Although secondhand tobacco smoke (SHS) exposure is declining in the United States (U.S.), it remains a public health concern, particularly among children. The prevalence of SHS among U.S. nonsmokers declined substantially from 87.5% to 25.2% from 1988 to 2014 (1). In 2023, 11% of US adults currently smoked cigarettes, a historic low from its peak prevalence of 42% in 1965 (2). In contrast, SHS exposure remained stable between 2011 and 2012 (25.3%), 2013–2014 (25.2%), and more recently from 2011 to 2018 (25.3%–24.6%) (1, 3). Tobacco smoke is a known human carcinogen also associated with other diseases, such as heart disease (4), and premature death in nonsmoking adults and children (5). In the U.S., an estimated 41,000 deaths among adults each year are caused by SHS (4). Among 3–11, 12–19, and ≥20-year-olds, the 3–11-year-old age group, which aligns with the 7–9-year-old age range represented in the Marietta Communities Actively Researching Exposures Study (CARES) cohort, has the highest percent of nonsmokers with serum cotinine between 0.05 and 0.10 ng/ml (1).

Smoking rates declined faster in urban areas compared with rural areas (6, 7), such as in Central Appalachia America. Additionally, easier access to tobacco products in rural discount stores and higher exposure to SHS in public and private areas make smoking cessation especially difficult in rural vs. urban settings (8). Moreover, Appalachian regions have some of the highest rates of poor sleep in the U.S. According to the 2020 Behavioral Risk Factor Surveillance System, crude and age-adjusted prevalence of short sleep in the highest quintile were counties clustered in Southeastern U.S. and along the Appalachian Mountains (9). According to the National Survey of Children's Health (2020–2021), the prevalence of short sleep among children ages four months to 14 years was highest in Appalachian states, with Ohio reporting a prevalence of 35.9 (95%CI:32.8–39.2) (10).

An increasing number of epidemiologic studies have examined the link between SHS exposure and poor sleep outcomes, including sleep disruption, reduced overall sleep quality, and increased daytime sleepiness (11–13). Among nonsmokers, exposure to SHS results in the inhalation of 60%–80% of airborne nicotine, leading to the absorption of nicotine concentrations similar to those of active smokers (14). Children may be uniquely vulnerable because the proportion of SHS deposited in a child's lungs is greater than in adults due to differences in the diameters and configuration of the airways (15). Nicotine, the primary active compound in tobacco smoke, is a stimulant that promotes arousal and alertness by activating cholinergic neurons in the basal forebrain and disrupting the balance of neurotransmitters involved in sleep regulation (16). These effects can contribute to difficulty falling and staying asleep. SHS exposure, as indicated by elevated cotinine levels in nonsmokers, has been associated with adverse sleep-related outcomes in both children and adults (17–19).

More than 70% of nicotine metabolizes into cotinine, which serves as the best objectively measured biological indicator for assessing SHS exposure and body burden (20, 21). The established range of 0.05–10 ng/ml for serum cotinine has historically been used to define SHS exposure. However, a more recent study analyzed samples using an expanded range of 0.015–10 ng/ml, potentially resulting in higher estimates of the population-level burden of SHS (3). Currently, a level at which serum cotinine has not been associated with adverse health effects has not been identified (1, 22). Even when objective measures are used to explore the relationship between serum cotinine and various sleep outcomes, conflicting results within and among studies have emerged. Some studies have identified significant associations when cotinine is measured continuously (17), while others have observed non-significant findings when using categorical cotinine thresholds (23). Further, higher odds of increased sleep problems among children were found when parent/caregiver report of SHS exposure was used (12). In contrast, findings by Włodarska et al. found more apnea-hypopnea events among children exposed to tobacco smoke when questionnaires were administered to parent/caregivers, but they found no statistically significant differences in waking, snoring, or sleep apnea (24). Variability in SHS-related sleep findings may reflect differences in how exposure is measured. Interviewer-administered questionnaires, adult respondents, and observational study designs yield more accurate SHS assessments, especially when paired with cotinine biomarkers (15).

