Jump to: Page Content, Site Navigation, Site Search,
You are seeing this message because your web browser does not support basic web standards. Find out more about why this message is appearing and what you can do to make your experience on this site better.
a Section of Epidemiology, Department of Population Health Sciences, National Institute of Public Health, PO Box 4404, 0403 Oslo, Norway, b Environmental Epidemiology Unit, Department of Public Health, University of Helsinki, Helsinki, Finland
Correspondence to: Professor Magnus
| Abstract |
|---|
|
|
|---|
Objectives: To review repeated surveys of the rising
prevalence of obstructive lung disease among children and young adults and determine whether
systematic biases may explain the observed trends.
Design: Review of published reports of repeated
cross sectional surveys of asthma and wheezing among children and young adults. The repeated
surveys used the same sampling frame, the same definition of outcome variables, and equivalent
data collection methods.
Setting: Repeated surveys conducted anywhere in
the world.
Subjects: All repeated surveys whose last set of
results were published in 1983 or later.
Main outcome measures: Lifetime and current
prevalences of asthma and current prevalence of wheezing. The absolute increase (yearly
percentage) in the prevalences of asthma and wheezing was calculated and compared between
studies.
Results: 16 repeated surveys fulfilled the inclusion
criteria. 12 reported increases in the current prevalence of asthma (from 0.09% to
0.97% a year) and eight reported increases in the current prevalence of wheezing (from
0.14% to 1.24% a year). Changes in labelling are likely to have occurred for the
reporting of asthma, and information biases may have occurred for the reporting of wheezing.
Only one study reported an increase in an objective measurement.
Conclusions: The evidence for increased
prevalences of asthma and wheezing is weak because the measures used are susceptible to
systematic errors. Until repeated surveys incorporating more objective data are available no firm
conclusions about increases in obstructive lung disease among children and young adults can be
drawn.
|
Key messages
|
| Introduction |
|---|
|
|
|---|
There is widespread belief that the prevalence of asthma is increasing in industrialised societies, particularly in children. This belief stems from results of epidemiological studies. Changes in diagnostic labelling and the presence of selection or information bias can lead to false interpretation of changes in repeated prevalence studies. We reviewed published studies on recent secular trends in asthma in the light of these potential sources of error. We limited the review to repeated cross sectional studies of asthma in children and young adults in which the last set of results was published in 1983 or later.
| Materials and methods |
|---|
|
|
|---|
Inclusion of studies
We included only repeated cross sectional studies that used the same methodology
(definition of variables, wording of questions, and methods of data collection) on samples of
children or young adults in the same geographic area. Furthermore, both the sampling frame and
the sampling method had to be identical within studies. Surveys based on attendance at health
care centres (for example, hospital admissions or consultations in general practice) were excluded
because time trends for these data depend on the selection of subjects as well as on organisational
efficiency, diagnostic methods, and data registration practices over time. Because the mortality
from asthma is low and mortality depends on both the incidence of asthma and the case fatality
rate we did not include mortality studies.
We excluded some reports because of non-equivalent methods of data collection or content of information, non-equivalent sampling, or lack of specification of sample sizes. For instance, the NHANES studies changed the wording of the question on asthma from "Did a doctor ever tell you that you had asthma" to "Has [name of child] ever been treated for the following? Asthma (Yes, No)."1 Peat et al changed questionnaires between surveys.2 A study from Aberdeen reported changes between 1964 and 1989 but was excluded because the first data collection was by interview and the second by questionnaire, and the wording of questions was changed.3 In addition, studies of Israeli4 and Finnish conscripts5 were excluded because sample sizes were not given. We also excluded studies in which the sampling frame was not equivalent on the two occasions.6 7 8
To identify studies for inclusion, abstracts of all studies referenced in Medline from January 1983 to April 1996 were extracted if they included "asthma" or "wheezing" combined with words such as "prevalence," "occurrence," "incidence," "increase," or "trend." Sixteen repeated cross sectional studies fulfilled the criteria for inclusion.9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Six studies were from the United Kingdom, four from New Zealand or Australia, two from Israel, two from Scandinavia, one from Taiwan, and one from the United States (table 1). Twelve studies concerned children and four were based on young adults.
