Article Text

Download PDFPDF

How much asthma is really attributable to atopy?

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

In recent decades it has become routine to describe asthma as an atopic disease. A theoretical paradigm has evolved in which allergen exposure produces atopic sensitisation and continued exposure leads to clinical asthma through the development of airways inflammation, bronchial hyperresponsiveness, and reversible airflow obstruction. As Martinez1 notes, this paradigm has been used with particular insistence with regard to house dust mite allergens,2 ,3 but other allergens (cat, cockroach, dog) are also believed to be important. The importance of atopy is most widely accepted for asthma in children whereas, among adults, asthma has traditionally been divided into “extrinsic” and “intrinsic” asthma, although this also has been challenged.4 It is acknowledged that not all cases of asthma fit this paradigm—for example, some occupational causes of asthma do not appear to involve atopy—but these are regarded as interesting minor anomalies that do not threaten the dominant paradigm.

In this review we assess the extent to which the development of asthma is attributable to atopy (we do not consider the separate issue of the extent to which the development of atopy itself is attributable to allergen exposure, although this is also the subject of debate1). We start by considering definitions of asthma and atopy and then review evidence on their association in random population surveys. We do not intend to argue that atopy does not play an important role in the development of a significant proportion of asthma cases. However, our concern is that the proportion of asthma cases attributable to atopy may have been overestimated, and that other possible aetiological mechanisms and risk factors for asthma may therefore have been neglected.

Definitions of asthma and atopy

The definition of asthma is still controversial but an appropriate definition is a precondition for addressing the issues considered in this paper. The term “asthma” encompasses a disparate group of disorders which produce similar clinical effects—that is, variable airflow obstruction5—and this has formed the basis of the definition of asthma.6 ,7 Some current definitions also emphasise the importance of airways inflammation,8although the relationship between airways inflammation and variable airways obstruction is not straightforward.9-12 In some studies asthma has been defined more restrictively in terms of the immunological or pathophysiological mechanisms by which variable airflow obstruction is presumed to have occurred—for example, atopy or bronchial hyperresponsiveness (BHR). However, this leads to a significant proportion of asthma cases being excluded, and the relationship between “asthma” and atopy or BHR then becomes merely tautological. Thus, asthma is best defined in terms of the phenomena involved—that is, variable airflow obstruction—without making any restrictions based on possible aetiological considerations.13 ,14 For these reasons we have focused on studies that used physician diagnoses of asthma or self-reported asthma or asthma symptoms. We did not use definitions of asthma based on BHR since this would also lead to a considerable proportion of asthma cases being excluded,15 and because BHR is part of the causal model that we are assessing (in fact, atopy is more strongly associated with BHR than it is with airflow variability16).

“Atopy” has previously been used as a poorly defined term to refer to allergic conditions which tend to cluster in families, including hay fever (allergic rhinitis), asthma, eczema, and other specific and non-specific allergic states.17 More recently, atopy has been characterised by the production of specific IgE in response to common environmental allergens,18 and skin prick testing provides a convenient test for atopy in epidemiological studies.19 However, it has been suggested that total serum IgE provides an overall estimate of the allergic component in asthma,4 and that total serum IgE is associated with asthma independently of specific IgE levels.20 In this review we therefore focus on studies of skin prick test positivity, but we also consider studies of total serum IgE levels.

To assess the association of atopy with asthma in individuals and in populations we conducted a Medline search from 1980 onwards for English language publications of studies that contained at least one of the key words “hypersensitivity, immediate”, “hypersensitivity”, “IgE”, or “skin tests”. From these we selected epidemiological studies on asthma (“asthma” combined with “cross-sectional studies”, “case-control studies”, “longitudinal studies” or “prevalence”; or “respiratory tract disease/epidemiology”; or “asthma/epidemiology”; or “bronchitis/epidemiology”). We then selected only population based studies with a source population of at least 600 subjects (in some instances these were prevalence case-control studies and the total number of cases and controls was less than 600, even though the source population was larger than 600). For the reasons noted above we excluded studies that used BHR in their definition of asthma. We also excluded studies which did not report the proportions with atopy or with raised total serum IgE levels among cases and non-cases, and studies using IgE levels measured before the age of one year. For studies with multiple publications we only used one report.