Given the conflicting methods for assessing SHS exposure and its relationship with sleep in the literature, and the importance of understanding how different assessment approaches relate to reported sleep disturbances in children, this study aimed to investigate the association between SHS exposure using three different methods of SHS assessment and parent/caregiver report of sleep disturbances in a rural Appalachian American pediatric cohort. We hypothesized that SHS exposure, whether measured by (1) ln serum cotinine levels continuously (ng/ml), (2) dichotomously (≥.05 vs. <.05 ng/ml), or by (3) parent/caregiver report of ≥1 smoker in the home, would be associated with shorter sleep duration among children. Secondary analyses explored associations between cotinine and other sleep characteristics, including night awakenings and snoring.

CARES consists of a longitudinal cohort of 404 rural Appalachian children living in Marietta and Cambridge, Ohio, and surrounding areas. Children were recruited at ages 7–9 years from 2008 to 2013. Their biological mothers resided in the study catchment area since the 16th week of pregnancy (25). Study design, recruitment, and data collection and compilation methods have been described in detail elsewhere (26–28). All parents/caregivers signed informed consent and all child participants signed informed assent forms approved by the Institutional Review Board of the University of Cincinnati (UC IRB #2011-2542).

This paper focuses on three methods for measuring SHS exposure: continuously measured ln serum cotinine concentrations, dichotomized ln serum cotinine levels (≥0.05 vs. <0.05 ng/ml), and parent/caregiver report of ≥1 smoker in the home. Methods of collecting and analyzing child serum cotinine in the CARES cohort have been described in detail elsewhere (26–29). Briefly, venous whole blood was collected by trained phlebotomists from the antecubital vein into a standard red-top tube (no preservative) and allowed to clot at room temperature for 15–25 min prior to centrifugation to obtain serum off the clot. Serum samples were frozen at a temperature of −20°C until shipped monthly for analysis to the New York State Department of Health's Wadsworth Center, Albany, New York. Serum cotinine analysis was performed using liquid chromatography/tandem mass spectrometry similar to that described by the Centers for Disease Control and Prevention (CDC) for the National Health and Nutrition Examination Survey (NHANES) (30) and New York State Wadsworth Laboratories for the New York City Health and Nutrition Examination Survey (31). The method limit of detection (LOD) was 0.05 ng/ml (26). Instrumental values below the LOD were provided by the laboratory and included for statistical analyses; values below 0 were imputed with the LOD divided by the square root of 2 for statistical analyses (32, 33).

Natural log transformation of serum cotinine concentrations aids in normalizing the data distribution and provides a quantitative measure of recent exposure to nicotine, reflecting a higher precision than self-reports. Dichotomizing serum cotinine levels at a threshold of 0.05 ng/ml (i.e., the LOD), although subject to some data loss, serves as a practical approach to classify individuals into exposed vs. non-exposed groups, simplifying the analysis and interpretation of SHS exposure effects on sleep outcomes. Finally, parent/caregiver-reported measures of ≥1 smoker in the home, while subject to reporting bias, nevertheless offer valuable information on chronic SHS exposure in household settings and are vital for assessing the social and behavioral contexts of exposure.

Twenty-six percent of children had a parent/caregiver report of ≥1 sleep disturbance. In fully adjusted models using serum cotinine as a continuous measure, prevalence of short sleep duration (35%) attenuated but remained significant. Frequent snoring was not significant, with having ≥1 sleep disturbance (11%) reaching borderline significance. In age-adjusted models, children were 3.6 times more likely to have short sleep duration with serum cotinine levels ≥.05 ng/ml (vs. <.05 ng/ml) and nearly 2.5 times more likely to have short sleep duration with ≥1 smoker (vs. none) living in the home. Increasing ln serum cotinine levels were associated with a significant 5.5-min decrease in average sleep duration in age-adjusted models. Parent education had a significantly positive association with average sleep duration, independent of other model covariates.

