Normal Values in Pulmonary Function Testing: Why the Reference Set Your Software Uses Changes Everything

Apr 9, 2026

The reference equations embedded in your pulmonary function testing software are not a background detail. They are the foundation of every clinical decision that flows from a test result. Whether a patient is classified as having obstruction, restriction, or normal lung function depends almost entirely on which predicted values the system uses to compare against. Get the reference set wrong, and clinically significant findings can be missed or misclassified, not because the test was performed poorly, but because the benchmark was inappropriate.

TL;DR

Lung function predicted values are derived from reference equations applied to patient demographics, and the choice of equation set directly affects clinical outcomes.
The GLI reference equations are the current global standard and represent a significant improvement over older population-specific datasets.
Interpreting results using fixed thresholds like 80% predicted can misclassify patients, particularly at the extremes of age and height.
Z-scores and the lower limit of normal (LLN) provide a more statistically robust method of identifying abnormality.
The software platform your lab uses determines which reference equations are available, how they are applied, and how easily they can be updated.

About the Author: This article was written by the Rezibase team, respiratory reporting specialists with over 37 years of combined experience building and supporting pulmonary function reporting systems for clinical physiology labs across Australia, New Zealand, the United Kingdom, and Ireland.

What Are Lung Function Predicted Values and Why Do They Exist?

Predicted values in pulmonary function testing are the expected results for a given individual based on their age, sex, height, and ethnicity. No two people have identical lungs, so a raw spirometry number like an FEV1 of 2.8 litres tells a clinician very little in isolation. The clinical meaning comes from comparing that result against a statistically derived prediction for someone with the same biological characteristics.

These predictions are generated by reference equations, which are built from spirometry data collected from healthy non-smoking populations. The key metrics most commonly assessed include:

FEV1 (Forced Expiratory Volume in 1 second): the volume of air forcefully exhaled in the first second
FVC (Forced Vital Capacity): the total volume of air that can be forcefully exhaled
FEV1/FVC ratio: the proportion of the FVC exhaled in the first second, used to detect obstruction [4]

A result is generally considered within the normal range if it falls above 80% of the predicted score, though this threshold has important statistical limitations [1].

What Are the GLI Reference Equations and Why Are They the Current Standard?

The GLI reference equations, developed by the Global Lung Function Initiative, represent the most widely adopted and statistically robust set of reference equations currently available for spirometry. They were designed to address a longstanding problem: older reference datasets were derived from specific, often narrow population groups, making them poorly applicable across diverse patient populations.

Key advantages of the GLI equations include:

Coverage across a continuous age range from childhood through to older adulthood
Inclusion of multiple ethnic groups with ethnicity-specific adjustments
Use of the Lambda-Mu-Sigma (LMS) statistical method, which allows for accurate z-score calculation across the full age spectrum
Endorsement by the American Thoracic Society (ATS) and European Respiratory Society (ERS)

Modern spirometry software, including platforms like Rezibase, incorporates a regularly updated Normal Values Library that includes GLI equations as a standard configuration. This matters because the equations your system uses are not always visible to the end user, yet they shape every interpretation output the system generates [2].

Why Does the 80% Predicted Threshold Fall Short?

The 80% predicted rule is intuitive and easy to apply, but it introduces systematic bias at the extremes of the demographic distribution. Consider two patients, both with FEV1/FVC values that sit just above 70%:

A 30-year-old tall male: his predicted FEV1/FVC is likely well above 70%, meaning his result may still represent meaningful decline.
A 75-year-old female: for her age and sex group, a value just above 70% may still fall within the statistically normal range.

Applying a universal fixed threshold to both patients produces different rates of misclassification depending on age and size [3].

This is why z-scores and the lower limit of normal (LLN) have emerged as the preferred interpretive framework. A z-score expresses how many standard deviations a result sits from the predicted mean for that individual. The LLN is typically set at the 5th percentile of the reference population, corresponding to a z-score of -1.645. Using this approach, approximately 5% of healthy individuals will fall below the LLN, compared to the much higher false positive rates that the 80% threshold generates in older populations [3] [5].

