In this large population based study with reports of family eczema-history at two different points of time, we found that having either parent with eczema was significantly associated with reported eczema at the age of 2 years. The associations were consistent when reported at 6-weeks as well as 1 year. Having one or several siblings with eczema, with or without either of the parents with eczema reported at 1 year, was also strongly associated with reported eczema at 2 years and significantly stronger for siblings only.
We have data on exposure reported at 6 weeks and 1 year. Although the reporting of exposure was at two different points of time, the two cohorts are comparable. Some 77% of those who answered Q2 (6 weeks) also answered Q3 (1 year) as well as Q5 (2 years). The two cohorts are generated from the same geographic area (the city of Trondheim), during the same time-period, and by the same midwifes and GPs.
Although eczema is a relatively prevalent disease among children in the western world, most children have mild degree of disease, making the diagnosis and differentiating between different phenotypes of the disease as well as mild cases challenging [21–23]. We found the prevalence of reported eczema 2 years to about 15% both among those who reported the family eczema history at 6-week and those who reported at 1 year, indicating a high reliability of the questions. Any misclassification of eczema cases might therefore be non-differential, and if so may have diluted the associations.
Several studies describe a maternal line of inheritance concerning eczema [5, 6, 22]. This maternal line of inheritance has led investigators to hypothesize that environmental influences operating in utero or in early infancy may be essential in determining disease expression . In this study eczema in the index child was significantly associated with eczema in mothers as well as fathers. This finding is in accordance with several others [8, 9, 25] and not supportive to the hypothesis of paternal genomic imprinting. However, parental recall bias should be taken into consideration also here. Any one parent who followed the child was asked to participate in the PACT-study. One limitation in this study is the fact that we don't know which parent filled in the questionnaire. Since many of the women were included during pregnancy and during their child's first year of living, we assume that mothers most likely have accepted to participate in the study when visiting her GP or midwife. It is therefore most likely that mothers have filled in the questionnaires. In both cohorts more boys than girls reported eczema at age 2 years. Despite this, more mothers than fathers reported to have ever eczema. When studying recall bias in parental questioning, a German group found that mothers tended to report more atopic diseases in a second questionnaire than in the first, whilst fathers were influenced by their child's development of atopic disease . In families without childhood eczema the sensitivity for mothers reporting paternal eczema was lower than in families with at least one child with eczema. The specificity was about the same . Although effect size is small in a population setting, filaggrin haploinsufficiency is a highly penetrant trait, and associated with increased eczema severity . It is therefore likely that severe eczema in any one of the parents represent a greater risk of eczema in the offspring. It seems also likely, that mothers more often would report a positive paternal history of eczema if the disease was severe or persisted into adulthood. The latter is supported by a Swedish study, who found that recall of childhood eczema history among adults was influenced by several factors such as high prevalence of eczema after the age of 15, more visits to the physician after the age of 15, more hand eczema and more sick-leave due to eczema . Since mothers most likely have filled in our questionnaires, an overestimation of the association between paternal eczema and eczema in index child is possible. This might be due either to more severe eczema in fathers or increased awareness due to development of eczema in one of the children. Both could explain the lack of a maternal line of inheritance in this study.
When family history of eczema is studied, allergy related disorders in siblings are seldom accounted for in the risk analysis. We found that, although having any one parent with eczema was associated with eczema in index child at 2 years, having one or several siblings with eczema together with mother and/or father with eczema was strongly associated with eczema in index child 2 years when reports were collected at 1 year. This association was also seen when eczema was reported only in sibling(s). There are different interpretations of these findings. One possible explanation could be that this is a reflection of a parental genetic disposition with incomplete penetrance . Mutations in the gene encoding filaggrin (FLG) have been identified as a strong predisposing factors for eczema  and especially severe phenotypes of the disease , but other candidate genes are also under investigation . Different environmental factors can alter the expression of different genes, as have been showed with the exposure to cat within the first year of life in those who carry mutations of FLG . Other environmental factors such as early colonisation from maternal microbial flora as well as shared environment among siblings with the same genetic predisposition may act in a similar way and explain the "eczema-sibling-effect" in this study.
The "eczema-sibling-effect" was not significant when reported at 6 weeks. Awareness of disease in first child with mild disease might be absent until same kinds of symptoms as e.g. dry skin is observed in second child. Also, in mild cases of eczema a significant proportion of the children are disease free by the age of 3 years . In this study, some 28.5% of parents with eczema in index child at 2 years changed their answer from "no" when reported at 6 weeks to "yes" when reported at 1 year regarding the question on whether or not siblings had eczema, as opposed to 6.4% among those without eczema in index child. These findings could be interpreted as an increase in awareness since some 80% of index children had developed symptoms on eczema at age 1 year. Another interpretation is that some of the older siblings have not yet been diagnosed with eczema. This is however less likely, since a majority of children with eczema starts with their disease during their first year of life .
In a German study of children 9-11 years old having two first degree relatives with the same atopic disease was highly associated with eczema . Diepgen & Blettner found a stronger correlation between siblings than between siblings and parents for all atopic diseases, also atopic dermatitis . Eczema in an older sibling was also found to be an independent risk factor for eczema among 4-years old in Sweden .
When only some 60% of those finishing the questionnaires on exposure were answering the questionnaires on health, one could argue that the data are prone to self-selection. In both cohorts the lost-to-follow-up group contained more current smoking mothers and fewer homeowners, indicating a lower socio-economic status (data not shown). However, other exposure data, including reported family eczema-history did not differ among those who followed up and those who did not.
Another limitation of the study was the reporting eczema in parents and siblings without a clinical verification. The question "have you, the child's father or any of your joint children ever had eczema" is a core ISAAC questions, but is most often used in combination with a question on rash located in typical places or diagnosis verified by a doctor. To the best of our knowledge, the question has not been validated in an adult population. However, the almost similar phrasing "have you ever had childhood eczema" was validated among adults in a Swedish population. The sensitivity and specificity of this question was 89.9% and 70.7%, respectively , and an overestimation of the reported prevalence of childhood eczema among adults is therefore likely. Regarding the use of this question among children (siblings), we have validated this in a former publication . This question alone, without a question identifying rash on typical location, gave a sensitivity of 96.8% and a specificity of 68.0% when validated against the UK Working Party Criteria. Adding a question of rash on typical locations decreased the sensitivity whereas the specificity increased. This gives reason to believe, that there might be an overestimation of reported eczema among the siblings, since other forms of dermatitis in young children, such as seborrhoeic dermatitis as well as nappy dermatitis might have been included. However, in case of such a misclassification, there is no reason to believe that this has changed from 6 weeks to 1 year, and could therefore not explain the differences in reporting of eczema in siblings in the two cohorts.
The strengths of this study are the large number of unselected participants as well as the prospective design. The consistency of reported eczema indicates a high reliability of the questions and the prevalence of reported eczema is well in line with the prevalence found in the PACT endpoint-study . In addition, the focus is on eczema-groups only, since other studies have showed that parental eczema may be a better marker for eczema in the offspring than other parental atopic diseases [8, 33].