Commentary: Developing best-practice guidelines for the reporting of E-values

VanderWeele, Tyler J.; Mathur, Maya B.

doi:10.1093/ije/dyaa094

Cited by 81 publications

(81 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For Study 3, I calculated E- values for learning disability identification. Consistent with best practice (Trinquart et al, 2019; VanderWeele & Mathur, 2020), I report LB 95% CIs for the E -values. Table 1 displays the originally reported ORs or RRs from Studies 1, 2, and 3, the LB CIs, and resulting E -values based on RRs of 1.0 and 2.0.…”

Section: Methodsmentioning

confidence: 77%

“…I examined for one source of bias (i.e., confounding) using one type of method. There are other types of sensitivity analyses for quantifying bias (Lash et al, 2009), although the E -value is considered particularly straightforward (VanderWeele & Mathur, 2020). Measurement error, misclassification, and selection are other sources of potential bias (Lash et al, 2009).…”

Section: Discussionmentioning

confidence: 99%

“…An E- value provides a conservative estimate of the confounder’s potential bias of an observed relation (VanderWeele, Ding, & Mathur, 2019). E- value uses and interpretations are currently being debated (Fox et al, 2020; Groenwold, 2020; Ioannidis et al, 2019; VanderWeele, Mathur, & Ding, 2019) and best-practice reporting guidelines as well as extensions are still emerging (Blum et al, 2020; Cusson & Infante-Rivard, 2020; VanderWeele & Mathur, 2020). Because studies reporting under-identification have adjusted for measured confounders (e.g., economic disadvantage) likely related to unmeasured confounders (e.g., lead exposure), the residual confounding may be relatively small (Morgan et al, 2015, 2017).…”

Section: Discussionmentioning

confidence: 99%

See 2 more Smart Citations

Unmeasured Confounding and Racial or Ethnic Disparities in Disability Identification

Morgan

2021

Educational Evaluation and Policy Analysis

View full text Add to dashboard Cite

Students who are Black or Hispanic are now reported to be less likely to be identified as having disabilities than similarly situated students who are White. Although repeatedly replicated, this finding is often characterized as in error. I use a new statistical technique, the E-value, to quantify the likelihood that unmeasured confounding explains observed associations between race or ethnicity and disability identification. Results based on calculations across three population-based studies using extensive statistical controls suggest that unmeasured confounding is an unlikely explanation for the observed associations. Unmeasured confounding that would result in levels of overidentification consistent with federal law and regulation is especially unlikely.

show abstract

Section: Methodsmentioning

confidence: 77%

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Unmeasured Confounding and Racial or Ethnic Disparities in Disability Identification

Morgan

2021

Educational Evaluation and Policy Analysis

View full text Add to dashboard Cite

show abstract

“…Like all sensitivity analyses, the E‐value should be reported and interpreted thoughtfully 38 . In particular, its interpretation depends on context, particularly on the measured confounders that have been adjusted in analysis 28 .…”

Section: A Framework For Choosing Sensitivity Analyses For Unmeasuredmentioning

confidence: 99%

“…Then the first study could be considered more robust to unmeasured confounding because in that study, unmeasured confounder(s) would have to be associated with both the treatment and the outcome by relative risks of 2.5 each, above and beyond all of the confounders that have already been measured and adjusted in the analysis. Further guidance and recommendations on the practical interpretation of the E‐value, and discussion of its limitations, is provided elsewhere 38 …”

Section: A Framework For Choosing Sensitivity Analyses For Unmeasuredmentioning

confidence: 99%

Assessing the impact of unmeasured confounders for credible and reliable real‐world evidence

Zhang

Stamey

Mathur

2020

Pharmacoepidemiology and Drug

View full text Add to dashboard Cite

Purpose: We review statistical methods for assessing the possible impact of bias due to unmeasured confounding in real world data analysis and provide detailed recommendations for choosing among the methods. Methods: By updating an earlier systematic review, we summarize modern statistical best practices for evaluating and correcting for potential bias due to unmeasured confounding in estimating causal treatment effect from non-interventional studies. Results: We suggest a hierarchical structure for assessing unmeasured confounding. First, for initial sensitivity analyses, we strongly recommend applying a recently developed method, the E-value, that is straightforward to apply and does not require prior knowledge or assumptions about the unmeasured confounder(s). When some such knowledge is available, the E-value could be supplemented by the rule-out or array method at this step. If these initial analyses suggest results may not be robust to unmeasured confounding, subsequent analyses could be conducted using more specialized statistical methods, which we categorize based on whether they require access to external data on the suspected unmeasured confounder(s), internal data, or no data. Other factors for choosing the subsequent sensitivity analysis methods are also introduced and discussed, including the types of unmeasured confounders and whether the subsequent sensitivity analysis is intended to provide a corrected causal treatment effect. Conclusion: Various analytical methods have been proposed to address unmeasured confounding, but little research has discussed a structured approach to select appropriate methods in practice. In providing practical suggestions for choosing appropriate initial and, potentially, more specialized subsequent sensitivity analyses, we hope to facilitate the widespread reporting of such sensitivity analyses in non-interventional studies. The suggested approach also has the potential to inform pre-specification of sensitivity analyses before executing the analysis, and therefore increase the transparency and limit selective study reporting.

show abstract

Changes in Attributable Fractions and Causal Inference for Schizophrenia and Other Psychiatric Outcomes

VanderWeele

2021

JAMA Psychiatry

View full text Add to dashboard Cite

Commentary: Developing best-practice guidelines for the reporting of E-values

Cited by 81 publications

References 14 publications

Unmeasured Confounding and Racial or Ethnic Disparities in Disability Identification

Unmeasured Confounding and Racial or Ethnic Disparities in Disability Identification

Assessing the impact of unmeasured confounders for credible and reliable real‐world evidence

Changes in Attributable Fractions and Causal Inference for Schizophrenia and Other Psychiatric Outcomes

Contact Info

Product

Resources

About