INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: paternal smoking; environmental tobacco smoke; pulmonary function; children; China

Maternal smoking has been associated with reduced pulmonary function in children (1, 2). Studies suggest that pulmonary function decrement in school-aged children is a result of combined early life (including in utero) and current exposures to maternal smoking (3, 4). However, studies on the effects of paternal smoking on children's pulmonary function have yielded inconsistent results. Studies in the United States (5), Great Britain (6), and Australia (7) found that decreased FEF_25-75% in children was associated with maternal smoking, but not paternal smoking. In contrast, studies in China (8) and Turkey (9) found associations of paternal smoking with decrements in children's FEF_25-75%.

Quantification of the effects of paternal smoking on children's pulmonary function is difficult, in part because children may be exposed to maternal smoking both prenatally and postnatally. The relative contributions of paternal versus maternal smoking to a child's total exposure vary with the amount of time each parent spends with the child, the number of cigarettes each parent smokes, and the degree to which each parent smokes inside or outside the home. Misclassification of exposure and confounding by maternal smoking may be important reasons for inconsistent results in studies of the effects of paternal smoking on children's pulmonary function.

This report is part of a large-scale study on environmental and genetic determinants of asthma in rural communities in China (10). The purpose was to investigate the effects of paternal smoking on the lung function of children. This rural population possessed unique advantages for evaluating the effects of paternal smoking on children's pulmonary function. Whereas approximately 75% of fathers were current smokers, few mothers had ever smoked. There was no major ambient air pollution and the types of homes were similar in these rural communities. A small proportion of households used coal for heating or cooking, a source of indoor air pollution.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Site and Population

This study was conducted in rural counties of Anqing, China, located on the northern bank of the Yangtze River in Anhui Province. Our protocols were approved by the Brigham and Women's Hospital and Harvard School of Public Health Institutional Review Boards for Human Studies and informed written consent was obtained from each subject. Asthma index families were identified by local physicians based on the following criteria: the presence of an asthma patient who was at least 8 yr old; availability of both parents for the study; the presence of at least one sibling, at least 8 yr old; and no more than one parent with asthma. Diagnosis of asthma was made by local physicians according to the following criteria: the patient had a history of repeated onset of wheeze with dyspnea, but was without symptoms between two events; and the obstructive symptoms could be improved significantly by a bronchodilator.

Our population of children was obtained from the asthma index families using the following inclusion criteria: (1) age  8 and  15 yr; (2) child had never smoked; (3) mother had never smoked; (4) siblings  20 yr of age had never smoked; (5) coal was not used for either heating or cooking in the home. Children with formerly (but not currently) smoking fathers were also excluded from the analysis. Of 1,904 children age 8 to 15 yr who were available from the asthma index families, 1,718 met our inclusion criteria (860 girls and 858 boys).

Procedures

Detailed descriptions of the data collection methods have been reported elsewhere (10, 11). Briefly, the survey was conducted between 1995 and 1998 by a team comprised of researchers from Anhui Medical University and the Anqing health institutes and locally hired interviewers who were fluent in the local dialect. All local physicians were polled, and an attempt was made to identify all asthma index families in each selected rural community. Following the guidelines of the National Institutes of Health (NIH) Collaborative Agreement on Asthma Genetics, pulmonary function tests (spirometry) were performed and a standardized questionnaire (modified American Thoracic Society- Division of Lung Disease [ATS-DLD]) was administered assessing respiratory history and symptoms, occupational and smoking histories, home environment, family history of asthma, and other chronic diseases. Weight and height were measured after subjects had removed their shoes and outerwear. Height was measured to the nearest 0.1 cm using a portable stadiometer. Weight was measured to the nearest 0.1 kg with the subject standing motionless on the scale.

Pulmonary Function Tests

Standardized pulmonary function tests were performed with the subject seated while wearing a nose clip and using equipment (Schiller, Dietikon, Switzerland) that conformed to the "Snowbird" guidelines set forth in the ATS statements on the standardization of spirometry (12, 13). The best FEV₁ and FVC from any acceptable curve were used for analysis. An acceptable FVC had a minimum duration of 6 s, measurement within 5% or 200 ml of other measurements (whichever was less), and judgment by a trained technician as an adequate maneuver.

