Anomalous Differences Between the Global Lung Function Initiative 2023 and 2012 Spirometry Reference Values.

Abstract

In 2012, the Global Lung Function Initiative (GLI) collected a large international database of spirometry measurements in healthy people from which predicted spirometric values were derived based on sex, age, and height for people from 4 different geographic ancestral groups. In 2023, a single set of predicted spirometry values was developed for the entire data set, designed to be independent of ancestry. What impact do differences in the proportion of participants from the 4 ancestral groups at each age cohort of the GLI database have on differences between predicted values using the 2012 and 2023 equations? Because 77.4% of participants in the GLI spirometry database are of broadly European ancestry, the GLI 2023 spirometry reference values used an inverse probability weighting to approximate an equal contribution of each of the 4 different ancestral groups to develop predicted global spirometric values. This study examined the consequences of using a constant inverse probability weighting factor for each ancestral group over the entire age range from 3.5 to 95 years. The proportional number of participants from each ancestry group in each age cohort varied considerably. The contribution of the European ancestry group, which had the largest predicted lung volumes, was highest in younger children and in older adults. The net effect was that when using GLI 2023 spirometry reference values, predicted lung volumes in older adults were higher than the lung volumes predicted for any of the 4 individual ancestral groups using GLI 2012 spirometry reference values. The opposite effect was found in children and adolescents. It was concluded that the inverse weighting strategy used to develop the GLI 2023 spirometric reference values introduced a systematic bias that skewed the predicted values, producing progressively overestimated predicted lung volumes with increasing age in adults and progressively underestimated predicted lung volumes with increasing age in children and adolescents.