Longitudinal predicting equations for lung function between 8 and 20 years of age

Introduction and aims: Normal values of lung function are usually derived from cross-sectional studies using height, age and gender as predictors. Longitudinal studies with multi-visits provide previous lung function and growth in height. The aim of this study is to determine whether these information could improve the precision of predicting lung function. Methods: A cohort of 8-10 year old children (n=718), living in Belmont, NSW, were studied at two yearly intervals from 19821992. Among them, 632 subjects attended at least two visits. A normal subgroup (n=496) was chosen from these subjects by excluding subjects who had airway hyper-responsiveness or recent whee at two or more visits. Regression analysis was used to derive cross-sectional predicting equations for lung function using height, age and gender as predictors. An auto-regressive model was used to derive longitudinal predicting equations for lung function using height, age, gender and, in addition, previous lung function and growth in height as predictors. Results: Concurrently measured height, age and gender explained 84% of the variation in FEVj and 86% of the variation in FVC in cross-sectional analysis. The additional information of previous lung function and growth in height increased the proportion of variance explained by 6% for FEV] and 7% for FVC. The standard deviation of the residuals, representing the standard deviation of the predicted values, was 0.38L for FEVj and 0.41L for FVC in the cross-sectional analysis. These values decreased to 0.30L and 0.29L, respectively, in the longitudinal analysis. The lower limit (5 percentile) of the normal range for FEV| is 80.2% predicted using cross-sectional predicting equation and 84.3% predicted using longitudinal prediction. For FVC the lower limit of normal is 80.5% predicted based on crosssectional analysis and 85.9% based on longitudinal predicting equation. Conclusions: Information about previous lung function and corresponding growth in height improves the precision of predicting equations for spirometric function. These longitudinal prediction equations could be used to generate nomograms for further use in clinical settings. :.