Introduction

Tobacco smoke exposure (TSE), defined as exposure to smoke from combustible tobacco products in the form of second- and/or thirdhand smoke, has significant health consequences for children, including increased lifetime risk of developing cancer, cardiovascular, and pulmonary diseases, and higher incidence of ear and upper respiratory infections and sudden infant death syndrome.1 Despite recent improvements in reducing pediatric TSE in the United States (U.S.), it remains prevalent among many populations, including children growing up in racially- and ethnically-minoritized and lower socioeconomic status communities.2,3,4 Non-Hispanic Black children, for example, have higher levels of TSE2,4 and have not experienced the same reductions in exposure seen in other pediatric populations.3,4,5 Many structural factors have contributed to TSE disparities in minoritized communities, including targeted tobacco industry advertising, increased tobacco retailer exposure, and health care access barriers.6 Linguistic, cultural, and health care navigational barriers are additional barriers to tobacco prevention and control for immigrant populations.6

Clinician-driven parental tobacco screening, brief intervention, and referral to treatment is an American Academy of Pediatrics-endorsed approach to reducing pediatric risk of household TSE.7 Identifying and treating parent or caregiver (hereafter parent) tobacco use during their child’s visit may increase parental access to tobacco treatment and their likelihood of cessation,8,9 yet levels of TSE screening in pediatric primary care are typically at or below 50% over a period of 3 months to one year.10,11

Multiple factors appear to influence the likelihood of TSE screening during pediatric visits. Parents who identify as Black,12,13,14 Hispanic and/or who speak Spanish as their preferred language,10,12,14 those less than 30 years old,12 and those who actively smoke12 are more likely to be screened for tobacco use during their child’s appointment. Pediatric patients who have one or more high-risk medical diagnoses that could be exacerbated by TSE (i.e., asthma or prematurity) also have higher levels of screening.10 The influence of a child’s age and insurance status (public or private) on TSE screening is less clear.10,12 At the clinic level, TSE screening may be more likely to occur during well/preventive health visits.12 Screening levels among pediatricians and family medicine physicians are similar, though pediatricians are more likely to document TSE in a patient’s chart.11,15 However, with one notable exception, the majority of these studies have been conducted in predominantly White populations with limited preferred language information,10 and no prior studies have examined how interpreter use might influence the likelihood of TSE screening during pediatric visits.

This study aimed to characterize the distribution of sociodemographic and clinic visit-related factors by household TSE screening in pediatric primary care within a large health system serving racially and ethnically diverse populations. Identifying which patients are more or less likely to be screened may identify opportunities to more equitably integrate pediatric TSE screening into routine pediatric primary care for diverse patient populations.

Methods

Study setting

We used electronic health record (EHR) data from one Midwestern U.S. metropolitan healthcare system affiliated with a university health center. The health system serves a racially and ethnically diverse population (including approximately 5% each of Somali and Hmong patients—two regional immigrant populations). Adult combustible cigarette smoking prevalence in the state where this study took place was 13% in 2021, higher than the U.S. average prevalence of 11.6%, but lower than the 14–15% prevalence in surrounding states.16 This study was reviewed and approved by the University’s institutional review board.

Analytic sample

Our sample included all pediatric clinic visits within family medicine and pediatric primary care clinics (N = 78) from November 2021 to October 2022. Patients were included if they did not opt out of their EHR data being used in research, were 0 to 18 years old, and presented for a preventive or problem-based visit with an advanced practice practitioner (APP, including physician assistants and nurse practitioners) or a physician (i.e., MD, DO, or MBBS). After extraction and de-identification of EHR data, 145,617 eligible pediatric visits were identified. Patient-level identifiers were used to link visit-level data from the EHR. If a pediatric patient had multiple visits within the study period, we included only the first visit, resulting in a final sample of 62,161 patients.

Measures

EHR data included documentation of TSE screening, visit characteristics (i.e., clinician type, specialty type, and the reason for the appointment – preventive or problem-based), patient demographics, and relevant high-risk clinical diagnoses (i.e., asthma, a combined measure of reactive airways disease (RAD)/wheezing/bronchiolitis, and history of prematurity) in diagnostic codes or problem lists (see Table 1). Our definition of high-risk medical diagnoses included chronic pediatric medical conditions that are often exacerbated by TSE and may increase a clinician’s likelihood of TSE screening, as previously defined by Sharifi et al.10 We defined positive TSE screening as documentation of parental and/or household tobacco use in the patient’s EHR problem list as an active problem or a diagnosis code (i.e., ICD-10 codes P96.81, “Exposure to tobacco smoke in the perinatal period,” and Z77.22, “Contact with or suspected exposure to environmental tobacco smoke, whether acute or chronic”) billed and/or any documentation of passive smoke exposure screening in the patient’s EHR social history under either “smoking tobacco use” or “tobacco use status” dropdown questions in the study period. The screening question used by the system during the study period was: “Is your child exposed to any type of tobacco smoke in their usual environment?.” Those patients without this evidence of TSE screening were classified as having no documented screening. Patient demographics included age (continuous and categorical), sex (female, male, other), race (American Indian/Alaskan Native, Asian, Black/African American, Native Hawaiian/Other Pacific Islander, and White, in any combination), ethnicity (Hispanic/Latino or not), country of birth (U.S. or other), interpreter use (yes or no), preferred language (English, Karen, Somali, Hmong, or other), insurance status (yes or no) and insurance type (public, private, or both).

