• risk factors;
  • birth weight;
  • atopic dermatitis;
  • breast feeding


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conflict of interest
  7. References


To analyze the effects of pregnancy and early events in the newborn on the risk of subsequent atopic dermatitis (AD) during the first year of life.

Patients and Methods

This is a prospective multicenter cohort study of newborns during the first year of life. Newborns identified on random days in three obstetrics departments in the area of Bergamo, Lombardy, Northern Italy, were eligible. At baseline, the mothers were interviewed by medical staff during their stay in hospital after delivery. At 6 and 12 months after delivery, a postal questionnaire was sent to the parents. Relative risks were calculated with and without adjustment by multiple regression analysis.


A total of 1081 newborns entered the study: 796 (74%) parents answered the 12-month questionnaire. Hundred and thirty-eight (17%) reported a diagnosis of AD at 6 months and 222 (28%) at 12 months. Parental history of AD and/or asthma was associated with an increased risk of AD (RR 1.5, 95%CI 1.1–2.0). Birth weight was slightly associated with an increased risk of AD: RR 1.04, 95%CI 1.001–1.08 (continuous variable, increment of 100 grams). No association emerged between breast feeding, smoking, and risk of AD.


This study in an Italian offspring cohort points to family history of atopic diseases and body weight at birth as relevant risk factors. The study was unable to document associations with other perinatal factors particularly breast feeding and parental smoking in the perinatal period.

Atopic dermatitis (AD) is the first step in the progression of atopic disorders in infants, the so-called atopic march, from AD to allergic rhinitis and asthma. Recent studies support the role of skin barrier defects (i.e., loss of function of filaggrin, an epidermal barrier protein) as facilitating entry for pathogens, allergens, and other environmental insults (toxins, irritants, pollutants). Therefore, genetic predisposition together with skin barrier defects and environmental exposure are now considered the main mechanisms of development of AD [1, 2].

The prevalence of AD has risen steeply in the last five decades, reaching about 20% in the first 2 yr of life in many industrialized countries, and becoming the most common inflammatory skin disease in childhood [3].

Thus, exposure during pregnancy or during the first weeks of life may greatly influence the risk of AD. Breast feeding and antibiotic use in pregnancy have all been considered in relation to the risk of AD, with variable results [4].

To analyze the effects of pregnancy and early events in newborns on the risk of AD during the first year of life, we conducted a prospective cohort study. Most published studies on the epidemiology of AD have been conducted in North European and American countries, and few data are available from southern European populations for very young children. This study was conducted in the Bergamo area, Lombardy, Italy.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conflict of interest
  7. References

This was a prospective multicenter cohort study of newborns during the first year of life. Newborns identified on random days in the three obstetric departments in the area of Bergamo, Lombardy, Northern Italy, were eligible. Subjects from families not speaking Italian or not resident in the Bergamo area were not included.

The mothers were identified and interviewed by medical interviewers during their stay in hospital after delivery. The investigator enrolled all the consecutive newborns observed from the start of the study who met the inclusion criteria on selected random days.

Enrollment started in February 2007 and was concluded in May 2009 when the number of patients established by the protocol (see below) was reached. Three questionnaires were planned: a face-to-face questionnaire with trained interviewers and two postal questionnaires sent to the parents 6 and 12 months after delivery. The baseline questionnaire collected information about the main demographic characteristics and personal habits of the parents (smoking, alcohol consumption), parents' history of allergic diseases, number of brothers and sisters of the newborn, course of pregnancy and mode of delivery, antibiotic treatment, and streptococcus prophylaxis during pregnancy, gestational age at delivery, birth weight, details of the child's home.

The follow-up questionnaires collected information about breast feeding, the home, baby's diet, selected diseases and the signs and symptoms associated with them, any hospital admission, discharge diagnosis, other health problems of the baby, parents' smoking habits during the baby's first year of life.

Diagnosis of AD

Diagnosis of AD was established when there was a history of itching skin rash [5], and a diagnosis of ‘atopic dermatitis’ or ‘atopic eczema’ made by a pediatrician or a dermatologist.

Statistical analysis

The main objective of the study was to analyze the 12-month incidence of AD in the study population and the risk factors for the development of AD. Adjusted and unadjusted relative risks (RR) and 95% confidence interval were calculated with the classical Mantel–Haenszel analysis of case–control data modified for the analysis of cohort studies [6]. In the model with adjustment, we included gender, mother's and father's age and education, family history of AD (or mother's history of AD and father's history of AD or mother's history of asthma and father's history of asthma, alternatively included in the model because of the correlation), cats, or dogs in the house at birth (or cats or dogs in the house after birth), disease in pregnancy, gestational age at delivery, mode of delivery, birth weight, use of antibiotic during pregnancy, streptococcus treatment during pregnancy, antibiotic use in the newborn, breast feeding (or duration of breast feeding or duration of breast feeding/other milk).


