Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Making causal inferences about treatment effect sizes from observational datasets

Making causal inferences about treatment effect sizes from observational datasets In the era of big data and cloud computing, analysts need statistical models to go beyond predicting outcomes to forecasting how outcomes change when decision-makers intervene to change one or more causal factors. This paper reviews methods to estimate the causal effects of treatment choices on patient health outcomes using observational datasets. Methods are limited to those that model choice of treatment (propensity scoring) and treatment outcomes (instrumental variable, difference in differences, control function). A regression framework was developed to show how unobserved confounding covariates and heterogeneous outcomes can introduce biases to effect size estimates. In response to criticisms that outcome approaches are not systematic and subject to model misspecification error, we extend the control function approach of Lu and White by applying Best Approximating Model technology (BAM-CF). Results from simulation experiments are presented to compare biases between BAM-CF and propensity scoring in the presence of an unobserved confounder. We conclude no one strategy is ‘optimal’ for all datasets, and analyst should consider multiple approaches to assess robustness. For both observational and randomized datasets, researchers should assess how moderating covariates impact estimates of treatment effect sizes so that clinicians can understand what is best for each individual patient. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Biostatistics & Epidemiology Taylor & Francis

Making causal inferences about treatment effect sizes from observational datasets

Download PDF

Making causal inferences about treatment effect sizes from observational datasets

Abstract

In the era of big data and cloud computing, analysts need statistical models to go beyond predicting outcomes to forecasting how outcomes change when decision-makers intervene to change one or more causal factors. This paper reviews methods to estimate the causal effects of treatment choices on patient health outcomes using observational datasets. Methods are limited to those that model choice of treatment (propensity scoring) and treatment outcomes (instrumental variable, difference in...
Loading next page...
 
/lp/taylor-francis/making-causal-inferences-about-treatment-effect-sizes-from-EcP0Eb0i2y
Publisher
Taylor & Francis
Copyright
This work was authored as part of the Contributor's official duties as an Employee of the United States Government and is therefore a work of the United States Government. In accordance with 17 U.S.C. 105, no copyright protection is available for such works under U.S. Law.
ISSN
2470-9379
eISSN
2470-9360
DOI
10.1080/24709360.2019.1681211
Publisher site
See Article on Publisher Site

Abstract

In the era of big data and cloud computing, analysts need statistical models to go beyond predicting outcomes to forecasting how outcomes change when decision-makers intervene to change one or more causal factors. This paper reviews methods to estimate the causal effects of treatment choices on patient health outcomes using observational datasets. Methods are limited to those that model choice of treatment (propensity scoring) and treatment outcomes (instrumental variable, difference in differences, control function). A regression framework was developed to show how unobserved confounding covariates and heterogeneous outcomes can introduce biases to effect size estimates. In response to criticisms that outcome approaches are not systematic and subject to model misspecification error, we extend the control function approach of Lu and White by applying Best Approximating Model technology (BAM-CF). Results from simulation experiments are presented to compare biases between BAM-CF and propensity scoring in the presence of an unobserved confounder. We conclude no one strategy is ‘optimal’ for all datasets, and analyst should consider multiple approaches to assess robustness. For both observational and randomized datasets, researchers should assess how moderating covariates impact estimates of treatment effect sizes so that clinicians can understand what is best for each individual patient.

Journal

Biostatistics & Epidemiology – Taylor & Francis

Published: Jan 1, 2020

Keywords: Causal modeling; control functions; propensity scoring; endogeneity; moderator analyses