Statistical Methods

The outcome variables in this study were FEV₁ and FVC. Previous studies have shown that the major determinants of pulmonary function are age, sex, and body size (14). It is noted that during the adolescent growth spurt, the relationship between height and pulmonary function deviates from a linear relationship (15, 16). This poses a challenge in choosing an appropriate model for pulmonary function during puberty. We evaluated various modeling approaches by examining the functional relationship between pulmonary function and important covariates and by comparing the squared Pearson correlation coefficients (r²). Our final model included sex, age, weight, height, and square of height. We also included father's education and child's asthma status in the models.

The effects of paternal smoking on children's FEV₁ and FVC were estimated using multiple linear regression analysis, with adjustment for important confounders. Because our population of children included siblings from the same families, we adjusted for intraclass correlation within families using generalized estimation equations (GEE) with SAS procedure GENMOD assuming an exchangeable correlation matrix structure (17). To further investigate the exposure-response relationship between paternal smoking amount and FEV₁ and FVC deficit in children, children of smoking fathers were subdivided into two groups: father smoked < 30 cigarettes/day and  30 cigarettes/ day. We compared each subgroup with children of never-smoking fathers (a reference group). Our analyses were also stratified by children's sex and asthma status to examine potential interactions between paternal smoking and these variables. To be comparable with previous studies, we presented effect estimates both in terms of absolute reduction (ml) and percentage change in pulmonary function.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This report includes a total of 1,718 children ranging in age from 8 to 15 yr. The characteristics of these children by father's smoking status are presented in Table 1. Compared with children of nonsmoking fathers, children of smoking fathers were slightly older, taller, heavier, had slightly higher values of FEV₁ and FVC, and had a higher prevalence of asthma.

When children's age, sex, weight, height, asthma status, and father's education were adjusted in the regression analysis, children of smoking fathers showed small, but detectable deficits in FEV₁ (36 ml, SE = 20, p = 0.075, or 1.8% change) and FVC (37 ml, SE = 22, p = 0.089, or 1.6% change). As shown in Table 2, when children of smoking fathers were divided into two subgroups (father smoked < 30 cigarettes/day and  30 cigarettes/day), we found that children whose fathers smoked  30 cigarettes/day had the largest deficits in both FEV₁ (79 ml, SE = 30, p = 0.009, or 4.0% change) and FVC (71 ml, SE = 34, p = 0.036, or 3.0% change). As shown in Tables 3 and 4, such a monotonic exposure-response relationship (consistently decreasing mean pulmonary function with increasing exposure) remained in all strata when we further stratified our analyses by children's sex and asthma status. We also examined the effects of paternal smoking on the ratio of FEV₁ to FVC, but did not find an important association (data not shown).

As shown in Tables 3 and 4, our data also suggested that the relationship appeared greatest among nonasthmatic girls for both FEV₁ ( 141 ml, SE = 49, p = 0.004, or  7.0% change) and FVC ( 161 ml, SE = 51, p = 0.002, or  6.9% change). We included interaction terms between paternal smoking and sex in the regression models, but found the interaction terms were not statistically significant for either FEV₁ or FVC (data not shown). A similar analysis was performed to test for heterogeneity of effects by asthma status, but no evidence was found of interaction between paternal smoking and asthma (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Studies of the effects of paternal smoking on children's pulmonary function have been inconsistent. This large cross-sectional study in rural Chinese communities demonstrated that paternal smoking was associated with deficits in FEV₁ and FVC of these fathers' children. We also found that children whose fathers smoked more cigarettes had larger deficits in both FEV₁ and FVC than either children whose fathers smoked less or children with nonsmoking fathers. Such a monotonic exposure-response relationship (consistently decreasing mean pulmonary function with increasing exposure) remained in all strata when we further stratified our analysis by children's sex and asthma status.

The low prevalence of maternal smoking in this population offered a unique opportunity to study the effects of paternal smoking in families with never-smoking mothers. When both mothers and fathers smoke, the effects of exposure to maternal smoking might make discrimination of paternal smoking effects more difficult. Maternal smoking during pregnancy results in fetal exposures to the constituents of tobacco smoke that are at least 20 times higher than exposure levels resulting from environmental tobacco smoke (ETS) (18). After birth, exposures to maternal smoking might be more frequent than exposures to paternal smoking if the mother spends more time with the child than does the father. In North America, Europe, and Australia where the prevalence of smoking among women is higher than in rural China, studies have included families in which the mother smoked (5). Because maternal smoking is a major risk factor for pulmonary function effects in children and difficult to quantify precisely, confounding by maternal smoking might be an important reason these studies have not detected an effect of paternal smoking on the pulmonary function of children. Relative to these Western countries, the number of cigarettes smoked by fathers in China might also more accurately reflect the exposures of their children. There are no cultural prohibitions against smoking indoors in these rural Chinese communities whereas a portion of men in North America, Europe, and Australia might step outdoors to smoke.

