INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: asthma; incidence; risk factors

The incidence of asthma has been measured in only a few geographical areas, using a variety of definitions (1). In a retrospective, cross-sectional multicenter study of adult asthma, the European Community Respiratory Health Survey (ECRHS), a similar generational increase in asthma incidence was observed in most countries irrespective of the prevalence of asthma (14). This study could not differentiate whether the increase was due to a change in diagnostic practices or to an actual increase in incidence. It is unknown whether incidence rates are proportional to prevalence. Prevalence depends on both incidence rates and duration. Risk factors and preventive strategies for incidence may differ from those for duration. This is the first study providing incidence data and its determinants for Spain, an area with low to medium prevalence of asthma (from 2.1 to 6.3%) (15).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A repeated survey of the participants in the second phase of the ECRHS study (15, 16) was conducted in 1998-1999, using identical questions about asthma (Figure 1). The average time between both surveys was 6.75 yr (range, 5.3 to 7.9 yr). The ethics committee of the institution approved both surveys. All participants gave informed written consent.

View larger version (22K): [in this window] [in a new window]   Figure 1.   Flow chart showing the number of subjects and the participation rates in the different phases of the study.

We estimated cumulative incidence of asthma between 1991-1993 and 1998-1999. Incident cases were defined as those at risk (free of the disease) at baseline, who gave a positive answer to the question "have you ever had asthma" in 1998-1999. We defined the population at risk in 1991-1993 in two ways. In the first definition the population at risk (n = 1581) comprised individuals who gave a negative answer to the question "have you ever had asthma" in 1991-1993. In the second, the population at risk (n = 1557) was defined as described above, but was further restricted by using information provided in 1998- 1999. Individuals reporting in 1998-1999 a year of onset 5 yr before 1991-1993 were excluded. We set the limit of 5 yr considering that a discrepancy of less than 5 yr could be due to recall bias when reporting age of onset.

Both these definitions were further restricted by taking into account the prevalence of bronchial hyperresponsiveness in 1991-1993 or, alternatively, the prevalence of asthma-type symptoms in 1991- 1993 (i.e., attack of asthma in the last 12 mo, being woken by an attack of shortness of breath in the last 12 mo, or taking asthma medication), or the prevalence of other respiratory symptoms (i.e., attacks of wheezing or whistling when not having a cold, having woken up with a feeling of tightness in the chest at any time in the last 12 mo, or having been woken by an attack of coughing in the last 12 mo).

Yearly incidence was estimated by the number of first occurrences of the disease over total person-time accumulated during the follow-up. The time contributed by each individual that reported asthma ever in 1998-1999 was fixed at the middle of the follow-up period. When information about age of asthma onset was used, the time contributed by each individual was computed until the day that asthma started. For 10 subjects (29% of all incident cases) whose asthma started less than 5 yr before 1991-1993, but who reported never having had asthma in the 1991-1993 survey, time of follow-up was set at 1 d.

Total IgE and specific IgE were measured by the CAP method (Pharmacia, Uppsala, Sweden) in 1991-1993 and described elsewhere (14). Atopy was defined as any specific IgE level greater than 0.35 kU/L to cat, Cladosporium, Dermatophagoides pteronyssinus, Parietaria, or timothy grass. Bronchial hyperresponsiveness was defined as a 20% or more fall in FEV₁ with respect to the highest postdiluent FEV₁ during the methacholine challenge in 1991-1993 with an accumulated dose of 5.117 µmol or extrapolated to 8 µmol of methacholine. A job exposure matrix (JEM) described elsewhere (17) was used to assess occupational exposures in the 1991-1993 job.

Determinants of the incidence were calculated by Poisson regression, with Stata statistical software, release 6.0 (StataCorp, College Station, TX).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 1,640 individuals of the 1,931 who participated in 1991-1993 (85%) had complete data in both interviews. Nonresponders differed with regard to sex (more males did not respond in 1998-1999) and area in comparison with respondents, but not regarding age, smoking, or any of the respiratory symptoms (Table 1).

The incidence of asthma was 5.53 per 1,000 person-years (Table 2), ranging from 3.50 in Galdakao to 7.87 in Huelva (p = 0.40). The incidence rate decreased to 3.28 when we excluded from the population at risk those individuals reporting in 1998- 1999 a year of asthma onset five or more years before the 1991- 1993 survey, despite not having reported asthma in the 1991- 1993 survey. After excluding subjects who reported respiratory symptoms in 1991-1993, the incidence rate dropped to 1.97 and, excluding those with bronchial hyperresponsiveness, the rate decreased to 1.50 per 1,000 person-years (Table 2).

Incidence was higher in those who had reported symptoms of asthma or other respiratory symptoms in a previous questionnaire and in women, those with bronchial hyperresponsiveness in 1991-1993, and those with high titers of total IgE and of specific IgE to timothy grass and Parietaria (Table 3). There were no statistically significant differences in incidence rates by age; area; atopic status to house dust mite, cat, or any allergen; smoking; occupational exposures to medium-high levels of biological or mineral dust or gases/fumes; or maternal asthma. Incidence did not differ significantly between areas.

