In a recent issue of this journal Riley et al. [1] presented evidence that in adults with asthma the forced mid-expiratory flow (FEF25-75%) is associated with clinical outcomes including previous ICU admission, persistent symptoms, nocturnal symptoms, blood eosinophilia and bronchial hyperreactivity independent of the forced expiratory volume at 1 s (FEV1), the forced vital capacity (FVC) or FEV1/FVC, indices that are classically used to identify pathological airflow limitation. This is somewhat surprising in the context of our earlier findings [2] that FEF25-75 does not contribute information after taking FEV1, FVC and FEV1/FVC into account; these findings have since been corroborated by other studies [3-4]. As such, the findings of Riley deserve careful consideration. The authors did not describe how they handled the age and height dependency of FEF225-75%, but from the context it seems that like the FEV1 and FVC it was normalized by converting to percent of predicted. We suspect that the use of percent predicted, although ingrained in respiratory medicine, may provide an explanation of their findings. As early as 1979 Sobol and Sobol wrote about the use of percent predicted for lung function indices [5]: "It implies that all functions in pulmonary physiology have a variance around the predicted, which is a fixed per cent of predicted. Nowhere else in medicine is such a naive view taken of the limit of normal." Riley et al. converted lung function indices to percent predicted based on Hankinson's regression equations, which have been derived from the NHANES III study. To explore potential bias using the percent of predicted for FEF25-75, we applied these equations to 607 healthy male European-Americans, aged 20-80 years, from the NHANES III study and calculated predicted values for FEF25-75% at ages 20-80 years and stature 178 cm.

Figure 1 – The scatter around the predicted FEF25-75% is not proportional to the predicted value (panel A), so that the coefficient of variation increases steeply with age (B). Expressing the measured FEF25-75% as percent predicted leads to a considerable age bias (C); the lower limit of normal (LLN) is also quite age dependent (D). Dotted lines are at ages 26 and 49 years, the range in Riley’s study.

NOTE: PLOS ONE does not have facilities to display Figure 1, which quintessentially illustrates our concern. However, the full text including the illustration can be downloaded from www.spirxpert.com/Letter_....

As shown in figure 1 the scatter does not increase in proportion to the predicted value, leading to a coefficient of variation (100*standard deviation/predicted value) which increases steeply with age. Expressing FEF25-75% as percent predicted introduces a considerable age bias, and the lower limit of normal expressed as a percentage of the predicted value declines steeply with age (figure 1). We wonder whether these biases have influenced the results in the study by Riley et al. It is preferable to transform indices to z-scores, which express how many standard deviations the measured value differs from the predicted one; they are thus standardised for age, height, sex and ethnicity and therefore free of bias, unlike percent predicted. It appears that the FEV1/FVC ratio was not standardised for these confounders, and that would introduce an additional bias.

Philip H. Quanjer, Depts of Pulmonary Diseases and Paediatrics-Pulmonary Diseases, Erasmus Medical Centre, Erasmus University, Rotterdam, The Netherlands.
Daniel J. Weiner, Children’s Hospital of Pittsburgh, Pittsburgh, PA, USA.

1. Riley CMSC, Wenzel SW, Castro M, et al. (2015) Clinical implications of having reduced mid forced expiratory flow rates (FEF25-75), independently of FEV1, in adult patients with asthma. PLOS ONE 10 (12): e0145476.
2. Quanjer PH, Weiner DJ, Pretto JJ, Brazzale DJ, Boros PW. (2014) Measurement of FEF25-75% and FEF75% does not contribute to clinical decision making. Eur Respir J 43: 1051-1058.
3. Boutin B, Koskas M, Guillo H, Maingot L, La Rocca MC, Boule et al. (2015). Forced expiratory flows' contribution to lung function interpretation in schoolchildren. Eur Respir J 45(1): 107-115.
4. Lukic KZ, Coates AL. (2015). Does the FEF25-75 or the FEF75 have any value in assessing lung disease in children with cystic fibrosis or asthma? Pediatr Pulmonol 50(9): 863-868.
5. Sobol BJ, Sobol PG. (1979) Per cent of predicted as the limit of normal in pulmonary function testing: a statistically valid approach. Thorax 34, 1-3.
6. Hankinson JL, Odencrantz JR, Fedan KB. (1999) Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med 159: 179-187.