Eczema is a common chronic disease which has significant morbidity and costs for children and their families. It is a common skin disease characterised by an itchy rash, often located in the flexural areas (e.g. folds of the elbows and knees), which frequently first occurs at a young age (less than 2 years) and follows a chronic relapsing course [1-3]. While many children with eczema experience remission as they grow up, approximately 20-40% of those affected will continue to experience eczema as adults [4, 5]. The high direct and indirect costs of chronic eczema to families and health care systems [6], and significant morbidity and impairment to quality of life in individuals [7] have been documented.

Concern that the prevalence of childhood eczema (as well as the prevalence of asthma and rhinitis) was increasing in affluent countries led to the establishment of the International Study of Asthma and Allergies in Childhood (ISAAC) in 1991 [8]. ISAAC used standardised methods to measure the prevalence of symptoms of asthma, rhinitis and eczema among children and adolescents in order to allow valid comparisons between centres, both within and between countries [9]. Six New Zealand centres (Auckland, Bay of Plenty, Hawke's Bay, Wellington, Nelson and Christchurch, all in 1993) participated in ISAAC Phase One [10] with prevalence of current symptoms of eczema ranging from 12.0% to 16.6% and 12.1% to 13.8% for the 6-7 year and 13-14 year age groups respectively [11]. In ISAAC Phase Two, of the New Zealand centres, only Hawke's Bay participated and the prevalence of current symptoms of flexural eczema was 14% detected by questionnaire and 8.2% detected by skin examination [12].

In 2002-3 ISAAC Phase Three took place, including five centres in New Zealand, repeating the methodology of Phase One [13] so that time trends in prevalence of symptoms could be assessed [14]. Prevalence of symptoms of asthma in New Zealand has been shown in phases One and Three to vary by ethnicity with high prevalence among Māori children and adolescents [15, 16]. However little is known about variation in prevalence by ethnicity and time trends for symptoms of eczema in New Zealand. An environmental questionnaire was added with questions about a variety of potential protective and risk factors.

Our hypotheses were that the prevalence of eczema in New Zealand is increasing, that it is more common among Māori and Pacific than European children and adolescents, and that there are some avoidable risk factors. The research objective was to answer these three questions: is eczema prevalence changing over time?; are there ethnic differences in prevalence?; and what are the risk factors for eczema?

ISAAC Phase One (1993) and Phase Three (2001-3) in New Zealand surveyed parents of 6-7 year old children and 13-14 year old adolescents concerning symptoms of asthma, rhinoconjunctivitis and eczema (core questionnaire), and, in Phase Three, added questions about environmental factors which may be associated with these diseases. Detailed information concerning the Phase One and Phase Three study design and questionnaires has been reported previously [9, 13]. Ethics Committee approval was obtained for each centre and centres obtained their own funding.

The questionnaire included demographic information concerning the age, date of birth, gender and ethnicity of participants. The ethnicity question included options for 'European/Pakeha', 'Māori', 'Pacific Island' and 'Other' ethnicities and multiple selections were accepted. In cases where multiple options were selected a single ethnicity was allocated using a prioritisation of 'Māori', 'Pacific Island', 'Other' and 'European/Pakeha' as for Phase One [15].

Schools were selected randomly from a defined sampling frame (except for Nelson where all schools in the study area participated). An information letter about the study was provided to parent/guardians of all eligible children. The questionnaires were completed at home by the parent/guardian of children, and self-completed by the adolescents at school. Passive consent was implied by return of the completed questionnaire to school by the children or failure to object to participation of the adolescents. Passive consent is associated with increased response rates and is acceptable for non-invasive questionnaire-based research in community settings [17].

The questions used in the ISAAC core questionnaire concerning symptoms of eczema were:

The main outcome measures of interest reported here are current symptoms of eczema (those participants who answered 'Yes' to questions 2 and 3), and current symptoms of severe eczema (those participants with current symptoms of eczema who also answered 'One or more nights per week' to question 6). Results are also presented for question 7 ('eczema ever') to provide information about use of the label 'eczema'. Early onset eczema was defined as participants with current symptoms of eczema who also replied "Under 2 years" for question 4 above, used only for the 6-7 year age group.

The environmental questions concerned weight, height, a range of dietary factors, physical exercise, cooking and heating fuels in the home, paracetamol use, antibiotic use, number of siblings, birth in New Zealand, years lived in New Zealand, maternal education, truck traffic in street of residence, birth weight, breastfeeding, exposure to pets and farm animals, and exposure to environmental tobacco smoke in the home. Several questions concerning early life exposures were omitted from the 13-14 year age group questionnaire as adolescents could not be expected to provide valid responses.

