guidelines on standardized challenge testing for airway ...issn 0903 - 1936 guidelines on...
TRANSCRIPT
Eur Respir J, 1993, 6, 1415-1416 Printed in UK - all rights reserved
CORRESPONDENCE
Copyright ©ERS Journals Ltd 1993 European Respiratory Journal
ISSN 0903 - 1936
Guidelines on standardized challenge testing for airway hyperresponsiveness
To the Editor:
I am very satisfied, as I am sure many other "lung doctors" all over Europe are, with the guidelines on lung function testing recently published in the European Respiratory Journal [1]. I am particularly interested in the section concerning airway responsiveness and am convinced this will help to standardize methods and facilitate comparison of the results of various studies.
Nevertheless, I am seriously worried about the statement in item 3.2.4. (page 63) of the Guidelines: "The challenge is stopped ..... when there is a 20% reduction in FEV1, or a doubling (100% increase) in sRaw".
The authors refer to the previous SEPCR guidelines [2, 3], in which a 40% increase in specific airways resistance (sRaw) is considered a positive response to bronchial challenge [2]. In the recommendations edited by QuANJER [3] the variance coefficient of specific airways conductance (sGaw) (reciprocal of sRaw), according to literature data, is 14-39%. What is the reason behind accepting a doubling of specific airway resistance
REPLY
From the authors:
We thank Gutkowski for his constructive comments on the recent recommendations by the European Community for Steel and Coal and the European Respiratory Society regarding the measurement of airway responsiveness [1]. However, we certainly do not share his concern that doubling specific airways resistance (100% increase in sRaw) is a dangerous level of airways obstruction during a challenge test, if the recommended precautions [1] have been taken.
The response obtained by forced expiratory volume in one second (FEY1) and by resistance measurements during challenge tests have been compared in a number of studies [2-4]. It appears that a 20% fall in FEY1 is usually reached at a 2-3 fold increase in lung resistance. This has led to the adoption of a provocative concentration of histamine or methacholine that increases sRaw by 100% (PC100sRaw) as an index of airway responsiveness in normal subjects [5], as well as in patients with asthma [6]. In fact, the recommendation of this threshold appears to be mther conservative, since the PC100sRaw is usually lower than the provocative concentration producing a 20% fall in FEY1 (PCufEY1) [2, 4]. The sRaw owes part of its great sensitivity to the fact that it combines the effect of an increase in airflow resistance and any associated increase in functional residual capacity (FRC) during bronchoconstriction.
as the positive result of the test? Moreover, a 20% reduction in forced expiratory volume in one second (~V1 ) is certainly not comparable with 100% increase in sRaw.
It seems even dangerous to recommend achieving a 100% increase of airway resistance during bronchial challenge. Pourquoi done?
References
1. Standardized lung function testing. Official statement of the European Respiratory Society. Eur Respir J 1993; 16 (Suppl.). 2. Eiser NM, Kerrebijn KF, Quanjer PhH. - Guidelines for standardization of bronchial challenges with (nonspecific) bronchoconstricting agents. Bull Eur Physiopathol Respir 1983; 19: 495-514. 3. Quanjer PhH. - Standardized lung function testing. Bull Eur Physiopathol Respir 1983; 19 (Suppl.): 1-95.
P. Gutkowski Lung Function Unit, Child Health Centre, Warsaw, Poland.
Consequently, the use of sRaw rather than Raw increases the sensitivity [7, 8] and, thus, the safety of the test; it also eliminates any error in the measurement of alveolar pressure by the plethysmographic technique [9]. The 100% change in sRaw is also appropriate in terms of signal to noise ratio. It is far beyond the 95% confidence interval of repeated measurements. Reproducibility is almost invariably reported as the coefficient of variation, which may not be appropriate, as in lung function tests the scatter is not usually proportional to the mean [10]. Reported variabilities range between 3-7.5% [7, 9, 11], which is remarkably low. Assuming a normal distribution, a change in sRaw in excess of 15% would indicate a significant change. This leaves a healthy margin between signal and noise, without risk.
Therefore, the PC100sRaw is a suitable index of sensitivity to inhaled bronchoconstrictors in man. The level of bronchoconstriction achieved does not reflect excessive airway narrowing [12] and can, therefore, be regarded as absolutely safe if the measurements have been standardized [10] and the usual safety precautions have been taken [ 1].
References
1. Sterk PJ, Fabbri LM, Quanjer PhH, et al. - Airway responsiveness. Standardized challenge testing with pharmacological, physical and sensitizing stimuli in adults. Eur Respir J 1993; 6(Suppl. 16): 53-83.
