INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Passive smoking in childhood poses an important risk factor for respiratory disorders such as bronchial asthma. Because bronchial hyperresponsiveness (BHR) is a hallmark of childhood asthma, a number of studies have addressed hyperreactivity in order to investigate the relevance of environmental tobacco smoke (ETS) to asthma. There are numerous procedures for ascertaining BHR, and most of them have resulted in similar ranges of sensitivity and specificity regarding clinical asthma (1). In most of these studies, passive smoke exposure of the population was ascertained by questionnaires (4, 5). However, natural exposure can vary from day to day (6), and in tests hyperreactivity has been shown to be inconsistently present even in people with asthma (7). Thus, ascertaining exposure and its effects within a narrow time window might be a prerequisite to detecting short-term consequences. Experimental ETS exposure methods have been applied in the case of asthma (8). However, in a large population with a healthy subpopulation, this approach is not feasible. In some epidemiologic studies on respiratory effects of passive smoking in children, cotinine excretion has been used to estimate the relevant exposure (6, 9, 10). Yet only a few of these studies compared the objective measurement of exposure with the objective measurement of bronchial responsiveness tested on the same occasion (11). Our paper presents cross-sectional and longitudinal analyses of the relationship between urine cotinine excretion and variability of peak expiratory flow rate (PEFR) in a community-based sample of school-age children.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study population. Data was collected in a longitudinal study designed to assess risk factors for allergy and asthma in a cohort of community-based children (12). All parents in three predominantly urban areas of southwestern Germany (Freiburg, Lörrach/Weil, Kehl) whose children started school in 1989 (n = 2,604) were invited to participate, and of these 1,812 took part (70%). The mean age at the beginning of the study was 7.3 yr (90% interval, 6.7-8.1 yr).

Questionnaires were distributed and completed by the parents between February and March 1991, and in the same months in 1992. In the two winters, homes were visited, questionnaires and mini-Wright peak flow meters were distributed, and they were later collected together with spot urine samples for cotinine analysis. The study groups for this analysis were restricted by two requirements: first, access to private homes (1990-1991, n = 1,303; 1991-1992, n = 1,206) and second, completed peak flow protocols and resources for cotinine analysis. Due to the latter condition, a random sample of children was chosen for in-home visits (n = 417), which formed the actual study group. The study was approved by the local ethical committee and also by local school authorities.

Questionnaires. A history of doctor's diagnosis of asthma and recurrent wheezy bronchitis (asthma), of asthmatic complaints in the preceding 12 mo (current asthma), and of parental atopy (maternal or paternal history of asthma, hay fever, or eczema) was ascertained in the first and second questionnaires. Current smoking by the mother, the father, or another person within the home was also recorded on both questionnaires. The first inquired additionally about educational level (highest level achieved at school by father or mother), maternal smoking habits during pregnancy, and the gender of the child.

Collection of urine samples and data on peak expiratory flow rate. Visits to homes took place between December and April, 1990- 1991 and 1991-1992. On both occasions, relative humidity in the child's bedroom was measured (Testoterm; Testoterm, Lenzkirch, Germany). At the beginning of both visits, children and parents were instructed in the correct use of the mini-Wright peak flow meter. Tests were performed in a standing position. Parents were asked to record twice daily (7:00 A.M.-9:00 A.M. and 4:00 P.M.-7:00 P.M.) the best of three readings their children achieved. Measurements were recorded on a sheet that reinforced the instructions. The importance of the correct time for measurement was emphasized, and parents were encouraged to insert blanks rather than inaccurate data. Peak expiratory flow rates (PEFR) were recorded over seven consecutive days in 1990-1991 and over 14 consecutive days in 1991-1992. The parents were also asked to report on their child's coughing or shortness of breath during the week (current respiratory symptoms). At the end of this observation period, the peak flow meters were collected and morning urine samples were taken.