Table 1 Pediatric patient and visit characteristics (N = 62,161)
Full size table

Data analysis

Descriptive statistics summarized the analytic sample’s characteristics by documentation of TSE screening (Yes vs. No). Categorical variables were presented as frequency (percentage), whereas continuous variables were presented as mean (SD). We compared the frequency of visits with and without TSE screening by subgroups of sociodemographic and visit characteristics as detailed above using chi-square tests. We used unadjusted and multivariable modified Poisson regression models17 with robust SE to explore the associations between the sample’s characteristics with TSE screening. These models estimated prevalence ratios and 95% confidence intervals. Results were interpreted based on the magnitude, direction, and precision (95% CI) of effect estimates, rather than statistical significance. All analyses were performed in Stata 18.0 MP.18

Results

Our sample included 62,161 pediatric patients, of whom 52,326 (84%) had documented TSE screening (Table 1). Overall, 74% of visits occurred with a physician, 59% with Family Medicine clinicians, and 13% with an interpreter. Patients in the sample had a mean age of 7.8 years ( ± 5.4), and were 50% female, 52% White, 25% Asian, 18% Black, 5% mixed race, and 6% Hispanic.

In unadjusted analyses (Table 2), TSE screening was more likely among patients who were 3 years of age or older, those who identified as White, those with any insurance, and those with a diagnosis of asthma or RAD/bronchiolitis/wheeze. For example, children ages 3 to 6 years (PR 1.35, 95% CI 1.34, 1.38), 7 to 12 years (PR 1.33, 95% CI 1.31, 1.35), and 13 years or older (PR 1.41, 95% CI 1.39, 1.43) were more likely to be screened for TSE relative to 0- to 2-year-old patients. TSE screening was less likely among those whose primary language was Hmong relative to English (PR 0.94, 95% CI 0.92, 0.96). While the PRs for use of an interpreter, country of birth, specialty type, and visit type were within the 95% confidence intervals, the effect estimates were close to the null. This suggests that if an association exists, it is likely small and may not be clinically meaningful. Screening levels were similar by patient sex, Hispanic ethnicity, clinician type, and those with a previous preterm birth diagnosis.

Table 2 Unadjusted prevalence ratio (PR) and 95% confidence intervals for the association between patient characteristics and documentation of household tobacco smoke exposure screening
Full size table

In our multivariable adjusted analyses (Table 3), patients who were 3 years or older remained more likely to have documented TSE screening than those who were 2 years or younger. Household TSE screening was also more likely among patients with any insurance (aPR 1.05, 95% CI 1.03, 1.08) and those with a diagnosis of RAD/bronchiolitis/wheeze (aPR 1.07, 95% CI 1.06, 1.08) relative to those without these diagnoses. In our adjusted analyses only, patients with a history of preterm birth (aPR 1.04, 95% CI 1.03, 1.06) were more likely to have been screened for TSE than those without this diagnosis. The aPRs for patient-identified race, those with a diagnosis of asthma, Karen language, country of birth, specialty type, and visit type were within the 95% confidence intervals; however, given the effect estimates near the null, these are unlikely to be clinically meaningful.

Table 3 Adjusted* prevalence ratio (PR) and 95% confidence intervals for the association between patient characteristics and documentation of household tobacco smoke exposure screening (n = 45,359)
Full size table

Discussion

Documented household TSE screening levels were high in this pediatric primary care sample in one large health system. However, we identified that patients who were older (3 years or more), had any insurance, and who had a history of certain high-risk medical diagnoses that could be worsened by TSE, such as reactive airways disease, bronchiolitis, and/or wheeze, were more likely to have documented TSE screening.

Our observed association between a high-risk medical diagnosis and TSE screening mirrors that of another study that found higher levels of screening among pediatric patients with at least one TSE-susceptible medical diagnosis.10 The presence of certain diagnoses may increase a clinician’s likelihood of considering and screening for TSE. We also found that patients who were three years or older were more likely to be screened than younger patients. The relationship between TSE screening and age has been mixed in prior studies, with one study finding higher levels of screening in patients older than two years,10 whereas another study demonstrated higher levels of screening in patients less than one year-old.12 Our results are notable given that younger patients typically have more primary care visits per year and, thus, more screening opportunities. Furthermore, a child’s early life is an especially important time to act given that mothers who quit smoking during pregnancy are susceptible to relapse in the postpartum period19,20 and that mothers may be more motivated to quit smoking while breastfeeding.21 Finally, our finding that patients with any insurance coverage had higher levels of TSE screening adds evidence to an ill-defined relationship in the literature, with one study finding no relationship between TSE screening and insurance status10 and another reporting higher screening levels among underinsured patients.12 Insurance status may impact the frequency of clinic visits and screening opportunities.

