Biomarker Validation Tests Accuracy of Maternal Reports on Infant Smoke Exposure

A new biomarker-based analysis is shining a spotlight on a subtle statistical trap in studies that try to reconstruct infants’ exposure to secondhand smoke. The work focuses on maternal reports given either concurrently—around the time exposure occurs—or retrospectively, after months or years have passed. Because such evidence often informs public-health guidance, the reliability of “did your baby breathe smoke?” answers matters.

The central finding is not that every retrospective report is wrong. Instead, the threat lies in a specific pattern of error: mothers who deny that their infants were exposed may be unreliable in a nontrivial subset of cases. In other words, misclassification is asymmetric. Missing exposures are more likely to be missed than falsely reported.

This asymmetry can quietly distort prevalence estimates. If true exposure is undercounted, studies may conclude that fewer infants were harmed—or that harms are milder—than reality suggests. The bias does not just shift numbers; it can also mask developmental risks that researchers are trying to detect in early-life populations.

Using biomarker validation as a reference point, the study evaluates how closely self-report aligns with measurable evidence. Biomarkers provide a more objective snapshot of exposure, allowing researchers to estimate which reporting modes perform better and under what conditions errors emerge. The analysis argues that retrospective designs are particularly vulnerable when researchers treat “no exposure” answers as equally trustworthy.

The implications are broad for long-term cohort studies, where exposure histories often rely on memory. If denial-based misclassification is common, then the analysis framework needs to acknowledge differential error. Otherwise, confidence intervals may look narrow while the underlying exposure distribution is wrong.

The authors suggest that accuracy-improving strategies could reduce bias, including adding contextual measures that corroborate reporting (such as household smoke-related factors). When objective verification is unavailable, triangulating self-report with supporting indicators may offer a practical substitute.

Ultimately, the study reframes a methodological question into a surveillance one: not whether retrospective recall is imperfect, but whether the specific direction of error systematically favors underestimation. That distinction could help reconcile conflicting findings across studies and sharpen estimates of early developmental harm.

If the results generalize, future research should incorporate models that explicitly account for asymmetric misclassification, rather than assuming symmetric reporting error. Doing so may help ensure that policies respond to the exposures infants truly face, not just the exposures researchers can reliably measure through memory.

Micalizzi, L., Sokolovsky, A.W., Murphy, C.M. et al. Insights into the validity of concurrent and retrospective maternal report of infant smoke exposure using biomarker-based validation.
Pediatr Res (2026). https://doi.org/10.1038/s41390-026-05238-3