The American Thoracic Society ’ s spirometric criteria alone is inadequate in asthma diagnosis

1Division of Respirology; 2Division of Rhematology, Department of Medicine, Royal University Hospital, University of Saskatchewan, Saskatoon, Saskatchewan Correspondence: Dr John A Gjevre, Division of Respirology, Department of Medicine, Royal University Hospital, 103 Hospital Drive, University of Saskatchewan, Saskatoon, Saskatchewan S7N 0W8. Telephone 306-966-1808, fax 306-966-8694, e-mail john.gjevre@usask.ca JA Gjevre, TS Hurst, RM Taylor-Gjevre, DW Cockcroft. The American Thoracic Society’s spirometric criteria alone is inadequate in asthma diagnosis. Can Respir J 2006;13(8):433437.

The American Thoracic Society's spirometric criteria alone is inadequate in asthma diagnosis Les seuls critères spirométriques de l'American Thoracic Society sont insuffisants pour diagnostiquer l'asthme A sthma is an inflammatory disease of the airways with generally reversible air flow obstruction and airway hyperresponsiveness causing episodic respiratory symptoms (1).In contrast, chronic obstructive pulmonary disease (COPD) is defined by the presence of air flow limitation that is not fully reversible after inhaled bronchodilators (2).The current standard for asthma diagnosis is based on the typical clinical features in conjunction with objective documentation of airway dysfunction, either with a significant change in forced expiratory volume in 1 s (FEV 1 ) after bronchodilator administration or with airways hyperresponsiveness.Alternatively, a significant change in forced vital capacity (FVC) postbronchodilator may be used.A postbronchodilator spirometric response by itself is nonspecific, and the diagnosis relies on physician confirmation.The level of bronchodilator response considered to be diagnostically 'significant' has been the subject of controversy (3,4).The definition of the degree of airway reversibility varies widely (5,6).The American Thoracic Society (ATS) has developed criteria which suggest a significant postbronchodilator FEV 1 and/or FVC response of 200 mL or greater and 12% improvement from baseline (7).The ATS has recently updated these criteria in collaboration with the ©2006 Pulsus Group Inc.All rights reserved

ORIGINAL ARTICLE
European Respiratory Society (8).To our knowledge, these criteria have never been fully validated in a North American population by testing against the clinical diagnosis of asthma.In the present study, patients at the Royal University Hospital (Saskatoon, Saskatchewan) who met the ATS bronchodilator spirometric criteria were examined for bronchodilator responses in reference to clinical diagnosis.

METHODS
All spirometry tests performed on patients referred to the Royal University Hospital adult pulmonary function testing laboratory between September 1999 and September 2004 were reviewed.All pulmonary function testing was performed in accordance with standard ATS protocol with the patient seated, the use of nose clips and the best of three spirometry flow readings recorded.Testing was performed using the Vmax 22 devices (SensorMedics Yorba Linda, USA).The Morris equation was used for predicted FEV 1 and FVC (9).A significant bronchodilator response was defined as an increase in FEV 1 (or FVC) of 12% (and 200 mL or greater) (5).If a patient had multiple spirometric tests, the most recent test results were recorded.Because access to medical records was required for clinical correlation, only adult patients (older than 18 years of age) who had been seen in outpatient clinics by staff physicians (n=10) of the respirology division were included in the present study.The total spirometry tests and delineation of patients meeting the study criteria are outlined in Figure 1.
All medical records were reviewed by a single investigator using a standard data collection form.The data recorded included age, sex, height, weight, FVC pre-and postbronchodilator, FEV 1 pre-and postbronchodilator, forced expiratory flow at 25% to 75% (FEF [25%-75%] ) pre-and postbronchodilator, referring physician diagnosis and staff respirologist's diagnosis.Diagnoses were recorded as asthma, COPD, bronchiectasis and 'other'.The 'other' category included chronic cough, dyspnea, pleural diseases, pneumonia, interstitial lung disease, cardiac diseases, sinus diseases, sleep-disordered breathing and miscellaneous lung diseases.While each separate diagnosis in the 'other' category was recorded in the database, these diagnostic codes were grouped together for data analysis to assess the spirometry bronchodilator response in comparison with the respirologist's clinical diagnosis of asthma.Because of the patient population, the study observed COPD and bronchiectasis, and thus this led to the four diagnostic categories.For the purposes of the present study, all patients who had either an isolated diagnosis of asthma or a concomitant diagnosis of asthma were considered to have asthma for data analysis.The standard for asthma diagnosis was the staff respirologist's recorded diagnosis based on an extensive consultation assessment (history, complete physical examination and spirometry) in the outpatient respirology clinic.A standard questionnaire was not used by the 10 respirologists because the present study was a retrospective reallife comparison of physician diagnosis with respect to ATS spirometry criteria.All 10 staff respirologists consented to the review of their patients' data.Approval for the present study was obtained from the University of Saskatchewan Medical Research Ethics Committee.