The study was conducted in accordance with the Declaration of Helsinki and the ICH-GCP guidelines. The study protocol was approved by the ethics committees of the participating centers. Written informed consent was obtained from the mothers before the study.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conflict of interest
  7. References

A total of 1081 newborns entered the study: 285 parents (26%, 95% CI 24–29%) did not respond to the postal questionnaire or phone interview at 12 months (Fig. 1). Thus, these analyses are based on 796 (74%, 95% CI 71–76%) newborns followed up for 12 months. The characteristics of the newborns lost to follow-up were largely similar to those for which follow-up data were available (Table 1).

Table 1. Comparison of characteristics between subjects lost to follow-up and those included in the analysis
 Lost to follow-up (12 months)p-value
Yes (n = 285)No (n = 796)
Na% (95% CI)Na% (95% CI)
  1. a

    In some cases, sum does not add up to the total because of missing values.

  2. b

    Low level: mother's and father's secondary school or high school education. High level: mother's and father's high school or university.

Male12446.3 (40.3–52.3)37748.8 (45.2–52.3)0.4799
Female14453.7 (47.7–59.7)39651.2 (47.7–54.8)
Mother's and father's age (years)
Mother <32 and father <358339.3 (32.7–45.9)18430.2 (26.6–33.9)0.0843
Mother <32 and father ≥352110.0 (6.0–14.1)609.9 (7.5–12.3)
Mother ≥32 and father <353014.2 (9.5–18.9)9215.1 (12.3–17.9)
Mother ≥32 and father ≥357736.5 (30.0–43.0)27344.8 (40.9–48.8)
Mother's and father's education
Low levelb12648.1 (40.4–55.8)30239.6 (36.1–43.1)0.0166
High levelb13651.9 (44.2–59.6)46060.4 (56.9–63.9)
Parity of the mother
012844.9 (39.1–50.7)40450.7 (47.3–54.2)0.2349
112644.2 (38.4–50.0)31239.2 (35.8–42.6)
≥23110.9 (7.2–14.5)8010.1 (8.0–12.1)
Birth weight (grams)
<2500176.3 (3.4–9.2)354.5 (3.05–6.0)0.3200
2500–400024690.8 (87.3–94.2)70691.1 (89.1–93.1)
≥400083.0 (0.9–5.0)344.4 (2.9–5.8)
Gestational age at birth (weeks)
<3793.9 (1.4–6.4)355.1 (3.5–6.8)0.6424
37–4017575.8 (70.2–81.3)52376.5 (73.3–79.6)
>404720.3 (15.1–25.6)12618.4 (15.5–21.3)
Family history of AD and/or asthma
No14450.5 (44.7–56.4)36746.1 (42.6–49.6)0.1996
Yes14149.5 (43.6–55.3)42953.9 (50.4–57.4)
Mother's smoking habits
Non-smoker17361.4 (55.6–67.1)52766.3 (63.0–70.0)<0.0001
Ex smoker7024.8 (19.8–29.9)24230.4 (27.2–33.6)
Current smoker after childbirth3913.8 (9.8–17.9)263.3 (2.0–4.5)
Father's smoking habits
Non-smoker10838.9 (33.1–44.6)41752.7 (49.2–56.2)<0.0001
Ex smoker6423.0 (18.0–28.0)23930.2 (14.4–19.7)
Current smoker after childbirth10638.1 (32.3–43.9)13517.1 (27.0–33.4)
Cats or dogs in the house at birth
No21174.0 (68.9–79.2)55269.4 (66.1–72.6)0.1361
Yes7426.0 (20.8–31.1)24430.7 (27.4–33.9)
Cats or dogs in the house after birth
No28399.3 (98.3–100.0)59975.3 (72.2–78.3)<0.0001
Yes20.7 (0.0–1.7)19724.8 (21.7–27.7)

Figure 1. Flow of subjects in the study population.

Download figure to PowerPoint

A total of 222 newborns reported a diagnosis of AD, 138 (62%, 95% CI 56–69%) within 6 months, and an additional 84 (29%, 95% CI 31–44%) between 6 and 12 months. Therefore, incidence estimates were, respectively, 17% (95% CI 15–20%) and 28% (95% CI 25–31%) at 6 and 12 months. Table 2 shows the distribution of study subjects according to the diagnosis of AD at 12 months and selected factors. A family history of AD and/or asthma was associated with an increased risk of AD in the newborns (RR 1.5, 95% CI 1.1–2.0). Maternal history of AD (RR 1.8, 95% CI 1.1–2.9) and paternal history of asthma (RR 1.9, 95% CI 1.2–2.8) were significantly associated with an increased risk of AD in the newborns.