Confidence intervals for the prevalence estimates were adjusted for the cluster sampling methodology. Logistic regression was used to examine whether there was more variation in prevalence of current symptoms of eczema, current symptoms of severe eczema and 'eczema ever' between centres than would be expected by chance, adjusted for gender, age, ethnicity, school decile (a surrogate measure of socio-economic status), and the cluster sampling methodology. Odds ratios for associations between the main outcome measures and environmental factors were also estimated using logistic regression, adjusted for gender, ethnicity, school decile, and the cluster sampling methodology. All analyses were carried out using SAS version 9.2 (SAS Institute Inc., Cary, NC, USA).

ISAAC Phase Three data was collected in Auckland and Wellington in 2001, Bay of Plenty in 2002, and Nelson and Christchurch in 2003. Four centres had response rates for both age groups greater than 80%: 6-7 year age group total 10,873 children, mean response rate of 85.2%, 13-14 year age group total 13,317 adolescents, mean response rate 89.2%. The response rate for the children in Wellington was low (47.2%) and data from this age group in Wellington was excluded from the analysis and the time trends analysis.

The Phase One and Phase Three results by centre, ethnicity and gender (including change per year) are presented in Table 1. The combined (all centres) prevalence of symptoms in children and adolescents respectively for current eczema was 15.0% and 8.8%; severe eczema 1.8% and 1.3%; 'eczema ever' 31.5% and 26.1%.

There was greater variation between centres than would be expected by chance for all outcomes for both age groups (p < 0.001) with the exception of current symptoms of severe eczema for the adolescents (p = 0.092). Auckland had the highest prevalence of severe eczema but lowest prevalence of 'eczema ever' in both age groups. Christchurch had the highest prevalence of current symptoms of eczema and 'eczema ever' in children. Wellington had the highest prevalence of current symptoms of eczema and 'eczema ever' in adolescents and Christchurch the lowest. Nelson had the lowest prevalence in children of current symptoms of eczema and severe eczema and in adolescents of severe eczema.

For the children, there was generally little change in prevalence by centre between Phase One and Phase Three for current symptoms of eczema and current symptoms of severe eczema but increases in prevalence for 'eczema ever' were observed across all centres for this age group. There was a different pattern for the adolescents, with decreases for current symptoms of eczema and, to a lesser extent, current symptoms of severe eczema, and little change in 'eczema ever'.

European/Pakeha and other participants had consistently lower prevalence than Māori who were lower than Pacific participants in both phases for current symptoms of eczema and current symptoms of severe eczema than, for Phase Three illustrated in Fig. 1. In contrast, for 'eczema ever', the pattern was reversed, with highest prevalence values found in European/Pakeha and Māori participants and lower values found for Pacific and other participants. There was no difference from the whole group in the pattern of change over time by ethnicity except for adolescents where decreases in prevalence for current symptoms of eczema and current symptoms of severe eczema were statistically significant only for European/Pakeha and Māori participants, and there were significant increases in prevalence of 'eczema ever' for European/Pakeha and Pacific participants.

All eczema symptoms were more common in girls than boys for both phases and both age groups. For the children, both genders showed a similar pattern of change between Phase Ones and Three as for the overall data. For the adolescents, the changes were similar for both genders for current symptoms of eczema (decrease) and current symptoms of severe eczema and 'eczema ever' showed little change.

Many environmental factors showed no association with eczema symptoms (Supplementary Tables 1 and 2). Positive associations (risk factors) with current symptoms of eczema, onset under 2 years, symptoms of severe eczema or 'eczema ever' were found for truck traffic in street of residence, current paracetamol use, paracetamol or antibiotic use in the first year of life. Inverse associations (potentially protective) were found for being resident in New Zealand fewer than 5 years, consumption of milk, and in the children only, current dog in the home, consumption of seafood, and eggs.

This study confirmed that eczema is common among children and adolescents in New Zealand. The prevalence of current eczema symptoms remains high, with New Zealand ranking 9th out of 60 countries among children and 39th among 96 countries among adolescents [18]. Current eczema symptoms are about twice as common in children (about 1 in 6) than in adolescents (about 1 in 12), although the difference in study methodology (parental report for the children vs. self-report for the adolescents) may be part of the explanation for this contrasting prevalence.

There was variation in prevalence between centres with inconsistent patterns except for Auckland which had the highest prevalence of current symptoms of severe eczema for both age groups but had the lowest prevalence of 'eczema ever' for both age groups. Nelson had the lowest prevalence of current symptoms of severe eczema for both age groups and also had a comparatively low prevalence of current symptoms of eczema in both age groups. Nelson had a lower prevalence of eczema in Phase One [11] and although various ideas for this lower prevalence have been put forward, there is no clear explanation.