Statistical analysis
SPSS version 12.0 (SPSS Inc, USA) was used for data entry and analysis.For the purpose of comparison, patients were divided into two categories -'asthma' and 'nonasthma'.The nonasthma group was further divided into 'COPD', 'bronchiectasis' and 'other'.Two group comparisons of continuous data were performed using independent two-tailed t tests.ANOVA was used for three or more group comparisons.The Bonferroni correction for multiple comparisons was used.Frequency data were evaluated by χ 2 testing.Comparison of varying levels of FEV 1 and FVC response between diagnostic groups allowed the plotting of a receiveroperator characteristic (ROC) curve that was used to further assess the value of these diagnostic criteria.

RESULTS
Of the 310 adult respiratory clinic patients included in the present study, 168 (54.2%) were male.The mean age was 63.2 years (range 22 to 89 years).The clinical diagnoses represented in this population, as extracted from the respiratory clinic medical record, were divided into four general categoriesasthma, COPD, bronchiectasis and 'other'.The referring physician's diagnosis correlated with the respirologist's diagnosis for asthma in 94 of 169 patients, for COPD in 61 of 85 patients, for bronchiectasis in six of 13 patients and for 'other' diagnoses in 41 of 43 patients.
The mean percentage improvement in FEV 1 postbronchodilator for the total population was 18.66% (median 16.67, SD 6.35).The mean percentage improvement for FVC postbronchodilator was 13.25% (median 12.03, SD 8.00).The mean percentage improvement in FEF (25%-75%) postbronchodilator was 28.42 (median 25.53, SD 25.51).The mean values for these variables in each of the diagnostic categories are outlined in Table 1.
The comparison of percentage improvements in these variables among these four diagnostic categories revealed significant differences between the groups.For percentage improvement in FEV 1 , no significant differences were observed by ANOVA among the four diagnostic groups (Figure 2).For percentage improvement in FVC, significant differences were observed among the asthmatic group and the 'other' group (P=0.035), and no significant differences in FVC were observed between the asthmatic group and the COPD or bronchiectasis groups (Figure 3).Comparison of percentage improvement in FEF (25%-75%) revealed significant differences between the asthmatic group and both the COPD and 'other' groups (P=0.002 and P=0.004, respectively) (Figure 4).When the nonasthmatic diagnostic categories were collapsed into a single disease category, a mean percentage FEV 1 postbronchodilator improvement of 17.82% (SD 5.51), mean percentage FVC improvement of 12.34% (SD 7.86) and mean percentage FEF (25%-75%) improvement of 27.53% (SD 25.89) was seen.When these values were compared with the asthma diagnostic group results, a difference between the percentage change in FEV 1 (P=0.031)was observed.However, this finding loses significance when corrected for multiple comparisons.No significant differences were observed between percentage changes in FVC or FEF (25%-75%) between these two groups.The population was then divided into those patients who met both the FEV 1 and the FVC criteria and patients who fulfilled just one of the two criteria.This comparison resulted in essentially equal numbers of patients in both the asthma and 'nonasthma' categories who met only one of the two ATS criteria.In the patients meeting both ATS criteria, 93 of 155 (60%) were diagnosed with asthma.This difference approached, but did not achieve, significance (P=0.06).
Then the diagnostic role of magnitude of bronchodilator responses was considered.Of the 310 patients meeting the 12% ATS criteria, 169 (54.5%) were clinically diagnosed with asthma.By increasing the requirement to 15% improvement  An ROC curve (Figure 5) was used to further evaluate the relationship between diagnostic sensitivity and specificity of the ATS criteria (200 mL or greater, and 12% improvement) within this particular study population.Recalling that when a test contributes no diagnostic information, the area under the curve would equal 0.5, and that when a test has 100% sensitivity and specificity, the area would then equal 1.0, the value obtained for the area under the curve for percentage improvements in FEV 1 was 0.569 (P=0.037), in FVC was 0.569 (P=0.036) and in FEF (25%-75%) was 0.517 (P=0.608).As shown in Figure 5, the ROC curve for each of these variables shadows the diagonal reference line.
To correct for (between-subject) variations in baseline values, the data expressing the changes in FEV 1 and FVC as percentages of the predicted value (rather than the percentages of baseline) were analyzed.Using the percentage prediction approach in patients who met strict criteria for airway obstruction at baseline (FEV 1 /FVC ratio of 70% or less, and FEV 1 less than 80% predicted, n=213), there was an increase in the ROC area under the curve for a percentage change in FEV 1 of 0.678 (P=0.000) and for a percentage change in FVC of 0.583 (P=0.038).For the whole population (n=310), there was a smaller magnitude of increase with a percentage change in FEV 1 of 0.640 (P=0.000) and a slight increase in percentage in FVC of 0.598 (P=0.003).In addition, using the percentage difference in predicted FEV 1 , there were significant differences between the asthma and COPD groups using posthoc comparisons with ANOVA (P=0.000).

