i) The average incidence of adult onset-asthma in northern Italy is 15.2/10,000/year, which ranks in the lower part of the ranges reported in longitudinal studies in other countries. Allergic rhinitis and asthma-like symptoms are strong and independent predictors of adult-onset asthma, which occurs more frequently in women and in recent generations.
ii) The presence of asthma attacks and nocturnal asthma symptoms (tightness and dyspnoea) in young adults is associated with a two-fold increase in the risk of dying in the subsequent seven years, compared to asymptomatic subjects.
Incidence
In our analysis of the incidence of asthma, the population at risk and the new cases of the disease were defined according to the occurrence or absence of a relevant clinical event, such as an asthma attack, which is less likely to be influenced by recall bias and more reproducible than the reporting of asthma symptoms [
15,
16]. It has also been reported that, at least in Italy, the large majority (85%) of subjects who self-reported an attack of asthma had been diagnosed by a doctor [
1]. Consequently, it is likely that our definition slightly underestimates the incidence of the disease compared to symptom-based definition [
17] on one hand; but then on the other, it guarantees better specificity [
18] and stability of our estimates than other definitions.
Our estimate of the average annual incidence of asthma in young adults in Italy was 15.2/10,000/year, ranging from 10.1/10,000/year, in the older generations, to 22.8/10,000/year in the younger generations. Our longitudinal study confirms that adult asthma occurs more frequently in women and in younger generations. The gender difference in the occurrence of asthma has been pointed out in several previous studies [
17,
19,
20] from different countries and it has been attributed mainly to the role of female sex hormones [
21]. The reasons for the generalized increase in asthma incidence in the youngest generations are still unknown. Changes in susceptibility to environmental stimuli leading to asthma (hygiene hypothesis) as well as changes in exposure to environmental factors (increase in air pollution) could have contributed to this increase [
22,
23]. Our finding that incidence of asthma is higher in urban than suburban areas support both the previous hypotheses.
The strongest independent predictor of the incidence of asthma was the presence of asthma-like symptoms, such as diurnal wheezing and nocturnal dyspnoea and tightness, in agreement with a recent longitudinal Spanish study [
4]. This finding indicates that there is a group of subjects for which the clinical phase of the disease starts with mild symptoms which are probably not recognized as asthma, but which will be diagnosed in the following years when the disease gets worse and the exacerbations are labelled as asthma attacks in the end.
The presence of allergic rhinitis at baseline, regardless of other asthma symptoms, was another strong predictor of asthma. This result concords with other longitudinal studies showing that allergic rhinitis is an independent risk factor for the onset of asthma [
24,
2].
The association between allergic rhinitis and asthma has been traditionally interpreted as the progression of a common airway disease associated with atopy [
25,
26]. It has also been reported [
24,
27,
28] that rhinitis is a significant risk factor for adult asthma in both atopic and non atopic subjects, suggesting that other mechanisms, besides atopy, should be invoked to explain the observed association [
29,
30]. Our study, based on questionnaire data and not on a measurement of atopy, cannot contribute to highlight this hypotheses. However, the strong association between early allergic rhinitis and onset of asthma in adulthood suggests that earlier treatment of allergic rhinitis symptoms [
31,
32] could modify the natural evolution of asthma and prevent the development of a more severe disease. Indeed, interventions such immunotherapy [
33], pharmacologic therapy [
34] and allergens avoidance [
35] for allergic rhinitis seem to be effective in preventing complications and the development of asthma [
29‐
31].
In agreement with other longitudinal studies [
36,
37], we found that active smoking does not increase the risk of asthma in adults, probably because susceptible people simply do not start smoking or quit smoking very quickly. This finding is in contrast with another longitudinal study on adolescents [
38], where active smoking was a risk factor for the incidence of asthma. In any case, whether active smoking is a risk factor for the incidence of asthma or whether it is a selection factor (healthy smoker effect) continues to be an open problem.
