Defining airflow obstruction.

Using a very large number of predominantly Chinese nonsmoking females aged 30–79 years, SMITH et al. [1] studied the relationship between airflow obstruction, household air pollution, household income, educational level and prior tuberculosis. They defined airflow obstruction as a ratio of forced expiratory volume in 1 s (FEV1) to forced vital capacity (FVC) of <0.7 or <5th percentile, and graded the severity of respiratory impairment using FEV1 80% of predicted as a cutoff point, according to Global Lung Function Initiative 2012 prediction equations [2]. There are two fundamental problems with these criteria that affect the interpretation of their findings. First, the fixed ratio of 0.7 does not represent the lower limit of normal for FEV1/FVC, because this ratio declines with age in healthy nonsmokers (table 1). Inevitably, the fixed ratio does not delimit mild airways obstruction and its use introduces an important age and sex bias [2–4]. As the normal FEV1/FVC ratio in most East Asians is somewhat higher and the scatter smaller than in Caucasians [2], use of a fixed ratio of 0.7 as a cutoff will result in more under-diagnosis in younger adults and less over-diagnosis in the elderly than in white people (table 1). Secondly, the use of per cent predicted for FEV1 is flawed: in adults the predicted FEV1 declines with age, but the scatter in absolute terms declines proportionally much less (table 1). Therefore, the lower limit of normal expressed as a percentage of the FEV1 declines with age. Consequently, the proportion of healthy nonsmoking East Asian females with an FEV1 below any fixed percentage increases progressively with age (fig. 1). The use of per cent predicted was introduced 50–60 years ago [5, 6] and adopted despite prompt and fundamental criticism [7, 8]. Despite lack of clinical evidence of its validity, its use has been canonised in countless international and national guidelines and has therefore become engrained in respiratory medicine, explaining its widespread and uncritical use. The use of per cent predicted discriminates against people over the age of 45 years (fig. 1). Misclassification due to the age bias will become a progressively larger problem as an increasing proportion of people remain healthy and fit to a ripe old age. Per cent predicted of a spirometric index does not represent a percentile. However, it is treated as such, leading to biases in research and clinical medicine. The bias arising from the use of 80% predicted …