
==== Front
Ann Am Thorac Soc
Ann Am Thorac Soc
AnnalsATS
Annals of the American Thoracic Society
2329-6933
2325-6621
American Thoracic Society

39212423
202405-555ED
10.1513/AnnalsATS.202405-555ED
Editorials
Testing the Limits: Alternative Definitions of Spirometry Thresholds
Brems J. Henry
https://orcid.org/0000-0003-1702-3201
McCormack Meredith C.
Division of Pulmonary and Critical Care Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland
1 9 2024
1 9 2024
1 9 2024
21 9 12471248
Copyright © 2024 by the American Thoracic Society
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is open access and distributed under the terms of the Creative Commons Attribution Non-commercial No Derivatives License 4.0. For commercial usage and reprints, please e-mail Diane Gern.

National Institutes of Health, National Heart, Lung, and Blood InstituteF32HL165771 National Institutes of Health, National Institute of Environmental Health SciencesP2C ES033415
==== Body
pmc

Spirometry is perhaps the most important tool that clinicians have to assess lung health and evaluate respiratory disease. Like many medical tests, however, selecting a cutoff between abnormal and normal is challenging, and it requires an inherent trade-off between the number of false-positive results and false-negative findings detected. Currently, American Thoracic Society and European Respiratory Society guidelines define the lower limit of normal (LLN) as the bottom fifth percentile of healthy individuals, a common but somewhat arbitrary threshold, and recognize that alternate definitions may be needed for certain populations or clinical scenarios (1). To delineate alternative thresholds, there is first a need to understand how various definitions of the LLN would impact identification of disease and prediction of outcomes. Such outcome-based assessments of LLN thresholds have been a key research gap.

In this issue of AnnalsATS, Cestelli and colleagues (pp. 1261–1271) provide an important assessment of how various LLN thresholds relate to respiratory symptoms and mortality among 26,000 young men from Norway (2). They examined thresholds at the 10th, 5th, 2.5th, and 1st percentiles when defined using 1) Global Lung Function Initiative (GLI) 2012 race-specific equations and 2) a “local” equation derived from the healthy subpopulation of their cohort. They found that lower thresholds have a stronger association with symptoms and mortality across a variety of spirometry metrics (i.e., forced expiratory volume in 1 s [FEV1], forced vital capacity [FVC], and/or FEV1/FVC below the LLN), an expected finding reflecting the association of lower lung function with these outcomes.

Notably, the authors found that the 2.5th percentile threshold for LLN was less strongly associated with mortality among asymptomatic patients than the 5th percentile was among symptomatic patients, which suggests a role for using varying thresholds on the basis of patient symptoms or risk factors. Furthermore, they found that when using the traditional fifth percentile as the LLN, individuals classified as “abnormal” by the local equation but as “normal” by GLI 2012 had increased symptom burden. Because the 5th percentile of the local equation is more akin to the 10th percentile of GLI 2012, this finding indicates a potential for higher LLN thresholds to increase sensitivity and identify more patients with disease.

By relating various LLN thresholds to respiratory symptoms and outcomes, this study raises critical questions about how we define the LLN for spirometry interpretation moving forward. Despite the common interpretive approach, spirometry is not dichotomous. The uncertainty inherent to any single threshold or even to any single interpretive strategy has long been recognized (3, 4). Although using an LLN at the fifth percentile rather than at higher thresholds certainly improves our capacity to identify individuals with increased risk of respiratory symptoms or death, the ideal threshold may depend on the patient’s risk factors, the presence of symptoms, or the specific disease process being evaluated (5, 6). As such, future updates to LLN definitions may need to consider stratifying on the basis of pretest probability. Alternatively, the current binary approach could be exchanged for a more probabilistic one, such as defining zones for normal, borderline, and abnormal to reflect the inherent uncertainty in spirometry interpretation (1).

Despite these possibilities, more work remains to understand the relationship of spirometry thresholds with disease status and outcomes. Although a major strength of the present study was the large size of the cohort, the homogeneous population of young White men warrants caution in extrapolating these results to other populations. There is significant variability in lung function and its relationship to symptoms across age, sex, and height (7). Even further, the choice of reference equation impacts the performance of LLN thresholds in predicting disease, as highlighted in the authors’ study. Thus, although the authors investigated GLI 2012 race-specific equations, further research is needed to understand the utility of various LLN thresholds in diverse populations when using GLI Global, the recently recommended race-neutral reference equation (8). This need for research is further underscored by the fact that there is considerable variability in the degree (and the direction, in some cases) of change in z-scores when transitioning from GLI 2012 to GLI Global, even among patients of the same race (9).

More broadly, the authors’ findings demonstrate an inherent tension between local and global reference equations. Although locally derived reference equations may improve the precision of lung function estimates by comparing a patient with a reference set of individuals most similar to themselves, the internationally derived GLI equations promote standardization of spirometry interpretation between pulmonary function testing laboratories, which can facilitate the transferability of care, applicability of guidelines, and clinical trial eligibility (1). Interestingly, this study hints at a possibility that LLN thresholds, rather than reference equations, could be varied to optimize the detection of disease in a given population.

