European birth cohort studies on asthma and atopic diseases: II. Comparison of outcomes and exposures – a GA2LEN initiative

Authors


Thomas Keil, MD, MSc
Institute for Social Medicine, Epidemiology and Health Economics
Charité University Medical Center Berlin
Luisenstr. 57
D-10117 Berlin
Germany

Abstract

Background:  The Global Allergy and Asthma European Network (GA2LEN) is a consortium of 26 leading European research centres committed to establish a European research area of excellence in the field of allergy and asthma.

Aim:  One of the GA2LEN work packages was designed to identify and compare the existing European birth cohort studies on asthma and atopic diseases. The present review compares their subjective and objective outcomes as well as exposure variables.

Methods:  A common database was established to assess study characteristics of observational birth cohort studies designed to examine asthma and atopic diseases. Data were collected by visiting most of the participating research teams and interviewing all relevant study personnel. For each study, the type of objective/subjective outcome parameters and potentially influential factors were recorded precisely for every time point during follow-up.

Results:  Eighteen birth cohort studies on asthma and atopic diseases were identified in eight European countries. Thirteen studies collected data on specific immunoglobulin E (IgE) to various inhalant and food allergens, whereas 12 performed skin prick tests (many at several time points during follow up). Several studies measured lung function, but across the cohorts no comparable standard procedures were used. For subjective evaluation of asthma and allergic rhinitis most studies applied the ISAAC questionnaire (sometimes modified), whereas the assessment of eczema was rather heterogeneous across the studies.

Conclusion:  This GA2LEN initiative established a unique common database of 18 European birth cohorts on asthma and atopic diseases. For selected cohorts, it seems that pooling data and performing common analyses may be possible to examine associations between certain exposure variables (e.g. pet ownership, tobacco smoke exposure and day-care) and selected outcome measures for atopy, asthma or allergic rhinitis (e.g. sensitization assessed by IgE or skin prick tests, doctor's diagnosis of asthma, parental perception regarding asthma/wheezing or hay fever symptoms).

Background

The Global Allergy and Asthma European Network (GA2LEN) is a consortium of 24 leading European research teams and two European nongovernmental research societies [European Academy of Allergy and Clinical Immunology (EAACI) and European Federation of Asthma and Allergy Associations (EFA)] selected by the European Commission, and committed to establish a research area of excellence in the field of allergy and asthma. The network intends to harmonize diagnostic and therapeutic procedures, accelerate the application of research results to clinical practice, patients’ needs and policy development, as well as to promote training and integration between public and private sectors (1, 2).

One of the work packages within the GA2LEN project was designed to identify all European birth cohorts on asthma and atopic diseases, to create a common database of the different study characteristics, which would allow comparison of study methods including design and assessment of outcome as well as exposure parameters. The information gained from this process could be used as a basis for evaluating the possibility to pool data and perform common analyses as well as to recommend criteria for conducting future birth cohorts or follow-up assessments of existing cohorts.

A previous report (part I of the review) described the establishment of a common database of 18 European birth cohorts on asthma and atopic diseases assessing study design, study setting, target population, recruitment criteria and follow-up rates (3). The cohorts started at different times (between 1985 and 2004), consequently the children are of a range of ages and the amount of data collected within different cohorts reflects this. Most studies plan further follow up and the database will be updated accordingly. The present report (part II of the review) compares the assessment of objective and subjective outcome parameters as well as of relevant exposure parameters at each time point during follow up.

Methods

We created a common database to assess relevant study characteristics of European observational birth cohort studies on asthma and atopic diseases as described in detail elsewhere (3). The structure of the existing database allows to enter further information for existing as well as for new birth cohorts.

In summary, data for this database was collected by visiting most of the participating research teams and interviewing all relevant study personnel. For each study, the type of objective/subjective outcome parameters and potentially influential factors were recorded precisely for every time point during follow-up. We also assessed the numbers of children included in each of the following parameters:

  • (i) asthma, allergic rhinitis, eczema (validated or own questions);
  • (ii) food and other allergies;
  • (iii) diet (validated or own questions);
  • (iv) medication for asthma or atopic diseases;
  • (v)  lung function tests;
  • (vi)  skin tests;
  • (vii)  immunoglobulin E;
  • (viii)  collected/stored urine and/or serum and/or DNA samples;
  • (ix)  dust samples;
  • (x)  environment (including housing, animal contact);
  • (xi)  siblings;
  • (xii)  day-care/kindergarten/school;
  • (xiii)  tobacco smoke exposure;
  • (xiv)  parental socioeconomic status;
  • (xv)  pet ownership;
  • (xvi)  vaccinations;
  • (xvii) weight/height;
  • (xviii) additional other data.

