|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
We used 3-wk peak expiratory flow (PEF) measurements (twice daily) made in the diary study of the
population-based Swiss Study on Air Pollution and Lung Disease in Adults to describe PEF-variability (PEFvar) (amplitude as a percent of the mean, PEF [i.e., difference between morning and evening values divided by the mean]) in the study population and in five subgroups (physician-diagnosed
asthma; current asthma, or physician-diagnosed asthma plus asthma attacks and/or medication; history of wheezing without a cold; hyperreactive; and nonsymptomatic). We assessed the performance
of PEFvar as a potential tool with which to screen for asthma. Alternatively, subjects with a PEFvar of
20%, 30%, and 50% on at least 2 d were considered to have high variability. The analyses
were conducted for subgroups with different pretest probabilities for asthma-related conditions. The
median PEFvar was 4.5%. Among asthmatic subjects, women had nonsignificantly higher PEFvar values
than did men. In all other groups, women had significantly lower PEFvar. Both in the entire population and in subgroups with a higher pretest probability for asthma-related conditions, screening performance of PEF was limited. A PEFvar of 20% on at least 2 d detected current asthma with a sensitivity of 36% (specificity = 90%; positive predictive value = 16.4%). Results were better among
subjects with a history of wheezing without colds (sensitivity = 40.4%; specificity = 83.6%; positive
predictive value = 45.2%). PEFvar, a useful measure both clinically and in epidemiology, is of limited value when unselected populations are screened for asthma-related conditions, since the overlap of
PEFvar distributions across subgroups is large.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
A major feature of asthma, considered to be an underdiagnosed inflammatory disease (1, 2), is increased bronchial reactivity (BR) with intermittent airway obstruction. Peak expiratory flow (PEF) is a useful indicator of variability in airway caliber, which correlates with the changes in FEV1 (3) and with airway hyperresponsiveness (AHR) (4). The use of diurnal variation of PEF to assess intermittent airway obstruction is popular in the monitoring of asthma patients (5). Differences between normal and asthmatic subjects in diurnal variability in PEF (PEFvar) relate to susceptibility to increased bronchial responsiveness among the latter subjects (6). Furthermore, PEFvar may be influenced by external factors. These features render PEFvar a useful measure in occupational medicine and in epidemiologic studies (5, 7).
PEFvar has been mainly described in clinical groups. Few studies report population data for this measure, and do so among rather small samples of adults (8), resutling in inherent limitations in assessing distributions of PEFvar among population subgroups. As a consequence, these studies make inconsistent conclusions about the impact of sex, age, or disease status on PEFvar.
During a diary study of the large population-based Swiss Study on Air Pollution and Lung Disease in Adults (SAPALDIA) (11), extensive assessment of PEF was conducted. A total of 3,279 subjects had both morning and evening PEF measurements. The purpose of the present report is twofold. First, given the large size of the population in which PEFvar was measured, we present population distribution data for morning-evening PEFvar, both for nonsymptomatic "normal" subjects and for groups with defined respiratory conditions. The relationship with sex and age will be further elucidated. Second, we determine the validity of PEFvar as a population screening tool for detecting asthma-related conditions.
PEFvar can be easily measured, and the portable meters used for this are reliable, inexpensive and simple to use. Thus, diurnal variability of PEF may be considered a useful tool with which to screen adults for undetected asthma. So far, the validity of PEFvar has been controversially interpreted. Whereas Higgins and colleagues consider PEFvar unjustified for distinguishing asthmatic from nonasthmatic individuals (8), Quackenboss and coworkers (9) refer to the high sensitivity of their test definition. Den Otter and associates compared the screening performance of PEF and histamine challenge among a selected group of symptomatic subjects (12). To our knowledge, no formal assessment has been conducted of the potential use of PEFvar as a screening tool for asthma in adult population subgroups with different pretest probabilities for asthma.
