Raking and regression calibration: Methods to address bias from correlated covariate and time‐to‐event error

Oh, Eric; Shepherd, Bryan E.; Lumley, Thomas; Shaw, Pamela A.

doi:10.1002/sim.8793

Cited by 17 publications

(22 citation statements)

References 29 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Oh et al. (2019), however, proposed an approximation,

{\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false)

, as the auxiliary variable, motivated by settings involving correlated measurement error in covariates and a censored event‐time only.…”

Section: Construction Of Better Auxiliary Variablesmentioning

confidence: 99%

“…Thus, the linear working model underlying the estimator from Oh et al. (2019) is given by

{\overset{ℓ}{true}}_{0} false(X_{i}, Z_{i}, U_{i}, Δ_{i} false) = γ_{0} + γ_{1} {\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false) + ε_{i}

. To assess whether the linear fit is appropriate, we plot

{\overset{ℓ}{true}}_{0} false(X_{i}, Z_{i}, U_{i}, Δ_{i} false)

against

{\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false)

from simulated data for various measurement error scenarios.…”

Section: Construction Of Better Auxiliary Variablesmentioning

confidence: 99%

“…Although less attention has been given to errors in an outcome of interest, there has been some recent work looking at errors in binary outcomes (Magder & Hughes, 1997; Edwards et al., 2013; Wang et al., 2016), discrete time‐to‐event outcomes (Hunsberger et al., 2010; Magaret, 2008; Meier et al., 2003), and to a lesser extent, continuous time‐to‐event outcomes (Gravel et al., 2018; Oh et al., 2018). There has been even less work to understand the impact of errors in both covariates and a time‐to‐event outcome, but it has recently been shown that ignoring such errors can cause severe bias in estimates of effects of interest (Boe et al., 2020; Giganti et al., 2020; Oh et al., 2019).…”

Section: Introductionmentioning

confidence: 99%

“…Instead, one can use a two‐phase design, which involves reviewing and correcting only a subset of the records, to obtain consistent estimates of effects of interest. There have been some methods proposed recently that employ this framework to incorporate the large error‐prone data with the smaller validated data to improve statistical inference, including regression calibration (Boe et al., 2020; Oh et al., 2019), multiple imputation (Giganti et al., 2020), and generalized raking (Oh et al., 2019). Generalized raking in particular has been shown to be robust to the structure of the measurement error, which can be quite complex for EHR data (Han et al., 2019; Oh et al., 2019).…”

Section: Introductionmentioning

confidence: 99%

“…There have been some methods proposed recently that employ this framework to incorporate the large error‐prone data with the smaller validated data to improve statistical inference, including regression calibration (Boe et al., 2020; Oh et al., 2019), multiple imputation (Giganti et al., 2020), and generalized raking (Oh et al., 2019). Generalized raking in particular has been shown to be robust to the structure of the measurement error, which can be quite complex for EHR data (Han et al., 2019; Oh et al., 2019). Specifically, generalized raking estimators use the error‐prone data as auxiliary variables to improve the efficiency of the analysis of the validated data without having to model the error structure, making them appealing for EHR settings where the true structure is likely unknown.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Improved generalized raking estimators to address dependent covariate and failure‐time outcome error

Shepherd

Lumley

et al. 2021

Biometrical J

Self Cite

View full text Add to dashboard Cite

Biomedical studies that use electronic health records (EHR) data for inference are often subject to bias due to measurement error. The measurement error present in EHR data is typically complex, consisting of errors of unknown functional form in covariates and the outcome, which can be dependent. To address the bias resulting from such errors, generalized raking has recently been proposed as a robust method that yields consistent estimates without the need to model the error structure. We provide rationale for why these previously proposed raking estimators can be expected to be inefficient in failure‐time outcome settings involving misclassification of the event indicator. We propose raking estimators that utilize multiple imputation, to impute either the target variables or auxiliary variables, to improve the efficiency. We also consider outcome‐dependent sampling designs and investigate their impact on the efficiency of the raking estimators, either with or without multiple imputation. We present an extensive numerical study to examine the performance of the proposed estimators across various measurement error settings. We then apply the proposed methods to our motivating setting, in which we seek to analyze HIV outcomes in an observational cohort with EHR data from the Vanderbilt Comprehensive Care Clinic.

show abstract

“…Oh et al. (2019), however, proposed an approximation,

{\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false)

, as the auxiliary variable, motivated by settings involving correlated measurement error in covariates and a censored event‐time only.…”

Section: Construction Of Better Auxiliary Variablesmentioning

confidence: 99%

“…Thus, the linear working model underlying the estimator from Oh et al. (2019) is given by

{\overset{ℓ}{true}}_{0} false(X_{i}, Z_{i}, U_{i}, Δ_{i} false) = γ_{0} + γ_{1} {\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false) + ε_{i}

. To assess whether the linear fit is appropriate, we plot

{\overset{ℓ}{true}}_{0} false(X_{i}, Z_{i}, U_{i}, Δ_{i} false)

against

{\overset{ℓ}{true}}_{0} false(X_{i}^{★}, Z_{i}, U_{i}^{★}, Δ_{i}^{★} false)

from simulated data for various measurement error scenarios.…”

Section: Construction Of Better Auxiliary Variablesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Improved generalized raking estimators to address dependent covariate and failure‐time outcome error

Shepherd

Lumley

et al. 2021

Biometrical J

Self Cite

View full text Add to dashboard Cite

show abstract

Optimal multiwave validation of secondary use data with outcome and exposure misclassification

et al. 2023

Self Cite

View full text Add to dashboard Cite

Observational databases provide unprecedented opportunities for secondary use in biomedical research. However, these data can be error‐prone and must be validated before use. It is usually unrealistic to validate the whole database because of resource constraints. A cost‐effective alternative is a two‐phase design that validates a subset of records enriched for information about a particular research question. We consider odds ratio estimation under differential outcome and exposure misclassification and propose optimal designs that minimize the variance of the maximum likelihood estimator. Our adaptive grid search algorithm can locate the optimal design in a computationally feasible manner. Because the optimal design relies on unknown parameters, we introduce a multiwave strategy to approximate the optimal design. We demonstrate the proposed design's efficiency gains through simulations and two large observational studies.

show abstract

Nonparametric estimation of a survival function in the presence of measurement errors on the failure time of interest

Jin,

Liu,

Mao

et al. 2023

Can J Statistics

View full text Add to dashboard Cite

This article discusses nonparametric estimation of a survival function in the presence of measurement errors on the observation of the failure time of interest. One situation where such issues arise would be clinical studies of chronic diseases where the observation on the time to the failure event of interest such as the onset of the disease relies on patient recall or chart review of electronic medical records. It is easy to see that both situations can be subject to measurement errors. To resolve this problem, we propose a simulation extrapolation approach to correct the bias induced by the measurement error. To overcome potential computational difficulties, we use spline regression to approximate the unspecified extrapolated coefficient function of time, and establish the asymptotic properties of our proposed estimator. The proposed method is applied to nonparametric estimation based on interval‐censored data. Extensive numerical experiments involving both simulated and actual study datasets demonstrate the feasibility of this proposed estimation procedure.

show abstract

Raking and regression calibration: Methods to address bias from correlated covariate and time‐to‐event error

Cited by 17 publications

References 29 publications

Improved generalized raking estimators to address dependent covariate and failure‐time outcome error

Improved generalized raking estimators to address dependent covariate and failure‐time outcome error

Optimal multiwave validation of secondary use data with outcome and exposure misclassification

Nonparametric estimation of a survival function in the presence of measurement errors on the failure time of interest

Contact Info

Product

Resources

About