Association of atopy with asthma in individuals

The proportion of asthma cases that are “attributable” to atopy (defined as skin prick test positivity) can be estimated by the “population attributable risk”. If exposure has an odds ratio for asthma of R (the odds ratio is the appropriate measure to use in prevalence studies and prevalence case-control studies15), then the proportion of exposed cases that are attributable to exposure is (R—1)/R, and the proportion of all cases in the population that are attributable to exposure (population attributable risk) is P(R—1)/R where P is the proportion of all cases that are exposed.


The proportions of asthmatic and non-asthmatic subjects who are skin prick test positive vary considerably between different studies4 ,21-34 (table 1). The weighted mean of the estimates from these studies, mostly conducted in Western countries, indicates that overall about 58% of children and 54% of adults with asthma were skin prick test positive; however, about 29% of non-asthmatic children and 24% of non-asthmatic adults were also skin prick test positive. The proportion of cases attributable to atopy varied from 25% to 63% in cross-sectional studies exclusively or predominantly in children, with a weighted mean of about 38%; in studies exclusively or predominantly in adults, the population attributable risk varied from 8% to 55% with a weighted mean of 37%. Thus, the population attributable risk was similar in children and adults, but it should be emphasised that these studies were not all done in the same populations and, in fact, studies performed within a single population (for example, reference 4) found higher attributable risks in children than in adults.

Table 1

Percentage of asthma cases attributable to atopy (defined as at least one positive skin prick test) in population based studies

It might be argued that the definition of skin prick test positivity used in these studies is too weak or non-specific since it is based on “at least one positive skin test” from a range of allergens. We therefore considered data from Sears et al 22 because this publication included the necessary information and the population attributable risk of atopy for asthma in this study (34%) was similar to the average for all studies in children in table 1 (38%). Table 2 shows that, as the definition of atopy (in terms of the number of positive skin prick test responses) is strengthened, the association with asthma also strengthens—that is, the odds ratio increases—but the population attributable risk decreases from 34% to 3% because of the reduction in the proportion of the population that is “positive”. Thus, when a more “severe” definition of atopy is used—for example, four or more positive skin prick tests—the association with asthma (as reflected in the relative risk estimate) increases, but the proportion of cases that are atopic (according to the more severe definition) decreases and the population attributable risk decreases. Similar analyses for the allergen-specific test results (table 2) showed that some specific allergens had stronger associations with asthma, but that the highest estimate of the population attributable risk (34%) is obtained with “any positive skin test”.

Table 2

Percentage of asthma cases attributable to atopy based on number of positive skin prick tests and specific skin prick tests


An obvious limitation of these data is that skin prick testing uses a wide range of allergens believed to be predominant in the area under study, but it will not necessarily identify all cases of atopy. It is therefore important also to consider atopy as defined in terms of total serum IgE, since it has been suggested that it provides an overall estimate of the allergic component in asthma4 and because it is more readily comparable between studies. Table 3summarises the studies of total serum IgE4 ,20 ,28 ,33 ,35 ,36 and shows that the population attributable risk of atopy (defined as a raised total serum IgE level) for asthma varied from less than 0% (an inverse association) to 80% with a weighted mean of 33%.

Table 3

Percentage of asthma cases attributable to atopy (defined as a total serum IgE level of 100+ IU/ml) in population based studies

Once again, these findings depend on the cut off point that is used to define atopy, and Burrows et al 4 have argued that some type of IgE mediated process may be involved in almost all asthma cases, even when skin test reactivity to common allergens is not found. Thus, if a more liberal definition of atopy (in terms of raised serum IgE levels) is used, a higher proportion of asthmatic subjects might be considered to be atopic. Table 4 shows data from Burrows et al 36 with the attributable risk estimates that would be obtained using different cut off points for serum IgE. Once again this study was chosen because it reported the relevant information, and because the population attributable risk of atopy for asthma in this study (30%) was similar to the weighted average for all studies (33%). The proportion of cases attributable to atopy continues to increase as the definition of atopy is “loosened” but, even using the most liberal definition (in which 85% of non-asthmatic patients and 95% of asthmatic patients are considered to be atopic), only about two thirds of the asthma cases are attributable to atopy.