Cotinine and creatinine measurement. Cotinine was quantified by gas chromatography/mass spectrometry using the selected ion monitoring mode adapted from the method described by Jacob and colleagues (13). An HP5890 gas chromatograph with automatic sample injection, a split-splitless capillary inlet system, and a selective mass detector (HP5971; Hewlett-Packard, Böblingen, Germany) was used. The method of measuring is described in detail elsewhere (14). Data was stored and processed using an HP Vectra 486/33N (Hewlett-Packard) and an MS Chem Station HPG103 C (Hewlett-Packard). To 1 ml of urine, 57.1 ng (0.2 nmol) internal standard (D₄ cotinine [(S)-cotinine-4',4'-d₂ perchlorate] in distilled H₂O; D₄ cotinine kindly supplied by Dr. Peyton Jacob), about 0.1 g sodium chloride and 1 M NaOH were added to achieve pH 12. Two mass fragments (cotinine, 176; internal standard, 180) were monitored and quantification was achieved by integration of the ion chromatograms.

Using pooled urine sampled from three nonsmoking donors as solvent, a seven-point serial dilution standard curve (linear from 0.88 to 70 ng ml¹) was constructed by plotting peak area ratio 176/180 versus concentration of cotinine. The equation for the standard curve was y = 0.010773 + 0.035075x. Intra-assay variation was 1.5% in samples with 70 ng/ml (53 ng/ml, 4.3%; 35 ng/ml, 4.1%; 18 ng/ml, 3.1%; 7 ng/ml, 5.5%; 4 ng/ml, 16%; 2 ng/ml, 129%; 0.88 ng/ml, 96%). Cotinine was detected in 19 of 20 control samples with a concentration of 3.5 ng/ml (estimated lower limit of sensitivity). Since ion monitoring in 27 control samples without cotinine revealed only one value less than 0.06, urine specimens with a ratio 176/180  0.06 were put at zero (~ 1.4 ng/ml = specificity of the assay). All values exceeding 0.06 were related to the standard curve. Urinary creatinine concentrations were determined by the Jaffé reaction (15). Urine cotinine excretion (UCE) was expressed as the ratio of cotinine to creatinine. All assays were performed without knowledge of smoking habits.

Statistical analysis. The percentage ratio of the amplitude over the mean of daily PEFR (higher value  lower value)/mean PEFR × 100) was calculated for each day as a standard measure of the variability of the PEFR (16). For each survey, diurnal variability of PEFR was expressed as the average amplitude over the mean (AVAM) for the observation time of those children who had measured their PEFR twice daily for at least 5 d. Median, 5th and 95th percentiles of AVAM, as well as the number of children exceeding the 90th percentile, were calculated and stratified for variables of interest. A linear regression analysis of the relation of UCE and log₁₀AVAM controlling for gender, current respiratory symptoms, and relative humidity was performed for each survey. A pooled regression analysis for both surveys was carried out using the method proposed by Liang and Zeger (17) and employing the mixed procedure of SAS (18). For the repeated measurements, an unstructured residual covariance matrix had to be assumed. Linearity of the statistical relationship was checked using dummy variables for UCE. So that results between surveys can be compared, analyses are presented for those children for whom AVAM and UCE were available in both surveys.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study population comprised 417 children with data from questionnaires, urine samples, and PEFR documentation for both years. The prevalence of important variables does not differ substantially from the subpopulation that must be excluded (n = 1,395) due to a lack of cotinine measurements (Table 1). However, a high educational level of the parents is found more frequently in the population analyzed, owing to a higher proportion of access to private homes. In the total population (n = 1,812), passive smoking in the last year was reported in 54% of families with low educational level, in 47% of families with medium educational level, and in 37% of families with high educational level. In both surveys, increased AVAM (> 90th percentile) was significantly associated with current asthma and current respiratory symptoms (Table 2). Ten children showed increased AVAM in both testings.

