Impacts of transitioning to Global Lung Function Initiative (GLI) spirometry reference ranges on airflow classification and severity in Papua New Guinea

Abstract

The diagnostic sensitivity of pulmonary function tests (PFT) improves when spirometry reference equations (SRE) match the local population. However, normative PFT data are lacking for Papua New Guinea (PNG) despite the high prevalence of tobacco smoking (52.9%) and likely high, but unknown, COPD prevalence. We aimed to evaluate the fit of Global Lung Function Initiative (GLI) SRE in PNG and to quantify differences in airflow classification and severity between SRE models. PFT data were collected from non-smoking, healthy participants in Port Moresby, PNG, and interpreted using NHANES III and GLI SRE. Model “best fit” was evaluated using mean Z-score=0 &<0.5, SD=1, 90% Z-scores >-1.64 &<+1.64. Airflow classification and severity rating agreement were compared using kappa (κ) statistics. 360 participants (51% males) aged 14-61 years performed PFT. Z-scores for FEV1 and FVC were significantly different between GLI SRE models. GLI(2)-Black most closely matched best fit criteria (FEV1 Z-score males=0.27±1.16, %<LLN=4.3, females=-0.11±1.20, %<LLN=10.2; FVC Z-score males=0.42±1.22, %<LLN=2.2, females=-0.22±1.06, %<LLN=5.7).GLI(2) airflow classification agreement was slight (κ=0.37) compared to NHANES III, fair (κ=0.37) to GLI(1)-Caucasian, and moderate (κ=0.61) to GLI(5)-Other. Using GLI(2) reclassified airflow from restrictive to normal (GLI(1)=88%, GLI(5)=78%) and from mixed to obstructive (GLI(1)=69%, GLI(5)=50%) and decreased severity ratings (GLI(1) κ=0.27; GLI(5) κ=0.53). In PNG, poorly fitting SRE have low diagnostic sensitivity and over classify restrictive and mixed, and under classify normal and obstructive airflow patterns.