Causal inference with measurement error in outcomes: Bias analysis and estimation methods

Shu, Di; Yi, Grace Y.

doi:10.1177/0962280217743777

Cited by 28 publications

(49 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although various constraints may be considered, choosing a particular constraint should be guided by the nature of the data. A detailed discussion can be found in the works of White et al, Shu and Yi, and Yi and He …”

Section: Discussion and Extensionsmentioning

confidence: 99%

“…With other regression forms for the treatment model, a closed‐form expression for G (·) may not be available or G (·) does not even exist to make conditions and satisfied; a similar aspect was discussed by Stefanski for a different setting. In such circumstances, we propose an augmented simulation‐extrapolation (SIMEX) method that roots from a combination of the method developed by Shu and Yi and the SIMEX method proposed by Cook and Stefanski …”

Section: Approximate Methodsmentioning

confidence: 99%

“…In the presence of misclassification in Y alone , Shu and Yi developed consistent estimators

\overset{E}{true} false(Y_{1} false) = \frac{1}{n false(p_{11} - p_{10} false)} true \sum_{i = 1}^{n} \frac{T_{i} Y_{i}^{*}}{{\overset{e}{true}}_{i}} - \frac{p_{10}}{p_{11} - p_{10}}

and

\overset{E}{true} false(Y_{0} false) = \frac{1}{n false(p_{11} - p_{10} false)} true \sum_{i = 1}^{n} \frac{false(1 - T_{i} false) Y_{i}^{*}}{1 - {\overset{e}{true}}_{i}} - \frac{p_{10}}{p_{11} - p_{10}}

for E ( Y 1 ) and E ( Y 0 ), respectively, to correct for misclassification effects in the outcome variable, where, for i = 1,…, n ,

{\overset{e}{true}}_{i}

is an estimate of propensity score e i in that is expressed in terms of X i together with Z i , and the subscript i is added to X and Z to show the covariate subvectors for subject i .…”

Section: Consistent Estimationmentioning

confidence: 99%

“…Many correction methods for dealing with measurement error have been developed for regression settings, where describing the association relationship is the focus. There has been little work on causal inference with measurement error, although this area has recently received increasing attention with some methods available to handle error‐prone covariates, misclassified treatment, and mismeasured outcomes …”

Section: Introductionmentioning

confidence: 99%

“…There has been little work on causal inference with measurement error, although this area has recently received increasing attention with some methods available to handle error-prone covariates, [15][16][17] misclassified treatment, 18,19 and mismeasured outcomes. 20,21 Measurement error and misclassification arise ubiquitously in applications and present a considerable challenge to statistical inference. In the presence of measurement error in both covariates and outcomes, to our best knowledge, there has been no research on addressing measurement error effects on causal inference.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Weighted causal inference methods with mismeasured covariates and misclassified outcomes

Shu

2019

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

Inverse probability weighting (IPW) estimation has been widely used in causal inference. Its validity relies on the important condition that the variables are precisely measured. This condition, however, is often violated, which distorts the IPW method and thus yields biased results. In this paper, we study the IPW estimation of average treatment effects for settings with mismeasured covariates and misclassified outcomes. We develop estimation methods to correct for measurement error and misclassification effects simultaneously. Our discussion covers a broad scope of treatment models, including typically assumed logistic regression models and general treatment assignment mechanisms. Satisfactory performance of the proposed methods is demonstrated by extensive numerical studies.

show abstract

Section: Discussion and Extensionsmentioning

confidence: 99%

Section: Approximate Methodsmentioning

confidence: 99%

“…In the presence of misclassification in Y alone , Shu and Yi developed consistent estimators

\overset{E}{true} false(Y_{1} false) = \frac{1}{n false(p_{11} - p_{10} false)} true \sum_{i = 1}^{n} \frac{T_{i} Y_{i}^{*}}{{\overset{e}{true}}_{i}} - \frac{p_{10}}{p_{11} - p_{10}}

and

\overset{E}{true} false(Y_{0} false) = \frac{1}{n false(p_{11} - p_{10} false)} true \sum_{i = 1}^{n} \frac{false(1 - T_{i} false) Y_{i}^{*}}{1 - {\overset{e}{true}}_{i}} - \frac{p_{10}}{p_{11} - p_{10}}

for E ( Y 1 ) and E ( Y 0 ), respectively, to correct for misclassification effects in the outcome variable, where, for i = 1,…, n ,

{\overset{e}{true}}_{i}

is an estimate of propensity score e i in that is expressed in terms of X i together with Z i , and the subscript i is added to X and Z to show the covariate subvectors for subject i .…”

Section: Consistent Estimationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Weighted causal inference methods with mismeasured covariates and misclassified outcomes