DISCUSSION
For many years, the definition of a 'significant' bronchodilator response has been controversial.The threshold at which the spirometric response criteria are set determines the prevalence of airway obstruction in the general population.In a recent large cross-sectional study, Viegi et al (10) showed clear differences between 'clinical criteria' and both European Respiratory Society and ATS definitions of airway obstruction.In that study population, the ATS criteria had the highest sensitivity but lowest specificity for any respiratory symptom or disease.Likewise, there is a variation in the clinical assessment of FEV 1 bronchodilator response.In a recent Australian survey of laboratory testing practices for bronchodilator reversibility testing, Borg et al (11) found significant variation in methods used to assess and interpret bronchodilator response.Even with careful and consistent application of a single method for assessing bronchodilator response, there can be variation over time in the percentage improvement of FEV 1 (12).In the Inhaled Steroids in Obstructive Lung Disease (ISOLDE) study population of moderate to severe COPD patients, investigators found that bronchodilator responsiveness was a continuous variable and over 52% of patients changed responder status between trimonthly physician visits (13).This further emphasizes the need to use spirometric criteria as a guide but not as an unimpeachable gold standard with which to make a diagnosis of asthma.
In our study population, relying on spirometric criteria alone was found to be inadequate in asthma diagnosis because only 54.7% of 310 patients meeting ATS bronchodilator response criteria were thought to be clinically asthmatic.By raising the threshold of FEV 1 bronchodilator response from the 12% (provided it is 200 mL or greater) required by ATS criteria to 20%, there was a little improvement in the positive predictive value of the spirometric test results, increasing from 54.5% to 63.5%.This increased positive predictive value was associated with a low sensitivity of 36.1%.When the spirometry results of the asthmatic patients were compared with those from the COPD, bronchiectasis and 'other' disease categories, we saw no significant differences among groups for percentage improvement in FEV 1 , and only between the asthma group and 'other' group for percentage improvement in FVC.The ROC areas under the curve values for these variables, although statistically significant, are uncomfortably close to the diagonal reference line which represents a test that contributes no information.These study results suggest the current ATS criteria have relatively poor specificity for asthma in our population.Sensitivity of these current criteria was outside the scope of the present study because only patients actually meeting the pulmonary function testing criteria were included in the present study.Other epidemiologically based studies evaluating asthma diagnoses have found that a bronchodilator response gives a higher specificity but overall poor sensitivity (14,15).
It is notable that many patients who did not have a clinical diagnosis of asthma had what is considered a significant bronchodilator response.While it is well recognized that approximately 10% of the COPD population had a component of bronchodilator response (16) Because the present study population was drawn from a tertiary care hospital pulmonary function testing laboratory and respirology subspecialty clinical practice, these findings may not be generalizable to other populations.Furthermore, the patient population included in the present study was heterogeneous with both new and long-standing respiratory clinic patients.Accordingly, varying degrees of pharmacotherapeutic intervention will have been used, hopefully with asthma-controlling benefits.It has been previously reported that fewer than 10% of asthmatic patients attending a tertiary referral respiratory clinic fulfill the typical spirometric inclusion criteria (which indicate suboptimal asthma control) for asthma drug trials (17).Thus, this anticipated superior control may result in a lesser degree of postbronchodilator improvement in established asthma patients, which would influence our results.
The present study was a retrospective analysis comparing a recorded clinical diagnosis of asthma with spirometric results from September 1999 to September 2004.The recorded clinical diagnoses were performed by experienced respirologists at a single site through the review of patient history and physical examinations.The spirometric data were available to these physicians at the time of the clinic visit and may potentially have influenced the clinical diagnosis.However, if so, one would expect the bias to be in favour of a diagnosis of asthma in this population who all met the ATS spirometry criteria for asthma.There were 10 different respirologists contributing to this data pool.Interobserver variation in diagnosis was not determined because patients had seen a single respirologist.It has been previously recognized that significant variation can exist among respirologists, both in the classification of obstruction and in the determination of bronchodilator response.In an Argentinean study (18) of 30 respirologists, the degree of disagreement for response to bronchodilator was 24%.

SUMMARY
While the ATS FEV 1 and FVC criteria are helpful in the diagnosis of asthma, relying on spirometric criteria alone is inadequate.In the present study, only 54.7% of patients meeting the ATS bronchodilator improvement criteria were thought to have clinical asthma.The diagnosis of asthma depends upon a careful review of patient history and physical examination.Spirometric changes may be helpful in supporting the diagnosis, but the true significance of various levels of bronchodilator response remains unclear.

FUNDING:
The authors thank The Lung Association of Saskatchewan for providing research support to Dr Hurst (The Lung Association of Saskatchewan Professorship) and Dr Cockcroft (Ferguson Professor of the Division of Respiratory Medicine).

TABLE 1
, this does not explain the overall number of patient FEV 1 responders in the 'nonasthma' group of the study.It seems a reassessment of what is considered a normal versus a 'significant' FEV 1 bronchodilator response in diagnosis of reactive airway disease is required.Overall, in the