Table 4

Percentage of asthma cases attributable to atopy, using different cut off levels for total serum IgE

These findings are likely to be underestimates because non-differential misclassification of atopy and/or asthma will usually bias the relative risk estimate towards the null value.15 On the other hand, the association between total serum IgE and asthma may in part, at least in some cases, simply be an association rather than reflecting a causal link. In particular, Sunyer et al 20 have suggested that: co-inherited genetic factors could increase susceptibility both to asthma and to the production of raised serum IgE levels; total serum IgE levels could in part be a consequence of asthma itself and could be a marker of non-allergic inflammation; or that IgE could also express a humoral autoimmunity since specific reactivity against human proteins structurally similar to allergens has been described. In each of these situations the association between total serum IgE and asthma would not be entirely causal, and the findings presented here would therefore be overestimates.

Association of atopy with asthma in populations

It is also of interest to consider the associations of the atopy measures with asthma prevalence at the population level, particularly in light of the global increases37 and the substantial international differences38 ,39 in the prevalence of asthma.

Table 5 summarises studies, identified from the same Medline search used for table 1, in which asthma and atopy (skin prick test positivity) were measured in the same population at different times, or in two or more different populations at the same time. Although a few studies suggest an association between the prevalence of atopy and asthma (Charpin et al,40Wieringa et al 45), most studies do not. For example, Peat et al 42 found little or no association between the prevalence of atopy and diagnosed asthma in different parts of Australia.

Table 5

Prevalence of skin prick test positivity and asthma in population based studies comparing different populations or the same population over time

Similarly, Leung and Ho41 reported that asthma prevalence was high in Hong Kong (7% for asthma ever), intermediate in Malaysia (3%), and low in San Bu, China (2%), but atopy prevalence was similar in the three centres (58%, 64% and 49%, respectively).

Yemaneberhan et al 44 reported major differences in the prevalence of asthma between rural (1%) and urban (4%) populations in southwest Ethiopia. However, skin prick test positivity to house dust mites was more common in rural (12%) than in the urban (4%) areas; there was little difference in the prevalence of skin prick positivity to other allergens such as mixed threshings orAspergillus (not shown in table).

There is evidence of an association of the prevalence of atopy with the prevalence of asthma in the studies showing higher levels of both in Western than in Eastern Europe.24 However, no increase in the prevalence of asthma was observed among East German children between 1991 and 1996, although the prevalence of atopy increased from 19% to 27%.48 On the other hand, Peatet al 46 ,47 found marked increases in diagnosed asthma in Busselton, Belmont, and Wagga Wagga, Australia (there were similar but less dramatic increases in the 12 month period prevalences of wheezing, not shown in table), but there was little change in the prevalence of atopy in these three centres (table 5).

The European Community Respiratory Health Survey49 has not yet fully published its results, but the prevalence of atopy (defined as raised specific serum IgE levels) appears to be associated with the prevalence of subjects reporting asthma attacks38 at the country level. However, this association is mainly driven by the English speaking countries with other European countries showing only a weak association, and no association has been observed between the prevalence of asthma and total serum IgE levels.49


The available epidemiological evidence suggests that the population based proportion of asthma cases that are attributable to atopy is usually less than one half. Higher estimates (up to two thirds) can be obtained by using very low cut off levels of total serum IgE, but these should be interpreted with caution since such a definition of atopy has limited practical use, and these associations may not always be causal. Moreover, standardised comparisons across populations or time periods show only a weak and inconsistent association between the prevalence of asthma and the prevalence of atopy. These findings indicate that the importance of atopy as a cause of asthma in individuals may have been overemphasised. The danger is that overemphasis on a particular theoretical paradigm for which the evidence is less substantial than is commonly assumed may have led to an under-recognition of, and insufficient research into, other possible aetiological mechanisms for the development of asthma.


The Wellington Asthma Research Group is supported by a Programme Grant from the Health Research Council of New Zealand and by a major grant from the Guardian Trust (Trustee of the David and Cassie Anderson Medical Charitable Trust). Juha Pekkanen was supported by the Academy of Finland.


View Abstract