References

  • Medical graduates, truthful and useful analytics with big data and the art of persuasion
  • Prediction models for cardiovascular disease risk in the general population: systematic review
  • Risk prediction tools for cancer in primary care
  • Severity and variability of depression symptoms predicting suicide attempt in high-risk individuals
  • Genetic risk, number of previous depressive episodes, and stressful life events in predicting onset of major depression
  • Early blood product and crystalloid resuscitation: risk association with multiple organ dysfunction after severe blunt traumatic injury
  • Predictors of functional recovery one year following hospital discharge for hip fracture: a prospective study
  • Development of the revised opioid risk tool to predict opioid use disorder in patients with chronic nonmalignant pain
  • Predicting adverse drug reactions in older adults; a systematic review of the risk prediction models
  • Risk prediction models for hospital readmission: a systematic review
  • Improved prediction of medical expenditures and health care utilization using an updated chronic disease score and claims data
  • Predicting success in medical school: a longitudinal study of common Australian student selection tools
  • Statistics and causal inference
  • Causal inference using potential outcomes: design, modeling, decisions
  • An overview of relations among causal modelling methods
  • Are we growing the right health services research workforce of the future? Thoughts from a national delivery system
  • Moving from discovery to system-wide change: the role of research in a learning healthcare system – experience from three decades of health services research in the veterans health administration
  • Robustness checks and robustness tests in applied economics
  • A unification of mediation and interaction: a four-way decomposition
  • Making causing inferences from observational datasets
  • Statistical modeling methods: challenges and strategies
  • A novel approach for developing and interpreting treatment moderator profiles in randomized clinical trials
  • Doubly robust estimation in missing data and causal inference models
  • Bias-corrected matching estimators for average treatment effects
  • The central role of the propensity score in observational studies for causal effects
  • An introduction to propensity score methods for reducing the effects of confounding in observational studies
  • Constructing inverse probability weights for marginal structural models
  • Bayesian and classical approaches to instrumental variable regression
  • Using matching, instrumental variables, and control functions to estimate economic choice models
  • Local instrumental variables and latent variable models for identifying and bounding treatment effects
  • The landscape of causal inference: perspective from citation network analysis
  • Invited commentary: understanding bias amplification
  • Directed acyclic graphs, sufficient causes, and the properties of conditioning on a common effect
  • Using directed acyclic graphs to guide analyses of neighbourhood health effects: an introduction
  • Observational studies – should we simply ignore them in assessing transfusion outcome?
  • Randomized controlled trials vs. observational studies: why not just life together
  • Randomized clinical trials and observational studies: guidelines for assessing respective strengths and limitations
  • Are randomized controlled trials the gold standard? from clinical intelligence to prescriptive analytics
  • Beyond randomized controlled trials: a critical comparison of trials with nonrandomized studies
  • Alternatives to the randomized controlled trial
  • Nonexperimental versus experimental estimates of earnings impacts
  • Using experiments to assess nonexperimental comparison-group methods for measuring program effects
  • Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods
  • Three conditions under which experiments and observational studies produce comparable estimates
  • Comparison of a randomized and two quasi-experimental designs in a single outcome evaluation
  • Comparison of evidence of treatment effects in randomized and nonrandomized studies
  • A comparison of observational studies and randomized, controlled trials
  • Randomized, controlled trials, observational studies, and the hierarchy of research designs
  • The design versus analysis of observational studies for causal effects: parallels with the design of randomized trials
  • Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease
  • Bias in observational studies of prevalent users: lessons for comparative effectiveness research from a meta-analysis of statins
  • Reducing bias in the assessment of treatment effectiveness
  • Does psychological safety impact the clinical learning environment for physician residents?
  • Smearing estimate: a nonparametric retransformation method
  • The logged dependent variable, heteroscedasticity, and the retransformation problem
  • Moderation analysis using a two-level regression model
  • On a class of bias-amplifying variables that endanger effect estimates
  • Assessing the generalizability of randomized trial results to target populations
  • Impact of attention-deficit hyperactivity disorder treatment on smoking cessation intervention in ADHD smokers
  • A systematic approach to subgroup analyses in a smoking cessation trial
  • Improving the analysis and modeling of substance use
  • Response to letter regarding “A systematic approach to subgroup analyses in a smoking cessation trial”
  • On estimating treatment effects under non-compliance in randomized clinical trials: are intent-to-treat or instrumental variables analyses perfect solutions?
  • Assessing the effect of an influenza vaccine in an encouragement design
  • Clustered encouragement designs with individual noncompliance: Bayesian inference with randomization, and application to advance directive forms
  • ITT analysis of randomized encouragement design studies with missing data
  • Multiple imputation methods for treatment noncompliance and missing data
  • Identifiability and estimation of causal effects in randomized trials with noncompliance and completely nonignorable missing data (with discussion)
  • Methods for clustered encouragement design studies with noncompliance and missing data
  • Identifiability and estimation of causal effects by principal stratification with outcomes truncated by death
  • Causal analysis in multi-arm trials with truncation by death
  • Identification and estimation of causal effects with outcomes truncated by death
  • Substantial gains in bias reduction from matching with a variable number of controls
  • Comparison of multivariate matching methods: structures, distances, and algorithms
  • Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies
  • Nonparametric estimation of average treatment effects under exogeneity: a review
  • Balancing covariates via propensity score weighting
  • Using propensity scores to help design observational studies: application to the tobacco litigation
  • A comparison of the ability of different propensity score models to balance measured variables between treated and untreated subjects: a Monte Carlo study
  • For objective causal inference, design trumps analysis
  • Reducing bias in observational studies using subclassification on the propensity score
  • Matching using estimated propensity scores: relating theory to practice
  • Using matched substitutes to adjust for nonignorable nonresponse through multiple imputations
  • Squeezing the balloon: propensity scores and unmeasured covariate balance
  • Importance of covariate selection in controlling for selection bias in observational studies
  • How bias reduction is affected by covariate choice, unreliability, and mode of data analysis: results from two types of within-study comparison
  • A case study about why it can be difficult to test whether propensity score analysis works in field experiments
  • Combining propensity score matching with additional adjustments for prognostic covariates
  • Evaluating treatment effectiveness in patient subgroups: comparison of propensity score methods with an automated matching approach
  • Retrospectives: who invented instrumental variable regression?
  • Two-stage residual inclusion estimation: addressing endogeneity in health econometric modeling
  • Empirical-likelihood-based difference-in-difference estimators
  • Study the effects of ACGME duty hours limits on resident satisfaction: results from VA Learners’ Perceptions Survey
  • The role of a second control group in an observational study
  • Choosing among alternative nonexperimental methods for estimating the impact of social programs: the case of manpower training
  • Comparing clinical predictors of deep venous thrombosis vs. pulmonary embolus after severe injury
  • Bayesian model averaging: a tutorial
  • Shotgun stochastic search for large p regression
  • Comparing stochastic optimization methods for variable selection in binary outcome prediction, with application to health policy
  • Bayesian variable selection using cost-adjusted BIC with application to cost-effective measurement of quality of health care
  • Akaike’s information criterion and recent developments in information complexity
  • Collinearity in generalized linear regression
  • New directions in information matrix testing: eigenspectrum tests
  • Generalized information matrix tests for detecting model misspecification
  • Making correct statistical inferences using a wrong probability model
  • Statistical tests for comparing possible misspecified and nonnested models
  • Maximum likelihood estimation of misspecified models
  • Model selection and accounting for model uncertainty in graphical models using Occam’s window
  • Discrepancy risk model selection test theory for comparing possibly misspecified or nonnested models
  • Model selection: an integral part of inference
  • Frequentist model averaging estimation: a review
  • Frequentist model average estimators
  • Consequences of model misspecification for maximum likelihood estimation with missing data
  • Comparison of logistic regression versus propensity score when the number of events is low and there are multiple confounders