The measurement of cotinine was higher in children of parents who smoked (Survey 1: median = 3.2 ng/mg, 5th-95th percentile = 0-24.0 ng/mg; Survey 2: median = 2.9 ng/mg, 5th-95th percentile = 0-26.5 ng/mg) than in children of nonsmoking parents (Survey 1: median = 0, 5th-95th percentile = 0/4.4 ng/mg; Survey 2: median = 0, 5th-95th percentile = 0- 6.6 ng/mg). The median value for UCE of children with detectable cotinine in urine was 4.94 ng/mg (n = 148) in the first and 4.44 ng/mg in the second survey (n = 188). In the group of children with UCE at or above the median, AVAM was increased (Table 2). With regard to gender, no difference in median AVAM existed. However, girls had a higher proportion of increased AVAM values (90th percentile and more) than boys: 10.9 versus 6.8% in the first and 10.0 versus 7.8% in the second survey.

With increasing cotinine excretion, the log₁₀AVAM increased in the first and second survey as well as in the longitudinal analysis (Table 3), being of borderline significance for the first and statistically significant for the second survey and the longitudinal data set. Stratifying the longitudinal model for maternal smoking during pregnancy did not reveal a combined effect with UCE on log₁₀AVAM (no maternal smoking during pregnancy: estimate = 0.0028, p = 0.056, n = 358). A pronounced association was obvious in the stratum with no current respiratory symptoms (estimate = 0.003, p = 0.004; Table 4). A higher estimate of UCE occurred for asthmatics; however, statistical significance could not be shown (Table 4). The effect of UCE was only detected in boys (estimate in Survey 1: 0.005, p = 0.022; Survey 2: 0.0045, p = 0.078; longitudinally: 0.005, p = 0.0001) and not in girls (estimate in Survey 1: 0, p = 0.98; Survey 2: 0.001, p = 0.57; longitudinally: 0.001, p = 0.31; Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This longitudinal epidemiologic study was designed to assess risk factors for respiratory disorders in childhood (12). Objective measurements of BHR and ETS exposure were used. The results showed that cotinine excretion following a 7-14-d period of peak flow observation was associated with peak flow variability. The narrow time window for the measurement of variable exposure and variable effect facilitated the detection of a covariation of both indicators. For the subsample with cotinine measurements and peak flow ascertainment, a selection bias with regard to the educational level of the parents cannot be excluded. Parents with less education were less willing to give access to their private homes. In these families, parental smoking was more common. Thus, in the analyzed sample the proportion of parental smoking was also slightly reduced (Table 1). Because we did not know the diurnal peak flow variation in children of the excluded sample, it remained unclear whether the selection of fewer exposed individuals was relevant to the relationship between UCE and log₁₀AVAM. No indication of a selection bias was seen with regard to other important variables.

The measurement of cotinine was in accordance with data on parental smoking within the previous 12 mo (14). Inconsistencies were infrequent and could potentially be explained by the misreporting of parental smoking, by sporadic changes of exposure before urine sampling, and by measurement error regarding the cotinine assay. The method of cotinine measurement we used, however, has been shown to be reliable (13). Urine was collected at the end of the PEFR protocol. However, UCE represented an integrated average of ETS exposure over the previous days and was found to show little variation within a period of 1 mo (10). Reproducible levels of UCE over 1 yr merely indicated constant exposure conditions in our population (14). Thus, we assume that UCE may give information on ETS exposure also during peak flow ascertainment.

The measurement of AVAM in our study served as an indicator of peak flow variability (PEFV), which is associated with bronchial asthma in children (19). Furthermore, PEFV corresponds to BHR measured by pharmacological challenge tests (20, 21). Peak expiratory flow is informative only with regard to large airway size and lacks sensitivity in detecting small airway obstruction, which is particularly a disadvantage in a clinical setting. On the other hand, measurement of diurnal PEFV is possible in large numbers of subjects simultaneously, and it can therefore deliver important information. Peak flow variability can also occur in nonasthmatic children as an indicator of variation in the function of the central airways. It is not quite clear whether increased PEFV in nonasthmatics is of prognostic value. However, in a recent study, initial asymptomatic BHR was associated with an increase in airway responsiveness and with the development of asthma symptoms during a 3-yr follow-up (22). Even in nonasthmatic children with BHR, a reduction in FEV₁ was found in consecutive years (23). Because PEFV is a measure of bronchial responsiveness, an elevated diurnal peak flow variability must be regarded as a risk factor for impaired airway function later in life.