Table 2. Main characteristics of the study subjects according to AD diagnosis
 Diagnosis of AD
No (574)Yes (222)RRb95%CIRRc95%CIp-value
  1. RR, relative risk; CI, confidence interval.

  2. a

    Row percent.

  3. b

    Not adjusted.

  4. c

    Adjusted (model included all variables).

Mother's age (years)
Continuous variable    1.000.97–1.040.990.95–1.030.5241
Mean (SD,range)32.4 (4.2, 20–44)32.5 (4.0, 20–43)       
Father's age (years)
Continuous variable    1.020.99––1.070.1713
Mean (SD,range)35.3 (5.0, 23–59)35.8 (4.9, 24–50)       
Mother's education
High school31972.811927.20.970.70–1.360.970.66–1.420.8833
Father's education
High school27974.49625.60.890.66–1.190.890.64–1.230.4732
Parity of the mother  
Mother's smoking habits
Ex smoker16869.47430.61.140.86–1.511.100.51–2.390.8092
Current smoker after childbirth1973.1726.91.010.47––1.540.3843
Father's smoking habits
Ex smoker16769.97230.11.090.81–1.470.940.62–1.410.7528
Current smoker after childbirth10275.63324.40.890.60–1.311.130.83–1.550.4329
Family history of AD and/or asthma  
Mother's history of AD  
Father's history of AD  
Mother's history of asthma  
Father's history of asthma  
Cats or dogs in the house at birth  
Cats or dogs in the house after birth  

No association emerged between maternal and paternal age and education, parents' smoking habits, presence of cats and dogs in the child's house, gestational age at delivery, and the risk of AD.

Table 3 shows the relation between the risk of AD and pregnancy, delivery factors, and breast feeding.

Table 3. Pregnancy variables, breast feeding and diagnosis of AD
 Diagnosis of AD
 No (574)Yes (222)RRb95%CIRRc95%CIp–value
  1. RR, relative risk, CI, confidence interval.

  2. € For an increment of 100 grams.

  3. a

    Row percent.

  4. b

    Not adjusted.

  5. c

    Adjusted (model included all variables).

Disease in pregnancy  
Gestational age at delivery (weeks)
Mean (SD, range)39.1 (1.6, 32–42)39.5 (1.5, 33–42)  1.141.03––1.250.1016
Continuous variable         
Mode of delivery
Induced vaginal10067.14932.91.200.86–1.661.240.87–1.770.2276
Caesarean section11075.93524.10.890.61––1.550.8324
Birth weight (grams)
Mean (SD, range)3253 (452,1290–4550)3366 (436,1580–4750)       
Continuous variable    1.041.01––1.080.0412
Use of antibiotics during pregnancy  
During whole pregnancy5464.33035.71.330.90–1.951.340.90–2.000.1513
During the 1st trimester1254.51045.51.660.88–3.131.850.96–3.580.0675
During the 2nd trimester2266.71133.31.210.66––2.010.8965
During the 3rd trimester3266.71633.31.210.73––2.090.5142
Streptococcus treatment during  
Antibiotic use in the newborn  
Breast feeding  
Breast feeding only32569.914030.11.320.90–1.931.340.90–2.020.1528
Breast feeding/other milks13673.54926.51.160.74–1.811.130.71–1.800.6157
Duration of breast feeding only (months)
Continuous variable    1.020.96––1.100.4765  
5 or more20268.99131.11.360.91–2.031.460.95–2.250.0877
Chi-square for trend: p-value    0.12750.0883   
Duration of breast feeding/other milks (months)
Continuous variable    1.020.98––1.080.3404  
5 or more23569.110530.91.350.91–2.011.390.91–2.120.1264
Chi-square for trend: p-value    0.10150.1025   

Birth weight was slightly associated with an increased risk of AD: RR 1.04, 95%CI 1.001–1.08 (continuous variable, increment of 100 grams). The risk of AD was slightly increased with gestational age at delivery: RR 1.1, 95%CI 1.03–1.3 (continuous variable, increment of 1 wk), but with loss of significance after adjustment (RR 1.1, 95%CI 0.98–1.3).

