INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A plethora of names (wheezy bronchitis, bronchiolitis, cough variant asthma, recurrent wheeze or just simply "asthma") have been used in the past to describe wheezing-associated respiratory illnesses in early childhood (1). In the last few years, prospective studies have clarified that wheezing in young children is not a single disorder (2, 3). Martinez and colleagues (3) found evidence that, in most cases, infant wheeze is a transient condition, not associated with increased risk of asthma or allergies later in life. In these infants, a reduced airway caliber seems to be the predisposing factor to wheeze in association with viral infections. Children who continued to wheeze up to school age were more likely to have developed atopy and to have abnormal lung function tests at 6 yr of life than children with transient early wheezing or no wheezing at all. A third group of wheezing children did not have wheezing in the first years of life but started to wheeze later. This late-onset pattern of illness was also likely to be associated with atopy.

Different factors could play a role in determining the onset and the outcome of wheezing, but the relationships between genetic, pre-, peri-, and postnatal factors and wheezing are not fully understood. In particular, the recent challenging hypothesis that infections acquired early in childhood might protect against, rather than promote, allergic sensitization and possibly atopic asthma is worth further evaluation (4). We therefore studied factors possibly related to wheezing in a large multicenter cross-sectional population-based study. The enrolled children were divided using criteria derived from Martinez and coworkers (3) into transient early wheezers, persistent wheezers, and late-onset wheezers. Because it is becoming increasingly clear that these wheezing conditions have a different pathogenesis, we hypothesized that they also have different risk factors.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

The children under study were enrolled in the SIDRIA (Italian Studies of Respiratory Disorders in Chidhood and the Environment) project, an extension of the ISAAC (International Study on Asthma and Allergies in Childhood) study (9). Full details of the SIDRIA study have been reported elsewhere (10). Briefly, the study was carried out between November 1994 and January 1995, outside the main allergen seasons, in eight areas of northern and central Italy and involved a large sample of children of two age groups (6- to 7-yr-old and 13- to 14-yr-old). In this paper we report data on 6- to 7-yr-old children, for whom we thought there was a better recalling of events that had occurred early in life. Standardized questionnaires, including ISAAC core modules for asthma, rhinitis, and eczema (11), were filled in by parents. The validity of some key items of the health section of the questionnaire was evaluated in a pilot study.

Definitions

Of the children enrolled in the SIDRIA study, four mutually exclusive groups were included in the present analysis: (1) those who had had at least one lower respiratory tract illness with wheezing (bronchiolitis or asthmatic bronchitis or asthma) in the first 2 yr of life, and had no wheezing in the last 12 mo (children with transient early wheezing); (2) those who had had at least one lower respiratory tract illness with wheezing in the first 2 yr of life, and had wheezing in the last 12 mo (children with persistent wheezing); (3) those with no lower respiratory tract illness with wheezing during the first 2 yr of life but who had wheezing in the last 12 mo (children with wheezing of late onset); (4) those with no lower respiratory tract illness with wheezing during the first 2 yr of life, no asthma up to 6 to 7 yr of age, and no wheezing in the last 12 mo (control group). Children for whom data on respiratory diseases in the first to second year of life or on wheezing symptoms in the last 12 mo were lacking were excluded from the analysis. This group differed from the others in its lower degree of paternal education.

A number of potential risk factors for wheeze were examined. Familial variables included maternal and paternal history of hay fever, asthma, and chronic obstructive pulmonary disease (COPD), reported as chronic bronchitis or emphysema. Pre- and perinatal variables included birth weight (< 2,500 g versus  2,500 g), maternal age at the time of the child's birth (a priori categorized as < 23 yr [9.6% of the whole group], 23 to 35 yr [79.8%, reference group] and > 35 yr [10.6%]), and maternal smoking during pregnancy. The number of cigarettes smoked during pregnancy was also studied as a continuous variable. Postnatal variables included breast feeding (1 to 6 mo, and > 6 mo versus no breast feeding), day care attendance during the first 2 yr of life, presence of siblings, ownership of a furry pet, and father's educational level. Paternal education was chosen as the closest indicator of socioeconomic status (12) and graded as < 8 yr of schooling, 8 to 12 yr, and  13 yr. Eczema-related symptoms (itchy flexural rash) and allergic rhinitis symptoms (sneezing or runny or blocked nose accompanied by itchy, watery eyes apart from colds) in the last 12 mo of the child's life were also studied.