| |
METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
We used the morning and evening PEF measures collected in the SAPALDIA Diary Study during 1992 and 1993. The SAPALDIA study started in 1991 with a cross-sectional survey (11) addressing long-term effects of air pollution on respiratory health among adults (18 to 60 yr of age) across eight Swiss areas (13). A random population sample, drawn from the local study registries, was invited to the cross-sectional survey. Fifty-nine percent (n = 9,651) of eligible subjects participated in the standardized interview, spirometry, bronchial challenge with methacholine, and skin and blood tests used in the survey to assess atopy status. SAPALDIA applied the same methods as the European Community Respiratory Health Survey (ECRHS) (14). Subsequent to the cross-sectional study, a subsample of 3,279 subjects was selected for the second part of SAPALDIA, a 2-yr diary study. All subjects reporting a history of respiratory symptoms or with increased bronchial responsiveness in 1991 were selected for follow-up. In addition, in each of the eight survey regions, a random sample of 150 healthy, nonsmoking subjects were included in the diary study sample. Participants were asked to complete a symptom and medication diary and to register morning and evening peak flow. Across a period of 2 yr, subjects were intended to participate for six diary periods of 4-wk duration each. Diary schedules were assigned so as to guarantee a balanced number of participants and distributions of symptomatic and nonsymptomatic subjects on any given day. The present analysis of PEFvar was restricted to those subjects with at least three consecutive weeks of diary study participation. Data from the first week were disregarded in order to exclude potential training effects (9). We included only data from Weeks 2 to 4 of the first diary period, requiring a minimum of 14 d of twice-daily peak flow registration (n = 3,074).
Peak Flow Measurement
Each subject received a mini-Wright peak flow meter (Clement Clarke International Ltd., Harlow, UK) at the local research centers. Participants were instructed and trained to use the meters according to guidelines recently summarized by Britton (15). Correct use of the meter and the registration of results in the diary was taught by a fieldworker. Both in the morning and evening, peak flow had to be measured with the subject in the standing position, and registered three consecutive times. Subjects using bronchodilator therapy were instructed to measure PEF prior to medication. During the first week of the survey, the local fieldworker called the subjects to remind them to use the PEF meter every morning and evening. Also emphasized was not to fabricate data, but rather to leave an open space if PEF was not measured. Furthermore, the 1-wk diaries were regularly controlled, and participants were contacted when needed. We transformed the reported PEF values with a quadratic function to adjust for the nonlinearity of the peak flow meter. The equation used for this was derived from a calibration study conducted with a subsample of our peak flow meters. The formula was: PEFrecalibrated = ([PEFread + 230.37]/ 33.48)2.
PEFvar
Several definitions of diurnal PEFvar have been used. Because of its statistical performance and the simplicity of its clinical use, the amplitude as a percent of the mean PEF is the predominant measure that we used (5, 16). PEFvar was calculated for each day and each subject as:
|
where PEFm was the best of three consecutive morning readings, and PEFe was the best of three consecutive evening readings. For each subject, the mean and SD of the daily PEFvar across the 14- to 21-d study period was calculated to derive PEFvar distributions in different subgroups.
To address the screening performance of PEFvar, we divided the population into groups of "high" and "low" PEFvar. The definition in each case was based on the two highest amplitude as a percent of the mean values for each subject. We present the screening performance for three cutoffs to define a "high amplitude as a percent of the mean" PEF: namely > 20%, > 30%, and > 50% amplitude as a percent of the mean on at least 2 d. These cutoffs correspond to "moderate," "severe," and "very severe" asthma (7).
Screening Concept
The theoretical screening framework validated in this analysis assumes that peak flow meters may be given to a population with the request for at least three consecutive weeks of PEF monitoring in both the morning and evening. PEFvar would be taken as the measure used to screen for asthma-related conditions. Persons with high variability would be considered "positive," requiring further follow-up (e.g., diagnostic procedures, control visit, treatment, etc.). Those with low variability would get no further consideration. In this framework, both the screening measure and the health outcome require further definition, described in the following sections.
Working Definitions of Asthma
To test the performance of PEFvar for detecting asthma-related conditions, the outcome was defined with data from the SAPALDIA cross-sectional assessment, which occurred, on average, a half year before the PEF monitoring period. We show the screening performance of PEFvar for four nonexclusive definitions of the target condition, as follows:
- 1. Physician-diagnosed asthma (positive answer to both the questions of: "Have you ever had asthma" and "Was it confirmed by a doctor").
- 2. Current asthma, consisting of physician-diagnosed asthma and either a history of attacks of asthma in the prior 12 mo and/or current use of asthma medication.
- 3. History of wheezing without colds (positive answer for wheezing without a cold in the prior 12 mo).
- 4. Increased BR, defined as a 20% decrease in FEV1 after inhalation of
2 mg of methacholine.
Definitions 1 and 2 refer to the participants' prior contact with a physician; therefore, we mainly tested the screening performance for a condition detected with the existing health care system and diagnostic procedures. Definition 3 is solely based on the subject's reporting of a key symptom of asthma. Definition 4 is an objective measure of airway reactivity, conducted in SAPALDIA according to standardized methods (11). Although BR is not identical to asthma, airway hyperreactivity is a crucial feature of asthma, and in clinical settings is often used to diagnose the disease.