The effect estimates from our study are comparable to those from previous studies on parental or maternal smoking and children's FEV₁ (1, 19, 20). The results of previous studies on FVC have been inconsistent. Wang and colleagues (3) and Ware and colleagues (19) found significant associations between maternal smoking and increased FVC. In contrast, Gharaibeh (21) reported a 13.4% decrease in FVC in children who lived with at least one adult who smoked  1 pack/day.

Our analysis did not reveal a significant association between exposure to paternal smoking and a decrease in the ratio of FEV₁ to FVC. This finding suggests that the negative impact of paternal smoking was on childhood lung growth rather than airflow obstruction. Additionally, exposure to paternal smoking did not modify the functional relationship between height and age (data not shown). This suggests that the observed decrease in pulmonary function represented a lung-specific effect rather than a broader effect on somatic growth.

In contrast to our results, other studies have reported a significant decrease in the ratio of FEV₁ to FVC, indicating an obstructive effect of exposure to ETS (3, 22, 23). The discrepancies between these and our results possibly indicate dissimilar effects of ETS at different stages of childhood growth. Wang and colleagues (3) and Demissie and coworkers (23) both studied North American children, many of whom were likely to have been exposed to maternal smoking in utero. Statistically significant associations were found in both studies between exposure to ETS and increased FVC. Corbo and colleagues (22) studied Italian children living in households in which there were no current smokers, but who were exposed outside the home in an environment with high prevalence of smoking and few restrictions on smoking in public places. Although some parents might have smoked previously, it is likely that many of these children were exposed neither to ETS in the home during childhood nor to maternal smoking in utero. No associations were found in this study between exposure to ETS and changes in FVC. In contrast, children in our study were not exposed to maternal smoking in utero, but were exposed during childhood to father's smoking at home.

Previous studies have also reported differences between boys and girls in the effects of parental smoking on pulmonary function. Although most studies that allow a comparison of boys and girls have found that the effect is stronger in boys (24, 25), some studies have shown a stronger effect in girls (26, 27). A very large longitudinal study in six cities in the United States found no evidence of significant differences between boys and girls in the effect of maternal smoking on lung function (3). Using meta-analysis with a random effects model, Cook and coworkers (1) concluded the literature provided no statistically significant evidence (p = 0.06) of a difference between boys and girls in the effect of parental smoking on FEV₁ in children. In children without asthma, our analysis showed a greater effect among girls, whereas in those with asthma the effect was greater in boys. Although our data suggest the relationship was greatest among nonasthmatic girls, we did not detect a statistically significant interaction between paternal smoking and sex or between paternal smoking and asthma status.

Our study had the following methodological limitations. First, paternal smoking was measured by self-report. We analyzed paternal smoking as a categorical variable rather than a continuous variable to overcome potential digit preference in self-reporting. This cross-sectional analysis did not permit investigation of cumulative exposure to paternal smoking in relation to children's pulmonary function. Our data also did not permit investigation of the relative effects of prenatal versus postnatal exposure to paternal smoking. Because our children were obtained from asthma index families, they might represent a population with increased sensitivity to ETS. Therefore, we do not know if our findings are representative of general populations. Furthermore, although we excluded those families which used coal for cooking or heating, we were unable to determine the extent of indoor air pollution in these rural communities.

Because over 60% of men smoke and 70% of the population reside in rural areas (28), the effects observed in this study are likely to be widespread in China. In this population, little is known about the extent to which the effects of ETS exposure in childhood persist into adulthood. Longitudinal studies in the United States and Canada have shown that exposures to maternal smoking in utero and the first 5 yr of life cause persistent lung function deficits that are independent of later exposures and detectable in children 6 to 18 yr of age (3, 4). In China, both active and passive smoking in adulthood have been associated with losses in pulmonary function (29, 30). Unless the high prevalence of smoking among Chinese men decreases, most male children will become smokers in adulthood and adult females will be widely exposed to ETS. Longitudinal studies are needed to determine the lifelong impact of childhood exposure to ETS in a population with high prevalence of tobacco use among men. From a clinical perspective, the effect of paternal smoking on children's pulmonary function is relatively small. However, any assessment of the overall impact of these findings must recognize that for most Chinese citizens, these effects are only some of the early manifestations in a lifetime of exposure to tobacco smoke.