Shu

2019

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

show abstract

Calcitonin Gene–Related Peptide Monoclonal Antibodies and Risk of SARS-CoV-2 Infection and Severe COVID-19 Outcomes Among Veterans With Migraine Disorder

Wang

Fenton

Deng³

et al. 2023

JAMA Netw Open

View full text Add to dashboard Cite

ImportanceCalcitonin gene–related peptide (CGRP), a neuropeptide involved in migraine pathophysiology, is also a key neuroimmune modulator. CGRP antagonists may help mitigate the hyperinflammatory response observed in patients with COVID-19; however, findings from the literature are contradictory, and to date, no study has investigated the safety and effectiveness of CGRP antagonists against COVID-19.ObjectiveTo evaluate the association between CGRP monoclonal antibody (mAb) treatment and risk of SARS-CoV-2 infection and sequela hospitalization, requiring supplemental oxygen, use of mechanical ventilation, or death.Design, Setting, and ParticipantsThis retrospective cohort study analyzed the electronic health records of US veterans aged 18 to 65 years who were diagnosed with migraine disorder and were at risk of COVID-19 between January 20, 2020, and May 19, 2022.ExposureInitiation of CGRP mAbs.Main Outcomes and MeasuresThe main outcome was cumulative incidence of SARS-CoV-2 infection. Odds of 30-day hospitalization, requiring supplemental oxygen, use of mechanical ventilation, or death were secondary outcomes.ResultsAmong 8 178 652 eligible person-trials (354 294 veterans), 9992 (mean [SD] age, 46.0 [9.5] years; 53.9% male) initiated CGRP mAbs and 8 168 660 (mean [SD] age, 46.6 [10.2] years; 65.7% male) did not initiate CGRP mAbs. Over a 28-month follow-up period, 1247 initiators (12.5%) and 780 575 noninitiators (9.6%) tested positive for SARS-CoV-2. After censoring persons who deviated from treatment, the incidence was 7.4 cases per 1000 person-months among initiators and 6.9 per 1000 person-months among noninitiators. The inverse probability–weighted observational analogs of intention-to-treat and per-protocol hazard ratios were 0.95 (95% CI, 0.89-1.01) and 0.93 (95% CI, 0.86-1.02), respectively. No significant differences in the likelihood of hospitalization (odds ratio [OR], 0.93; 95% CI, 0.62-1.41), requiring supplemental oxygen (OR, 0.77; 95% CI, 0.45-1.30), use of mechanical ventilation (OR, 0.85; 95% CI, 0.26-2.84), or death (OR, 0.67; 95% CI, 0.09-5.23) were observed between CGRP mAb initiators and noninitiators who tested positive for SARS-CoV-2.Conclusions and RelevanceIn this cohort study, CGRP mAb treatment was not associated with positive SARS-CoV-2 test results or risk of severe COVID-19 outcomes, suggesting that CGRP mAbs may be used for migraine prevention during the COVID-19 pandemic. Given the few events of requiring supplemental oxygen, use of mechanical ventilation, and death, replication analysis in a larger sample of patients later in the course of disease is warranted.

show abstract

Inverse‐probability‐of‐treatment weighted estimation of causal parameters in the presence of error‐contaminated and time‐dependent confounders

Shu

2019

Biometrical J

Self Cite

View full text Add to dashboard Cite

Inverse-probability-of-treatment weighted (IPTW) estimation has been widely used to consistently estimate the causal parameters in marginal structural models, with timedependent confounding effects adjusted for. Just like other causal inference methods, the validity of IPTW estimation typically requires the crucial condition that all variables are precisely measured. However, this condition, is often violated in practice due to various reasons. It has been well documented that ignoring measurement error often leads to biased inference results. In this paper, we consider the IPTW estimation of the causal parameters in marginal structural models in the presence of error-contaminated and time-dependent confounders. We explore several methods to correct for the effects of measurement error on the estimation of causal parameters. Numerical studies are reported to assess the finite sample performance of the proposed methods. K E Y W O R D Scausal inference, inverse-probability-of-treatment weighting, logistic regression model, marginal structural model, measurement error, time-dependent confounding

show abstract

Causal inference with measurement error in outcomes: Bias analysis and estimation methods

Cited by 28 publications

References 29 publications

Weighted causal inference methods with mismeasured covariates and misclassified outcomes

Weighted causal inference methods with mismeasured covariates and misclassified outcomes

Calcitonin Gene–Related Peptide Monoclonal Antibodies and Risk of SARS-CoV-2 Infection and Severe COVID-19 Outcomes Among Veterans With Migraine Disorder

Inverse‐probability‐of‐treatment weighted estimation of causal parameters in the presence of error‐contaminated and time‐dependent confounders

Contact Info

Product

Resources

About