There is growing evidence that ETS early in life is a risk factor for asthma (24), reduced lung function (25, 26), and BHR in asthmatics (24). However, the relative importance of pre- and post-natal ETS is not sufficiently understood (25). Risk assessment regarding prenatal ETS exposure requires an observation period starting in pregnancy or earlier (27). Postnatal effects of ETS can be studied separately if the response variable, here PEFV, shows variation with time. This was the case in the current study, since AVAM values in both surveys were only slightly correlated (r_s = 0.15) and only a quarter of children with increased AVAM (> 90th percentile) in Survey 1 were also reactive in Survey 2. Thus, the effect of UCE in the two surveys did not refer to individual responsiveness of the same subpopulation. Furthermore, a postnatal effect of ETS on PEFV was suggested, since a separate analysis of children whose mothers did not smoke during pregnancy did not alter the results.

Decreased levels of pulmonary function have been reported when children are postnatally exposed to ETS (25, 28). In contrast, Cunningham and colleagues found no effect of postnatal ETS exposure on airway function in a large population aged 8-12 yr, although a clear association of prenatal passive smoke exposure and decreased lung function at rest could be demonstrated (26). Moreover, no effect of postnatal ETS exposure during the first 2 yr of life on changes in expiratory flow-volume rates could be shown (27). We report evidence for BHR due to postnatal ETS exposure, and it seems reasonable that prenatal exposure can cause impaired lung function at rest and that postnatal exposure can cause BHR. On the other hand, BHR can precede airflow obstruction at school age and must therefore be considered a risk factor for chronic disorders (23).

Besides susceptibility to ETS effects depending on age, bronchial asthma seems to predispose children to adverse effects from ETS. Even in children without current asthmatic symptoms, we found a significantly positive relationship between UCE and log₁₀AVAM (Table 4). Although we could not demonstrate a statistically significant association in our small subsample of children with asthma, a higher estimate value of UCE could be a hint that a stronger ETS effect exists in those children. With regard to this question, O'Connor and coworkers (25) found an effect of ETS on lung function at rest throughout the total study population, whereas the association of ETS and BHR could be found mainly in the asthmatic subpopulations. The latter association has also been reported in other investigations (29). In a community-based sample from Italy, Martinez and coworkers found an association of any smoking by parents and BHR to carbachol (30). Moreover, Casale and coworkers found an increased PEF rhythm amplitude in a small group of schoolchildren exposed to passive smoking (31). Measuring the internal dose of ETS, our data support these findings.

Finally, the ETS effect can also be modified by the child's gender. A stronger effect of prenatal ETS exposure on the lung function of male schoolchildren was demonstrated by Cunningham and colleagues (32), whereas an earlier study was unable to detect such a difference in infancy or a stronger effect in girls (33). It is possible that gender differences in the effects of early ETS exposure on lung function could occur later in life, since mechanical properties and postnatal growth and maturation may differ between boys and girls (34). Compared with girls of school age, boys starting with low lung function have shown a definite reduction in FEV₁ with age when exposed to parental smoking (35). In addition to pronounced ETS effects on the lung function of males, a stronger association with airway hyperresponsiveness has been reported (24). Current knowledge is not able to explain the susceptibility of boys to ETS effects. Because the causes of susceptibility could be important for prevention, studies on the reasons for the gender differences in childhood are needed.

In summary, we employed objective methods for the measurement of ETS exposure and airway reactivity in a community-based sample of children and were able to demonstrate an association of postnatal passive smoking and bronchial responsiveness. Because bronchial hyperresponsiveness has been shown to be a risk factor for chronic disorders, an increased diurnal peak flow variability due to ETS exposure may be interpreted as early sign of chronic airway damage. Our data also suggests that the enormous incidence rates of asthmatic episodes in children throughout the world might at least partly be a consequence of ETS exposure. To prevent airway disease, it is essential to put available strategies against passive smoking into action and to support the optimization of these strategies.