Few diseases have a relation to age which is as fascinating and complex as for asthma. It is a chronic but not necessarily lifelong condition. The incidence of asthma and wheezing illness peaks in very early childhood, but new incident cases occur throughout life.^1–4 In many affected subjects, particularly children, it disappears after some time.^2–4 However, in a substantial proportion of cases which have apparently lost the disease it will come back, often after many years.^3,4 Another intriguing feature is that the sex ratio changes with age. Most studies show that boys are affected more often by wheezing illnesses than girls, but this sex ratio usually reverses during or shortly after puberty, partly due to a higher incidence in females.^5–9

Most studies investigating the relationship between age and asthma have been performed in infants and children and, in fact, different age related phenotypes in childhood have been described which seem to have distinct causes and consequences.^2,10 Information on the relation to age in adults is scarce and has often been based on routine data or cross sectional studies.^1,6

In this issue of Thorax Chinn and colleagues report the findings of phase II of the European Community Respiratory Health Survey (ECRHS) which involved following more than 11 000 randomly selected adults (participants in ECRHS phase I stage 2 at which time they were aged 20–44 years) for a period of 5–11 years.¹¹ The investigators used the same standardised questionnaires at the start and end of the observation period, asking about the occurrence and severity of respiratory symptoms in the 12 months before the survey. The study is unique because of its international approach which included 29 study centres from 14 countries. Consistency of findings across countries argues strongly for the validity of the results.

Nevertheless, there are several methodological issues that should be considered before accepting the findings as valid. In particular, measuring asthma in populations is no easy task.¹² There is no single simple instrument by which cases can be identified. Instead, there is a whole battery of different measures, all of which have advantages and disadvantages. Chinn et al¹¹ report data collected by standardised symptom based questionnaires which are considered to be the standard method for measuring the prevalence of asthma in epidemiological studies.¹² However, since these do not measure the prevalence of asthma in individuals with complete accuracy, changes within individuals between repeated surveys may reflect measurement error rather than genuine changes in morbidity. Chinn et al therefore report the “net change” in symptom status for each centre rather than reporting separately the rate at which previously disease free subjects became symptomatic (incidence) and the rate at which previously diseased subjects became asymptomatic (remission). While this approach is perhaps regrettable from a clinical point of view, it is methodologically valid and provides findings that are of considerable interest.

The study showed no increase in the 12 month period prevalence of wheeze and more severe asthma symptoms during the follow up period which averaged about 8 years. However, there was a significant increase in the reported 12 month period prevalence of attacks, labelled as “asthma attacks”, and in the point prevalence of asthma medication use and nasal allergies.

How can these apparently contradictory patterns be reconciled? An increased prevalence of asthma attacks and medication use could be due to an increase in the prevalence of severe asthma, in recognition of symptoms by patients, in diagnostic labelling of wheezing illness by physicians, or in medical treatment of the condition. Each of these, in turn, could reflect temporal trends (changes over time) or age effects (changes with increasing age). However, Chinn et al also measured the prevalence of wheeze and severe asthma symptoms and these did not increase. It is therefore most likely that the observed increases reflect changes in diagnostic labelling and/or medical treatment for mild and/or moderate asthma.

Although the study by Chinn et al does not allow to disentangle potential age and period effects, it is interesting to review the current evidence for time trends in asthma prevalence. An increase in the prevalence of asthma and allergies in the late 20th century is generally accepted. While most studies reported increases in the prevalence of symptoms and diagnoses (and these are mostly in children), these observations are supported by studies which also included physiological markers.¹³ Recently, however, several studies have suggested that this increase may have come to an end—at least in some areas.

A study in adults over the period from 1972 to 1998 still observed increases in the prevalence of asthma symptoms and diagnoses which was more pronounced among those aged less than 40 years.¹⁴ Two other studies on adults examining time trends during the 1990s, however, found no increase in the prevalence of either asthma symptoms¹⁵ or bronchial hyperresponsiveness,¹⁶ but an increase in the prevalence of reported asthma diagnoses. An investigation of Swiss adolescents observed no further increase in the prevalence of asthma symptoms and allergic sensitisation in the late 1990s.¹⁷ Finally, Anderson et al,¹⁸ in a large survey in the UK, actually found a decrease in the prevalence of symptoms of asthma, hay fever, and eczema in 12–14 year olds between 1995 and 2002. During the same period there was an increase in the lifetime prevalence of diagnoses of the three disease entities under investigation. Robertson et al¹⁹ reported a similar reduction in the prevalence of asthma in children in Melbourne during 1993–2004. Thus, there is evidence from several countries that time trends in the burden from asthma may have levelled off—in some countries even reversed—while the rate of reported diagnoses continued to increase. It is important to note, however, that these reports come mostly from high income countries with prevalence rates ranking among the highest in the world.²⁰ The global burden of asthma, however, will be determined to a large extent also by what happens in low income countries. In this respect, the upcoming results of phase III of the International Study of Asthma and Allergies in Childhood (ISAAC), which has studied recent time trends in about 100 centres worldwide (including those of Anderson et al¹⁸ and Robertson et al¹⁹), will be of particular interest.²¹

If, in fact, time trends in asthma prevalence have levelled off, it is not clear which factors have determined this change. It is possible that the increase in asthma prevalence has reached a natural plateau in English speaking countries in which virtually all the “susceptibles” may have developed the condition. However, this speculation would not explain the apparent decrease in some countries. In this regard, it is likely that improved medical treatment, especially the use of inhaled steroids, has contributed.¹⁶ It has been argued that many patients do not benefit from new treatment because their disease is not diagnosed and/or treated adequately.^22,23 In this sense, the increased prevalence of diagnosed asthma reported by Chinn et al, in the absence of an increased symptom prevalence, could also be a reflection that medical care of asthma patients has changed for the better. Whatever the explanation, the findings of Chinn et al are in line with those of other recent studies and may, in fact, be good news.