Copyright 2005 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2005
The American Academy of Pediatrics (Elk Grove Village, Ill) has recommended that pediatricians assess their patients’ environmental tobacco smoke (ETS) exposure, but the specific questions most likely to identify children with high ETS exposure are not known. Cotinine is a nicotine metabolite, present in hair, that can be used to quantify months of ETS exposure.
To develop a brief screening tool that will accurately predict ETS exposure as defined by child hair cotinine level.
We compared the performance of a series of easily administered screening questions regarding home ETS exposure to child hair cotinine levels. Subjects were a convenience sample of healthy children aged 2 weeks to 3 years of both self-reported smokers and nonsmokers.
Hair samples and questionnaire data were obtained from 291 children. Based on clinical applicability and statistical significance, 3 questions (“Does the mother smoke?”, “Do others smoke?”, and “Do others smoke inside?”) were selected as a valid screening tool to determine children’s ETS exposure risk. Maternal report of smoking outside only or smoking few cigarettes per day had no impact on child hair cotinine levels.
It was possible to derive a simple, specific, and valid screening tool that can be used in pediatric offices to identify children at risk for ETS exposure. Further research is needed to test this tool prospectively.
Tobacco smoke exposure impairs children’s respiratory health.1- 5 Infants exposed to environmental tobacco smoke (ETS) have an increased incidence of pneumonia, bronchiolitis, asthma, and otitis media with effusion, with greater rates of hospitalization and a longer time to recovery from these conditions than unexposed children.6- 13 These findings are most striking for infants and young children exposed to maternal smoking because of the intensity and duration of the exposure.14,15
The American Academy of Pediatrics (Elk Grove Village, Ill) advised physicians that tobacco prevention and control activities should ideally begin at the first pediatric visit.7 Accordingly, they suggested that pediatricians assess tobacco use and ETS exposure in their patient’s extended family and environment, encourage smokers to smoke outside the home and consider quitting, and record this information in the patient’s medical record.7Bright Futures: Guidelines for Health Supervision of Infants, Children, and Adolescents,16 the consensus guidelines on preventive services for children and their families, also includes items prompting pediatricians to advise parents not to smoke in the home or in the car. In a comprehensive analysis of the pediatrician’s role in reducing tobacco exposure in children, Stein et al17 advocated that pediatricians assess factors related to tobacco exposure during all patient contacts, including ETS exposure for infants and children and tobacco use among parents and teens.
Despite these recommendations, simple, specific, and validated screening procedures for ETS exposure in the pediatric office setting have not been developed. For example, asking a simple yes/no question “Do you smoke?” directed at the primary caregiver does not capture information on all smokers in the home or whether smoking occurs in the presence of the children. Screening for ETS exposure needs to be simple and fast in order to be implemented in busy office settings. The specific question(s) most likely to identify children with high ETS exposure are not known.
Children’s tobacco smoke exposure can be assessed both by parental report and by quantification using cotinine, a nicotine metabolite and a biomarker of exposure.18- 21 Hair cotinine is not derived directly from ETS exposure but from absorbed nicotine, which is metabolized to cotinine and then deposited in growing hair.22,23 Hair cotinine is significantly correlated with average daily nicotine intake22,23 and is easy to collect in a clinical setting from infants and toddlers.20,24 Saliva, blood, and urine as specimens for cotinine level determination are used frequently in research but are difficult to collect in a clinical pediatric setting and measure short periods (days) of ETS exposure.25 In contrast, hair cotinine reflects serum cotinine, measures exposure over several months, and is not influenced by short-term exposures to ETS hours or even days prior to hair sampling. Hair nicotine is also used to quantify ETS exposure.26 However, nicotine in hair may be derived from ambient nicotine that adheres to hair. Cotinine in hair is derived only from what an individual has actually inhaled and metabolized, and therefore, it measures actual systemic exposure to nicotine. In a cross-sectional study, elevated hair cotinine levels have been linked to childhood asthma.27
While hair cotinine is an outstanding source for measuring child ETS exposure, it is often available only in academic settings and is expensive. Thus, it is not likely to be adapted as a useful screening tool. Hair cotinine level does, however, represent an objective standard with which questions about child ETS exposure can be compared. We hypothesized that a brief questionnaire could be developed to be used in the clinical setting to determine a child’s ETS exposure risk, using child hair cotinine levels as the gold standard of exposure.