Although our significant associations between SHS exposure and short sleep duration when using continuously or dichotomously measured ln serum cotinine levels were only observed in age-adjusted models, and attenuated in fully adjusted models, previous studies have found associations with increasing serum cotinine and risk of trouble sleeping in fully adjusted models. Yolton et al. found a higher odds ratio of sleep onset delay, parasomnias, sleep-disordered breathing (SDB), daytime sleepiness, and overall sleep disturbance with increasing serum cotinine among children 6–12-years, even after adjusting for covariates such as prenatal tobacco exposure, maternal depression, and household density (17). Likewise, Du et al. found significant correlations between high serum cotinine levels (.05–3.00 ng/ml) and shorter sleep duration among adolescents after fully adjusting for covariates such as poverty income ratio, physical activity, and caffeine intake (18). The contrast in findings may reflect differences in covariate selection across studies. While our models included age, sex, parent education, BMI, and history of breathing difficulty, variations in specific confounders and outcome assessment may influence the observed associations. In contrast to our non-significant results when using parent/caregiver reports of having ≥1 smoker in the home, a review and meta-analysis by Safa et al. found a pooled OR of 1.20 (95% CI: 1.09–1.33; p = .0003; n = 692,690 adolescents and adults) for short sleep duration among the SHS-exposed vs. not exposed groups after adjusting for covariates (19). Similar to our study, all studies in the meta-analysis assessed SHS exposure using self-reported questionnaires and interviews. It should be noted, as stated by the authors, that there was large heterogeneity between the studies (I² = 68%, p = .004), indicating variation in effect sizes across 7 studies that examined short sleep duration, and only 2 of the 7 studies included adolescents (19). Similarly, Reichenberger et al. found that SHS exposure was associated with reduced sleep duration at age 5, with an even greater reduction by age 9 in longitudinal analyses. Cross-sectional analyses at age 9 also showed a negative association between SHS exposure and sleep duration when assessed through caregiver interviews (13). In contrast, Davila et al. found no significant associations between SHS-exposed nonsmokers and reports of sleep disorders or symptoms [OR = 1.43 (95% CI: 0.79–2.57)] (22); however, their study was conducted in an adult population (mean age: 48.3 ± 0.29 years, n = 4,123). While our study aligned with several key conditions associated with higher sensitivity and specificity of SHS exposure assessment, including the use of interviewer-administered questionnaires in an observational context and responses provided by adults (15), we did not include repeat questionnaires or cotinine measurements. The absence of repeated measures may have contributed to the non-significant associations observed in our study, despite following other best practices in self-reported exposure assessment. We observed a significantly higher prevalence of children with ≥1 sleep disturbance when using continuously measured ln serum cotinine levels in age-adjusted models, nearly reaching significance after adjusting for all covariates. However, when assessing whether serum cotinine levels ≥0.05 ng/ml were associated with having ≥1 sleep disturbance, no significant associations were found in either age-adjusted or adjusted models. The inconsistencies in our findings, both within this study and compared to previous literature, highlight how results vary depending on the method used to measure SHS exposure.

Our study observed a significantly increased prevalence of frequent snoring among children with increasing ln serum cotinine levels in age-adjusted models; however, this relationship did not reach significance in fully adjusted models. Similar to our non-significant association between SHS exposure and snoring, regardless of the SHS measurement method used, a systematic review of the literature on SHS and SDB in children, characterized by habitual snoring [loud snoring ≥3 nights per week (40)], identified three studies that found no association (41). More notably, however, the review also reported 15 case-control studies that demonstrated a significant association between SHS exposure and SDB.

Associations between SHS exposure (regardless of measurement method) and frequent night awakenings were not statistically significant in either age-adjusted or fully adjusted models. Our findings diverge from the results of a recent cross-sectional study of more than 45,000 children in China, which reported that SHS exposure in utero and during the first two years of life was associated with increased odds of sleep disturbances, such as shorter sleep duration, night awakenings, snoring, and breathing problems, at ages 6–18 years (12). The significantly larger sample size in that study may have provided greater statistical power to detect associations that were not evident in our more modestly sized cohort.

