Wheezing, because of its tendency to recur, is one of the most common problems in the first months of life.1 The lack of specificity of the clinical manifestations, the variability of patient response to the existing treatments, and the association between wheezing and viral infections in this age group2 can have a significant impact upon the quality of life of both the infant and the family,3 with an increase in the use of healthcare resources,4 and an important economic impact.5

Many studies have addressed wheezing in infants, although most have focused on viral causes6–8 or on the influence of allergies.9,10 In turn, some authors have related wheezing to obstetric antecedents,11 early exposure to certain allergens,12 immune alterations caused by certain environmental exposures13,14 or the administration of certain medications during pregnancy.15

A group of Spanish coordinators of the International Study of Asthma and Allergies in Childhood (ISAAC)16 developed the International Study on Wheezing in Nursing Infants (Estudio Internacional de Sibilancias en Lactantes (EISL)) to determine the prevalence of wheezing, its severity and risk factors in the first 12 months of life.17 These and other previous as well as subsequent studies have indicated that there are other factors apart from viral infections that might interact and condition the presence of recurrent wheezing (RW) in the first months of life, such as infant gender, attending nursery school, diet, characteristics of the home, pets, maternal diet during pregnancy, smoking, the presence of siblings, pneumonia, etc.1,18–21 The objective of the present study was to identify risk factors that may be acted upon in order to modify the evolution of recurrent wheezing in the first months of life. In this respect, a predictive model was developed, based on certain risk or protective factors, capable of predicting the probability of developing recurrent wheezing in the first year of life – with the evaluation of variations in probability according to changes in one or more factors.

The study included a total of 1164 infants, and presented a response rate of 71% (603 males (51.8%) and 561 females (48.2%)). The infant birth weight was 3030±770g, with a height at birth of 49.48±3.46cm.

Table 1 shows the results referring to the risk factors studied in the infants that suffered recurrent wheezing (RW) in the first year of life, with the corresponding odds ratios and 95% confidence intervals.

Statistical significance as risk factors for the development of RW in the first year of life was identified for: a briefer duration of exclusive breastfeeding (OR 1.53 (95%CI 1.12–2.10)), nursery school attendance (2.09 (1.53–2.86)), the presence of eczema in the infant (2.69 (1.93–3.74)), a history of asthma in the mother (1.91 (1.16–3.13)), the regular consumption of so-called fast foods during pregnancy (3.96 (2.18–7.19)), and the presence of certain pets (birds and rabbits/hamsters) both at birth and after birth (1.86 (1.16–2.96)). Although no correlation to smoking was found throughout pregnancy (1.16 (0.75–1.79)), on separately considering smoking in the different trimesters of pregnancy, an association was identified when the mother smoked in the second and third trimesters (1.63 (1.07–2.50) and 1.82 (1.20–2.77), respectively), and after delivery (1.47 (81.04–2.10)). In contrast, a lower birth weight (0.41 (0.21–078)) and consumption of the Mediterranean diet during pregnancy (0.91 (0.87–0.95)) were identified as protective factors.

Taking into account the above findings, we developed the logistic regression model, considering all the aforementioned significant variables, as described in Material and methods. The results are shown in Table 2.

The following variables were excluded from the model: exclusive breastfeeding ≤3 months (p=0.092); smoking in the mother during the second trimester (p=0.863); and current smoking in the mother (p=0.457). This means that after considering the variable smoking during the third trimester, the other two variables related to smoking did not contribute significant information, in the same way as exclusive breastfeeding.

Attending nursery school, presenting eczema, an asthmatic mother, and maternal smoking during the last three months of pregnancy were significant risk factors for RW. Mediterranean diet during pregnancy in turn exerted a significant protective effect against RW (see OR and 95%CI values in Table 2).

The model for estimating the probability of RW is expressed as follows:

Figs. 1 and 2 show the results in the form of classification trees (that of the first figure being based on whether the mother suffered asthma or not).

Figure 1.

Classification tree with the different probabilities of developing recurrent wheezing during the first year of life, with the variables found to be significant in the multivariate analysis, for infants with asthmatic mothers. Abbreviations: AM: asthmatic mother, IE: infant eczema, SM: smoking mother, MD: Mediterranean diet, NS: attended nursery school.

(0.16MB).

Figure 2.

Classification tree with the different probabilities of developing recurrent wheezing during the first year of life, with the variables found to be significant in the multivariate analysis, for infants with non-asthmatic mothers. Abbreviations: AM: asthmatic mother, IE: infant eczema, SM: smoking mother, MD: Mediterranean diet, NS: attended nursery school.

(0.16MB).

In this regard, infants presenting eczema and attending nursery school, with a mother who has asthma, smoked during the last trimester of pregnancy, and did not follow a Mediterranean diet during pregnancy were found to have a 79.7% probability of suffering recurrent wheezing (Fig. 1).

In contrast, infants with mothers presenting none of these factors had a mere 4.1% probability of suffering recurrent wheezing (Fig. 2).