Subjects were a convenience sample of healthy children 2 weeks to 3 years of age recruited from the Columbus Children’s Hospital Primary Care Center (Columbus, Ohio). The Primary Care Center provides primary care to a low-income population, with nearly 85% of patients insured by Medicaid or uninsured. The racial distribution of the Primary Care Center patients is approximately 50% African American, 40% white, and 10% other. The upper age cutoff of 3 years for study eligibility was chosen because children younger than 3 years are those at the greatest risk for the diseases associated with ETS exposure.1,28,29 Children who were recruited for this study were also offered an opportunity to participate in another study establishing normal values for “adult-type” infant pulmonary function tests. For this reason, exclusion criteria (for both studies) were prematurity (<36 weeks’ gestation) and/or a diagnosis of asthma or other chronic pulmonary or cardiac disease.
Data were collected from June 1997 to July 2000. Children of both self-reported smokers and nonsmokers were recruited for the study. Primary caregivers of eligible children filled out a questionnaire that included items on demographics; their (the primary caregiver’s) smoking status; number of cigarettes smoked per day, if applicable; location where smoking occurred; the smoking status of other household members or visitors to the home; and the location where their smoking occurred, if applicable.
Primary caregivers who consented gave samples of their own and their child’s hair for cotinine determination. Ten shafts of hair or more, 2 to 5 cm in length, were cut at the root in the occipital area and were analyzed for cotinine by radioimmunoassay according to the methods described by Eliopoulos et al.22 The lowest sensitivity of the assay was 0.01 ng/mg of hair when 2 mg of hair was used. Hair concentrations of cotinine are expressed as nanogram per milligram of hair. The Columbus Children’s Hospital institutional review board approved of the project and the consenting procedures.
Primary caregivers were asked if they were smokers, defined as smoking 1 cigarette per day in the previous 7 days. Other household exposure was determined by asking the primary caregiver if other smokers live or visit with the child frequently. If so, the relationship of the smoker to the child was noted. Exposure from other household members and/or visitors will be referred to in the text and tables as “others” smoking.
The following variables were created from a set of questions asked in the survey: the location where the mothers or primary caregivers and/or others smoke and the presence of a household smoking ban. Mothers or primary caregivers who were self-reported smokers were asked to rate the 3 locations where they were likely to smoke, in order of frequency. If any location inside the home was selected as 1 of the 3 locations, the mother or primary caregiver was defined as an “indoor smoker.” If only sites outside the home were selected (eg, on the porch, at work, backyard) they were considered an “outdoor smoker.” Similar categorizations were done for the location where other household members or visitors smoked. If neither the mother (or primary caregiver) nor others were categorized as “indoor smokers,” the home was considered to have a “smoking ban.”
Because categorization into discrete exposure groups allows the results to be more clinically applicable, the hair cotinine values were used to empirically divide the children into 3 groups: low exposure (cotinine level <0.3 ng/mg), medium exposure (cotinine level 0.3-0.7 ng/mg), and high exposure (cotinine level >0.7 ng/mg). Univariate analyses comparing each of the categorical demographic and smoking factors with the level of the child’s hair cotinine were conducted using χ2 tests. Single-factor analysis of variance was used to find differences in the continuous factors for the different levels of the child’s hair cotinine. Multivariable, multinomial logistic regression modeling was used to find the most parsimonious subset of factors that best predicted a child’s smoke exposure as estimated by hair cotinine level. Predicted probabilities from this model were used to create a decision tree that health care professionals can use to assess an individual child’s risk.
Mothers or primary caregivers completed 453 questionnaires. Hair samples were collected from 291 children. The reasons for not collecting hair samples from every child were that the mother or primary caregiver refused to consent or the child did not have enough hair to sample. The excluded cases were not different from the included cases in terms of smoking status, race, and income. Children who were not included were on average 3 months younger than the included children (P<.01).
Since the child’s primary caregiver was the child’s mother for 92% of participants, the term mother will be used to refer to all caregivers. Of these, 171 (59%) were self-reported nonsmokers and 120 (41%) were self-reported smokers. For smokers, the mean (SD) number of cigarettes per day was 10.9 (7.6); 96 (80%) were classified as indoor smokers by the method previously described, and 23 (20%) smoked outside exclusively. The mean (SD) age of the mothers was 24.9 (6.09) years; and the children, 1.32 (0.77) years. A demographic description of the population can be found in Table 1.