Finally, in contrast to a recent paper [
39] that found an association between asthma prevalence and socioeconomic status, our results do not support that being in a low social class increases the risk of developing asthma. This suggest that previously reported associations, based on prevalence studies, reflect more the association of social class with the severity and/or the persistence of the disease rather than its association with the occurrence of asthma.
Mortality
To our knowledge, few studies have been published on the overall mortality risks of adults with asthma or asthma-like symptoms. This is one of the few, that prospectively investigated asthma related mortality risk in a large cohort of young adults. We ascertained life status for all subjects still resident (82%) in the same areas where they lived in 1991, while we assumed that people who were untraceable or who had moved before the second survey were censored alive halfway through the follow-up. As there are no reasons to expect that censoring could be related to the outcome or presence of asthma symptoms, our estimates of the mortality risk are not biased by potential selection in the initial cohort.
The presence of asthma attacks and/or nocturnal asthma symptoms was associated with an increased mortality risk from all causes (SMR = 2.05; 95% CI:1.06–3.58). Our estimate of the relative mortality risk for asthmatics agrees with those previously reported, which ranged from 1.5 [
9] to 2.4 [
5].
Although the excess death due to chronic obstructive diseases accounted for a substantial part of the mortality risk in other studies [
9,
5,
6], in our study it was mainly due to non respiratory causes of death, such as accidents, and to a lesser extent, to cardiovascular diseases and cancer, which concords with a French study [
8]. This discrepancy can be explained by the fact that our cohort included much younger individuals than previous studies, and it is well known that young people < 45 years are at a much lesser risk of dying from chronic diseases than older people.
Death from accidents is the leading cause of death in the general population aged 15–44 years, at least in Italy. Our results showed that asthmatics belonging to this age group had a 5- to 6-fold risk of dying from accidents than the general population; this suggests that a previous history of asthma attacks and nocturnal symptoms made them even more susceptible to this kind of hazard. This could be due to the fact that asthmatic symptoms, especially nocturnal ones, may interfere with diurnal attentiveness, which could cause fatal occupational accidents [
40] as well as car accidents [
41]. Furthermore, it has been reported that asthma patients have high rates of anxiety disorders and major depression [
42,
43], which can also make them more prone to accidents. Another aspect to take into consideration is the fact that many asthmatic subjects also take antihistamines for their symptoms [
45], which could have a sedative effect [
46] and alter their diurnal concentration. However, in our study, no association was found between the use of drugs reported in the first study and subsequent mortality.
The moderate increase in mortality from cardiovascular diseases and cancer has already been reported [
44,
9] and could reflect the greater susceptibility of asthmatics to traditional risk factors, which could be fully appreciated in cohorts older than ours. No association was found between wheezing at baseline and mortality likely because this symptom alone has either a low specificity for asthma [
18] or identifies a very mild form of the disease. Furthermore, nocturnal asthma-like symptoms or asthma attacks could be partially attributable to other diseases like psychiatric illnesses and heart diseases.
As the screening questionnaire used in the first ECRHS study did not include information on smoking habits, it was impossible to adjust for this factor in the mortality analysis. However, the sensitivity analysis carried out on a sizable sample showed that when smoking information was taken into account, the results were very similar, suggesting that the failure to adjust for this potential confounder could have biased our results only to a minor extent.
Nevertheless, some caveat in the interpretation of our mortality results should be taken into account due to the relatively short follow-up period (7 years) and the consequent low number of deaths. As a consequence, our analysis cannot establish a definite causal relationship between the presence of severe asthma symptoms and the subsequent overall mortality due to both the observational nature of our survey and the expected low number of deaths. The absence of a specific asthma related risk of death in our young cohort does not weaken our result: in fact, mortality from asthma is completely avoidable in this age group, and no deaths are expected to be found if health services are adequate [
47]. Consequently, the suggestion emerging from our study of an increased non specific mortality risk even in young subjects reporting severe asthma symptoms should be considered as a warning signal, indicating that asthma symptoms may involve a more general risk than normally expected, and should at least promote other studies dealing with the overall mortality pattern in asthmatics.