INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

When occupational asthma is suspected, the diagnosis should be confirmed by objective methods (1, 2). Serial monitoring of FEV₁ during specific inhalation challenges in the laboratory or in the workplace is the most accurate method to confirm the cause-effect relationship between a specific occupational exposure and asthma (2). However, these tests are not widely available. The use of serial peak expiratory flow (PEF) monitoring has been extensively evaluated as a simple and widely available method (3). This method has limitations: lack of compliance, possible falsification of results, and at times long recording periods (4).

FEV₁ is still regarded as the gold standard for measuring airflow caliber. This physiologic index reflects the caliber of both large and small airways, whereas PEF is more a reflection of the caliber of large airways (5). We have previously shown that FEV₁ is more sensitive than PEF in detecting late asthmatic reactions, a frequent occurrence in occupational asthma (6). A portable and relatively inexpensive apparatus that can record both PEF and FEV₁ could therefore represent a significant improvement over instruments that record PEF only in the investigation of occupational asthma. Burge and coworkers (7), in their pioneering studies, suggested retaining the best of two reproducible values of PEF, i.e., within 20 L/min of each other, for analysis and discarding nonreproducible ones (7). The effect of such a selection on the accuracy of PEF monitoring in the diagnosis of occupational asthma remains unclear. Becklake (10) has suggested that test failure resulting in nonreproducible spirometric results is more likely to occur in subjects who have airflow limitation. It can therefore be hypothesized that keeping these nonreproducible values may improve the assessment.

The aims of this prospective study were: (1) to compare the accuracy of PEF and FEV₁ monitoring in the diagnosis of occupational asthma, using specific inhalation challenge tests as the gold standard, and (2) to study the effects of selecting only reproducible values (best of all versus best of two reproducible values) on the interpretation of graphs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twenty consecutive subjects were referred to the Department of Chest Medicine at Hôpital du Sacré-Coeur, Montreal, or the Department of Medicine at the University of British Columbia, Vancouver, for investigation of possible occupational asthma. All had a history suggestive of occupational asthma, including asthma symptoms that worsened at work and improved away from work. Safety data sheets of all products present in the workplace were obtained for these subjects. The need for antiasthmatic medications was unchanged during the study period. Inhaled ₂-adrenergic agents were prescribed on an as-needed basis.

Baseline Evaluation

All subjects were assessed in the same way at the time they were referred to the Department. Spirometry was carried out according to the American Thoracic Society recommendations (11). Reference values for FEV₁ were obtained from Knudson and coworkers (12) and for PEF from Nunn and Gregg (13). A methacholine challenge test was performed with a Wright nebulizer (output, 0.14 L/min) at tidal volume breathing for 2 min using the procedure described by Cockroft and coworkers (14). The provocative concentration of methacholine that induced a 20% fall in FEV₁ (PC₂₀) was determined on a semilogarithmic scale by interpolation of the last two points. A PC₂₀ of 16 mg/ml or less was considered to represent the presence of significant bronchial hyperresponsiveness.

PEF and FEV₁ Monitoring

The Clement Clarke Ventilometer VM1 (Clement Clarke Inc., London, UK) was used to assess FEV₁ and PEF. This new apparatus consists of a slightly modified mini-Wright peak-flow meter with a sensor and electronic signal processing circuitry. The VM1 monitors the expired breath of the subject over a period of 10 s. A transducer monitors the pressure in the expired stream at a point close to the mouthpiece and supplies an electrical signal to the processing circuitry. The signal is digitalized and processed, and digital readings of FEV₁ and PEF are displayed. The mini-Wright peak-flow meter has its own pointer and approximate measuring scale, in order to check that the VM1 is functioning correctly.

Prior to the study, the reliability of this new apparatus was assessed in our laboratory. Briefly, 53 consecutive subjects with a wide range of airway calibers (mean FEV₁ ± SD, 2.16 ± 1.2 L; range, 0.4 to 4.8 L) attending an outpatient respiratory clinic were recruited prospectively. Forced expiratory maneuvers were performed according to the criteria of the American Thoracic Society (11), under the close supervision of a respiratory technician, either on the VM1 followed by a Vitalograph dry spirometer (Vitalograph, Buckingham, UK) or in the reverse order. For PEF, values measured with the digital reading were compared with values measured with the pointer of the enclosed mini-Wright. For FEV₁, values measured with the VM1 were compared with values measured with the Vitalograph. For each measurement, the best of three consecutive blows was kept for analysis. Means ± SD of PEF values assessed with the digital reading and with the pointer were 366.6 L/min ± 166.5 and 371.0 L/min ± 154.9, respectively (the t test for paired samples was not significant). Means ± SD of FEV₁ values assessed by the VM1 and the Vitalograph were 2.15 L ± 1.2 and 2.16 L ± 1.2, respectively (the t test for paired samples was not significant). The strength of association determined using simple linear regression was high, with Pearson's correlation coefficient of 0.99 for PEF and of 0.97 for FEV₁.