We also present results for a group of healthy subjects, defined as nonsmoking participants with no history of asthma or respiratory symptoms in the last 12 mo prior to the study, and with normoreactive airways.
Analyses
The analyses were conducted with the statistical package SAS/STAT (SAS Institute Inc., Cary, NC [17]) on a mainframe computer. Using the four definitions of asthma given earlier as an alternative "gold standard," we calculated the major statistical features of the screening test according to the schedule of Greenhalgh (18), as follows: sensitivity (i.e., the probability of detecting truly asthmatic individuals with the test), specificity (i.e., the probability that nonasthmatic individuals will have a low PEF variability), likelihood ratio of a positive test (i.e., the likelihood of a positive test result in an asthmatic person as compared with a nonasthmatic person) and of a negative test (i.e., the likelihood to find a negative test in an asthmatic person as compared with a nonasthmatic person), accuracy (i.e., the proportion of all tests with a correct result), and both positive predictive value (i.e., the proportion of asthmatic individuals among those with high PEFvar) and negative predictive value (i.e., proportion of nonasthmatic individuals among those with low PEFvar). Screening performance depends on the pretest probability (i.e., the prevalence of the screened condition). Therefore, we present the screening validity in population subgroups with different pretest probabilities for asthma-related conditions.
| |
RESULTS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
The characteristics of the study population are shown in Table 1 and are compared with those of the SAPALDIA cross-sectional population. According to the SAPALDIA Diary Study sampling strategy, symptomatic and hyperreactive subjects were significantly overrepresented in the PEF study population. A total of 205 Diary Study participants were excluded from this analysis because they had less than 14 d of PEF measurements during the first 4-wk period. This small group, however, did not differ from the PEF study-population (n = 3,074) in any of the characteristics presented in Table 1 (data not shown).
|
Table 2 confirms significant differences in lung function values for the nonsymptomatic subjects as compared with the others. Values are indicated as percent predicted based on the equations for the same population (19). Lung function values were lowest among subjects with current asthma. Age and height distributions did not differ across the subgroups.
|
As shown in Table 3, among subjects with asthma and/or
reactive airways, group mean PEF values were lower and the
diurnal PEFvar distribution was shifted toward higher values
than for the nonsymptomatic subjects. This was particularly
true for those with current asthma (during the prior 12 mo),
although there was considerable overlap (Figure 1). As expected, mean values were lower in women. The mean PEFvar
in men was 5.6% (SD = 3.7; median = 4.6%) and in women
was 5.2% (SD = 3.5; median = 4.4%). This sex difference remained unchanged in a model adjusting for age, height, smoking status, and study area. Both for amplitude as a percent of
the mean and for the log-transformed values, variability was
significantly lower among women. Further analysis revealed a
significant interaction of asthma status and sex. Among currently asthmatic men, mean PEFvar was 8.2% (SD = 5.6; median = 6.5%), as compared to asthmatic women, who had a
nonsignificantly higher PEFvar of 10% (SD = 7.2; median = 7.9%). In our study, age was unrelated to PEFvar in both crude
and adjusted models. The occurrence of days with high PEFvar
was significantly less frequent among nonsymptomatic subjects. In fact, an amplitude as a percent of the mean PEF of
50% was observed only among subjects with asthma-related conditions and was, again, most prevalent among those with
current asthma (3.6%).
|
|
The ability to distinguish between asthmatic subjects, defined in four different ways, and nonasthmatic subjects is
shown in Table 4, using three cutoffs for high PEFvar. The
highest sensitivity did not exceed 36% for any chosen cutoff
used to define high variability, nor for any of the four definitions of asthma. Therefore, the test would miss a large proportion of subjects with asthma-related conditions (high rate of
false-negative results). Although sensitivity was highest for a
20% cutoff, the overall performance (accuracy, positive
predictive value, likelihood ratio of a positive test) was better
for the 30% cutoff.
|
Screening validity data for PEFvar used to detect current
asthma, defining an amplitude as a percent of the mean PEF
of 20% on at least 2 d as positive test, are shown in Table 5.
The sensitivity for a positive test was highest among current
smokers (56.7%). The positive predictive value, however, was
low (11.1%). In the subpopulation of subjects with a history of
wheezing and parental asthma, the sensitivity reached 41.2%.
