Statistical inference for association studies in the presence of binary outcome misclassification

In biomedical and public health association studies, binary outcome variables may be subject to misclassification, resulting in substantial bias in effect estimates. The feasibility of addressing binary outcome misclassification in regression models is often hindered by model identifiability issues. In this paper, we characterize the identifiability problems in this class of models as a specific case of"label switching"and leverage a pattern in the resulting parameter estimates to solve the permutation invariance of the complete data log-likelihood. Our proposed algorithm in binary outcome misclassification models does not require gold standard labels and relies only on the assumption that outcomes are correctly classified at least 50% of the time. A label switching correction is applied within estimation methods to recover unbiased effect estimates and to estimate misclassification rates. Open source software is provided to implement the proposed methods. We give a detailed simulation study for our proposed methodology and apply these methods to data from the 2020 Medical Expenditure Panel Survey (MEPS).

[1]  L. Waller,et al.  Enhanced Inference for Finite Population Sampling-Based Prevalence Estimation with Misclassification Errors , 2023, The American Statistician.

[2]  J. Zelner,et al.  Identified vaccine efficacy for binary post-infection outcomes under misclassification without monotonicity , 2022, 2211.16502.

[3]  Tianxi Cai,et al.  On the global identifiability of logistic regression models with misclassified outcomes , 2021, 2103.12846.

[4]  G. Ziegler.  Binary Classification Tests, Imperfect Standards, and Ambiguous Information , 2020, 2012.11215.

[5]  Grace Y Yi,et al.  Genetic association studies with bivariate mixed responses subject to measurement error and misclassification , 2020, Statistics in medicine.

[6]  Francesca Molinari,et al.  Estimating the COVID-19 infection rate: Anatomy of an inference problem , 2020, Journal of Econometrics.

[7]  Alexandra Chouldechova,et al.  Fairness Evaluation in Presence of Biased Noisy Labels , 2020, AISTATS.

[8]  Bhramar Mukherjee,et al.  Statistical inference for association studies using electronic health records: handling both selection bias and outcome misclassification , 2019, Biometrics.

[9]  L. Lix,et al.  Comparing external and internal validation methods in correcting outcome misclassification bias in logistic regression: A simulation study and application to the case of postsurgical venous thromboembolism following total hip and knee arthroplasty , 2018, Pharmacoepidemiology and drug safety.

[10]  Sander Greenland,et al.  Separation in Logistic Regression: Causes, Consequences, and Control. , 2018, American journal of epidemiology.

[11]  Paul Gustafson,et al.  Bayesian inference for unidirectional misclassification of a binary response trait , 2018, Statistics in medicine.

[12]  Kristian Lum,et al.  An algorithm for removing sensitive information: Application to race-independent recidivism prediction , 2017, The Annals of Applied Statistics.

[13]  Alaa M Althubaiti,et al.  Information bias in health research: definition, pitfalls, and adjustment methods , 2016, Journal of multidisciplinary healthcare.

[14]  M. Budoff,et al.  Prevalence and Correlates of Myocardial Scar in a US Cohort. , 2015, JAMA.

[15]  Donal O’neill.  Measuring Obesity in the Absence of a Gold Standard , 2015, Economics and human biology.

[16]  W. Yao.  Label switching and its solutions for frequentist mixture models , 2015 .

[17]  Peter Szolovits,et al.  Improving the power of genetic association tests with imperfect phenotype derived from electronic medical records , 2014, Human Genetics.

[18]  Linda Valeri,et al.  The estimation of direct and indirect causal effects in the presence of misclassified binary mediator. , 2014, Biostatistics.

[19]  M. Verleysen,et al.  Classification in the Presence of Label Noise: A Survey , 2014, IEEE Transactions on Neural Networks and Learning Systems.

[20]  Stephen G. Walker,et al.  Label Switching in Bayesian Mixture Models: Deterministic Relabeling Strategies , 2014 .

[21]  R. Rekaya,et al.  Genome wide association studies in presence of misclassified binary responses , 2013, BMC Genetics.

[22]  Sarah Desmarais Jay Singh,et al.  Risk Assessment Instruments Validated and Implemented in Correctional Settings in the United States , 2013 .

[23]  Robert H Lyles,et al.  Validation Data-based Adjustments for Outcome Misclassification in Logistic Regression: An Illustration , 2011, Epidemiology.

[24]  Robert H Lyles,et al.  Sensitivity analysis for misclassification in logistic regression via likelihood methods and predictive value weighting , 2010, Statistics in medicine.

[25]  J. Mckinlay,et al.  Disparities in physicians' interpretations of heart disease symptoms by patient gender: results of a video vignette factorial experiment. , 2009, Journal of women's health.

[26]  S. Anderson,et al.  Zero inflation in ordinal data: Incorporating susceptibility to response through the use of a mixture model , 2008, Statistics in medicine.

[27]  Raymond J. Carroll,et al.  Measurement error in nonlinear models: a modern perspective , 2006 .

[28]  J. Neuhaus,et al.  Binomial Regression with Misclassification , 2003, Biometrics.

[29]  Dean M. Young,et al.  Parameter Subset Selection and Multiple Comparisons of Poisson Rate Parameters with Misclassification , 2002, Comput. Stat. Data Anal..

[30]  D Gianola,et al.  Threshold Model for Misclassified Binary Responses with Applications to Animal Breeding , 2001, Biometrics.

[31]  J. Neuhaus.  Bias and efficiency loss due to misclassified responses in binary regression , 1999 .

[32]  Paul S. Albert,et al.  Modeling Repeated Measures with Monotonic Ordinal Responses and Misclassification, with Applications to Studying Maturation , 1997 .

[33]  L. Magder,et al.  Logistic regression when the outcome is measured with uncertainty. , 1997, American journal of epidemiology.

[34]  J. Mckinlay,et al.  Non-medical influences on medical decision-making. , 1996, Social science & medicine.

[35]  S. Faraone,et al.  Measuring diagnostic accuracy in the absence of a "gold standard". , 1994, The American journal of psychiatry.

[36]  Ronald G. Ehrenberg,et al.  What Price Diversity , 1993 .

[37]  Alan Agresti,et al.  Categorical Data Analysis , 2003 .

[38]  Diane Lambert,et al.  Identifiability of finite mixtures of logistic regression models , 1991 .

[39]  R. Redner,et al.  Mixture densities, maximum likelihood, and the EM algorithm , 1984 .

[40]  S. Greenland,et al.  Correcting for misclassification in two-way tables and matched-pair studies. , 1983, International journal of epidemiology.

[41]  D. Rubin,et al.  Maximum likelihood from incomplete data via the EM - algorithm plus discussions on the paper , 1977 .

[42]  R Core Team,et al.  R: A language and environment for statistical computing. , 2014 .

[43]  Jean-François Dupuy,et al.  Maximum likelihood estimation in the logistic regression model with a cure fraction , 2011 .

[44]  J. Mckinlay,et al.  Patient characteristics and inequalities in doctors' diagnostic and management strategies relating to CHD: a video-simulation experiment. , 2006, Social science & medicine.

[45]  M. Stephens.  Dealing with label switching in mixture models , 2000 .

[46]  S. Aggrey,et al.  The Application of Clinical Genetics Dovepress Analysis of Binary Responses with Outcome- Specific Misclassification Probability in Genome-wide Association Studies Romdhane Rekaya , 2022 .