Analysis

Frequencies were computed without excluding missing answers, which were therefore counted as negative. Missing answers were always lower than 5%. Associations between several variables and each group of wheezing children (taking as the reference group those without wheezing) were assessed by polytomous logistic regression analysis (13), which allows the use of a unique comparison group for three mutually exclusive outcomes. Results are presented as odds ratios (ORs) and 95% confidence intervals (95% CI). In a first analysis, ORs were adjusted only for sex and geographic area. To test the hypothesis that the ORs for the three groups of children were the same, a log-likelihood ratio (LR) test was computed. In this test the deviance of the logistic model where the ORs are constrained to be the same across groups, is subtracted from that of a model in which the ORs are allowed to vary. If the LR test was significant, pairwise comparisons between groups were performed.

To evaluate differences in prevalence while controlling for mutual confounding factors, a multivariate polytomous logistic regression analysis was performed. Variables simultaneously included in the model were a priori defined as: sex, geographic area, parental history of asthma or COPD, birth weight, maternal age at birth, maternal smoking in pregnancy, day care attendance, siblings, paternal education, personal history of allergic rhinitis or eczema, person who filled in the questionnaire. To test the hypothesis of an effect modification by sex, we analyzed the association between the variables included in the multivariate analysis and wheezing separately in boys and girls. All the statistical analyses were performed using SAS and STATA.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Questionnaires were returned for 18,737 children (96.3%). Complete data for respiratory diseases in the first 2 yr of life and for wheezing symptoms in the last 12 mo were not available for 2,132 and for 272 children, respectively, leaving a total of 16,333 children (8,419 males and 7,914 females) for the present analysis. Only 2.6% of these children belonged to a different racial or ethnic background. Of the 16,333 children studied, 1,221 (7.5%) were classified as having had transient early wheezing, 671 (4.1%) as having had persistent wheezing, and 918 (5.6%) as having had late-onset wheezing, while 13,523 (82.8%) formed the control group.

Wheezing was significantly more frequent in boys than in girls, the ORs (95% CI) being 1.54 (1.37-1.74), 1.85 (1.57-2.18), and 1.50 (1.31-1.72) in children with transient early wheezing, persistent wheezing, and late-onset wheezing, respectively.

Tables 1, 2, and 3 show frequencies of exposures in the four groups according to parental history of airway diseases, pre- and perinatal variables, and postnatal variables. ORs were adjusted for geographic area and sex. The results of the LR test are also shown: the test is significant when the ORs of the three groups of wheezing children are not the same. The main results of pairwise comparisons between groups are reported in the text.

Parental history of airway diseases (hay fever, asthma, COPD) was associated with an increased risk of all categories of wheezing (Table 1). Maternal hay fever was significantly (p = 0.011) more associated with late-onset wheezing than with transient early wheezing. Maternal asthma, maternal COPD, and paternal COPD were significantly (p < 0.010) more associated with persistent wheezing than with transient early wheezing and with late-onset wheezing. In children with persistent wheezing, there was a tendency for the association with parental asthma and COPD to be stronger if the mother rather than the father was affected.

Children with wheezing were significantly more likely than control subjects to be of low birth weight (Table 2), but the ORs for the three groups were not significantly different. Children of mothers younger than 23 yr of age at the time of delivery were not more likely to wheeze than children of 23- to 35-yr-old mothers. Having a mother older than 35 yr of age was protective against transient early wheezing (Table 2). Exposure to maternal smoke during pregnancy was associated with an increased risk in all the categories (Table 2), the association being significantly (p < 0.001) stronger for persistent wheezing than for late-onset wheezing. For each increase in consumption of five cigarettes/day, the ORs (95% CI) were 1.18 (1.08 to 1.29), 1.35 (1.22 to 1.49), and 1.07 (0.96 to 1.20) for transient early wheezing, persistent wheezing, and late-onset wheezing, respectively.

Breast feeding and exposure to other children both inside (siblings) and outside the home (use of day care) appeared to have different effects on transient early wheezing and late- onset wheezing. In fact, breast feeding lasting longer than 6 mo was protective against transient early wheezing, whereas it was a moderate risk factor for wheezing of late onset (p = 0.007) (Table 3). On the contrary, having siblings and attending a day care center were both risk factors for transient early wheezing and protective for late-onset wheezing (Table 3). We did not find an exposure-response relationship between the number of siblings and different wheezing patterns. Both ownership of a furry pet and paternal education showed no significant association with any pattern of wheezing (data not shown).