Over the 9 year (average) time period somewhat different time trends were identified for the children (little change for current symptoms of eczema and severe eczema, and an increase for 'eczema ever') and adolescents (decreases for current symptoms of eczema and severe eczema, and little change for 'eczema ever').

Prevalence of symptoms of eczema was highest among Māori and Pacific participants for both phases. This contrasts with the Phase One asthma symptom results where European/Pakeha participants were intermediate between Māori and Pacific participants [15]. There was little evidence of changes in the burden of eczema symptoms between ethnic groups.

New Zealand children and adolescents have high rates of hospital admission for serious skin infection, which have been increasing over the last decade, have a clear social gradient and show marked ethnic gradients with Pacific 4.5 and Māori 2.8 times the European rate [19]. Infected eczema makes a modest contribution to these figures, and points to under-treatment of eczema being a factor, as well managed eczema and early treatment of infected eczema would not normally be followed by hospital admission. Moreover Māori and Pacific children and adolescents are disproportionately affected by socioeconomic disadvantage (poverty) and thus the direct and indirect costs of doctors visits and treatments for eczema may be unsustainable for some families [20].

Prevalence of all symptoms was higher among female participants which is consistent with many other studies [21]. Time trends in prevalence for both genders were generally similar, reflecting the overall patterns although there was an indication that the prevalence of severe symptoms decreased among females but not males in the adolescents.

Many potential aetiologic factors have been examined within New Zealand including antibiotics [22, 23], breastfeeding [23-25], family history [23, 24, 26], maternal smoking [23], migration [27], immunisation [23, 28], paracetamol [22] and pets [23]. We found few consistent associations between eczema symptoms and reported environmental factors. Potential risk factors of particular interest were truck traffic in street of residence, current paracetamol use, paracetamol or antibiotic use in the first year of life. Those which showed potentially protective associations with current symptoms of eczema, onset under 2 years, symptoms of severe eczema or 'eczema ever', were being resident in New Zealand fewer than 5 years, consumption of milk, and in the children only, current dog in the home, consumption of seafood, and eggs. Reverse causation could be the explanations for many of these findings: paracetamol or antibiotics could be given for eczema, and consumption of milk, eggs, seafood, and exposure to dogs avoided because of eczema.

The associations with birth in New Zealand and years lived in New Zealand are consistent with data presented by Waite et al. [27] regarding eczema among Tokelauan children living in New Zealand and Tokelau, and a study of immigrant children in Italy which reported lower prevalence and incidence of eczema [29], and are consistent with the ISAAC findings for asthma [30].

The association with truck traffic is also consistent with the global level ISAAC Phase Three results [31]. Positive associations have also been reported among children in Germany [32] and Taiwan [33], and among adults in Sweden [34]. The findings are concerning because of the increase in truck traffic in New Zealand roads.

We examined early onset of eczema (less than 2 years of age) in the children because mutations in the filaggrin gene (an important structural protein in the epidermis) are strongly associated with the early eczema phenotype [35, 36]. The pattern of eczema prevalence and its association with environmental factors was similar for all current eczema symptoms and early onset of eczema symptoms, showing no distinguishing features for early onset of eczema symptoms.

The strengths of this study include participation from centres from a number of regions within New Zealand and the standardised methodology, allowing valid comparisons between centres and ISAAC phases. The study was carried out to a high standard and achieved a high response rate. Rigorous data and methodology checks were used to reduce the risk of bias.

There are some limitations which should be acknowledged. Estimates of change per year in prevalence between Phase One and Phase Three have been presented in order to adjust for slightly different periods between phases among the participating centres. However, these values should not be interpreted as a consistent linear change. Recall bias may have affected the estimates of prevalence of 'eczema ever', and this may be worse for adolescents who may have difficulty accurately recalling early life events. For this study, eczema was defined by questionnaire, not examination by trained research staff. However an ISAAC Phase Two analysis comparing questionnaire and skin examination assessment of eczema has shown that the questionnaire is a valid measure in community-based studies [12]. The environmental questionnaire included a large number of questions (50 for the children and 38 for the adolescents) and the questions were intentionally simple in order to be useable and translatable throughout the world. However some environmental factors which would have been of interest could not be included and it was not possible to examine any factor in depth.

Finally, this study has shown that young New Zealanders continue to experience a significant burden of eczema and this burden is disproportionally higher among Māori and Pacific children and adolescents. It is vital that all New Zealand children and adolescents have timely, affordable access to optimal management of eczema.