In this group, the test had the highest positive predictive
value, of 58.3%. We observed the highest screening sensitivity
among smokers with a history of wheezing (69%). As shown
in Table 5, restriction of this subgroup to those without asthma
medication did not improve the screening performance of
PEFvar. In all the other subgroups, the measures of validity
were lower.
|
Table 6 presents the same parameters, but here used for the identification of subjects with bronchial hyperresponsiveness. The PEFvar test performed best among those with a positive history of wheezing and parental asthma. Nevertheless, sensitivity was low (33.3%). In contrast to screening for current asthma (Table 5), further stratification by smoking status did not improve the validity of PEFvar for detecting bronchial hyperresponsiveness (not shown).
|
Screening for hyperreactive airways would be of particular interest for detecting undiagnosed asthma. This is shown in the lower part of Table 6. Among these groups of subjects without current asthma, the sensitivity of PEFvar for hyperreactive airways remained low. Thus using PEFvar to detect BR would miss the majority of subjects with BR. The positive predictive value of high PEFvar remained low, at only 56.7% even among subjects with undiagnosed asthma but with an FEV1 below 80%.
| |
DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
We present population-based data for diurnal PEFvar distribution, assessed with mini-Wright PEF meters during a period of 14 to 21 d. With a total sample of 3,074 adults, these results are based on the largest adult population so far, allowing comparison of variability distributions among population subgroups. As suggested by Higgins and colleagues, we used amplitude as a percent of the mean PEF as a measure of PEFvar (8). In a random sample of 121 adults (18 to 75 yr of age) and 221 subjects with a history of wheezing, they showed the highest intraclass correlation and best separation of asthmatic from nonasthmatic individuals for amplitude as a percent of the mean PEF relative to other measures of PEFvar. Compared with their study, the median amplitude as a percent of the mean PEF was less than half as large in the SAPALDIA sample, both among nonsymptomatic subjects and those with asthma. This could be due to the study protocol used by Higgins and colleagues, calling for PEF measurements every 2 h, whereas SAPALDIA subjects had PEF measured in the morning and evening only. Lebowitz and coworkers have shown that in normal subjects, PEFvar increases with an increasing number of PEF readings used to derive PEFvar (5). The underestimation of variability may be particularly strong in asthmatic individuals (20). In community-based studies, however, the requirement for more than two PEF measurements per day may strongly limit compliance (15). Our PEFvar results were, however, very similar to those reported for the reference group in the study of Quackenboss and coworkers (9) (median value of amplitude as a percent of the mean PEF = 5.0%), as well as for the nonsymptomatic Canadian sample in a recently published cohort study (21) (median of 3.4%), and the Dutch sample (n = 511; age: 20 to 70 yr) of the ECRHS, with median values of the amplitude as a percent of the mean PEF of 3.7% (14, 22).
In our study, both the average PEF and PEFvar were clearly different for nonsymptomatic subjects and those with asthma-related conditions. This was particularly true for the subgroup with current asthma, which had the lowest absolute PEF values and the highest PEFvar, among both men and women (Table 3). The prevalence of subjects with high PEFvar on at least 2 d was strongly related to current asthma, with PEFvar above 20% at least six times more frequent in this group than in nonsymptomatic subjects. Among the latter, PEFvar above 50% could never be observed.
As in the study by Quackenboss and coworkers (9), PEFvar was independent of age in our study. Among the female Dutch ECRHS study sample, a higher PEFvar was observed in older women (22). However, the higher prevalence of asthma and chronic obstructive pulmonary disease in this female subgroup may partly explain this observation (10).
In contrast to other investigators, we observed lower PEFvar among women. As in the Dutch ECRHS study (22), Higgins found a higher PEFvar in women (16). Our analyses suggest that among asthmatic individuals, women may have higher PEFvar than men. Boezen and associates (22) did not provide sex differences within disease categories. As shown in a later analysis of the same ECRHS population, diurnal PEFvar was significantly related to symptoms such as dyspnea, which were more frequent in women (23). Furthermore, in the ECRHS study (22), women not only had higher PEFvar but also larger heterogeneity in this measure. This may also indicate that PEFvar is inherently different in yet undefined subgroups. Given the observed unexplained discrepancies, sex differences of PEFvar needs to be further addressed with particular emphasis on respiratory symptoms and age. The recent publication of the Pollution Effects on Asthmatic Children in Europe (PEACE) study among 6- to 12-yr-old children showed no gender difference (24).