The actual values for the children’s hair cotinine levels were highly skewed. Median and percentile values are presented. The median child hair cotinine value was 0.58 ng/mg (range, 0-8.18 ng/mg), with the 25th percentile = 0.17 ng/mg and the 75th percentile = 1.39 ng/mg. Ninety-nine children (34%) had low ETS exposure with a hair cotinine level less than 0.3 ng/mg, 68 children (23%) had medium ETS exposure with a hair cotinine level of 0.3-0.7 ng/mg, and 124 children (43%) had high ETS exposure with a hair cotinine level of more than 0.7 ng/mg. Seven subjects (2%) had hair cotinine levels less than 0.01 ng/mg; all of them had mothers who were nonsmokers.
Variables that were tested for association with child hair cotinine level category are shown in Table 2. Higher child hair cotinine level was associated with multiple measures of lower socioeconomic status. Maternal, but not paternal, smoking status was associated with higher child hair cotinine levels; child age and sex had no relationship with hair cotinine levels. African American children were more likely to have medium or high hair cotinine levels than white children. Living in a home where others smoke, where others smoke inside, or with no smoking ban were all associated with higher child hair cotinine levels. The number of cigarettes smoked per day by the mother (if she was a smoker) and the location the mother smoked (inside vs outside the home) had no impact on child hair cotinine level categorization.
The lack of relationship between outdoor vs indoor smoking was examined in more detail in Table 3. The potential exposure categories of the child are delineated into 3 categories, and the medians and the 25th and 75th percentiles for each category are shown for the entire sample, stratified by race. Table 3 demonstrates that there was no significant difference in hair cotinine levels for children who were exposed to maternal indoor vs outdoor smokers. Although African American children had higher hair cotinine levels than white children, the relationships within each racial group for each category remained the same.
Multivariate, multinomial logistic regression modeling was performed to find the set of clinically relevant variables that were best associated with medium- or high-risk ETS exposure. The variables selected for inclusion in the multivariate analysis were those that were statistically significant and deemed by the investigators to be reasonable and quick questions to be asked during a routine well-child visit. Demographic questions, such as family income or maternal education, were not included, since these would not be asked routinely in an office visit. Factors found to be significant in the model were exposure to mother smoking, exposure to others smoking, and where others smoke (inside or outside the home). From this model, a predicted probability of medium or high ETS exposure was calculated, which has been represented in the form of a complete decision tree for health care professionals (Figure 1). Children can be categorized based on the answers to the 3 questions in the model. For example, children of mothers who smoke, who are also exposed to others who smoke inside the home, had an 81% chance of having high ETS exposure. Alternatively, children of self-reported nonsmokers living in homes where others don’t smoke had a 64% chance of low ETS exposure.
Complete decision tree for health care professionals to predict probability of environmental tobacco smoke exposure.
Through a comparison of questionnaire information to the analysis of hair cotinine levels from almost 300 children younger than 3 years, it was possible to develop a 3-question model to determine a child’s ETS exposure risk category. This model was derived from questionnaire and hair cotinine data and has not yet been tested prospectively in clinical practice.
There are several limitations to our methods. The validity of the 3 exposure groups may be questioned since the categorization of child hair cotinine levels was based on an empirical division of the hair cotinine levels of the subjects, not on previously defined norms. However, the divisions used in the current study correspond to the findings of Adair-Bischoff and Suave30 using hair cotinine levels to validate self-report of parental smoking (n = 92). Median hair cotinine levels of 0.3 ng/mg, 0.48 ng/mg, and 0.6 ng/mg corresponded to 0, 1, and 2 parental smokers, respectively. These levels are similar to the cut points of the categories derived in the current study. Additionally, Knight et al27 found a statistically significant difference between the hair cotinine level of children with asthma (mean hair cotinine level, 0.70 ng/mg—the same level as the high-exposure category in the current study) vs control children (mean hair cotinine level, 0.39 ng/mg). Thus, we have reason to believe that the exposure categories defined herein are meaningful.
Our research definition of indoor vs outdoor smokers was more stringent than responses typically obtained in a clinical setting. A smoker who admitted to any indoor smoking, even if it was the least likely place where smoking occurred, was considered an indoor smoker for the purposes of this study. An outside smoker was narrowly defined as someone who never admitted to any indoor smoking. Because of this method of categorization, relatively few women (23) were included in the outside smoker group. Therefore, the power of the study to detect differences in hair cotinine levels in the children of inside vs outside smokers was less than desired (63%). However, in office-based pediatric settings, many women who would be labeled as inside smokers by our strict research definition would describe themselves as outside smokers. For this reason, we believe that our conclusion can be generalized to real-life clinical pediatric practice.