Examining health care navigational factors that could impact TSE screening, such as race/ethnicity, language, interpreter use, and foreign country of birth, was a central focus of this study. These factors may shape families’ interactions with the health care system and influence the consistency of TSE screening due to structural inequities and systemic racism in clinical care delivery.6 While our bivariate analyses showed higher levels of TSE screening among White patients relative to non-White or mixed-race patients, meaningful differences did not persist in our adjusted analyses. We also found no statistically significant associations between TSE screening and Hispanic ethnicity or language used during the visit. These results contrast with prior studies that demonstrated higher levels of TSE screening among Black and Hispanic parents relative to White parents,10,12,14 higher levels of screening among Spanish-speaking patients,10,14 and lower screening levels among Haitian Creole-speaking patients.10 Similarly, when we examined how interpreter use and country of birth related to screening levels, we found no statistically significant relationship. The similar levels of TSE screening levels across these health navigational factors in our multivariable analyses are notable, particularly given recent evidence showing lower levels of preventive medical care among non-English speaking families.22 Our findings may indicate that University-led health system efforts to standardize screening approaches in their workflows have likely led to more equitable delivery for patients who require additional linguistic and health navigational support.

Our observed TSE screening levels were generally similar across visit-level factors, including provider, specialty or visit type. Our results contrast with the prior literature showing that pediatricians had higher levels of documenting patient’s TSE in the EHR relative to family medicine physicians,15 though their rates of asking parents about household TSE have been found to be similar,23 and that TSE screening levels were higher during preventive health visits.12

Documented TSE screening levels were substantially higher in this health system than in other studies examining TSE screening levels among pediatric patients.10,11 This provides an example of how suggests previous health system efforts to integrate TSE screening questions into routine workflows can lead to higher screening levels. Further research, however, is needed to define which efforts lead to improved TSE screening and how to improve screening uptake in clinical practice without exacerbating screening disparities.

Our findings further highlight opportunities to strengthen TSE screening among younger pediatric patients and those who are uninsured. Leveraging problem-based visits in addition to preventive care to conduct systematic screening may help to address TSE in this latter group, particularly among minoritized populations who may present less frequently for preventive health care visits.22 EHR-embedded TSE screening and interventions to connect parents to tobacco treatment using clinician decision support (CDS) tools has shown promise as means to integrate screening into routine pediatric primary care.10,24,25 CDS approaches have been proposed as a means of improving equitable TSE screening and treatment among minoritized populations.26 Prior qualitative health systems research has identified potential facilitators to integrating TSE screening into pediatric care for minoritized populations, such as leveraging linguistically and culturally-adapted resources and culturally-congruent interpreters, clinicians, and staff to build trust and reduce stigma.26 Combining both EHR and team workflow changes with other focused efforts for populations with lower levels of health insurance coverage will be important to ensure that these groups do not fall behind as systems aim to improve TSE screening. Finally, TSE screening is the first step in the American Academy of Pediatrics’ clinical practice guideline to reduce childhood TSE. Future research should examine how enhanced and more equitable TSE screening approaches affect parental connection to effective tobacco treatments in diverse populations.

This secondary data analysis has several limitations. First, our outcome variable of documented household TSE screening, defined as timestamped documentation of passive smoke exposure in the patient’s chart or an active issue in the patient’s problem list, may not fully represent the actual household TSE levels during our study period. Several prior studies have used parent recall rather than EHR documentation,12,13,14 which may explain observed outcome differences. Our sample’s low level of positive TSE screens, well below epidemiologic levels of pediatric TSE4 and adult smoking levels in the study catchment area,16 suggest that our outcome measure may be overrepresenting screening levels or that the approach to collecting this information during the study period is not producing accurate results. Second, interpreter use and language use during a clinic visit are rough proxies for limited English proficiency but may not accurately identify patients who experience language-based barriers to care, suggesting that additional studies are needed. Third, we only controlled for variables available in our EHR for pediatric patients and may have missed other unmeasured confounders such as parent education level, generational status, parent age, and smoking status of the parents. Fourth, this data was collected during the COVID-19 pandemic and related changes in clinical workflows in the health system during that time may have influenced our findings. However, it is important to note that the study period occurred after vaccines became more widely available when many routine clinical workflows had begun to stabilize. Finally, our findings are representative of one diverse metropolitan health system and may not generalize to other contexts.

Conclusions

In a large health system with high levels of documented household TSE screening levels, we demonstrate lower levels of TSE screening among patients younger than 3 years of age and among those who are uninsured. These disparities in screening highlight the importance of systematically integrating pediatric TSE screening during routine pediatric primary care visits to ensure that clinicians are identifying and addressing household TSE in all patients. Future health system efforts should consider opportunities to incorporate standardized screening measures that are integrated within existing clinician and staff workflows to more seamlessly identify TSE and connect parents and caregivers who smoke with resources to quit and strategies to reduce TSE in their cars and homes.