There are limitations to our study. Our sample consisted of a relatively small proportion of children with frequent night awakenings (N = 17), which may contribute to our lack of significant associations. Additionally, other factors known to be associated with sleep were not included in the dataset: melatonin or other sleep aid use, caffeinated drink consumption, exposure to blue light and/or use of electronic devices in bed, neighborhood noise, and nighttime light exposure. Additionally, parent-reported child sleep measures may be subject to recall bias and may not accurately reflect objective sleep, as the study lacked daily sleep logs and electronic sleep monitoring. Bedtime is an important determinant of sleep duration for children (42). Parent/caregiver-reported sleep duration was assessed at a single time point for typical weekdays and weekends during both the school year and non-school year, limiting the ability to capture sleep patterns or changes over time. As noted above, we also did not measure screen time in bed, which is a well-documented factor associated with delayed sleep onset and reduced sleep duration in children and adolescents (43). Finally, since no formal p-value adjustment was applied (e.g., Bonferroni correction or false discovery rate), given the exploratory nature, findings should cautiously be interpreted as hypothesis-generating given multiple tests across related sleep outcomes.

Despite these limitations, this study has numerous strengths. One strength of this study is that the CARES cohort reflects the demographic composition of the Appalachian population. Given that Appalachia's population was 26.6 million in 2023, with non-White people comprising only 21.3% (national rate = 41.6%) and White people comprising 78.7% (national rate = 58.4%), our cohort is comparable to Appalachian communities (44). Moreover, smoking behaviors reported in our cohort align with regional trends: according to the CDC and the Appalachian Regional Commission, adult smoking prevalence in Central Appalachia was 25.2% in 2020 (6), compared to a national rate of 15.2% (45). In our sample, 36.6% of children lived in households with ≥1 smoker, reflecting the elevated household and community-level tobacco burden consistently documented across Appalachian populations (6). Another strength is the use of validated sleep questions and serum cotinine as an objective SHS biomarker, both collected during the same study visit. It should be noted, however, that although serum cotinine has a high specificity and sensitivity with a relatively prolonged half-life of 16–20 h in children (46–48), it provides a relatively short-term snapshot of recent exposure and does not reflect chronic or cumulative exposure (46, 47). Lastly, we utilized three methods for measuring SHS exposure to investigate its association with parent/caregiver-reported sleep disturbances in a rural Appalachian pediatric cohort. Although there have been several previous studies that have found significantly positive relationships between SHS as measured by parent/caregiver-reported tobacco use and sleep disruption among children and adolescents (11–13, 36, 37, 41), yet few have included serum cotinine as an objective internal measure of environmental tobacco smoke exposure in pediatric populations. Geometric mean serum cotinine levels rose with more household smokers; however, parent-reported smoking may underestimate SHS exposure, as 37% reported a smoker while 50% of children had detectable cotinine, indicating possible underreporting or unrecognized sources (27).

To our knowledge, this is the first study to report on the impact of SHS exposure on sleep disturbances using multiple SHS measurement methods in Appalachian children. Ln serum cotinine as a continuous measure was the only SHS measure associated with a sleep disturbance: a higher prevalence of short sleep duration after adjusting for covariates. Although causal inference cannot be established from this observational design, the consistency of findings across analytic approaches supports a plausible association between tobacco smoke exposure and children's sleep disruption. These findings highlight the need to maintain smoke-free environments where children live, play, and sleep. Parents and caregivers should avoid smoking indoors or in vehicles, and educators should promote awareness of secondhand smoke's impact on sleep and overall health. Community efforts to establish smoke-free homes, schools, and recreational spaces remain essential. Moreover, our findings underscore the importance of using a continuous measure of serum cotinine as an objective measure of SHS exposure when assessing sleep disturbance in child cohorts.