The New Zealand Asthma and Allergy Cohort Study (NZA²CS): Assembly, Demographics and Investigations

The incidence of atopy, allergic disease and asthma is increasing in industrialised nations, including New Zealand. Longitudinal birth cohorts such as in Tucson USA [1] Melbourne [2] and Perth [3] Australia, and the German multi-centre atopy study (MAS-90) [4], as well as cross-sectional studies such as Belmont [5] in Australia have identified a number of potential risk factors for the development of atopy, allergic disease and asthma. These include a genetic predisposition, foetal growth [6] and a variety of early life environmental factors.

Exposure to house dust mite allergens is strongly associated with sensitisation [7], and in turn sensitisation is strongly associated with asthma [8]. The relationships between allergen exposure and the development of asthma appear more complex [9]. New Zealand has amongst the highest levels of house dust mite allergen in the world, and is therefore well placed to further explore such relationships. Although a birth cohort established 34 years ago in Dunedin has reported on determinants of asthma and atopy [8, 10], early life exposure information from this cohort is sparse, preventing detailed examination of current hypotheses.

This paper describes the establishment of a New Zealand birth cohort, to explore environmental and genetic factors associated with the development of atopy, allergic disease and asthma. Specifically we wished to explore the influence of perinatal factors, anthropometric measurements, infant infections, family size, immunisation, diet, environmental allergen and environmental tobacco smoke exposure on the development of atopy, allergic disease and asthma. There are a number of reasons for mounting such a study in New Zealand, including high prevalence of allergic disease and asthma, high allergen levels, and low industrial pollution levels. In addition, the New Zealand population has been extensively studied in the context of other international work [11, 12], allowing comparisons with previous studies to be undertaken. There is growing recognition of the importance of longitudinal birth studies in trying to understand the true progression of asthma and atopic disease and to identify risk factors [13]. Equally it is important to recognise their limitations and to document the assembly and investigation of birth cohorts. This has also recently been recognised by the publication of the study designs of asthma and allergic disease birth cohorts in Europe [14].

In this paper we have therefore summarised the assembly and study of a birth cohort of infants in New Zealand to age eight years.

A prospective birth cohort of 1000 infants born in two metropolitan centres (Christchurch, population 417,000, and Wellington, population 420,000) in New Zealand was established in 1997. Planning for the study started in 1996. The two centres were chosen because there were established respiratory research organisations in both cities, and the requirements for large numbers of infants and potential recruitment challenges necessitated two-centre recruitment. Details of the timeline for all assessments are given in Table 1.

Mothers and infants were recruited through midwives between 1997–2000 in Wellington and 1998–2001 in Christchurch. Expectant mothers were approached by their midwives to give written informed consent to be enrolled in the study prior to delivery. Midwives were selected randomly from all midwives practising in Wellington and Christchurch and were trained and supported by trial coordinators. Midwives were encouraged to discuss the study and try and enrol all expectant mothers under their care, to reduce selection bias. No explicit attempt was made to recruit families with a history of atopy, allergic disease or asthma, indeed the importance of enrolling all mothers was stressed. There were no explicit exclusion criteria in the peri-natal period; in particular non-singleton pregnancies were able to be included. Midwives were asked to keep a record of all mothers that they approached and to collect basic demographic information on them in particular mothers who declined to participate. In addition, efforts were made to ensure Maori participation by ongoing contact with Maori health providers and consumer groups. All mothers gave written informed consent for their children to participate in this study to monitor their health and development and undergo regular assessments in the form of interviews, questionnaires and a variety of specific tests. The initial informed consent process covered assessments for the first five years, with subsequent consent for the six and eight year assessments being obtained prior to these assessments (including consent for DNA collection, storage and genetic analysis-see below). Verbal agreement was obtained from each child prior to the six and eight year assessments, as required by our local ethics committees; if a child refused any particular segment, this was not undertaken.

All assessments were administered by trained research staff, using a pre-agreed manual of procedures. Extensive pre-testing of assessment procedures undertaken on non-participant families prior to use, and ongoing inter-centre quality control of assessments and staff was undertaken through the study.

Ethics committee approval for the conduct of the study was obtained from the Wellington and Canterbury Regional Ethics Committees.

In 1993 and 1994 a pilot study of 154 new born infants was undertaken in Wellington [15]. These children were followed with visits at 15 months and 6 years. This pilot provided insights into the design and practical problems associated with large cohort studies, in particular problems with the skin prick testing. There was an initial response rate of over 90% with a 15 months follow-up rate of 97%.

The statistical rationale behind a cohort of 1000 children was based on this pilot data for the major risk factors, and the prevalence of asthma was based on New Zealand research available at the time. At 15 months, with a follow-up rate of 90%, a prevalence of atopy of 20%, a prevalence of the major risk factors of 20% (e.g. high house dust mite exposure) the study will have 95% power to detect a relative risk of atopy of 1.7. At 6 years, with a follow-up rate of 75%, a 12 month period prevalence of asthma symptoms or atopy of 20%, a prevalence of the major risk factors of 20%, the study will have >90% power to detect a relative risk of 1.7. These assumptions are conservative and are in keeping with the findings of the ISAAC study, which found the prevalence of asthma symptoms in six-to seven-year old children to be 25% [11]. Furthermore our follow-up rate at 6 years is greater than the 75% predicted.