We have reported a detailed assessment of the screening
validity of PEFvar. To our knowledge, this issue has been
partly addressed in adults in only three studies, which reached
conflicting conclusions. The first population-based study, by
Higgins and colleagues (8), compared the PEFvar of 121 randomly selected adults (18 to 75 yr of age) with that of a group
of 221 subjects reporting wheezing in the previous year (8), indicating no clear-cut distinction between the two groups. The
subjects recorded PEF over a period of 7 d, measuring PEF
every 2 h during waking hours. According to Higgins and colleagues, it is not justified to use PEFvar to distinguish these
groups (16). The second study, however, reported the amplitude as a percent of the mean PEF to be a sensitive tool for
detecting asthma in epidemiologic surveys (9). The Tucson population study was based on 207 adults (ages 15 to 65 yr)
and 186 children (ages 6 to 15 yr). The authors' main focus was
sensitivity, which was near 60% and clearly higher than in our
work or in the first population-based study (43%) (8). However, the sensitivity of PEFvar depends on the definition of the
PEFvar and on the selected cutoff for high variability. Quackenboss and coworkers calculated z-scores for each day across
all individual PEFvar values. The 95% range of the individual
97.5% upper limit of the z-scores was used to distinguish normal from high PEFvar. Our derivation of high variability relied
on more restrictive, clinically appealing cutoffs (7). Whereas
23.4% of nonasthmatic and 54.3% of asthmatic subjects had
high values for amplitude as a percent of the mean PEF in the
Tucson study, our least restrictive definition (PEFvar 20%
on at least 2 d) identified only from 6% (nonsymptomatic) to
36% (current asthma) of subjects as having BR (Table 3).
Thus, sensitivity is expected to be lower with our approach.
We also calculated validity for a 10% cutoff level on at least
2 d, in order to define high PEFvar, reaching a sensitivity level
close to 75% for current asthma and 58% for BR (data not
shown). As shown in Figure 1, overlap of the PEFvar distribution is large, underlying its very limited use for screening. Furthermore, in both Quackenboss and coworkers' and our own
study, positive predictive values of PEFvar were low. These results among adults are therefore similar to those reported for
children, for whom PEFvar had a sensitivity for diagnosed
asthma of 36% and a positive predictive value of 45% (25).
A third study, by Thiadens and associates (26), has recently
been published. They studied a total of 182 adult patients who consulted general practitioners for persistent respiratory symptoms, but had not previously been diagnosed as asthmatic. A 2-wk PEFvar assessment was used, and asthma diagnostic procedures were conducted. The authors concluded
that PEFvar may be helpful in screening for asthma among
subjects who have been coughing for at least 2 wk. The discrepancy between our interpretation and that of Thiadens and
associates relates to the different setting and focus of their
work. First, their study aimed at an optimal definition of high
PEFvar to maximize the probability of asthma given a positive
test. The probability that the test score indicates asthma increases with the number of days during which PEFvar is increased. In other words, their emphasis was on the positive predictive value of PEFvar, which depends heavily on the pretest probability of an asthma-related conditon (18). In a population survey it may be feasible to assign subjects to groups of
different pretest probabilities for asthma-related conditions.
As in our study, this may be based on prior questionnaire
data. As shown in Table 6 (e.g., among subjects with a history
of wheezing, particularly if combined with a parental history
of asthma), the positive predictive values steadily improved,
reaching values like those in the Dutch study, in which the patients had a high pretest probability for asthma (> 40%) (26).
Sensitivities of the test criteria in the Dutch study were generally low, and were comparable to our results, in which sensitivity reached 69% at best. During the Dutch 2-wk PEFvar
screening period, 30 of 69 asthmatic subjects did not have a
single day in which PEFvar 15%, and 44 never had PEFvar
20% (defined as amplitude in percent of the highest reading) (26). In the Dutch study, the highest sensitivity for a diagnosis of asthma did not exceed 40%, using at least 2 d with
PEFvar 15% or 20% (defined as amplitude in percent of
the highest reading) as a test criterion.
Detection of asthma-related conditions could be of particular interest among yet undiagnosed subjects (26). This has been shown with BR (lower part of Table 6) as an objective marker. The validity of a high PEFvar was generally low, with the best performance among subjects without (known) current asthma but with an FEV1 below 80% predicted. As shown by the Dutch Study group (4, 10), BR and PEFvar, although significantly correlated measures of airway variability, are not interchangeable. In fact, they may measure different aspects of asthma (7, 9, 27). Our observation of a limited sensitivity of PEFvar in predicting BR is in line with these earlier studies. One may argue that the limitations of the validity of PEFvar as a screening tool in our study were due to having well-treated asthmatic subjects as the target group. However, as shown in the subgroup analyses (Tables 5 and 6), the performance of PEFvar was only marginally improved in groups with the highest pretest probability for a high PEFvar: namely, among those with no treatment (no current asthma) but with a history of wheezing, a family history of asthma, current smoking, or with obstructive airways disease (reduced FEV1).
