Prevalence and Quantification of Secondhand Smoke Exposure Among Hospitalized Children <6 Years of Age

We prospectively enrolled children aged <6 years hospitalized at a single children’s hospital in the United States between 2010 and 2012 and included children with pneumonia or acute, nonrespiratory illnesses. The children with pneumonia represent a subset of children enrolled in the Etiology of Pneumonia in the Community study, which enrolled children with radiographically confirmed, community-acquired pneumonia and whose enrollment criteria has been previously published.¹¹  Concurrently, we enrolled children with acute, nonrespiratory illnesses (eg, trauma, septic arthritis). We excluded children hospitalized within the previous 7 days, those living in a long-term care facility, newborns who never left the hospital, those with significant immunosuppression, and those with a tracheostomy. We also excluded those in whom a urine sample could not be obtained. Caregivers were interviewed by study personnel to obtain sociodemographic characteristics and past medical history. Those with both public and private insurance were coded as public insurance. A urine sample was also obtained and immediately frozen at −80 °C. This study was approved by our institutional review board.

The primary outcome was SHS exposure quantitatively defined by using urinary cotinine.⁶  After absorption, ∼80% of nicotine is directly metabolized into cotinine,⁵  thus making it an ideal biomarker for SHS exposure. Urine samples were analyzed by tandem liquid chromatography–mass spectrometry at the University of California, San Francisco.⁵  Cotinine was analyzed as an ordinal variable indicating intensity of SHS exposure, defined as no exposure (below limit of detection, <0.025 ng/mL), light SHS exposure (0.025–0.25 ng/mL), and heavy SHS exposure (>0.25 ng/mL).⁶  To contrast with our objective measures, we also classified SHS exposure on the basis of information from the caregiver interviews. Caregivers were asked, “How many household members smoke (either indoors or outdoors)?” with responses classified as 0 or ≥1 household smokers, and “Is smoking allowed indoors?” classified as yes or no. Responses were organized into a composite variable of caregiver-reported home SHS exposure (no reported exposure, ≥1 household smokers but no smoking allowed in home, and ≥1 household smokers with smoking allowed in home).

Descriptive statistics included frequencies (percent) for categorical variables and median (interquartile range [IQR]) for continuous variables. We compared baseline characteristics across groups defined by intensity of SHS exposure on the basis of urinary cotinine level (no exposure, light, and heavy) using Wilcoxon rank and χ² tests. Similar analyses were performed analyzing demographic data among children without reported SHS exposure. Violin plots were generated by using log-transformed cotinine values for groups defined by the composite variable of caregiver-reported home SHS exposure. A multivariable proportional odds regression model was used to examine associations between sociodemographic characteristics and intensity of SHS exposure. We also examined associations between SHS exposure intensity and diagnosis (pneumonia versus nonrespiratory illness). For this analysis, diagnosis was the outcome of interest, and urinary cotinine was the primary exposure variable. All analyses were completed by using R version 3.5.1.

A total of 239 children were enrolled, including 149 (62%) with pneumonia (median age 15 months [5–31]) and 90 (38%) with nonrespiratory illness (median age 16 months [4–35]) (Supplemental Table 4).

Although reported home SHS exposure was common (36%), it substantially underestimated exposure detected by urinary cotinine (77% of children), including 17% with light exposure and 59% with heavy exposure (Table 1). Exposure intensity varied significantly with respect to parental education, race, and insurance type. There were no differences in exposure intensity by season. The highest urinary cotinine level was among children exposed to indoor smoking (7.78 ng/mL, IQR 2.93–18.65; P < .001) (Supplemental Fig 1). Even so, the majority of children with no reported home SHS exposure (64%) also had urinary cotinine levels above the limit of detection.

In multivariable analysis, intensity of SHS exposure did not vary by age or sex (Table 2). As compared with non-Hispanic white children, Hispanic children had significantly less SHS exposure (odds ratio [OR] 0.25, 95% confidence interval [CI] 0.10–0.61). Exposure intensity in non-Hispanic Black children did not differ from that of non-Hispanic white children. Exposure intensity varied by household educational attainment, with significantly lower SHS exposure among children whose caregivers attended college (OR 0.43, 95% CI 0.18–1.02), had a college degree (OR 0.19, 95% CI 0.08–0.47), or had an advanced degree (OR 0.11, 95% CI 0.03–0.37), as compared with children whose caregivers had a high school education or less. Private insurance was also associated with decreased SHS exposure intensity, as compared with public insurance (OR 0.38, 95% CI 0.19–0.78). In a multivariable analysis of children with no reported SHS exposure, exposure intensity as measured by urinary cotinine varied in a similar pattern, as reported in the overall study population (Supplemental Table 3).