Moving forward, certain trade-offs of various LLN thresholds will need to be considered. For example, lower thresholds among asymptomatic patients may decrease the rate of false-positive results, but it is unclear to what degree this will decrease the sensitivity to detect early disease or how these points should be weighed against each other. As the authors themselves note, further research will need to be done, including cost-effectiveness research, to understand the full implications of modifying the definition of the LLN in spirometry interpretation.

In considering the optimal approach to spirometry interpretation, alternative approaches to LLN-based definitions merit consideration. A fixed ratio (FEV1/FVC, <0.7) to define obstruction remains recommended by Global Initiative for Chronic Obstructive Lung Disease guidelines (10). However, fixed ratios and thresholds may underestimate disease among Black patients and may encode sex and age biases as well (7, 11–13). Alternatively, a metric such as FEV1Q, which is based on a posited lower limit of lung function necessary for survival, avoids the need for reference equations and better predicts survival than our current approaches, highlighting intriguing potential for future use in spirometry interpretation (14). Artificial intelligence may also improve future interpretive strategies, although its currently applicability is limited (15). Because of the numerous contexts in which spirometry is used, different interpretive strategies may ultimately be warranted when screening, assessing severity, or determining disability, for example. Finally, the focus on empirically derived interpretive strategies should not obscure the critical importance of comparing an individual’s lung function with their own historical values when possible.

In summary, this novel study demonstrates potential opportunities to improve the diagnostic utility of spirometry by refining LLN thresholds on the basis of their ability to predict disease-relevant outcomes. Although more work remains to determine optimal thresholds, clinicians must recognize the limitations inherent to our current use of the fifth percentile LLN as a binary cutoff. Whether with our current or with potential future definitions of the LLN, spirometry is best interpreted in the clinical context.

Supported by National Institutes of Health, National Heart, Lung, and Blood Institute, grant F32HL165771 (J.H.B.) and National Institutes of Health, National Institute of Environmental Health Sciences, grant P2C ES033415 (M.C.M.).

Author disclosures are available with the text of this article at www.atsjournals.org.
==== Refs
References

1 Stanojevic S Kaminsky DA Miller MR Thompson B Aliverti A Barjaktarevic I et al. ERS/ATS technical standard on interpretive strategies for routine lung function tests Eur Respir J 2022 60 2101499 34949706
2 Cestelli L Stavem K Johannessen A Gulsvik A Nielsen R Outcome-based definition of the lower limit of normal in spirometry: a study of 26,000 young adult men Ann Am Thorac Soc 2024 21 1261 1271 38656819
3 Culver BH How should the lower limit of the normal range be defined? Respir Care 2012 57 136 145 22222132
4 American Thoracic Society Standardization of spirometry, 1994 update Am J Respir Crit Care Med 1995 152 1107 1136 7663792
5 Tan DJ Lodge CJ Walters EH Bui DS Pham J Lowe AJ et al. Can we use lung function thresholds and respiratory symptoms to identify pre-COPD? A prospective, population-based cohort study Am J Respir Crit Care Med 2024 209 1431 1440 38236192
6 Lambert A Drummond MB Wei C Irvin C Kaminsky D McCormack M et al. Diagnostic accuracy of FEV1/FVC Z-score in patients with asthma J Allergy Clin Immunol 2015 136 649 653.e4 25863976
7 Agustí A Noell G Brugada J Faner R Lung function in early adulthood and health in later life: a transgenerational cohort analysis Lancet Respir Med 2017 5 935 945 29150410
8 Bhakta NR Bime C Kaminsky DA McCormack MC Thakur N Stanojevic S et al. Race and ethnicity in pulmonary function test interpretation: an official American Thoracic Society statement Am J Respir Crit Care Med 2023 207 978 995 36973004
9 Brems JH Balasubramanian A Raju S Putcha N Fawzy A Hansel NN et al. Changes in spirometry interpretative strategies: implications for classifying COPD and predicting exacerbations Chest 2024 10.1016/j.chest.2024.03.034
10 Agustí A Celli BR Criner GJ Halpin D Anzueto A Barnes P et al. Global Initiative for Chronic Obstructive Lung Disease 2023 Report: GOLD Executive Summary Am J Respir Crit Care Med 2023 207 819 837 36856433
11 Miller MR Quanjer PH Swanney MP Ruppel G Enright PL Interpreting lung function data using 80% predicted and fixed thresholds misclassifies more than 20% of patients Chest 2011 139 52 59 20522571
12 Çolak Y Afzal S Nordestgaard BG Vestbo J Lange P Young and middle-aged adults with airflow limitation according to lower limit of normal but not fixed ratio have high morbidity and poor survival: a population-based prospective cohort study Eur Respir J 2018 51 1702681 29449425
13 Regan EA Lowe ME Make BJ Curtis JL Chen QG Cho MH et al. Use of the spirometric “fixed-ratio” underdiagnoses COPD in African-Americans in a longitudinal cohort study J Gen Intern Med 2023 38 2988 2997 37072532
14 Balasubramanian A Wise RA Stanojevic S Miller MR McCormack MC FEV1Q: a race-neutral approach to assessing lung function Eur Respir J 2024 63 2301622 38485146
15 Topalovic M Das N Burgel P-R Daenen M Derom E Haenebalcke C et al. Pulmonary Function Study Investigators Artificial intelligence outperforms pulmonologists in the interpretation of pulmonary function tests Eur Respir J 2019 53 1801660 30765505