There was a positive association betwen personal history of eczema and allergic rhinitis at 6 yr of age and the different patterns of wheezing. The ORs (95% CI) for eczema were 1.82 (1.46 to 2.27), 3.89 (3.09 to 4.91), and 2.96 (2.40 to 3.66) for transient early wheezing, persistent wheezing, and late-onset wheezing, respectively. The corresponding ORs (95% CI) for rhinitis were 1.99 (1.54 to 2.57), 10.02 (8.16 to 12.30) and 8.16 (6.77 to 9.83). For both variables the association was stronger (p < 0.001) for persistent and late-onset wheezing than for transient early wheezing.

The associations between variables of interest and different categories of wheezing mutually adjusted by logistic regression analysis are displayed in Table 4. The results are similar to those already shown in the first analysis with ORs adjusted only for sex and geographic area. No effect of modification by sex was found when analysis was run separately for boys and girls.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The identification of determinants for wheeze in early childhood has been the subject of several investigations (14). However, no attempt was made, as in our study, to separate transient, persistent, and late-onset wheezing, three wheezing conditions first identified by Martinez and colleagues in their prospective study of 826 infants born in the mid-1980s in Tucson, Arizona (3). Our retrospective study was conducted on a much wider population and in a very different environment and took into account a broader set of risk factors, including a number of postnatal variables. We found that the extent andespecially for postnatal variableseven the sign of the association between wheezing and several risk factors were often different in the three categories of wheezing children. Although the observed associations were not always identical to those obtained by Martinez, our findings strengthen the hypothesis that wheezing in childhood is a heterogeneous condition with several distinct patterns associated with different pathogenesis and risk factors (1, 3).

A family history of lower airways disease (asthma or COPD) was a risk factor for all three groups of wheezing children. Such a familial predisposition was not unexpected for children who were wheezing at 6 yr of age (both persistent and late-onset wheezers). Twin and family studies have clearly shown a familial aggregation in asthma (19, 20), and recent studies have suggested that up to 75% of the variation in propensity to asthma is explained by a genetic effect (21). A result at variance with the study by Martinez and colleagues (3) was the finding that having a parent with asthma or COPD also raised the risk of transient early wheezing. In the last few years two studies (22, 23) have observed a greater frequency of parental asthma or wheezing in infants who developed recurrent lower respiratory tract infections with wheeze in the first months of life; these infants had a preexisting airway function impairment possibly owing to a reduction of the airway caliber. Furthermore, an increased risk of wheezing lower respiratory illnesses has been observed in children of parents with a history of bronchiolitis or asthma early in childhood (24). No distinction was made between transient early and persistent wheeze in these studies (22); however, the knowledge that most infants who wheeze have transient conditions, and the finding of a congenital impairment in lung function in the group of early wheezers as a whole, suggest that there might be a familial component also to transient early childhood respiratory wheezing.

We found a positive association between exposure to maternal smoking in pregnancy and wheezing with onset at any time. The risk was significantly greater for persistent wheezers. Our findings are fairly consistent with those of the Tucson study, where, however, the risk for late-onset wheezers failed to reach statistical significance. Previous studies have shown that smoking in pregnancy is associated to a dose-related reduction in lung function (25). Such a reduction in lung function has been observed during the first few days or weeks of life, i.e., before significant postnatal smoking exposure occurred (25). The influence of in utero smoking exposure on lung growth and maturation has also been demonstrated in animal studies (28). Prematurity and low birth weight have also been associated with a greater predisposition to wheeze in childhood (14, 15, 17, 29). The association with low birth weight was confirmed in the present study, and remained significant in multivariate analysis when corrected for maternal smoking in pregnancy, which is known to affect fetal growth and gestational duration. Evidence for a direct relationship between birth weight and forced expiratory flow indices in children and adults, irrespective of neonatal respiratory illness and of in utero smoking exposure (30, 31), suggests that the effect is again possibly due to reduced airway function related to abnormal intrauterine development that might persist in adult life.

Young motherhood (< 20 yr) has been found (15, 32) to be a risk factor for wheezing respiratory illnesses in the first years of life. We were not able to confirm these findings probably because of the low number of mothers younger than 20 yr (2.1%) included in our study. We found instead that having a mother older than 35 yr at the time of child's birth was protective against transient early wheezing; such a protective effect remained after adjustment for other factors, possibly reflecting a different pattern of child care by older mothers.

