Pulmonary lung-function testing plays an important role in surveillance programs for occupational respiratory disorders.
Spirometry is usually utilized by applying preset cut-off values to discriminate between healthy and unhealthy subjects.
This article demonstrates the usefulness of decision analysis techniques to arrive at an optimal diagnosis.
The diagnostic performance of FEV1 and FEV1/FVC was evaluated by relative operating characteristics curves (ROCs) applied to data of a cohort gathered in 1965.
Both parameters showed quite similar ROCs, with a maximal sensitivity of 40% at a specificity of 95% relative to the physician's diagnosis of respiratory disorder.
The area under the curves was. 75 for both FEV1 and FEV1/FVC, illustrating that misclassification of 25% of the subjects is likely to occur.
Regarding the consequences of a false-positive and a false-negative decision as of equal importance, the 5% - percentile (FEV1 residual less than - 1.2 L) would be the optimal cut-off An FEV1 residual below the lower 5% percentile was six times more likely to appear in subjects with chronic nonspecific lung disease (CNSLD) than in subjects without.
The post-test probability of CNSLD was three to four times the pre-test probability.
In occupational or public health practice, however, false-positive results need to be avoided, even at the expense of a higher false-negative rate.
In those situations, a more rigid cut-off between normal and abnormal values may be warranted.
Mots-clés Pascal : Médecine travail, Exposition professionnelle, Appareil respiratoire pathologie, Homme, Surveillance biologique, Fonction respiratoire, Spirométrie, Diagnostic, Expiration forcée, Respiration
Mots-clés Pascal anglais : Occupational medicine, Occupational exposure, Respiratory disease, Human, Biological monitoring, Lung function, Spirometry, Diagnosis, Forced expiration, Respiration
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 96-0367673
Code Inist : 002B03L01. Création : 10/04/1997.