Results

Outcome parameters (objective)

Immunoglobulin E (IgE).  Thirteen of the 18 birth cohorts on asthma and atopic diseases investigated the development of specific IgEs to various aero- and food allergens. MAS (Germany) and DARC (Denmark) assessed sensitization at 8 time points (particularly in early childhood twice or more per year), Odense 1985 (Denmark) at 5, ECA (Norway), BAMSE (Sweden), MAAS (UK) and PIAMA-NHS (The Netherlands) at 3 time points during follow up. Yearly, between 3 and 6 studies performed IgE tests resulting in data for 1100–4200 children combined for each year of the first 6 years of life. BAMSE (Sweden) and LISA (Germany) collected the largest amount of IgE data, each for >2000 children (Table 1). Using the Pharmacia CAP system (4, 5), most studies determined specific IgE to mite, cat, dog, grass, birch, milk and egg, some studies additionally to moulds, peanut, wheat, soy, fish (cod, trout), horse or cockroach. Cord blood was collected in 14 studies from over 11 000 children to perform IgE measurements as well as further analyses (e.g. specific IgG, CD14), in 9 studies the cord blood was stored (data not shown).

Table 1.   Number of children with data on sensitization (IgE) after birth in European birth cohorts on a asthma and atopic diseases
Study nameTime points (years)
inline image*inline image–1−2−3−4−5−6−7−8−9−10−11−12−13−14−15
  1. *Additionally, cord blood IgE data are available from most studies.

Odense 1985276 232  198     188    162
Isle of Wight           953     
MAS 802676628 654526687   579  428  
ECA  490       1019     
AMICS-Ashford     389   454       
BAMSE  320 2614    2442      
MAAS 223 186 521  ongoing       
PIAMA-NHS     416   ongoing       
GINI-B      945         
AMICS-Barcelona                
AMICS-Menorca     369           
LISA  2176             
SEATON                
Amnio                
DARC441388329296  352         
AMICS-PAULA 410              
GEPSII Rome                
CO.N.ER Bologna                
Total, estimated700140042001100380014001800687>100024422700428162

Skin prick testing.  Twelve birth cohorts performed skin prick tests, 5 of them examined the study children at least 4 times during follow up: Odense 1985, DARC (both in Denmark), Isle of Wight, MAAS and Amnio (all in the UK). Up to the age of 10 years there was data available in 2 to 7 studies for almost each year of follow up. Total number of children ranging from 1000 to >2500 per year (Table 2). Using skin prick solutions of different companies, most studies tested for grass and tree pollen, mite, cat, dog, milk and egg allergens, while other allergens (such as from other foods) were considered less frequently (data not shown).

Table 2.   Number of children with data on sensitization (skin prick tests) in European birth cohorts on asthma and atopic diseases
Study nameTime points (years)
inline imageinline image–1−2−3−4−5−6−7−8−9−10−11−12−13−14−15
  1. *Only symptomatic children were tested.

Odense 1985  232  198     188    162
Isle of Wight 344*436* 981     1036     
MAS     700          
ECA  526       1011     
AMICS-Ashford      552 548       
BAMSE  320             
MAAS 386 813 814  ongoing       
PIAMA-NHS        ongoing       
GINI-B                
AMICS-Barcelona                
AMICS-Menorca                
LISA                
SEATON   151  700          
Amnio 198 181 170 150           
DARC511468 495 522  401         
AMICS-PAULA     ongoing          
GEPSII Rome                
CO.N.ER Bologna                
Total, estimated5111400230015001100>25001000>10002200162