The positive predictive value of PEFvar was generally poor,
indicating that among those subjects tagged as positive by the PEF test, only a minority would in fact be asthmatic (e.g., with an amplitude as a percent of the mean PEF of 30% on at
least 2 d defined as high variability, the positive predictive
value was only 27% for current asthma, reaching 61% for BR
[data not shown]). As shown by Thiadens and associates, the
probability of asthma strongly increases with the number of
days of high variability. However, the downside of a high positive predictive value is the drastic reduction of sensitivity.
It has to be emphasized that the interpretation of our results and of other screening data, such as those of Thiadens and associates, depends on the purpose and setting of the screening procedure. From a clinical perspective, sensitivity may be emphasized to ensure that all subjects with disease are detected and get treatment. In accordance with Thiadens and associates, we confirm that PEFvar may not serve this purpose, given its low sensitivity. From a diagnostic perspective within a defined selected clinical setting (e.g., general-practice patients presenting with persistent symptoms [26]), one may accept low sensitivity while maximizing the positive predictive value of a particular measure. In this case, PEFvar may be useful, and the strategy shown by Thiadens and associates may be applied. From a health-care-system perspective, consequences of both the sensitivity and positive predictive value of a screening program ought to be taken into account. Before implementing screening with sensitive (but not specific) tests, the procedures to be applied to subjects with positive results should be clarified in order to identify false positives in a second screening step (e.g., as a consequence of a positive predictive value of about 20%, as in our study or the one by Quackenboss and coworkers, 80% of subjects with high PEFvar would undergo further procedures despite being unlikely to have an asthma-related condition). The selection of the screening population strongly influences the validity of a particular measure, which depends on pretest probabilities. In a general population, the screening performance of PEFvar may even be worse than reported in the present study. First, to derive our population from the cross-sectional random SAPALDIA population (11), we oversampled subjects with respiratory symptoms and reactive airways (Table 1). The performance of a PEF test in an unselected random population is likely to be even less valid than in our total sample. Second, compliance of the screened populations with the measurement of PEF has to be taken into account. As shown by Quirce and coworkers (28) in an occupational population, more than 20% of reported PEF values were invented rather than measured. We may assume that our participants, derived from the SAPALDIA cross-sectional study, represented a rather compliant selection of the population, and the screening validity of PEFvar in an unselected general population is expected to be lower.
In conclusion, PEFvar can be easily measured twice a day over a 3-wk period in large populations. It may be of great use as a marker of variability in airway caliber, in both epidemiologic and clinical studies. Tests of PEFvar may be useful in clearly defined clinical settings as part of the overall diagnostic procedure. However, their use in population screening for asthma-related conditions has serious limitations because of the large overlap of PEFvar distribution between asthmatic and nonasthmatic individuals and the low sensitivity of PEFvar even in subgroups with an increased pretest probability of having asthma-related conditions.
| |
Footnotes |
|---|
Correspondence and requests for reprints should be addressed to Nino Künzli, M.D., Ph.D., Institut für Sozial- und Präventivmedizin der Universität Basel, Steinengraben 49, 4051 Basel, Switzerland. E-mail: kuenzlini{at}ubaclu.unibas.ch
(Received in original form July 2, 1998 and in revised form February 8, 1999).
This work was published in abstract form in American Journal of Respiratory and Critical Care Medicine 1997;155:A78.SAPALDIA Team: P. Leuenberger, Study Director; U. Ackermann-Liebrich, Program Director; P. Alean, K. Blaser, P. Bolognini, J. P. Bongard, O. Brändli, P. Braun, C. Bron, M. Brutsche, C. Defila, G. Domenighetti, S. Elsasser, L. Grize, P. Guldimann, P. Hufschmid, W. Karrer, H. Keller-Wossidlo, R. Keller, N. Künzli, J. C. Luthi, B. W. Martin, T. Medici, C. Monn, A. G. Peeters, A. P. Perruchoud, A. Radaelli, C. Schindler, J. Schwartz, G. Solari, M. Schoni, J. M. Tschopp, B. Villiger, B. Wuthrich, J. P. Zellweger, and E. Zemp.