Children hospitalized with pneumonia were more likely to have ≥1 comorbidity compared with children hospitalized with nonrespiratory illnesses; otherwise, there were no significant baseline differences (Supplemental Table 4). In multivariable analyses, there were no detectable differences in SHS exposure intensity between children with pneumonia and children with acute, nonrespiratory illness (OR 1.13, 95% CI 0.521–2.44 for heavy exposure versus no exposure; OR 1.18, 95% CI 0.48–2.88 for light exposure versus no exposure).

In our study of children aged <6 years hospitalized with pneumonia or nonrespiratory illnesses, more than three-quarters of children had objective evidence of SHS exposure. Caregiver report substantially underestimated SHS exposure in this population. Higher intensity of SHS exposure was associated with non-Hispanic ethnicity, lower caregiver educational attainment, and public insurance.

Our finding that most hospitalized children aged <6 years had objective evidence (urinary cotinine level >0.025 ng/mL) of SHS exposure is consistent with a previous study that found an exposure prevalence of 76% among adolescents admitted to an urban general hospital that was not attributable to active smoking on the basis of objective urinary cotinine levels (urinary cotinine level <30 ng/mL).⁶  Compared with contemporaneous estimates, the frequency of exposure in these studies of hospitalized young children and adolescents is more than twice that from a national sample that defined SHS exposure on the basis of urinary cotinine levels.⁸  Although previous studies have revealed associations between respiratory illness severity and SHS exposure,^3,4  our study did not. We note, however, that our study population was restricted to hospitalized children, a group that likely differs from the general population in several ways. Thus, although much is known about the deleterious effects of SHS exposure on acute and chronic respiratory health,⁴  our findings suggest that all hospitalized children are at increased risk for SHS exposure and its harmful effects.

The findings of this study also reveal the shortcomings of hospital intake screenings for SHS because most children with no reported home SHS exposure had biochemical evidence of exposure. There are several potential reasons for this, most notably underreporting of home SHS exposure and exposure outside of the primary home environment. Other factors known to contribute to increased SHS exposure that may contribute to underreporting include exposure to e-cigarettes (vaping)¹²  and living in multiunit housing.⁹  Regardless of the reasons, subjective reporting of SHS exposure is imperfect and has the potential to substantially underestimate true exposure.^7,13  Of note, we observed that a cotinine level < 0.25 ng/mL appears to predict the absence of home SHS exposure, which could serve as a useful screening tool.

The high prevalence of SHS exposure in our study population confirms that more interventions are needed to reduce its burden on the health of young children. Pediatric hospitalizations have been investigated as an opportunity for smoking cessation counseling for caregivers.^14,15  Most caregivers were open to counseling regarding their smoking during their child’s hospitalization,¹⁴  but these studies revealed questionable efficacy of these interventions,¹⁵  suggesting more nuanced and personalized approaches may be needed. Promoting more stringent tobacco control policies is another important opportunity to curtail SHS among children. Previous work has revealed that policies aimed at discouraging smoking (eg, taxes, smoke free–air laws, and media campaigns) are effective and have been linked to substantive reductions in smoking prevalence after implementation.¹⁶ 

Our study had several limitations. First, the single institution design may limit the direct extrapolation of our findings to other settings. Second, the respiratory group was narrowly defined as children with pneumonia, excluding children with diagnoses like asthma or bronchiolitis. Third, data were not collected on a matched, nonhospitalized group of children from a similar region to provide insight on how the hospitalized population differed from the general pediatric population. Fourth, the data were collected between 2010 and 2012, and it is possible the patterns of smoke exposure have changed. It will be important to confirm these findings in larger, contemporary, and more heterogenous cohorts of children.

Among hospitalized young children, reported home SHS exposure was common but substantially underestimated when compared with urinary cotinine levels. The highest urinary cotinine levels were among children exposed to indoor smoking. Future public health interventions, as well as more robust SHS exposure screenings on hospital admission, are needed to reduce the prevalence of SHS exposure among young children.