No association emerged between a history of gestational diseases (diabetes, hypertension, liver diseases) in pregnancy, mode of delivery, and maternal use of antibiotics during pregnancy, and risk of AD. No relationship emerged between breast feeding and risk of AD.

A sample size of 796 subjects with a proportion of AD diagnosis of 27.9% (222 cases) achieve at least 80% power to detect an RR = 1.5 if the proportion of exposure in the control group is assumed to be between 0.2 and 0.5. (Test statistic used: two-sided Z test with pooled variance; significance level of the test: 0.05.) [7].

Fig. 2 shows the relative risk and 95% CI of selected factors on a logarithmic scale.


Figure 2. Relative risks (RR) and 95% confidence intervals (CI) of selected factors.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conflict of interest
  7. References

In this study, the cumulative incidence of physician-confirmed diagnoses of AD was about 28% at 1 yr. This is higher than the prevalence estimated in other studies [8-10]. For example, a cross-sectional study conducted in Italy among children aged 9 yrs reported a lifetime prevalence of AD of about 15%. [10].

In a study conducted in Verona (Northeastern Italy) and published in 2008, an incidence of 18.1% was detected among preschool children [11].

AD is a common condition in infancy and generally disappears at about 3 yrs of age in a significant proportion of children.

This study confirmed the main recognized risk factors for AD such as family history of AD and asthma [12, 13]. We found the RR of developing AD during the first year of life was 50% higher in children of parents with a history of AD than in children with no history.

The results suggested that birth weight was associated with the risk of AD. No association emerged between risk of AD and parental age, maternal and paternal smoking during pregnancy, history of gestational diseases in pregnancy, mode of delivery, treatment with antibiotics during pregnancy or delivery, and breast feeding compared with other milks feed.

The relation between birth weight and the risk of AD has rarely been analyzed [14]. In a US cohort study of pregnant women and their offspring [15], birth weight was not associated with the risk of AD. A more recent cross-sectional study [16] and a case–control study [17] of AD in children older than 1 yr found a positive association between AD and obesity. A Swedish study also reported a positive relation between fetal growth and childhood atopic eczema [14].

Biologic explanations for the associations have been put forward. Impaired immune tolerance is considered to be a sequel of immune changes due to the activity of adipokines, bioactive molecules secreted in white adipose tissue. Further, obesity is related to the high levels of circulating IL6, leptin, and TNF-alpha, which in turn are associated with the risk of AD [18][19-21]. Birth weight has been suggested to be related with the risk of obesity [22].

Whether environmental tobacco smoke increases the risk of allergic diseases is still debated. No significant association was observed between perinatal tobacco smoke exposure and the development of AD in a prospective cohort study in Japan [23], but other studies showed positive associations [24]. We did not find any association between paternal and maternal smoking and risk of AD.

Finally, as recent studies suggest, there was no evidence of a protective effect of prolonged breast feeding (5 months or more) against AD among children at age of 12 months and it might even be a risk factor, particularly for children without family history (as an expression of genetic predisposition) [25, 26]. A recent meta-analysis did not find any strong evidence of a protective effect against AD of exclusive breastfeeding for at least 3 months, even among children with a family history of atopy [27].

In the interpretation of these findings, potential limitations of the study should be considered.

First of all, 26% (95% CI 24–29%) of women were lost to follow-up. This is a high proportion, but we have to consider that this study included a non-selected population identified in a first level center at delivery, and similar dropout rates have been reported in other studies with a similar design. Moreover, the characteristics of women and newborns lost to follow-up were similar to the general and clinical characteristics of the whole study population at study entry (Table 1).

Unfortunately, no information was available on the women asked to participate that declined to enter the study.

A further concern may regard the criteria adopted for identifying new cases. We relied on a medical diagnosis by a pediatrician or a dermatologist. It is however possible that parents misunderstood the diagnosis or that a few cases had not been referred to a physician. The Italian healthcare system offers universal coverage, and primary care pediatricians are required to vaccinate all children in their area. This point is however an important limitation of this study.

Another limitation is the fact that all considered factors were self-reported. In particular, reproductive and neonatal information was self-reported by the mother. The validity of maternally reported pregnancy and delivery information has been however reported of high reliability [28].

In general, any misclassification of exposures and outcomes should tend to reduce observe associations.

The strengths of this study include its prospective nature and the possibility of qualifying exposure precisely as it was not affected to any large extent by recall bias.

This study in an Italian offspring cohort points to family history of atopic diseases, gestational age, and body weight at birth as main risk factors for AD. Other perinatal factors seemed much weaker, and no associations were found with parent education, breast feeding, and parental smoking in the perinatal period.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conflict of interest
  7. References