The demographics of the cohort are shown in Table 2. A total of 1105 children were recruited. The retention rate at three months was 96.3% and at fifteen months was 91.4%. As of April 2006, a total of 421 children have reached age six years, and have undergone the six-year assessment in Wellington; 264 children have been assessed in Christchurch (which started recruitment later and will finish six-year assessment in 2007). 127 children have to date completed the eight-year questionnaire in Wellington. 15.2% of the children at recruitment have been identified as Maori. A positive family history of asthma, eczema or hay fever has been identified in 84% of children.

The gender ratio in the cohort is 561:544 (50.8%:49.2%) (Male:Female), with 45.9% being first-born, 35.5% being second-born, and 18.6% being third-born or higher. Maternal age ranges from 15.5 years to 46.7 years.

Cohort studies always present a considerable challenge to recruit and retain a representative sample of a population. We have assembled a population based birth cohort of over 1000 children in Wellington and Christchurch with good retention rates. We specifically sought to obtain a representative population sample rather than to recruit families with a history of atopy, allergic disease or asthma in order to provide greater generalisability of the results. We were reliant on midwives for recruitment of mothers in their care. We randomly sampled midwives from a variety of maternity care facilities including local hospital birthing units, secondary care facilities and a number of independent midwife practices. We met with the midwives regularly and asked them to approach all mothers in their care and to keep a record of all mothers that declined to participate. This proved very difficult in practice and we have no reliable information on how many mothers declined, although we believe that in almost all cases midwives did approach all mothers in their care. We are confident however that the cohort sample is largely representative of the wider New Zealand population, based upon our comparisons with national demographic statistics. For example, in the NZ Census of Populations and Households 2001, 14.7% of the population identified as Maori. Overall, in Christchurch and Wellington, 15.2% of the cohort children have been identified as Maori. Also parity in the cohort is not dissimilar from the NZ National Statistics [34]. We had a slightly higher proportion of 1^st born children (46% cf 41% nationally), the same proportion of second born children (36%), and a slightly lower proportion of children with higher sibships (19% cf 24% nationally). The national statistics were based on births within marriages which may explain the difference. Recent birth cohort studies have tended to recruit from University or academic teaching hospitals where high risk pregnancies or deliveries are likely to be over represented and may therefore differ from the general population. The issues raised by non-random sampling for these birth cohorts are discussed by Keil and co-workers who review recent European asthma and allergy birth cohort studies as part of the GA²LEN (Global Allergy and Asthma European Network) initiative. [14].

Another concern in cohort assembly is participation bias, the inevitable increased participation of subjects who have a family history of the conditions being studied. This cannot be prevented but the extent to which it is present can be assessed by collecting information on non-responders and those not wishing to take part. In assembling this cohort such information was not available for the reasons discussed above. To assess participation bias we have compared a family history of asthma, eczema or hay fever amongst the cohort infants with the participants in the Dunedin multidisciplinary cohort study [8]. We defined family history as either parent having a history of asthma, eczema or hayfever. Using this definition in the most recent round of the Dunedin study (response rate 96%), when subjects were aged 32 years, (similar to the average age of mothers in our cohort) 70% of female cohort members reported a history of one or more allergic diseases and 64% of males, (personal communication Dr R. Hancox, Deputy Director, Dunedin Multi-disciplinary cohort study) showing that 89% of random pairs of the Dunedin cohort would have a history of allergic disease. This suggests that our cohort is a reasonably unbiased sample with no obvious evidence of participation bias due to a family history of allergic disease.

The initial decision to use the Quintest equipment for the fifteen-month skin test was made after initial piloting suggested difficulties in performing SPT's using standard prick methods in 15 month old infants. It was considered that the Quintest would offer a number of advantages including ease of administration, reducing distress because of quicker testing times; standardisation of technique; less effect from child movement, and the simultaneous completion of all tests. During the course of the study however we became concerned that the Quintests were producing smaller wheals than would have been expected using standard techniques. In 2000, we conducted a study to examine the Quintest and compare it to a standard prick-through application. Not only did we confirm smaller wheal sizes from the Quintest, amongst atopic adults, but we also we showed that this would lead to a 25% misclassification of atopic status, when using a mean wheal diameter of 3 mm as a cut off [35]. The Quintest also tended to produce more positive responses to the negative control. We also found that there were significant differences between two commonly used sources of commercial allergen preparations, with allergens from one source consistently producing smaller wheals regardless of the methods used to apply them. Once again the source of allergens also led to a 25% misclassification of atopy. In order to reduce the impact of the smaller wheal sizes from the Quintest we have used the histamine to wheal ratio to define atopy. Thus for each allergen after subtraction of the negative control, atopy was defined as a histamine allergen ratio ≥0.5. This ratio assessment has been recommended by a number of investigators particularly in young children [22] and reduces the impact of smaller wheal sizes due to the Quintest device. It is important to recognise that the assessment of atopic status in infancy and early childhood can only ever be a partial assessment. Many children will exhibit transient food sensitisation and many others will not yet have developed sensitivity to environmental allergens.

A number of birth cohorts have been established internationally to study a wide variety of factors associated with the development of atopy, allergic disease and asthma [14]. Given the recent hypotheses that the development of these conditions involves a complex interaction between genetic factors, and timing and magnitude of different environmental factors, there are significant advantages to investigators sharing testing protocols and information at early stages, in order to test hypotheses under different environmental conditions. The GA(2)LEN project funded by the European Union is an example of this approach. The high prevalence of allergic disease and asthma in New Zealand, with high allergen levels and low industrial pollution levels, allows testing of hypotheses generated by both cross-sectional and longitudinal studies in a unique environment. We present information about our cohort to contribute to this process and encourage interested researchers to contact the corresponding author for further information.