Apparent but not real increase in asthma prevalence during the 1990s

Asthma prevalence appears to have increased dramatically in industrially developed countries over the last two decades as judged by hospital admission rates, diagnoses, use of medication, and certified deaths. An increase in disease prevalence has been supported by a number of studies in which the same population has been surveyed more than once at intervals of ≤25 yrs 1–4. Observed rates of change for those affected within UK populations have been >6% per annum for 8–13-yr-old children 1, 2, 7% per annum for university entrants 3, and 8% per annum for 12-yr-old children 4. These findings were based on questionnaire responses, and so may have been influenced by increased awareness of asthmatic symptoms and an increased likelihood of reporting them. There is a possible source of confusion in expressing the magnitude of change in terms of percentage; if prevalence itself is quantified as a percentage (as in the present study) and increases from, for example, 4.5–4.8, the net change of 0.3 represents a 7% increase in the number of affected subjects.

Few studies have included objective features of asthma severity or prevalence, such as airway responsiveness measurements, and no clear picture has emerged from those that have. Burr et al. 4 reported a greater than two-fold increase in “current asthma” amongst schoolchildren between 1973 and 1988, but there was only a 15% increase in the proportion who showed a bronchoconstrictor response to exercise (decrement in peak expiratory flow of >15%). By contrast, Peat et al. 5 showed that an 87% increase in “current asthma” amongst 8–10-yr-old children between 1982 and 1992 was paralleled by an 83% increase in the prevalence of airway hyperresponsiveness to histamine 5. However, the type of nebuliser used for the inhalation challenge tests was changed over the period of their study (with a probable change in aerosol output) and this might have confounded the result 6. The same group showed a relatively modest 21% increase in “current asthma” between 1981 and 1990 in 18–55 yr olds, and found that the prevalence of airway hyperresponsiveness fell over the period of the study 7. These discrepancies raise the possibility that recent increases in symptom reporting, physician diagnoses, and medication (which involve countries as widely separated as Australia, Finland, and the USA 5, 8, 9) do not reflect a true increase in the pathophysiological abnormalities of asthma. They led a recent reviewer to conclude that the evidence for an increase in asthma prevalence is weak and that further studies of defined populations using objective data are necessary 10.

The current authors have conducted surveys of the adult population of Newcastle-upon-Tyne in the north east of England in 1992–1993 and 1998–1999 using the same methodology and equipment. In these surveys, objective measurements of spirometry, airway responsiveness, and atopy were made in addition to questionnaire assessments of symptoms, physician diagnoses, and medication. The perception of bronchoconstriction reported by participants undergoing methacholine tests was also assessed.

Methods

The study followed closely, but not exactly, the design of the European Community Respiratory Health Survey (ECRHS) 11. An initial survey was carried out in 1992–1993 and a repeat survey in 1998–1999. On each occasion a postal questionnaire was sent to 6,000 adult residents of Newcastle-upon-Tyne aged 20–44 yrs, randomly selected from the Health Authority register and stratified equally by age and sex. Each subject received up to three mailings of a short questionnaire comprising nine items about asthmatic symptoms and treatment, and a further item related to hay fever (table 1⇓). All mailings were carried out between July and September. To assess response bias attributable to an incorrect address, a random sample of nonresponders was investigated by comparing details on the Health Authority register with those on the Electoral Roll.

A random sample of responders was then invited to attend the laboratory for an interviewer-administered questionnaire, measurements of height and weight, skin-prick tests with six common aeroallergens, spirometric tests, and a methacholine-challenge test. The aim was to recruit 600 subjects for these laboratory studies of each survey. The methods were the same in the two surveys and have been published in detail elsewhere 12, 13. All laboratory studies took place between October and April to avoid the pollen season.

Methacholine was administered by nebuliser in doubling cumulative doses over the range 3.125–6,400 µg. Airway responsiveness was expressed both as the dose/response slope (DRS) and as the dose of methacholine provoking a 20% fall in forced expiratory volume in one second (FEV₁, PD₂₀) 14. Using this methodology, PD₂₀ values <200 µg are usually associated with active asthma, values 200–1,000 µg are sometimes associated with active asthma, and values >1,000 µg are seldom associated with active asthma 12, 15. The precision of the method is described fully elsewhere 13.

To assess perception of bronchoconstriction, all subjects were asked: whether they were aware of any abnormal respiratory sensation at the point of maximal bronchoconstriction at the end of the methacholine test, even if the decrement in FEV₁ was of a negligible degree; whether the sensation could be described as wheeze, chest tightness, or breathlessness; and whether it had ever been experienced before 16. Local ethics approval was obtained and all subjects gave written informed consent.

