Quanjer et al. (GLI-2012)
About this reference
Quanjer and colleagues received more than 160,000 records from 72 centres in 33 countries. After cleaning, 97,759 records of healthy nonsmokers remained; 74,187 people in four sufficiently coherent population groups contributed to the fitted equations, including 57,395 in the group the paper labels "Caucasian." The study used LMS regression to model FEV1, FVC, FEV1/FVC, and other indices continuously by age, height, sex, and the paper's ethnicity groups. GLI-2012 was an influential international reference set, but its group-specific approach is no longer current ATS/ERS guidance. The societies now recommend a race-neutral approach, with GLI Global as the available average equation (Bhakta et al. 2023). This site preserves a clearly labelled reproduction of the GLI-2012 coefficient set called "Caucasian," evaluated at fixed NHANES median heights (the survey-weighted median of measured standing height in NHANES 2015-2018, computed from the public examination files; see NHANES 2015-2018 median heights); it is not offered as a current clinical calculator.
Known limitations
- The site tables use only the source-labelled Caucasian coefficient set. Current ATS/ERS guidance recommends race-neutral interpretation rather than choosing a race- or ethnicity-specific GLI-2012 equation
- Each displayed age band is a point calculation at the bracket midpoint and one NHANES median height, not a percentile distribution pooled from everyone in that bracket. A person's expected value changes continuously with exact age and height
- The 80+ row is evaluated at age 85, while GLI-2012 is defined only through age 95. Just 0.8% of the cleaned source records were aged 80 or older, and the supplement explicitly advises caution above age 80
- The reference sample comprised healthy lifelong nonsmokers. The equations describe a selected reference population rather than the distribution in a general population that includes smoking and respiratory disease
- The fitted equations used 74,187 records. The larger 97,759 figure in the abstract is the cleaned pool before groups that could not be combined into the final four-group model were excluded