The mathematical model developed exhibited a true-positive rate of 81.48% and a specificity of 99.5%. Overall, the correct classification rate was 89.7% (efficacy of the model). Both the false-positive and false-negative rates were low (10.11% and 18.52%, respectively).

Asthma is the most common chronic disease in childhood, and wheezing is often detected in the first months of life. This high prevalence suggests that there are different subgroups of infants exhibiting different inflammatory responses to a range of triggering factors, and this has led to many studies on the subject in recent years.

In this sense, many authors have examined the presence of wheezing in nursing infants. The recently published data of the EISL,1,18,20,21,24,25 as well as those of our own group,19 have revealed great variability in the statistical significance of the risk factors for wheezing and RW in different parts of the countries and in different centres.26

Some authors have investigated the relationship between certain nutrients during pregnancy and the development of RW or asthma in the first months of life. In this respect, Miyake et al.27 found an increased intake of dairy products, calcium, vitamin D, alpha-linolenic acid and docosahexaenoic acid during pregnancy to be associated with a decrease in RW at 16 and 24 months of age. Castro et al. in turn observed an association between olive oil consumption during pregnancy and a lesser incidence of RW in infants under one year of age.28 In our study we found a statistically significant relationship between consumption of a Mediterranean diet on the part of the mother during pregnancy and the development of RW in the infant. The association was strong enough to enter the predictive model. Such an association has also been reported in older children,29 in some cases relating a Mediterranean diet and obesity.30

The relationship between RW in infants and a family history of atopy is known, although our group only found an association between RW and maternal asthma – in coincidence with the observation of other authors.28,31

Regarding breastfeeding, and in concordance with other investigators,18,32–35 we observed a lesser proportion of RW among the infants exclusively breastfed for more than three months, compared with those infants breastfed for three months or less – although the model suggests this protective effect against infections of exclusive breastfeeding during the first year of life to be partially cancelled when the infants attend nursery school, since the latter factor prevails due to its strong influence in the final model.

A notable observation in our study was the association between low birth weight and a lesser incidence of RW. These results must be viewed with caution, however, since only 11 infants with RW weighed less than 2500g at birth, and none weighed less than 2000g at birth.

Although many studies have described statistically significant correlations to different risk or protective factors,36–39 our review of the literature yielded no studies relating multiple factors in the context of a predictive model of RW in infants according to exposure to risk or protective factors as was done in our study.

Based on the proposed model, our study in Salamanca has shown that infants with all the risk factors included in the model have a 79.7% probability of developing RW, while those infants that present no such factors have a risk of only 4.1% of developing RW. However, the model also describes the probability of suffering RW with any other combination of risk factors. The variables are represented in the model on a dichotomic basis: 0 (absent) or 1 (present). As an example, if the mother was asthmatic, the infant suffered eczema, the mother smoked during the third trimester of pregnancy, the mother followed a Mediterranean diet, and the infant did not attend nursing school, the probability of suffering RW dropped to 40.8% (i.e., a reduction of 38.9%) (Fig. 1).

All the possible alternatives, with their respective probabilities, according to whether the mother suffered asthma or not, are reflected in Figs. 1 and 2.

As has been commented above, the results published by the EISL show great variability.1,18–21,24 The mathematical model is based on the assumption that the factors which may enter the model vary according to the centre, country or geographical zone involved. In this regard, the proposed model must be adjusted to the results obtained in each zone, and future genetic studies will probably afford more precise information capable of explaining the great variability of the results obtained in the different geographical settings in which the EISL has been carried out.

As possible limitations of our study, mention must be made of the fact that the data were supplied by the parents on answering the EISL wheezing questionnaire, rather than by the attending paediatricians – although the mentioned questionnaire has been duly validated.22 On the other hand, when there were doubts regarding some item of the questionnaire, the parents were called by telephone or the reference paediatrician was consulted. It must also be noted that this is a cross-sectional study, and seasonal variations due to the relationship between RW and viral infection in this particular age group may have constituted a confounding factor – particularly in the winter months. This limitation was minimized by including infants born in all seasons of the year during a period of two years.

Lastly, the main strong points of the study are its sample size based on the general population, and the fact that infants from all primary care centres (both urban and rural) in the province of Salamanca were included, with the inherent cultural and socioeconomic differences involved.

The present study has allowed us to identify different risk or protective factors capable of interacting with each other, and which can determine whether or not RW will develop during the first year of life in the population studied. It would be important to conduct a prospective study based on the data of the EISL, with a view to designing a model for calculating the probability of suffering RW in a given population depending on the risk or protective factors to which it is exposed.

This project has received financial support from the Gerencia Regional de Salud de Castilla y León, Spain. GRS 287BA/08.

A grant was received from the Fundación Ernesto Sánchez-Villares of the Sociedad de Pediatría de Asturias, Cantabria, Castilla y León.

The authors declare that they have no conflicts of interest.