Prenatal Tobacco Smoke Exposure Increases Hospitalizations for Bronchiolitis in Infants

Marcello Lanari; Silvia Vandini; Fulvio Adorni; Federica Prinelli; Simona Di Santo; Michela Silvestri; Massimo Musicco


Respiratory Research. 2015;16(152) 

In This Article


Study Subjects

This study was promoted by the Italian Society of Neonatology with the National Research Council and involved thirty neonatology units, registering 1,000 births or more each year, in the northern, central and southern areas of Italy. The study was approved by each one of the ethical committees of the participating centres. Parents in the study provided their signed informed consent to participate in the study; details on the modalities of recruitment and of data collection have been described elsewhere.[17,18] Briefly, all consecutive newborns at 33–34 weeks gestational age (wGA) seen in a recruitment period lasting 1 year from the first enrolment by the participating centres were enrolled. For each enrolled newborn of 33–34 wGA, two newborns of the same sex and with the nearest date of birth were enrolled: one of 35–37 wGA and one of >37 wGA. The enrolment was carried out according to GA with the aim to analyse potential differences related to GA. Exclusion criteria were: life expectancy shorter than 6 months; participation in clinical studies on pharmacological or surgical interventions; haemodynamically significant congenital heart diseases or chronic lung diseases; programmed or administered RSV prophylaxis with a humanized monoclonal antibody (palivizumab). Excluded infants were those not residing in the geographical catchment area of the enrolling neonatology unit hospital.

Data Collection

From November 1st 2009 to December 30th 2012, 2,314 newborns (1,113 females and 1,201 males) were enrolled in the study. At time of birth, parents were interviewed with a structured questionnaire on their demographics, health and socio-demographic status, living conditions and TSE. In addition, information on pregnancy, delivery and newborn conditions given by the parents were integrated, when necessary, with the information contained in the clinical record form. The same physician in each center collected the data using a standard record form. After discharge, two structured follow-up phone interviews with the child's parents were carried out by trained interviewers. The first interview took place during the infective respiratory epidemic season (in Italy from November to March[19]), the second at the 12 month after birth. The interviews collected data on possible environmental risk conditions for bronchiolitis, including household health and crowded living conditions, daily contact with siblings or other children and day care attendance and lack of or early interruption or no breastfeeding.

Assessment of Pre- and Postnatal TSE

Exposure to tobacco smoke during pregnancy was assessed by asking the mother whether she smoked during pregnancy, and if so, how many cigarettes per day and whether the father or other people smoked regularly in her presence. The information collected was then collapsed into a single variable with the following four categories for prenatal TSE: no; only passive; active with 1–15 cigarettes per day and active with 16 or more cigarettes per day.

Exposure to tobacco smoke after birth was assessed during the two follow-up interviews by asking the parents if the mother, father or other people in the household smoked inside and/or outside of the child's living environment in the first year of life. Similarly as for prenatal TSE, the information collected was collapsed into a single variable; postnatal TSE with the following three categories: no; outside the child's living environment and in the child's living environment. Finally, a third variable was constructed: any tobacco smoke exposure that was comprehensive of any prenatal or postnatal exposure.


The main outcome of this study was hospitalization and/or death due to bronchiolitis during the first year of life, as classified by the ICD-9 code 466.1 (codifying for acute bronchiolitis).[20]

During the two follow up interviews, parents were asked about hospitalizations of the child. If the child was referred as having been hospitalized, further confirmation was obtained by retrieval of the hospital record forms or by direct contact with the physician who was responsible of the infant during hospitalization.

Statistical Analysis

The cumulative time-dependent risks of hospitalization for bronchiolitis during the first year of life of the infants were calculated with survival analysis. Relative risks were estimated as hazard ratios (HR) with 95 % Confidence Intervals (CI) calculated from the standard errors derived from Cox's proportional hazard model.[21] Multivariate analyses were carried out to estimate the independent contribution of each considered factor to the risk of bronchiolitis.

In order to identify further potential predictors of bronchiolitis hospitalization besides exposure to tobacco smoke, we used a two-step approach. Firstly, we performed a multivariate analysis within the prenatal and the neonatal variables considered as potential risk factors for bronchiolitis (Table 1 of the Additional file 1 The variables significantly associated with the outcome (the level of statistical significance was set at p-value = 0.05), served for defining the exposure to pre and perinatal risk factors of bronchiolitis. Children were considered prenatally exposed (father suffering from respiratory diseases, no recourse to assisted reproductive technologies and use of corticosteroid therapy for lung maturation) and/or perinatally exposed (male sex, singleton delivery and surfactant therapy), when at least one of these significant risk factors was present. Pre and perinatal exposures were then entered in the final multivariable analysis as single dichotomous variables. Although non statistically significant in the first step of the analysis, age of the mother (continuous variable), parents' level of education in years of completed school (both less or equal to 8 years, at least one more than 8 and less or equal to 13 years, both more than 13 years) and weeks of gestational age were considered as potential confounding variables and were entered in the final multivariable analysis.

We considered also well-known environmental risk factors (RFs) of bronchiolitis as single covariates: exposure to epidemic season (defined as children living for at least one of the first 3 months of life during the calendar period between November to March); no breastfeeding (i.e. feeding of the newborn without breast milk beyond the age of 1 month); presence of siblings or children (less than 10 years old) sharing the same living environments; crowded living conditions (i.e. the presence of five or more people older than 10 years in the living environment of the newborn) and day care attendance.

Sample size was calculated by fixing a predefined precision of the estimate of the absolute risk of hospitalization and/or death for bronchiolitis induced or not by RSV during the first year of life. Assuming that the risk of hospitalization for bronchiolitis during the first year of life was about 7 % a sample size of 2,500 newborns could provide a 95 % confidence interval (95 % CI) of 6.0 to 8.1 % which is largely consistent with a random error of less than 20 %.

The level of statistical significance was set for all the analysis at p = 0.05.

The software package used was IBM SPSS Statistics for Windows version 21.0 (Armonk, NY, IBM Corp.).