Lung function tests.  Eleven birth cohorts assessed lung function in their study children, up to 4 times during follow up, some only in subgroups of the total study population. For the age groups 4–6 years and 9–10 years most data of lung function from the European birth cohorts are available (Table 3). There is only little data in very young children reflecting the fact that children of this age often cannot perform forced expiratory manoeuvres reliably. However, the ECA study (Norway) did lung function tests (e.g. tidal flow volume in the awake state) in over 800 newborns, of whom 614 were re-examined at 10 years of age. Across the studies and age groups there was no standard procedure of measuring lung function. Mostly spirometry was performed (although with many different devices across the studies), sometimes also as part of a bronchial challenge test which was used to determine bronchial hyperresponsiveness to either cold/dry air, metacholine, histamine, or exercise. A few studies tested reversibility by the application of bronchodilators, but using different substances. Furthermore, some birth cohort studies measured airway resistance using the interrupter technique (Rint), several others also examined exhaled NO (data not shown).

Table 3.   Number of children with data on lung function tests at different time points in European birth cohorts on asthma and atopic diseases
Study nameTime points (years)
inline imageinline image–1−2−3−4−5−6−7−8−9−10−11−12−13−14−15
Odense 1985     256     226    219
Isle of Wight           981     
MAS       799   638  640  
ECA.803 316       1019     
AMICS-Ashford       573 565       
BAMSE    2927    2612      
MAAS 38 536 782  Ongoing       
PIAMA-NHS     484   Ongoing       
GINI-B      2183         
AMICS-Barcelona                
AMICS-Menorca                
LISA                
SEATON     486          
Amnio                
DARC       401         
AMICS-PAULA                
GEPSII Rome                
CO.N.ER Bologna                
Total, estimated80138316536340015003200799>100026122800640219

Outcome parameters (subjective)

For the assessment of symptoms regarding asthma (wheezing) and allergic rhinitis, most birth cohorts used standardized questions (sometimes modified) from the ISAAC questionnaire. To further assess allergic respiratory symptoms, almost all studies developed additional own questions. Some studies also had a special focus on possible allergic reactions to food: Odense 1985, DARC (both in Denmark), Isle of Wight, MAAS (both in UK), GINI-B, LISA (both Germany), GEPSII Rome, CO.N.ER Bologna (both in Italy). To assess eczema signs/symptoms most cohorts developed their own questions or used the ISAAC method (e.g. PIAMA-NHS, NL). To determine current severity of eczema, six studies used the standardized and validated scoring system SCORAD during a physical examination: MAS, GINI-B, LISA, AMICS-PAULA (all in Germany), ECA (Norway) and DARC (Denmark). Only a few studies, such as MAS (Germany) evaluated quality-of-life of the children, using a standardized general-health related instrument, KINDL (6), as a further subjective outcome parameter.

Exposure variables

Across the studies a broad spectrum of exposure parameters was assessed (exposure in the epidemiological sense, meaning any objective and/or subjective factors that can potentially influence disease).

Biological factors.  To examine possible genetic determinants of asthma or atopic diseases 13 birth cohorts collected blood, either at baseline or during follow up, to extract DNA (in some studies the extraction of DNA is pending). In some cases buccal swabs were collected for DNA extraction with whole genome amplification. All studies combined, DNA is available from over 13 000 children (Fig. 1).

Figure 1.

 Number of children with blood or buccal samples for DNA (n = 13232) in the European birth cohorts on asthma and atopic diseases with a total of 28 277 recruited children.

Most studies sequentially measured height and weight allowing calculation of body mass index. Other biological samples were collected such as urine, cord blood and also peripheral blood mononuclear cells for detailed immunological assessments. Many studies have serum and/or plasma stored for future measurements of inflammatory markers or for proteomic studies.

Environmental and psychosocial factors.  During the first year of life, 11 studies collected dust samples to analyse exposure to mite or pet allergens as a risk factor for sensitization or allergic diseases. Six studies did so between the second and the sixth year. Some studies, such as MAS, LISA, AMICS-PAULA (all in Germany), MAAS (UK) and PIAMA-NHS (the Netherlands) measured endotoxin in dust samples, others have stored the samples for possible analyses of allergens or endotoxin (e.g. DARC, Denmark). Assays for the detection of indoor allergens in house dust used monoclonal antibodies to major allergens by ELISA technique (e.g. ALK, Horsholm, Denmark; Indoor Biotechnologies, Manchester, UK). The methods were not standardized across studies: samples were taken (i) from varying locations, such as the children's and/or parents’ mattress, bedroom and/or living room floor or the sofa; (ii) by study team members or by parents; (iii) by supplied or by own devices; (iv) some studies sieved the dust while others did not (Table 4).