In our study, postnatal variables were highly informative in characterizing the different groups of wheezing children as summarized in Figure 1. Having siblings and attending a day care center were risk factors for transient early wheezing, while breast feeding lasting for at least 6 mo was slightly protective. It is well known that in the first years of life wheezing illnesses are associated with viral infections; these are more frequent the greater the crowding at home or the use of day care facilities (33, 34). On the other hand, breast feeding might offer some protection especially in categories at risk (33, 35). In infants with transient early wheezing, narrower or smaller airways may be more likely to become obstructed when infected by viruses, thus leading to wheezing.

View larger version (8K): [in this window] [in a new window]   Figure 1.   Breast feeding, siblings, and day care attendance as risk factors for wheezing in young children. ORs and 95% CI. Solid circles: transient early wheezing; gray circles: persistent wheezing; open circles: late-onset wheezing.

Contrary to what was clearly seen for transient early wheezing, late-onset wheezing was inversely associated with siblings and the use of day care facilities. As recently hypothesized in several studies (4), a possible explanation for these findings is that infections acquired early in life could downregulate the immune system against an allergic response to environmental antigens. In infants predisposed to atopy, a lower frequency or severity of early childhood infections might be responsible for the reduced stimulus to T helper cell, type 1 (Th1) cytokine production, leading to dominance of the proinflammatory cytokines involved in asthma as well as in other atopic diseases.

The role of breast feeding on subsequent atopic diseases has been the subject of a 50-yr-long controversy, still unresolved (36). It has recently been suggested that breast feeding offers substantial protection against development of respiratory allergy in children (37). On the contrary, we found a tendency to an increased risk of late-onset wheezing. Popular understanding of the relationship between breastfeeding and childhood respiratory illnesses possibly induced mothers with a history of hay fever or asthma to breast feed their infants in a higher proportion (76.2%) than mothers with such a negative history (72.3%). Selective use of breast feeding by atopic mothers, however, is unlikely to have influenced the association with late-onset wheezing because an increased risk was also found among children with no maternal history of atopy, and because no interaction was found between a positive maternal history of asthma or hay fever and breast feeding or duration of breast feeding (data not shown). Our findings are consistent with the increased risk of eczema and hay fever among breast-fed infants reported in three different studies in the United Kingdom (4, 6, 38). Although these data need to be interpreted with caution, they too might be explained by the protective effect of breast feeding on infections acquired early in life, particularly those transmitted by the fecal-oral route (39).

No clear association was found either between variables indicating a greater and more precocious contact with other children (having siblings, attending day care) or breast feeding and development of persistent wheezing. As can be seen in Figure 1, persistent wheezers behaved as a population in which these factors, which have an opposite sign in early versus late-onset wheezers, appear to neutralize each other.

Our study was population-based and included a sufficiently large group of subjects to ensure adequate statistical power. However, the retrospective design could have resulted in a reduced or biased recall of both symptoms and risk factors. To increase specificity, in particular for wheezing in the first 2 yr of life, we asked parents to report episodes of bronchiolitis or asthmatic bronchitis rather than simply wheezing symptoms. We also checked our results by restricting the analysis to children who had more severe wheezing (defined as at least three episodes of bronchiolitis or asthmatic bronchitis in the first 2 yr of life). Despite the reduced size of the cases (transient early wheezing = 568, persistent wheezing = 500), we obtained very similar results (data not shown). On the other hand, we may have included in the control subjects or in late-onset wheezers a number of children who had milder symptoms in the first 2 yr of life, or we may have missed mild symptoms occurring in the 12 mo preceding the questionnaire, leading to a misclassification among early, persistent, and late-onset wheezers. Nondirectional misclassification, however, would have biased the ORs toward the null value. With regard to the risk factors investigated, selective recall would have induced parents of the children with current wheezing (persistent and late-onset wheezers) to over or underreport conditions known to be associated with the disease, in particular smoking history or breast feeding practice. However, we found a positive association between maternal smoking in pregnancy and wheezing only in early lifea condition for which information bias is less likely. A better recall of breast feeding practice could have distorted the risk estimates for persistent and late-onset wheezers. However, the ORs of the two outcomes are not consistent with each other.

In conclusion, our work shows that in childhood both pre- and postnatal risk factors are differently associated with different wheezing phenotypes. Future epidemiologic studies on asthma in childhood should consider the time of appearance of wheezing disorders as a key factor to better disentangle the role of various determinants.