of incident dementia in patients receiving ADT for advanced PCa.

Sign in / Sign up

Export Citation Format

Share Document