Table 4.   Number of children with exposure data from dust samples in European birth cohorts on asthma and atopic disease
Study nameTime points (years)
inline imageinline image–1−2−3−4−6>6–9>9–10>10
  1. f, floor; mc, mattress child; mp, mattress parents.

Odense 1985         
Isle of Wight         
MAS    969 (f)  718 (f)234 (f) 531 (f) 755 (f)   
ECA  2000 (f)
 516 (mc)
    950 (mc) 
AMICS-Ashford   618 (f, mc)
   128 (mp)
        
BAMSE  4089 (mp)   2554 (mc)    
MAAS  947 (f, mc, mp) 850 (f, mc)  834 (f, mc)   
PIAMA-NHS 1041 (f, mc, mp)    503 (f, mc)    
GINI-B         
AMICS-Barcelona   487 (mc)        
AMICS-Menorca   482 (mc)        
LISA 2166 (f, mc, mp)        
SEATON         
Amnio  196 (f, mc, mp)       
DARC   454 (mc) 388 (mc) 329 (mc)450 (mc)  325 (mc)   
AMICS-PAULA   548 (f, mc, mp)        
GEPSII Rome         
CO.N.ER Bologna         
Total, estimated  5200 (f)
10000 (mc)
  8000 (mp)
1100 (f)
1500 (mc)
1100 (mp)
2700 (f)
800 (mc)
1100 (f)
1300 (mc)
1000 (f)
3000 (mc)
1550 (f)
1200 (mc)
950 (mc)

To varying degrees, all studies collected subjective information regarding socioeconomic status, tobacco smoke exposure, pet ownership, day-care, siblings or breast feeding. Some studies extensively assessed the home environment, e.g. by personal visits from environmental inspectors or trained study team members and/or by detailed questionnaires: Isle of Wight, MAAS (both UK), ECA (Norway), BAMSE (Sweden), LISA (Germany), and the four AMICS-studies (UK, Spain and Germany). A few studies collected detailed information on psychological factors, others used very detailed questionnaires on dietary intake with measures of maternal vitamin levels (for SEATON, UK, the influence of diet on the risk of allergies was the primary study objective).

Discussion

We have previously reported on the construction of a database of 18 European birth cohort studies specifically designed to investigate influential factors for the development of asthma and atopic diseases, and have compared characteristics of the various study designs (3). The present review (part II) compares the assessment of outcome measures and exposure parameters (both subjective as well as objective) in the 18 European birth cohorts.

Objective outcome parameters

Seventeen of the 18 birth cohorts objectively assessed atopic diseases by measuring specific IgE and/or performing skin prick tests to various allergens. Specific IgE-tests to common allergens in Europe such as mite, cat, dog, grass, birch, milk and egg were measured in many studies at different time points during follow up. Taking the children's age at blood sampling into account, for some of the cohorts pooling of data and performing common analyses seem possible. It could be of particular interest to combine data for allergens that are less common causes of sensitization. Such allergens have not been studied well in the past due to insufficient statistical power of single studies. A possible limiting factor for common analyses of specific IgE results could be the lack of standardized laboratory methods, including heterogeneous allergen extracts. However, at least for IgE to pollen, concordance of results in different assays seem to be satisfactory (7–11). Ideally, sub-samples of stored serum could be used to determine whether significant differences of IgE-values exist between the laboratories involved in the European birth cohorts.

For the diagnosis of allergic sensitization many studies also used skin prick tests, as they are easy to perform, relatively inexpensive and the results are immediately available. However, different standard panels of allergens are used across Europe, as published recently in an overview of the methods applied for skin prick testing in 29 allergy centres within GA2LEN (12). For several of the birth cohort studies included in the present paper, common analyses of skin prick tests seem possible at least for common allergens (cat, dog, mite, grass and tree pollen, cow's milk or hen's egg). Most studies evaluated results after 15–20 min as recommended (13, 14) and recorded crude mean wheal diameters (although the definition of a positive test varied). As for IgE, a possible limiting factor for comparing skin prick test results between studies could be the lack of standardized allergen extracts which contain a variety of different proteins, glycoproteins and polysaccharides. This could be the reason for a poor correlation between two different allergen extracts in a study from the Netherlands examining 22 adults (15). However, particularly for children, There is a lack of data are available on the comparison of extracts from different manufacturers.