The data from the two surveys were pooled and Chi-squared tests, unpaired t-tests, and stepwise multiple linear and logistic regression techniques were used to identify predictors of participation in the study, symptoms, and airway responsiveness.

Results

Questionnaires

The postal and laboratory questionnaires identified substantial increases from 1992–1993 to 1998–1999 in the prevalences of asthmatic symptoms, physician diagnosis, and use of medication (table 1⇑). The prevalences of symptoms increased by a mean of 4.3% on the postal questionnaire and 4.8% on the laboratory questionnaire. There were very similar prevalence increases in physician-diagnosed asthma (4.2%) and the use of medication (4.8%, weighted mean from both questionnaires), giving an average overall change of 4.4% over the 6 yrs. The use of inhaled corticosteroids rose from 5.0% to 9.3% (p=0.003), and inhaled β-agonists from 10.5% to 14.1% (p=0.054). No dosage information was obtained. In subjects with a physician-diagnosed asthma, inhaled corticosteroid use increased from 37.2% to 52.9% (p=0.035), but there was no change in β-agonist use (68% versus 64%, p=0.6).

The increases were more marked in participants aged <30 yrs (overall 8.7%) compared with those who were older (2.7%). For the laboratory study, in which corresponding measurements of airway responsiveness were obtained, the increases in positive questionnaire responses from these younger participants were especially striking for “asthma attacks in last year” (7.2%, p=0.025), “physician-diagnosed asthma” (13.7%, p<0.0001), and “current medication” (fig. 1⇓).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Percentages of laboratory populations reporting current asthma treatment in 1992–1993 (□) and 1998–1999 (└): effect of age.

Airway responsiveness

Methacholine tests were carried out in 591 subjects (302 males and 289 females) in 1992–1993, and 580 (303 males and 277 females) in 1998–1999. The remaining 68 participants in the laboratory studies (from both surveys) either declined or were considered unsuitable because of pregnancy, cardiac disease, epilepsy or baseline FEV₁ <60% of the predicted value.

The means for the baseline measurements of FEV₁ exceeded 100% pred and did not differ significantly between the two surveys (table 2⇓). The baseline FEV₁ is not, therefore, likely to have differently influenced PD₂₀ measurements between the two surveys. A positive methacholine test (i.e. PD₂₀ <6,400 µg) was obtained from 37.2% of subjects in 1992–1993 compared with 31.6% in 1998–1999, indicating that airway hyperresponsiveness was less prevalent in the later survey (table 2⇓ and fig. 2⇓). The geometric mean DRS lessened in parallel from 0.00527 to 0.00379 (p=0.004). Multiple regression analyses showed that this association with survey year was of borderline statistical significance after adjusting for the effects of sex, smoking, and baseline FEV₁ expressed as a percentage of the predicted value (table 3⇓). The OR for a positive methacholine test in 1998–1999 compared with 1992–1993 was 0.77 (95% CI 0.58–1.01). Similar results were obtained for the analysis of PD₂₀<1,000 µg (OR 0.64, 95% CI 0.49–0.90), PD₂₀<200 µg (OR 0.79, 95% CI 0.49–1.26), and log DRS (t=−3.17, p=0.002).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.—

Distribution of airway responsiveness (dose/response slope (DRS)) in the population samples of 1992–1993 (—) and 1998–1999 (---). Geometric mean DRS 1992–1993=0.00525, 1998–1999=0.00377, p=0.002.

View this table:

• View inline
• View popup

Table 2—

Lung function, airway responsiveness levels, and atopy

View this table:

• View inline
• View popup

Table 3—

Logistic regression analysis of airway responsiveness: provocative concentration causing a 20% fall in forced expiratory volume in one second (FEV₁, PD₂₀) <6400 µg methacholine

For the younger participants, aged <30 yrs, the prevalence of airway responsiveness at a level consistent with active asthma (PD₂₀<1,000 µg) fell from 26.9% to 22.7%, despite the increases in positive responses to the questionnaires.