Our findings are applicable to children from low-income homes at highest risk for ETS-associated morbidity because of their young age.1,28,29 By limiting our sample to children without chronic cardiac and respiratory diseases, our population is closer to a community-based cohort than it would have been if children with these conditions had been included.
This report neither proves nor disproves the advisability of public health efforts such as the Environmental Protection Agency’s suggestion that parents of children with asthma “take it outside” (smoke outdoors) since we did not measure the impact of mothers actually smoking outside on child hair cotinine level. Specifically, we found that maternal report of their own smoking location in the context of pediatric care was not related to child hair cotinine levels.
Our finding that maternal report of their own smoking location was not related to child hair cotinine level was different from that of Matt et al31 who reported that maternal report of number of cigarettes smoked in the presence of a child correlated well with the child’s urine cotinine and ambient air nicotine levels. However, Matt et al studied a self-selected group of low-income women who volunteered to learn more about health habits, including smoking, and who were aware that air monitors for nicotine would be placed in their homes. All of these factors may have changed their smoking behavior and their reporting of such behavior. Additionally, urine cotinine is a short-term measure of ETS exposure, so measurements reflected behaviors while the research was ongoing. In the current study, parents had no opportunity to change their smoking behavior or reporting since the questionnaire and hair specimen were obtained at the same time. Therefore, these results of these studies are not comparable.
However, other research supports our finding regarding location where smoking took place. Al-Delaimy et al32 reported that location where smoking occurred (indoor vs outdoor) had no impact on child hair nicotine levels. We speculate that the social desirability of reporting outdoor smoking is the reason for our finding that maternal report of where smoking occurred was not related to child hair cotinine levels. In a study of self-reported modifications of smoking behavior, Winkelstein et al33 reported that 25 of 29 parents endorsed “outside-only smoking” along with mutually exclusive categories such as opening windows or smoking in another room. Social desirability of reporting to be a nonsmoker occurs frequently in the context of prenatal visits, where the misclassification rate regarding regular tobacco use varies between 5% and 25% depending on the setting.34- 36 Maternal report of cigarettes per day and indoor vs outdoor smoking may be similarly unreliable in the context of pediatric care.
Based on our data, our practical clinical recommendations are as follows: health care professionals can obtain an ETS risk assessment for a child younger than 3 years by asking the mother “Do you smoke?” Information spontaneously offered by the mother at that point—that she smokes outdoors or smokes few cigarettes per day—is not relevant to the child’s ETS risk categorization. If the mother reports that she is a nonsmoker, then 2 more questions can be asked at that point: “Do others who live or frequently visit with you smoke?” and “Do they smoke indoors?” In a busy office setting, simplifying the model even further may be preferable to the health care professional. This can be done by merging the medium- and high-exposure categories. A graphic depiction of this simplified model is shown in Figure 2. If the mother is a nonsmoker, children with medium and high ETS exposure can be identified by simply asking if smokers live or frequently visit the home.
Simplified decision tree for health care professionals to predict minimum probability of environmental tobacco smoke exposure.
In conclusion, it was possible to develop a simple screening tool to be used in the office setting to define children at highest risk for ETS exposure. Although in clinical practice, mothers frequently report smoking few cigarettes per day and smoking outside, these factors did not prove to be significant determinants of their child’s exposure. Further research is needed to test and refine this tool in the practice setting.
Correspondence: Judith A. Groner, MD, Section of Ambulatory Pediatrics, Department of Pediatrics, The Ohio State University College of Medicine & Public Health, 700 Children’s Dr, Columbus, OH 43205 (firstname.lastname@example.org).
Accepted for Publication: July 22, 2004.
Funding/Support: This study was funded by the American Lung Association, New York, NY, and Columbus Children’s Hospital Research Institute, Columbus, Ohio.
Groner JA, Hoshaw-Woodard S, Koren G, Klein J, Castile R. Screening for Children’s Exposure to Environmental Tobacco Smoke in a Pediatric Primary Care Setting. Arch Pediatr Adolesc Med. 2005;159(5):450-455. doi:10.1001/archpedi.159.5.450