In a multivariate model, only bronchial hyperresponsiveness and specific IgE against timothy grass reached statistical significance, after adjusting for female sex and symptoms of asthma (Table 4). The inclusion in the model of having had respiratory symptoms in 1991-1993 did not modify the associations observed for bronchial hyperresponsiveness or timothy grass. The effect of total IgE on incidence disappeared after adjusting for bronchial hyperresponsiveness.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of the present study provide information about the incidence of adult asthma in a population aged 26 to 50 yr in Spain during the 1990s. Incidence was related to having had bronchial hyperresponsiveness and having IgE titers against timothy grass but not atopy defined by house dust mite, cat, or any allergen. Incidence rates varied from 5.53 to 1.50, depending on the definition of who was free of asthma at baseline. These rates are similar to those reported in other areas (1), although comparability is limited because of different criteria used to define who is free of the disease at baseline and because of differences in the definition of the disease, as such.

Measuring incidence for a chronic intermittent disease such as asthma is difficult. Conflicting data between the two questionnaires (i.e., reporting an age of onset of asthma before 1991-1993 when contacted in 1998-1999, but reporting never had asthma when contacted in 1991-1993) may occur as a result of difficulties in recall of asthma, in part due to its complex intermittent natural history. To overcome the problem of defining asthma by using questionnaire data and defining the population at risk for incident asthma, we performed a sensitivity analysis. When age of onset was not taken into account, 58 new cases of asthma were found in about 7 yr (rate = 5.53 per 1,000 per year). This rate was much higher than the incidence of asthma for the whole period between birth and 1991- 1993 (n = 59, rate = 1.13), which might be explained by recall bias related to asthma symptoms and age of onset in the population at risk (18). Strachan and co-workers (18) showed an underestimation of one-third of recall wheezing in surveys. We also computed incidence rates excluding individuals who had reported asthma-type symptoms or other respiratory symptoms in one of the previous questionnaires. This conservative approach could correct the effect of changes in diagnostic practice over time, resulting in an underestimation of the incidence rates.

A minor problem that did not affect the estimation of the incidence density was due to the potential remission of 13 subjects who reported ever having had asthma in 1991 but not in 1998. These subjects had probably had asthma some years ago but have not had asthma attacks in more recent years.

Incidence was significantly higher in women, which agrees with studies looking at prevalence (19), and with a retrospective study of incidence using cross-sectional multicenter data of the ECRHS (22). This suggests that incidence is higher not only just after adolescence in females, but also later on (23). Differences in incidence of asthma between sexes were not confounded for bronchial hyperresponsiveness.

We found that bronchial hyperresponsiveness (BHR) was the strongest predictor of asthma incidence, being independent of the presence of asthma symptoms. BHR is a physiological characteristic associated with asthma and may be involved in many of the pathways through which asthma occurs, although BHR is not specific for asthma (24). Similarly, the presence of previous symptoms of asthma was associated with a higher incidence, although part of this association was explained by the presence of bronchial hyperresponsiveness. These findings probably indicate that there exists a pool of subjects with symptoms of asthma or bronchial hyperresponsiveness, who are probably undiagnosed and therefore without knowledge of having asthma, and are more likely to be diagnosed in subsequent years. When we excluded from the population at risk those with bronchial hyperresponsiveness or asthma-type symptoms in 1991-1993, the incidence rate was reduced by 50%.

Incidence of asthma was similar in nonatopic and atopic people, defining atopy as reactivity to any allergen. This contrasts with studies carried out in children (25), and also with the association between cat allergen and new-onset airway hyperresponsiveness in older men (29), although among healthy men from the Normative Ageing Study of Boston (30) a positive reaction to any skin test was not associated with incident cases of wheezing. We found that having IgE against timothy grass predicts incident asthma, which agrees with a previous study that related hay fever and incidence of adult asthma in Finland (21).

Incident rates were within the range of those observed in other studies in areas with higher prevalence (1, 2, 5, 9, 13, 14, 19, 20). The similar incidence in adults of countries with different prevalence could be due to the fact that factors related to adult onset of asthma differ from factors related to duration of the disease. These findings indicate that factors related to asthma onset in adulthood are probably common in all areas, whereas factors related to duration or childhood onset may differ. Factors potentially related to asthma duration are probably related to immunological development, and the interaction of the immunological system with airway epithelial damage due to viral infections (31, 32). Factors connected with asthma onset in adulthood are less known and only occupation (33) or sex hormones in women (34) appeared as distinct etiologies compared with childhood.

The small number of incident cases could explain some of the nonstatistically significant associations. Misclassification of asthma at follow-up, due to the lack of reliability of asthma definition by questionnaire, is a common drawback in epidemiological studies of general populations. Nonresponse, although small (16%), may have biased the estimated incidence rates, but it is unlikely that rate ratios were biased given that bronchial hyperresponsiveness and atopy were similar in participants and nonparticipants in the follow-up (data not shown). Finally, we have not been able to measure changes in bronchial hyperresponsiveness and atopy during the follow-up, and the analysis relies on a single baseline measurement.

In summary, we used for the calculation of incidence different definitions of the population at risk, which allows a comparison with other studies. These definitions resulted in a change of incidence rates from 5.53 to 1.50. These rates are within the range of those observed in areas with higher prevalence, suggesting that factors related to incidence of asthma in adulthood could not explain geographical variations in prevalence. In our study bronchial responsiveness and atopy to grass, independent of respiratory symptoms, are the main determinants for new-onset asthma in adulthood.