Perception of bronchoconstriction

Younger age, female sex, the presence of symptoms on the laboratory questionnaire, the degree of bronchoconstriction during the methacholine test, a lower band of PD₂₀ measurement (<200 µg, 200–1,000 µg, 1,001–6,400 µg, >6,400 µg methacholine), atopy, and involvement in the 1998–1999 survey were all predictors of the ability to perceive (and report) methacholine-induced bronchoconstriction when the results from both surveys were combined (table 4⇓). In 1992–1993, 62.5% of subjects reported the development of an abnormal respiratory sensation at the end of the methacholine test compared with 77.2% in 1998–1999 (p<0.001, fig. 3⇓). The associations with younger age, symptoms, and the degree of methacholine-induced bronchoconstriction were apparent within both the 1992–1993 and the 1998–1999 populations separately, but the association with sex was only significant in the 1992–1993 population. The association with atopy was of borderline statistical significance in 1992–1993 (OR 1.75, 95% CI 0.99–3.08) but was clearly significant in 1998–1999 (OR 1.99, 95% CI 1.15–3.43). There were no differences between the two surveys in the terms used to describe the respiratory sensation induced by bronchoconstriction, or in the proportion of subjects who reported having experienced this previously.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.—

Perception of bronchoconstriction during methacholine test in 1992–1993 (□) and 1998–1999 (└). FEV₁: forced expiratory volume in one second.

View this table:

• View inline
• View popup

Table 4—

Predictors of perception of bronchoconstriction during methacholine tests

Discussion

The current authors have identified significant increases relevant to asthma in the prevalences of symptoms, physician diagnosis, and use of medication in the young adult population of Newcastle-upon-Tyne between 1992–1993 and 1998–1999. The rates of increase are substantial and equivalent to a ∼5% increase per annum in the number of subjects affected, but they are similar to those reported in other studies in which the same population has undergone repeated investigation 1–5. The rates of increase are more striking in subjects aged <30 yrs. A recent study of general practice consultation rates suggested that the UK asthma “epidemic” might have peaked around 1993 17, but the findings of that study could be explained by organisational changes or by changes in treatment. Other data from UK general practice suggest an increase in asthma prevalence between 1994 and 1996, similar to that found in the current study 18.

The increases in asthmatic symptoms in the current study were paralleled by an increase in the prevalence of atopy, as judged by skin-prick test results and questionnaire reporting of hay fever (table 1⇑). The greatest effect was on the skin test response to Timothy grass pollen, which might be the most labile from year to year due to variations in the extent of pollen exposure. However, similar though smaller changes were seen with D. pteronyssinus and cat, which suggests a real increase in atopy similar to that found in other studies 1, 2, 19. As with the asthmatic symptoms, there was a tendency for the increase in atopy to be more marked in the younger subjects, after adjusting for sex and smoking.

Although asthma and atopy were associated in both 1992–1993 and 1998–1999 surveys, the increases in asthmatic symptoms were not much greater in the atopic than the nonatopic groups, and the proportion of physician-diagnosed asthma that could be attributed to atopy fell slightly from 42% to 38% over the period of the study. Other investigations have identified a similar lack of association between changes in asthma and changes in atopy. For example, the prevalence of atopy among Leipzig schoolchildren increased from 19 to 27% between 1991 and 1996 but there was no change in the prevalence of apparent asthma or airway responsiveness over the same period 19. Conversely, Australian schoolchildren showed a more than two-fold increase in asthma prevalence over the 1980s with no significant changes in atopy prevalence 5, 7. Asthma and atopy are not as closely associated as is commonly supposed, even amongst children 20, and these epidemiological studies suggest that factors other than increased atopy or increased exposure to common environmental aeroallergens have been making important contributions to the changing epidemiology of asthma.

Upton et al. 21 recently reported that increased diagnostic awareness could have been entirely responsible for a 2.5-fold increased asthma prevalence in nonatopic subjects between 1972 and 1996. They found no evidence that a similar bias influenced the diagnosis of atopic asthma, but the extent to which the increases in estimates of asthma prevalence generally are related to increased awareness and increased willingness to make the diagnosis is uncertain. The present authors' failure to identify increases in airway responsiveness in parallel with the increased prevalence of symptoms supports the possibility that these factors have exerted an important influence.

The current findings are not likely to have been due to a measurement artefact. The dosimeter and nebulisers were identical in the two surveys, the dosimeter was calibrated for aerosol output before each survey using an ion tracer method, and there was considerable overlap in the research personnel 22. Furthermore, the increase in symptom prevalence noted from the questionnaires in 1998–1999 was accompanied by an increased, not decreased, reported awareness of bronchoconstriction in association with the methacholine tests. The increase was seen only with decrements of FEV₁ <20% and was most marked at the lowest level of bronchoconstriction. Thus, reported perception of FEV₁ decrements <5% increased in prevalence from 22.9 to 55.1%, while for decrements of 16–20% the increase was from 74.7 to 86.5% (fig. 3⇑). Decrements in FEV₁ of <5% indicate levels of airway responsiveness well below those conventionally associated with asthma, and they suggest that symptoms are now being reported after relatively trivial changes in airway calibre.