Daily Diary Completion

Detailed oral and written instructions concerning the proper use of the VM1 were provided. We emphasized the importance of not storing the instrument in cold places, especially in a car during the winter term. Subjects were asked to record their values four times daily (each value of both FEV₁ and PEF from three consecutive blows) (15). Days at work and days off work were identified. Any inhalation of short-acting ₂-adrenergic agent was also recorded. Recording could start during a period at work or away from work, depending on which was most convenient.

Diagnosis of Occupational Asthma Based on Diary Data

Data were entered into personal computer files and verified. PEF and FEV₁ data were separately plotted according to the method of Burge and coworkers, which produced graphs of maximum, minimum, and mean daily values (9). Graphs of individual data were also drawn. Working periods and any inhalation of rescue medication were specified. We aimed at having at least two periods of 5 d at work followed by 2 away from work (a full 2 wk). However, a shorter period at work occurred in five instances (Subjects 4, 8, 9, 13, and 15) (Table 1), the subjects being removed from the workplace because they were too symptomatic and could not carry on with their usual work. In three instances, the period away from work was 3 d and not 4 (Subjects 7, 8, and 16) (Table 1), the reason being that the period at work did not result in any significant fluctuation, therefore not even justifying a period away from work. In all instances, the graphs were judged to show trends that were sufficient for suggesting a diagnosis.

For the purpose of this study, two separate sets of graphs were plotted: first, the best of three consecutive blows was used, and the same number of PEF and FEV₁ readings plotted on the graphs; second, only the best of two reproducible blows, that is, ± 20 L/min for PEF, and ± 5% for FEV₁, was kept for analysis. The value of ± 20 L/ min for PEF corresponded to ~ 5% as the mean baseline PEF value was 475 L/min (Table 1). Graphs were interpreted independently by three readers (M.C., J.L.M., and L.P.) according to previous proposals (16, 17): they were visually analyzed for the presence or absence of occupational asthma in a blind randomized fashion. Readers first interpreted the graphs plotted with the best of two reproducible blows. Three months later, the graphs plotted with the best of three consecutive blows were interpreted. The lowest value of PC₂₀ assessed during the monitoring, either while at work (one to five values obtained depending on the subject) or while away from work (one to 11 values obtained), was the only additional information given to these experts.

Diagnosis of Occupational Asthma Based on Specific Inhalation Challenge or Work-shift Spirometry

A specific inhalation challenge test was carried out in the laboratory (n = 17) or at work (n = 3) by exposing the subject to the relevant occupational agent as previously described (1). Specific inhalation challenge tests took place within 2 wk after completion of PEF and FEV₁ monitoring. Inhalation challenge test was considered positive when there was a sustained fall in FEV₁ reaching at least 20% from the preexposure value after exposure to the specific agent in the absence of significant (< 10%) changes on a control day.

Data Analysis

The following indices were also calculated for both PEF and FEV₁: Index 1, amplitude percent mean (highest reading  lowest/mean × 100); Index 2, amplitude percent highest (highest  lowest/highest × 100); the latter index was selected from previous studies as it demonstrated the best diagnostic performance among other calculated indices (16, 18). For each of these indices, positive criteria for occupational asthma were defined as follows: at least three or more days at work than away from work with values above 12 or 20%.

The results of visual analyses of both FEV₁ and PEF graphs were studied using the kappa coefficient of agreement among the three readers (19). Sensitivity, specificity, and predictive values were calculated for the results of the visual analyses and indices, using specific inhalation challenge tests as the gold standard. In the case of the visual analyses, the presence of occupational asthma required the agreement of at least two of the three readers (17).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Some of the baseline anthropometric, clinical, and functional results for the study subjects, as well as the duration of the monitoring period at work and away from work, are shown in Table 1. Eight subjects used ₂-adrenergic agents alone on an as-needed basis, whereas 10 subjects used inhaled steroids regularly and ₂-adrenergic agents when needed. All subjects had been exposed at work for at least 1 yr and had been symptomatic for at least 6 mo. Baseline FEV₁ was 80% predicted or more in 11 subjects; PC₂₀ was 16 mg/ml or less in 17 subjects. Occupational asthma was diagnosed in 11 subjects (55%) whereas nonoccupational asthma was diagnosed in six subjects (30%). The remaining three subjects were found not to have asthma. Incriminated agents at work were high-molecular-weight agents in six cases and low-molecular-weight agents in the other five cases.

A total of 4,590 values of PEF and FEV₁ were registered by the 20 subjects. In the case of PEF, 1,227 readings (26.7%) did not satisfy the criteria set by Burge and co-workers, i.e., the best value was different from the second best by more than 20 L/min. There were 1,501 FEV₁ values (32.7%) for which the difference between the two best values was more than 5%. FEV₁ failed to meet the reproducibility criteria more often than PEF using the VM1 instrument (p < 0.001).

The results of visual analyses of PEF and FEV₁ of the three independent readers are shown in Table 2 (all best values) and in Table 3 (best of two reproducible values).