The SAPALDIA study is supported by Swiss National Science Foundation grant 4026-028099. Drs. Künzli and Zemp have an Advanced Research Fellowship of the Swiss Science Foundation (Nos. 32-048922.96/1; 32-35450.92).
Acknowledgments: The authors are grateful to the study populations and the SAPALDIA technicians in all eight study centers.
| |
References |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
|
|---|
1. Barnes, P.. 1992. Poorly received asthma. Thorax 47: 408-409 [Medline].
2.
Siersted, H.,
J. Boldsen,
H. Hansen,
G. Mostgaard, and
N. Hyldebrandt.
1998.
Population based study of risk factors for underdiagnosis of
asthma in adolescence: Odense schoolchild study.
Br. Med. J.
316:
651-655
3. Paggiaro, P., G. Moscato, D. Giannini, A. Di Franco, and G. Gherson. 1997. Relationship between peak expiratory flow (PEF) and FEV1. Eur. Respir. J. 10(Suppl. 24):39s-41s.
4. Brand, P., D. Postma, H. Kerstjens, G. Koeter, and The Dutch CNSLD Study Group. 1991. Relationship of airway hyperresponsiveness to respiratory symptoms and diurnal peak flow variation in patients with obstructive lung disease. Am. Rev. Respir. Dis. 143: 916-921 [Medline].
5. Lebowitz, M., M. Krzyzanowski, J. Quackenboss, and M. O'Rourke. 1997. Diurnal variation of PEF and its use in epidemiological studies. Eur. Respir. J. 10(Suppl. 24):49s-56s.
6. Lebowitz, M. D.. 1991. The use of peak expiratory flow rate measurements in respiratory disease. Pediatr. Pulmonol. 11: 166-174 [Medline].
7. Enright, P., M. Lebowitz, and D. Cockroft. 1994. Physiologic measures: pulmonary function test. Am. J. Respir. Crit. Care Med. 149: S9-S18 .
8. Higgins, B., J. Britton, S. Chinn, D. Trevor, D. Jenkins, P. Burney, and A. Tattersfield. 1989. The distribution of peak expiratory flow variability in a population sample. Am. Rev. Respir. Dis. 140: 1368-1372 [Medline].
9. Quackenboss, J., M. Lebowitz, and M. Krzyzanowski. 1991. The normal range of diurnal changes in peak expiratory flow rates: relationship to symptoms and respiratory disease. Am. Rev. Respir. Dis. 143: 323-330 [Medline].
10. Boezen, H., D. Postma, J. Schouten, H. Kerstjens, and B. Rijcken. 1996. PEF variability, bronchial responsiveness and their relation to allergy markers in a random population (20-70 yr). Am. J. Respir. Crit. Care Med. 154: 30-35 [Abstract].
11.
Martin, B.,
U. Ackermann-Liebrich,
P. Leuenberger,
N. Künzli,
E. Zemp,
Stutz,
R. Keller,
J. P. Zellweger,
B. Wuthrich,
C. Monn,
K. Blaser,
G. Bolognini,
J. P. Bongard,
O. Brandli,
P. Braun,
C. Defila,
G. Domenighetti,
L. Grize,
W. Karrer,
H. Keller-Wossidlo,
T. C. Medici,
A. Peeters,
A. P. Perruchoud,
C. Schindler,
M. H. Schoeni,
B. Villiger, and
SAPALDIA Team.
1997.
SAPALDIA
Methods and participation in the cross-sectional part of the Swiss Study on Air Pollution
and Lung Disease in Adults (SAPALDIA).
Soz. Präv. Med.
42:
67-79
.
12. den Otter, J., G. Reijnen, W. J. H. M. van den Boesch, C. van Schayck, J. Molema, and C. van Weel. 1997. Testing bronchial hyper-responsiveness: provocation or peak expiratory flow variability? Br. J. Gen. Pract. 47: 487-492 [Medline].
13. Ackermann-Liebrich, U., P. Leuenberger, J. Schwartz, C. Schindler, C. Monn, G. Bolognini, J. Bongard, O. Brändli, G. Domenighetti, S. Elsasser, L. Grize, W. Karrer, R. Keller, H. Keller-Wossidlo, N. Künzli, B. W. Martin, T. C. Medici, A. P. Perruchoud, M. H. Schoni, J. M. Tschopp, B. Villiger, B. Wuthrich, J. P. Zellweger, and E. Zemp. 1997. Lung function and long-term exposure to air pollutants in Switzerland. Am. J. Respir. Crit. Care Med. 155: 122-129 [Abstract].