The prevalence of inhaled corticosteroid use increased from 5.0 to 9.3% over the period of the current study. Such medication can be expected to increase mean PD₂₀ values 3–4 fold 23, 24, and is therefore likely to be the chief explanation for the observed decrease in the measured mean level of airway responsiveness in 1998–1999. In an attempt to adjust for these differences in corticosteroid use, the data was re-analysed after lowering the PD₂₀s of those using inhaled corticosteroids by a factor of four. Thus, the levels of airway responsiveness that would have been present at the times of both surveys had it not been for corticosteroid use were estimated. There were then similar proportions of the population in the PD₂₀<200 µg category in 1992–1993 (9.6%) and 1998–1999 (9.1%). The corresponding figures for the PD₂₀ category <1,000 µg were 22.8% in 1992–1993 and 17.4% in 1998–1999. Similar outcomes were observed when the atopics and nonatopics were analysed separately. These analyses strengthen the current authors' beliefs that the diminished level of airway responsiveness in 1998–1999 can be largely explained by the increased use of inhaled corticosteroids, and that this increase in corticosteroid use did not mask an actual increase in airway responsiveness. The findings are consistent with other recent studies. Peat and colleagues 5, 7 noted an increase in asthma symptoms but a nonsignificant decrease in the distribution of airway responsiveness in young adults in Australia, and, in a case-controlled study, Richter et al. 25 showed almost a doubling of asthma diagnoses in Hamburg from 1990–1991 to 1995–1996 but a slight decrease in airway responsiveness 25.

The overall importance of selection biases to the present findings is difficult to estimate. They are potentially important, as the presence of respiratory symptoms is a recognised determinant of participation in such studies 26. To the present authors' knowledge, the current study is the first repeat survey of a population following the ECRHS template and so the repeatability of the methodology is unknown. The response rates in these surveys were within the range reported by centres participating in the ECRHS, but they were low and the response rate in the second survey was lower than the first. The method of Drane 27 was used to estimate overall symptom prevalences in the population. This assumes a linear relationship between symptom prevalence and the probability of responding to a postal questionnaire, and it extrapolates results from a series of mailings to estimate the prevalence in nonresponders. It did not lead to any reduction in the estimates of the prevalence increases in asthma symptoms, physician diagnosis, or treatment. There were also no systematic differences in response to the postal questionnaire between those who did and did not take part in the subsequent laboratory studies. The critical point is that recruitment bias is not likely to explain the discrepant findings of increased symptoms yet decreased airway responsiveness.

Van Schyack et al. 28 demonstrated that individuals with poor perception of bronchoconstriction are less likely than others to seek medical attention. The converse is also likely to be true, and the increased perception of bronchoconstriction reported in the current population over 6 yrs in association with the methacholine tests is likely to have contributed to the increased prevalences of asthmatic symptoms and physician-diagnosed asthma. Little is known about the mechanisms that underlie the perception of bronchoconstriction and the explanation for the change in this population is unclear. Bronchoconstriction may not have been the true cause in all the subjects responding positively, since measured decrements in FEV₁ were of trivial degree in some. It is likely, however, that there were more subtle physiological changes, perhaps at the small airway level. The authors have consequently used the term “bronchoconstriction” somewhat loosely to indicate the maximum effect generated by the administration of methacholine.

Females, younger subjects, and atopics are recognised to have an increased ability to perceive bronchoconstriction 29 and these relationships were apparent in the present study. The increase in the prevalence of atopy was not, however, a major factor because this increased ability to perceive methacholine-induced bronchoconstriction was seen almost equally amongst both atopics and nonatopics, and there was no interaction between atopic status and the influence of the second survey of 1998–1999. Treatment with inhaled corticosteroids has been reported to increase the perception of bronchoconstriction 30 but this is unlikely to have had an effect as substantial as was seen in the present study population. Psychological factors can influence the perception of most sensory stimuli and it is likely that increased education and awareness of asthma among the current population was responsible. About 15% of the young adult population have intermediate levels of airway responsiveness (PD₂₀ between 200 and 1,000 µg using the current authors' techniques) but in only ∼20% of these have physician-diagnosed asthma 16. These 20% show few, if any, differences from the 80% majority in terms of airway physiology, so changes in the ability to perceive bronchoconstriction (and report it) have the potential to make a substantial contribution to estimates of asthma prevalence.

The current authors believe that the results presented here are consistent with the hypothesis that the current rise in asthma symptom recognition, diagnosis, and treatment in adults is largely a consequence of better education, increased awareness of symptoms, and/or increased readiness to report them, and an increased willingness of physicians to make the diagnosis and initiate treatment. No objective evidence of any increase in the prevalence or severity of the pathophysiological features of asthma was found.