Assessing Sensitivity to Unmeasured Confounding Using a Simulated Potential Confounder

Carnegie, Nicole Bohme; Harada, Masataka; Hill, Jennifer

doi:10.1080/19345747.2015.1078862

Cited by 73 publications

(115 citation statements)

References 70 publications

Supporting

Mentioning

113

Contrasting

Order By: Relevance

“…In the real case, we performed additional sensitivity analyses to assess the impact on ATT estimates of unobserved confounding. We used R software V. 3.2.4 to generate and balance data and to perform sensitivity analyses . All further analyses were performed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA), and the codes are available in Appendix S3.…”

Section: Methodsmentioning

confidence: 99%

Propensity score methods and regression adjustment for analysis of nonrandomized studies with health‐related quality of life outcomes

Cottone

Anota

Bonnetain

et al. 2019

Pharmacoepidemiology and Drug

View full text Add to dashboard Cite

Purpose The aim of this study was to investigate the potential added value of combining propensity score (PS) methods with multivariable linear regression (MLR) in estimating the average treatment effect on the treated (ATT) in nonrandomized studies with health‐related quality of life (HRQoL) outcomes. Methods We first used simulations to compare the performances of different PS‐based methods, either alone or in combination with further MLR adjustment, in estimating ATT. PS methods were, respectively, optimal pair (OPM) and full (OFM) PS matching, subclassification on the PS (SBC), and the inverse probability of treatment weighting (IPTW). We simulated several scenarios, according to different sample sizes, proportions of treated vs untreated subjects, and types of HRQoL outcomes. We also applied the same methods to a real clinical data set. Results OPM and IPTW provided the closest Type I error to the nominal threshold α = 0.05 across all scenarios. Overall, both methods showed also lower variability in estimates than SBC and OFM. SBC performed worst, generally providing the highest levels of bias. Further MLR adjustment lessened bias for all methods, however providing higher Type I error for SBC and OFM. In the real case, all methods provided similar ATT estimates except for one outcome. Conclusions Our findings suggest that for sample sizes up to n = 200, OPM and IPTW are to be preferred to OFM and SBC in estimating ATT on HRQoL outcomes. Specifically, OPM performed best in sample sizes of n ≥ 80, IPTW for smaller sample sizes. Additional MLR adjustment can further improve ATT estimates.

show abstract

Section: Methodsmentioning

confidence: 99%

Propensity score methods and regression adjustment for analysis of nonrandomized studies with health‐related quality of life outcomes

Cottone

Anota

Bonnetain

et al. 2019

Pharmacoepidemiology and Drug

View full text Add to dashboard Cite

show abstract

“…None of the methods perform particularly well when U is omitted. However, semi‐parametric treatSens does perform better than the linear methods of and in the nonlinear/nonlinear simulation scenario. This is due to BART's ability to flexibly model arbitrary nonlinear response surfaces.…”

Section: Simulation Studymentioning

confidence: 98%

“…We use the parametric two‐sensitivity‐parameter model presented in as a foundation. Specifically, the original model for binary treatment variables underlying linear treatSens is as follows:

Y ∣ X, U, Z \sim normalN ((), X β^{y} + ζ^{y} U + τZ, σ_{normaly}^{2}),

\begin{matrix} Z ∣ X, U & \sim Bernoulli (Φ (X β^{z} + ζ^{z} U)), \\ U & \sim Bernoulli (π^{u}), \end{matrix}

where Φ denotes the standard normal cumulative distribution function or probit link.…”

Section: Sensitivity Analysis Framework and Assumptionsmentioning

confidence: 99%

A flexible, interpretable framework for assessing sensitivity to unmeasured confounding

et al. 2016

Self Cite

View full text Add to dashboard Cite

When estimating causal effects, unmeasured confounding and model misspecification are both potential sources of bias. We propose a method to simultaneously address both issues in the form of a semi‐parametric sensitivity analysis. In particular, our approach incorporates Bayesian Additive Regression Trees into a two‐parameter sensitivity analysis strategy that assesses sensitivity of posterior distributions of treatment effects to choices of sensitivity parameters. This results in an easily interpretable framework for testing for the impact of an unmeasured confounder that also limits the number of modeling assumptions. We evaluate our approach in a large‐scale simulation setting and with high blood pressure data taken from the Third National Health and Nutrition Examination Survey. The model is implemented as open‐source software, integrated into the treatSens package for the R statistical programming language. © 2016 The Authors. Statistics in Medicine Published by John Wiley & Sons Ltd.

show abstract

“…Marginal structural models can be used to address confounding by time‐dependent variables and has recently been applied to EHR in Sperrin et al Techniques for reducing and eliminating confounding often assume that the potential confounders are measured. When key confounders are not measured, sensitivity analyses and related statistical methods can be used to explore the impact of and to correct for potential unmeasured confounding …”

Section: Statistical Issues Related To Biobank Researchmentioning

confidence: 99%

The emerging landscape of health research based on biobanks linked to electronic health records: Existing resources, statistical challenges, and potential opportunities

Beesley

Salvatore

Fritsche

et al. 2019

Statistics in Medicine

View full text Add to dashboard Cite

Biobanks linked to electronic health records provide rich resources for health‐related research. With improvements in administrative and informatics infrastructure, the availability and utility of data from biobanks have dramatically increased. In this paper, we first aim to characterize the current landscape of available biobanks and to describe specific biobanks, including their place of origin, size, and data types. The development and accessibility of large‐scale biorepositories provide the opportunity to accelerate agnostic searches, expedite discoveries, and conduct hypothesis‐generating studies of disease‐treatment, disease‐exposure, and disease‐gene associations. Rather than designing and implementing a single study focused on a few targeted hypotheses, researchers can potentially use biobanks' existing resources to answer an expanded selection of exploratory questions as quickly as they can analyze them. However, there are many obvious and subtle challenges with the design and analysis of biobank‐based studies. Our second aim is to discuss statistical issues related to biobank research such as study design, sampling strategy, phenotype identification, and missing data. We focus our discussion on biobanks that are linked to electronic health records. Some of the analytic issues are illustrated using data from the Michigan Genomics Initiative and UK Biobank, two biobanks with two different recruitment mechanisms. We summarize the current body of literature for addressing these challenges and discuss some standing open problems. This work complements and extends recent reviews about biobank‐based research and serves as a resource catalog with analytical and practical guidance for statisticians, epidemiologists, and other medical researchers pursuing research using biobanks.

show abstract

Assessing Sensitivity to Unmeasured Confounding Using a Simulated Potential Confounder

Cited by 73 publications

References 70 publications

Propensity score methods and regression adjustment for analysis of nonrandomized studies with health‐related quality of life outcomes

Propensity score methods and regression adjustment for analysis of nonrandomized studies with health‐related quality of life outcomes

A flexible, interpretable framework for assessing sensitivity to unmeasured confounding

The emerging landscape of health research based on biobanks linked to electronic health records: Existing resources, statistical challenges, and potential opportunities

Contact Info

Product

Resources

About