When all best values were kept for analysis (Table 2), the interpretation of PEF graphs yielded a sensitivity above 70%, together with a 100% specificity for two of the three readers. Agreement between readers, expressed by the mean of the Kappa value, was satisfactory. When the same analyses were used on the interpretation of FEV₁ graphs, both diagnostic performances and Kappa values were less satisfactory.

The results of interpretation of graphs plotted with the best of two reproducible values are shown in Table 3. PEF readings gave a high specificity, but a lower sensitivity; agreement between readers also decreased slightly. Analyses of FEV₁ graphs resulted in a low sensitivity, but a higher specificity, compared with analyses of FEV₁ graphs using all best results.

Diagnostic performances of visual analysis and quantitative indices when the graphs were plotted with all best values, as they compared with the results of the specific inhalation challenges, as shown in Table 4. When agreement of at least two of the three readers was required to define occupational asthma, the sensitivity and specificity of PEF recording were 73 and 100%, respectively, for PEF, and of FEV₁ readings, 55 and 89%, respectively. The diagnostic performances of quantitative indices were less satisfactory. The results of similar analyses of graphs using the best of two reproducible values are shown in Table 5. PEF recordings have a lower sensitivity and a similar specificity. Performances of calculated indices were poor for both PEF and FEV₁.

PEF and FEV₁ recordings of all best results and of the best of two reproducible values in the same subject are shown in Figure 1. This figure illustrates the two points made above: PEF and recording of all best results are more accurate.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Monitoring of the best of two reproducible and of all best results for PEF (upper panels) as well as of the best of two reproducible and of all best results for FEV1 (lower panels) for days or portions of days at work (horizontal lines) and away from work in the same subject. PEF and recording of all best results are more accurate. Vertical arrow = use of inhaled 2-adrenergic agent.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Two conclusions can be drawn from this study of unsupervised PEF and FEV₁ monitoring in the diagnosis of occupational asthma. First, PEF appears to be more accurate than FEV₁ in serial assessment of airway caliber in the diagnosis of occupational asthma. Second, analysis of graphs using the best value yields better diagnostic accuracy than does analysis of graphs using the best of two reproducible values.

The interpretation of PEF graphs by the three readers had higher specificity than sensitivity for diagnosing occupational asthma. Although no criterion was developed to help the three readers while "eye-balling" the graphs, it is likely that they interpreted the graphs positive for occupational asthma when they observed a typical pattern of change, as described by Burge (20). The results of this study are consistent with those of our previous study in which we found PEF monitoring to be more useful in confirming than in excluding occupational asthma (4). On the other hand, "stable" PEF graphs do not necessarily exclude occupational asthma.

One striking finding of this study is in the lower diagnostic accuracy of serial FEV₁ monitoring in the diagnosis of occupational asthma. This differs from the findings of Bérubé and coworkers (6) who compared the accuracy of PEF and FEV₁ in detecting late asthmatic reactions in 88 subjects after specific inhalation challenges in the laboratory: changes in PEF were far less sensitive than changes in FEV₁. Prior to this study, we found that FEV₁ measured with the VM1 apparatus compared favorably to FEV₁ measured with a Vitalograph. One possible explanation for the the lower diagnostic performances of FEV₁ in this study is the difficulty of obtaining FEV₁ properly in an ambulatory setting without the supervision of a respiratory technician; this is reflected in the lower reproducibility of FEV₁ in this setting. It is indeed conceivable that supervision of the expiratory maneuver by a technician is more important for FEV₁ (the procedure has to last for at least 1 s) than for PEF.

To our knowledge, the effect of screening data for reproducible values has not been previously studied. Since the early evaluations by Burge and coworkers of subjects exposed to colophony or isocyanates (7, 8), further investigations on PEF monitoring in occupational asthma have been conducted: Perrin and coworkers (16) selected the best of two reproducible values before plotting the graphs, whereas Côté and coworkers (17) kept the best of all values. In the latter study, interestingly, sensitivity and specificity, assessed through visual analysis of graphs obtained from sawmill workers exposed to red cedar, reached 87 and 90%, respectively. Moreover, in an epidemiologic setting, it has been shown that the application of rigid repeatability criteria may bias findings by excluding subjects with accelerated loss of lung function (21). Becklake (10) therefore suggested that test failure in itself carries a greater chance of unfavorable outcome. By analogy, the poorer diagnostic accuracy of the visual analysis of graphs plotted with best of two reproducible values may be explained by overscreening of low values, resulting in a "smoothed" graph (Figure 1).

The results of this study may have practical implications in the diagnosis of occupational asthma. PEF monitoring may be unnecessarily specific at the expense of unacceptably low sensitivity. The major pitfall is the lack of sufficient sensitivity, so "negative readings" of PEF graphs do not exclude occupational asthma, especially when graphs are plotted with reproducible data. Secondly, ambulatory nonsupervised FEV₁ monitoring is not better than PEF monitoring. It would therefore seem nonproductive to use devices that monitor both FEV₁ and PEF in place of instruments that record and store PEF.