|
Variables
We extracted three measures of disease prevalence from the studies: lifetime occurrence
(cumulative incidence) of asthma, current asthma, and current wheezing. For current wheezing,
studies asked either for symptoms during a defined retrospective period (12 months or three
years) or for the presence of symptoms recently or occasionally. Data collection methods for
children were primarily based on questionnaires completed by parents whereas studies of young
adults also relied on physical examination. Operational definitions of asthma and wheezing were
usually not specified, and the implicit definitions (wording of questions) differed between
studies. Table 1) briefly describes the data collection
methods, outcome measures, and definitions of asthma and wheezing.
Statistical methods
For each study the absolute changes in prevalence (percentage change yearly) were
estimated from the published figures. The 95% confidence intervals for these estimates
were based on weighted linear regression (if more than two years were studied) or on the Poisson
distribution for calculating the standard error of the difference between two prevalences. The
relative increase over time was studied by the ratio of the prevalence rates at the first and last
surveys (with 95% confidence intervals calculated by Taylor series) for both asthma and
wheezing.
| Results |
|---|
|
|
|---|
Asthma
Table 2) shows the absolute prevalences for each
survey as well as the yearly increase in prevalence. The prevalence of lifetime occurrence of
asthma ranged from 5.5% to 31.8% whereas the prevalence of current asthma
showed less variability. For lifetime asthma the absolute increase in prevalence ranged from
0.35% to 2.08% yearly. For current asthma the increase ranged from 0.09%
to 0.97%. All studies showed increasing prevalences. The increases were in general lower
for young adults. Figure 1) shows the absolute yearly
increases in the prevalence of current asthma by country and time period. The studies from the
United Kingdom found very different degrees of increase. The two studies with the most recent
start found the largest yearly increases.
|
|
Wheezing
Only studies in the United Kingdom, Australia, and New Zealand reported current
wheeze. The prevalence of current wheezing varied from 4.6% to 25.4%. The
absolute yearly increase in the prevalence of current wheezing ranged from 0.14% to
1.24%. Studies started after 1980 showed the largest absolute yearly increases in
prevalence of wheezing (fig 2).
|
The five studies that included both current asthma and current wheezing are listed in table 3), which shows the relative increase in both measures. In the New Zealand study current asthma increased by 60% and current wheezing by 50%. In that study and the study from Australia asthma and wheezing increased proportionally. In contrast, all three studies from the United Kingdom showed a larger increase in the prevalence of current asthma than in the prevalence of current wheezing.
|
| Discussion |
|---|
|
|
|---|
Apparent increases in the prevalence of asthma and wheezing over time was found in all the reviewed studies. The degree of increase differed substantially between the studies, even those within the same country. In Australia and New Zealand the increases in asthma and wheezing were proportional whereas in the United Kingdom asthma increased comparatively more than wheezing. Before concluding that real increases in obstructive lung disease among children and young adults have been taking place in these populations we need to assess critically the informational content of the data.
Asthma
A main concern is whether the content of the asthma diagnosis is changing. In 1983 a
study from Newcastle showed a substantial underdiagnosis of asthma.26 The increase in parent reported asthma observed during the
1980s may be a consequence of changes among physicians in making use of the asthma
diagnosis and giving better information to parents.16
Better treatment options with a focus on early introduction of drugs and a change in the criteria
for applying the diagnosis to children27 may also have
increased the prevalence.28 Changing criteria for asthma
may also have relevance for the studies of conscripts in Israel,21 22 in which physicians
examined subjects with a wheezing history, and for the Taiwan study,13 in which paediatricians revised the questionnaires. Physical
examination for the determination of asthma adds a further layer of complexity, as the
examination is subjective and the diagnosis of asthma may depend on the period in which the
person is examined.
