Background Assessment of early-life exposure to environmental tobacco smoke (ETS) often relies on retrospective maternal report. We evaluated the validity of such reports up to 17 years later in dyads with prenatal and/or household smoke exposure. Methods Baby’s Breath was a randomized controlled trial of a smoke avoidance intervention among 738 pregnant women and their infants followed through 6 months postpartum (2006-2009). Mothers reported their own smoking and infant ETS exposure at 6 months postpartum; bioverified with salivary cotinine. Pathways recruited 109 dyads from Baby’s Breath when children reached adolescence (2021-2026). Mothers retrospectively reported infant ETS exposure. Positive and negative predictive value, sensitivity, specificity, Cohen’s kappa, and cotinine levels across self-report categories were computed for maternal self-report of smoking and both concurrent and retrospective report of infant ETS exposure. Results Maternal self-reported smoking was concordant with bioverified cotinine whereas both concurrent and retrospective maternal reports of infant ETS exposure were not. The predictive value of retrospective reporting depended entirely on how exposure was characterized. Conclusion In this cohort of families with smoke exposure, postpartum maternal reports of their own smoking closely aligned with bioverified cotinine, whereas concurrent and retrospective reports of infant ETS exposure showed poor concordance. Impact <UnorderedList Mark="Bullet"> <ItemContent> Longitudinal validation of maternal retrospective reports of infant ETS exposure, using bioverified cotinine data, revealed systematic recall bias. </ItemContent> <ItemContent> Mothers accurately recalled confirmed exposure up to 17 years later but frequently denied or underestimated it when biomarkers indicated otherwise. </ItemContent> <ItemContent> Reports were sensitive to acknowledged exposure but showed poor specificity for non-exposure, increasing misclassification risk. </ItemContent> <ItemContent> These findings urge caution when using retrospective reports in etiologic research and underscore the importance of biomarker validation. </ItemContent> </UnorderedList>

Insights into the validity of concurrent and retrospective maternal report of infant smoke exposure using biomarker-based validation

期刊论文
出版模式：

开放获取
发表日期：

2026年7月17日

查看全文

Lauren Micalizzi,
Alexander W. Sokolovsky,
Cara M. Murphy,
Valerie S. Knopik,
Patricia Markham Risica

摘要

Background

Assessment of early-life exposure to environmental tobacco smoke (ETS) often relies on retrospective maternal report. We evaluated the validity of such reports up to 17 years later in dyads with prenatal and/or household smoke exposure.

Methods

Baby’s Breath was a randomized controlled trial of a smoke avoidance intervention among 738 pregnant women and their infants followed through 6 months postpartum (2006-2009). Mothers reported their own smoking and infant ETS exposure at 6 months postpartum; bioverified with salivary cotinine. Pathways recruited 109 dyads from Baby’s Breath when children reached adolescence (2021-2026). Mothers retrospectively reported infant ETS exposure. Positive and negative predictive value, sensitivity, specificity, Cohen’s kappa, and cotinine levels across self-report categories were computed for maternal self-report of smoking and both concurrent and retrospective report of infant ETS exposure.

Results

Maternal self-reported smoking was concordant with bioverified cotinine whereas both concurrent and retrospective maternal reports of infant ETS exposure were not. The predictive value of retrospective reporting depended entirely on how exposure was characterized.

Conclusion

In this cohort of families with smoke exposure, postpartum maternal reports of their own smoking closely aligned with bioverified cotinine, whereas concurrent and retrospective reports of infant ETS exposure showed poor concordance.

Impact

Longitudinal validation of maternal retrospective reports of infant ETS exposure, using bioverified cotinine data, revealed systematic recall bias.

Mothers accurately recalled confirmed exposure up to 17 years later but frequently denied or underestimated it when biomarkers indicated otherwise.

Reports were sensitive to acknowledged exposure but showed poor specificity for non-exposure, increasing misclassification risk.

These findings urge caution when using retrospective reports in etiologic research and underscore the importance of biomarker validation.