1416 CORRESPONDENCE
2. Habib MP, Par(; PD, Engel LA. - Variability of airway responses to inhaled histamine in normal subjects. J Appl Physiol: Respirat Environ Exercise Physiol 1979; 47: 51-58. 3. Gonsinor E, Krilger M, Meier-Sydow J. - Influence of physiological method and criterion of evaluation on results of bronchial antigen provocation tests. Prog Respir Res 1980; 14: 104-108. 4. Wheatley JR, Pare PD, Engel LA. - Reversibility of induced bronchoconstriction by deep inspiration in asthmatic and normal subjects. Eur Respir J 1989; 2: 331-339. 5. Orehek J, Gayrard P, Smith AP, Grimaud C, Charpin J. - Airway response to carbachol in normal and asthmatic subjects. Distinction between bronchial sensitivity and reactivity. Am Rev Respir Dis 1977; 115: 937-943. 6. Kreit JW, Gross KB, Moore TB, Lorenzen TJ, D'Arcy J, Eschenbacher WL. - Ozone-induced changes in pulmonary function and bronchial responsiveness in asthmatics. J Appl Physiol 1989; 66: 217-222. 7. Lloyd TL Jr, Weight GW. - Evaluation of methods used in detection changes of airway resistance in man. Am Rev Respir Dis 1962; 87: 529-538.
8. Higenbottam T, Clack TJH. - A method for standardizing airway resistance for variations in lung volume. Clin Sci 1979; 57: 397-400. 9. Eiser NM, Kerrebijn KF, Quanjer PhH. - Guidelines for standardization of bronchial challenges with (nonspecific) bronchoconstricting agents. Bull Eur Physiopathol Respir 1983; 19: 495-514. 10. Quanjer PhH, Tammeling GJ, Cotes JE, Pedersen OF, Peslin R, Yemault J-C. - Lung volumes and forced ventilatory flows. Eur Respir J 1993; 6 (Suppl. 16): 5-40. 11. Stein M, Tanabe G, Rege V, Khan M. - Evaluation of spirometric methods used to assess abnormalities in airway resistance. Am Rev Respir Dis 1966; 93: 257-263. 12. Sterk PJ. - The determinants of the severity of acute airway narrowing in asthma and COPD. Respir Med 1992; 86: 391-396.
P.J. Sterk Lung Function Laboratory, University Hospital, NL-2300 RC, Leiden, The Netherlands. Ph.H. Quanjer Department of Physiology, Leiden University, NL-2300 RC Leiden, The Netherlands.
Models for analysis of longitudinal data
To the Editor:
In his editorial comment on the paper by SHERRILL et al. [1] WARE (2] discusses the use of efficient statistical models for longitudinal data, and how improved efficiency and validity results from proper model specification.
He fails to point out, however, that these models are a special case of a more general class of "multilevel" models, which explicitly take account of hierarchically structured data. In longitudinal repeated measures designs we have a basic two-level hierarchy, where measurement occasions are nested within subjects. In fact, however, the paper by SHERRILL et al. [I] analyses data which come from a three-level hierarchy, since individuals are further nested within households. Since the SHERRILL et al. [1] model ignores this third level, it is incomplete and is potentially invalid if there is a sizeable betweenhousehold variation in respiratory symptoms.
General models for hierarchically structured data have been studied by a number of authors, not mentioned by WARE [2], and general purpose software packages are widely available. In particular, the model studied by SHERRILL et al. [1] together with the extension to incorporate between-household variation as well as general
REPLY
From the author:
The comments of Goldstein and colleagues provide a useful addition to the discussion of the paper by SHERRILL et al. [ 1]. Those working to develop methods for the analysis of longitudinal data have tended to ignore other types of clustering in the data, in part because within-subject correlation tends to be much larger than correlation from other types of clustering. For example, intrafamily corre-
continuous time autoregressive structures can be fitted using one of these packages, ML3 [3].
This package has been successfully used to describe the development of maximum oxygen consumption in young athletes [4]. The hierarchical model was used to assess the effects of training by separating this effect from those of normal growth and development.
References
1. Sherrill, DL, Lebowitz MD, Knudson RJ, Burrows B. -Longitudinal methods for describing the relationship between pulmonary function, respiratory symptoms and smoking in elderly subjects: the Tucson study. Eur Respir J 1993; 6: 342-348. 2. Ware JH. - Analysis of longitudinal data: choosing and interpreting regression models. Eur Respir J 1993; 6: 325-327. 3. Prosser R, Rasbash J, Goldstein H. - ML3: software for 3-level analysis. Users guide. Institute of Education, University of London, 1991. 4. Baxter-Jones A, Goldstein H, Helms P. - The development of aerobic power in young athletes. J Appl Physiol 1993; (in press).
H. Goldstein, Institute of Education, University of London, London, UK. A. Baxter-Jones, P. Helms, Dept of Child Health, University of Aberdeen, Scotland, UK.
lation of pulmonary function measurements will be much smaller than within-subject correlation. Nevertheless, it would be helpful to have the capacity to explore these issues empirically. In the past, efforts to do so have been handicapped by the limitations of available software. Perhaps the next generation of software, including ML3, will be more flexible.
J.H. Ware Harvard School of Public Health, Boston, USA.