In addition to increased professional awareness, a public awareness bias may be present if allergies and asthma have been extensively discussed in public. After reviewing the many mechanisms that can influence an asthma diagnosis Anderson concluded that "the presence of a diagnosis of asthma is of little use epidemiologically."29 On the basis of these considerations we cannot conclude from repeated studies that sought only the presence of asthma that asthma has increased.
Wheezing
If we cannot trust the asthma diagnosis for time trend studies we are left with the increase
in wheezing (table 2). The eight studiesfrom
New Zealand, Australia, and the United Kingdomshowed variation in the yearly increase
in prevalence of current wheezing from 0.14% in London17 to 1.24% in two Australian towns.12 A remaining question is whether the everyday meaning of the
word wheezing has been constant over time. Do better educated parents more easily use this
word for symptoms in their children? Is the tolerance of mild respiratory symptoms lower than
it previously was? Have public health campaigns to increase the awareness of wheezing as a sign
of asthma or media reports of increasing rates of asthma led to increased parental awareness of
symptoms in their children? These questions cannot be answered from the studies reviewed. In
the London study less disability in connection with wheezing was reported on the second
occasion (1991).17 This may be explained by an increase
in the recognition and reporting of symptoms rather than by effects of treatment, which
presumably should reduce disability and symptoms to the same degree.
Lack of objective measures
Wheezing and other symptoms of obstruction are clinical expressions not easily captured
by objective measurements in cross sectional studies. However, both bronchial hyperreactivity
and positive skin prick test reactions are correlated with asthma,13 though none of these measures provides a standard. But we
should expect that the prevalence of these findings would increase in the general population of
children if the prevalence of asthma really had increased.
Only two of the 16 studies included objective measurements on both occasions. In a study from Wales15 peak expiratory flow rates were measured before (PEFR 1) and after (PEFR 2) six minutes of free running and the outcome analysed as 100x (PEFR 2/PEFR 1). In both surveys (conducted in 1973 and 1988) the mean value was 96% (calculated from midpoint values in table 3). The prevalence of airway hyperresponsiveness induced by exercise (more than 15% decrease) was 6.7% (SE 0.9%) in 1973 and 7.7% (0.9%) in 1988. In an Australian study comparing schoolchildren in 1982 and 1992 there was no change in skin prick test positivity to five allergens.12 The proportion of children who had a positive bronchial response to histamine challenge was twice as high in 1992 as in 1982. The type of spirometer used differed between the two occasions. An unknown point is whether the increased use of drugs might increase the response to histamine.
Selection bias
Within each country the samples consisted of schoolchildren, students, or conscripts
drawn from the general population. There is a lack of repeated cross sectional studies of children
below school age. Response rates are generally high. The important question is whether the
repeated samples within studies are sufficiently comparable. We were careful to exclude studies
that did not use the same geographic sampling frame (same schools or same regions).
Nevertheless, the effects of selective migration between surveys cannot be controlled in detail
and may influence some of the smaller studies based on samples of schoolchildren in cities. The
studies based on conscripts21 22 24 or on nationwide
sampling14 20 were
more reliable in this respect. The studies were conducted in several countries representing large
differences in geographic location and culturenamely, the United Kingdom, the United
States, Australia and New Zealand, Taiwan, Sweden, Finland, and Israel. To complete the picture
studies are needed from other countries, though it is unlikely that trends will be noticeably
different. In our opinion, selection bias is unlikely to explain the observed trends.
Conclusion
The prevalence of asthma is difficult to follow over time owing to changes in diagnostic
practice. The argument for an increase in obstructive lung disease among children rests on
parental or self reports of symptoms in three countries (United Kingdom, New Zealand, and
Australia) and finding an increased prevalence of bronchial hyperresponsiveness to a histamine
challenge in one study. Information bias may explain the trends. It is encouraging that new
epidemiological studies are under way with standardised questions on severity and objective
measurements.30 Until such studies have been performed
on more than one occasion in the same population we believe that the evidence for an increasing
trend in obstructive lung disease among children and young adults is weak.
| Acknowledgements |
|---|
Funding: None.
Conflict of interest: None.
| References |
|---|
|
|
|---|