14. Burney, P., C. Luczynska, S. Chinn, and D. Jarvis. 1994. The European Community Respiratory Health Survey. Eur. Respir. J. 7: 954-960 [Abstract].
15. Britton, J. 1997. Measurement of peak flow variability in community populations: methodology. Eur. Respir. J. 10(Suppl. 24):42s-44s.
16. Higgins, B. 1997. Peak expiratory flow variability in the general population. Eur. Respir. J. 10(Suppl. 24):45s-48s.
17. SAS Institute Inc. 1989. SAS/STAT User's Guide, Version 6, 4th ed. SAS Institute Inc., Cary, NC.
18.
Greenhalgh, T..
1997.
How to read a paper: papers that report diagnostic
or screening tests.
Br. Med. J.
315:
540-543
19. Brändli, O., C. Schindler, N. Künzli, R. Keller, A. Perruchoud, and the SAPALDIA Team. 1996. Lung function in healthy never smoking adults: reference values and lower limits of normal of a Swiss population. Thorax 51: 277-283 [Abstract].
20. D'Alonzo, G., V. Steinijans, and A. Keller. 1995. Measurements of morning and evening airflow grossly underestimate the circadian variability of FEV1 and peak expiratory flow rate in asthma. Am. J. Respir. Crit. Care Med. 152: 1097-1099 [Abstract].
21.
Laprise, C., and
L. Boulet.
1997.
Asymptomatic airway hyperresponsiveness: a three-year follow-up.
Am. J. Respir. Crit. Care Med.
156:
403-409
22. Boezen, H., J. Schouten, D. Postma, and B. Rijcken. 1994. Distribution of peak expiratory flow variability by age, gender and smoking habits in a random population sample. Eur. Respir. J. 7: 1814-1820 [Abstract].
23. Boezen, H., J. Schouten, D. Postma, and B. Rijcken. 1995. Relation between respiratory symptoms, pulmonary function and peak flow variability in adults. Thorax 50: 121-126 [Abstract].
24.
Timonen, K.,
J. Nielsen,
J. Schwartz,
A. Gotti,
V. Vondra,
C. Gratziou,
P. Giaever,
W. Roemer, and
B. Brunekreef.
1997.
Chronic respiratory
symptoms, skin test results, and lung function as predictors of peak
flow variability.
Am. J. Respir. Crit. Care Med.
156:
776-782
25. Siersted, H., G. Mostgaard, N. Hyldebrandt, H. Hansen, J. Boldsen, and H. Oxhoj. 1996. Interrelationships between diagnosed asthma, asthma-like symptoms, and abnormal airway behavior in adolescence: the Odense Schoolchild Study. Thorax 51: 503-509 [Abstract].
26. Thiadens, H., G. De Bock, F. Dekker, J. Huysman, J. Van Houwelingen, M. Springer, and D. Postma. 1998. Value of measuring diurnal peak flow variability in the recognition of asthma: a study in general practice. Eur. Respir. J. 12: 842-847 [Abstract].
27. Josephs, L., I. Gregg, M. Mullee, and S. Holgate. 1989. Nonspecific bronchial reactivity and its relationship to the clinical expression of asthma: a longitudinal study. Am. Rev. Respir. Dis. 140: 350-357 [Medline].
28. Quirce, S., G. Contreras, A. Dybuncio, and M. Chan-Yeung. 1995. Peak expiratory flow monitoring is not a reliable method for establishing the diagnosis of occupational asthma. Am. J. Respir. Crit. Care Med. 152: 1100-1102 [Abstract].
This article has been cited by other articles:
 |
|
 |
|
  P. L. Enright, R. L. McClelland, A. S. Buist, and M. D. Lebowitz Correlates of Peak Expiratory Flow Lability in Elderly Persons Chest, December 1, 2001; 120(6): 1861 - 1868. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. F. Goldstein, B. A. Veza, E. H. Dunsky, D. J. Dvorin, G. A. Belecanech, and I. C. Haralabatos Comparisons of Peak Diurnal Expiratory Flow Variation, Postbronchodilator FEV1 Responses, and Methacholine Inhalation Challenges in the Evaluation of Suspected Asthma Chest, April 1, 2001; 119(4): 1001 - 1010. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Proc. Am. Thorac. Soc. | Am. J. Respir. Cell Mol. Biol. |