How Does Your Reporting Software Determine Which Reference Equations Are Applied?

In practice, spirometers and pulmonary function test equipment have software that uses reference equations to calculate predicted values [2]. But the choice of which equations to apply, and whether they are updated when new standards emerge, is a function of your reporting platform, not your measurement device.

This creates a critical and often underappreciated gap. A lab may be using modern equipment performing technically excellent tests, while the reporting system quietly applies an outdated reference dataset from a narrowly defined population. The resulting fev1 fvc predicted values may not reflect the patient's demographic accurately at all.

The practical implications for labs include:

Older software with fixed, non-configurable reference sets cannot adapt as standards evolve
Vendor-locked systems may delay incorporating updated equations pending manufacturer updates
Labs operating across multiple sites or demographics need flexibility to select the appropriate equation set per patient

Platforms designed with clinical configurability in mind allow labs to select from a curated library of validated reference equations and apply them consistently across patient records.

What Does the Research Say About AI and Lung Volume Estimation?

An interesting development in the field comes from a 2025 study published in JMIR AI, which explored the use of artificial intelligence-based algorithms to estimate static lung volumes and capacities using spirometry measures [6]. The research highlighted the potential for AI to extend the interpretive value of spirometry data in settings where full body plethysmography is unavailable.

This is worth noting not as a replacement for robust reference equation selection, but as an indicator of the direction the field is heading: greater automation, smarter interpretation, and less reliance on single-threshold rules. The foundational requirement, however, remains the same. Any automated or AI-assisted interpretation layer still depends on the quality and appropriateness of the underlying reference data.

Frequently Asked Questions

What are the GLI reference equations used for?
They are used to calculate predicted values for spirometry results, allowing clinicians to assess whether a patient's lung function is within the normal range for their age, sex, height, and ethnicity.

What is a normal FEV1/FVC ratio?
In healthy adults, the FEV1/FVC ratio typically falls between 70% and 85%. A value below the lower limit of normal suggests obstructive lung disease such as COPD [4].

Why is the 80% predicted threshold problematic?
It applies a fixed cutoff regardless of age, sex, or height, which can over-diagnose obstruction in older patients and under-diagnose it in younger ones. Z-scores and LLN provide more statistically consistent thresholds [3].

Can different software systems produce different interpretations from the same test data?
Yes. If two systems apply different reference equations, the predicted values will differ, and the classification of results (normal, mildly abnormal, severely abnormal) may differ as well [2].

What is the lower limit of normal (LLN)?
The LLN is the 5th percentile of the distribution for a reference population. Results below this threshold are considered outside the normal range, regardless of whether they fall above or below a fixed percentage [5].

Does the spirometry handbook for primary care address reference equations?
Yes. The Spirometry Handbook for primary care, published by the National Asthma Council Australia, provides guidance for health professionals on both performing and interpreting spirometry, including the application of predicted values [7].

How often should reference equations be updated in a reporting system?
Reference equations should be updated whenever new validated datasets are published and endorsed by major bodies such as the ATS or ERS. A reporting platform that does not allow configuration updates creates compliance and clinical risk over time.

About Rezibase

Rezibase is Australia's most advanced cloud-based respiratory and sleep reporting platform, built by respiratory scientists for respiratory scientists. With over 37 years of experience in the field and deployments across more than 35 sites including NSW Health and the NHS in the United Kingdom, Rezibase understands the clinical and operational demands of modern pulmonary function labs. The platform includes a pre-configured, regularly updated Normal Values Library, ATS-aligned reporting tools, and a fully vendor-neutral architecture that lets labs work with any device without lock-in. For labs looking to ensure their reference equations, workflows, and reporting standards stay current, Rezibase provides the infrastructure to make that happen.

Ready to see how Rezibase handles normal values, reference equations, and clinical reporting in practice? Visit rezibase.com to explore the platform or start your 30-day free trial.

References

Inside Rezibase: Built by Respiratory Scientists, for Respiratory Scientists ›