Eleven studies have objective data on (allergic) asthma or respiratory disease from lung function tests, but only few in preschool children. As methods (and devices) to assess lung function vary considerably, pooling data from the different birth cohorts does not seem reasonable (16–19).

Subjective outcome parameters

Many studies assessed asthma and allergic rhinitis symptoms using standardized questions according to the ISAAC protocol. Therefore, selecting parental perception of asthma/wheezing or hay fever symptoms (‘ever’ or ‘last 12 months’) or ‘doctor's diagnosis of asthma’ could potentially include data from many of the GA2LEN birth cohorts for common analyses of these subjective outcome parameters.

Exposure variables

No standardized questions for family history of allergic diseases, pet ownership, socioeconomic status, tobacco smoke exposure, housing, day-care etc., were used. However, by some adaptations, it should be possible to examine these potentially influential factors in common data analyses.

Regarding indoor allergens from dust samples (mainly between birth and age 6 years), the different dust collection methods need to be considered. To minimize sampling device bias, Wickens et al. recommended that, allergen and endotoxin are expressed as a concentration, and that the bed is considered the major source of allergen exposure. To have confidence in comparisons of allergen and endotoxin reservoir levels between centres, standardization in the use of sampling equipment is important (20, 21).

Conclusions

  • 1The unique cooperation between the participating research teams of the GA2LEN project established a common database of 18 European birth cohorts on asthma and atopic diseases. Measurements have been made at different ages but within the 18 studies there is a degree of overlap which would facilitate collaborations, such as the opportunity to pool data and perform common analyses. Furthermore, the ‘younger’ birth cohorts could design if they so wished the future follow-up time points to coincide with some of the ‘older’ cohorts.
  • 2Prevalence data of asthma and/or atopic diseases might be difficult to compare, because study settings, sampling strategies and methods are quite different.
  • 3To examine possible associations between certain exposure variables and selected outcome measures for asthma, allergic rhinitis or atopy, we see a possibility to pool data and perform common analyses for some cohorts regarding the following endpoints:
  • objective outcome variables, such as sensitization assessed by IgE or skin prick tests to common indoor, outdoor and food allergens;

  • subjective outcome variables, such as specific ISAAC questions on asthma/wheezing or allergic rhinitis (‘ever’ and ‘last 12 months’) or doctor's diagnosis of asthma.

  • 4Exposure variables, such as family history, pet ownership, tobacco smoke exposure, day-care and many others, were assessed comprehensively in most studies but not in a standardized way. Therefore, in order to pool data and perform common analyses to examine cause–effect relationships, it would be necessary to define uniform categories for the exposure variables.
  • 5For the interpretation of effect estimates regarding associations between certain exposure variables and outcome measures it is important to investigate carefully the potential sources of heterogeneity between the birth cohorts.

Acknowledgments

The study was supported by the Global Allergy and Asthma European Network (GA2LEN) under the Sixth Framework Programme for Research of the EU. Collection of data from the UK studies was funded by Asthma UK. We express our special gratitude to the GA2LEN-office as well as to all principal investigators and contributors of the 18 included birth cohorts. In particular, we would like to thank H. Arshad (Isle of Wight study, UK); P. Cullinan (AMICS, Ashford, UK); G. Devereux (SEATON, Aberdeen, UK); G. Faldella (CO.N.ER, Bologna, Italy); F. Forastiere (GEPSII, Rome, Italy); O. Herbarth (LISA, Leipzig, Germany); D. Porta (GEPSII, Rome, Italy); B. Schaaf (LISA, Bad Honnef, Germany); A. Seaton (SEATON, Aberdeen, UK); J. Sunyer (AMICS, Barcelona, Spain); M. Torrent (AMICS, Menorca, Spain); E. von Mutius (AMICS-PAULA, Munich, Germany); J. Warner (Amnio study, Southampton, UK); H.-E. Wichmann (GINI-B and LISA, Munich, Germany) and A. Woodcock (MAAS, Manchester, UK).

Ancillary