Chest
Volume 139, Issue 1, January 2011, Pages 52-59
Journal home page for Chest

Original Research
Pulmonary Physiology
Interpreting Lung Function Data Using 80% Predicted and Fixed Thresholds Misclassifies More Than 20% of Patients

https://doi.org/10.1378/chest.10-0189Get rights and content

Background

Differences in COPD classification have been shown in population data sets when using fifth percentiles as the lower limit of normal (LLN) vs the current GOLD (Global Initiative for Chronic Obstructive Lung Disease) guidelines of FEV1/FVC < 0.70 for detecting airway obstruction and an FEV1 of 80% predicted for detecting and classifying the severity of COPD (GOLD/PP). Many lung function laboratories use 80% predicted to determine whether results are abnormal. Misclassification of the full range of lung diseases in large patient groups when using GOLD/PP criteria instead of the LLN has not been explored previously.

Methods

We determined the discrepancy rates in pulmonary function test interpretation between the GOLD/PP and LLN methods on prebronchodilator lung function results from a large number of adult patients from the United Kingdom, New Zealand, and the United States.

Results

In 11,413 patients, the GOLD/PP method misclassified 24%. Ten percent of patients who had normal lung function were falsely classified with a disease category, and 7% of patients were falsely attributed with emphysema. The GOLD/PP method gave false-positive classifications for airflow obstruction and restrictive defects to significantly more men (P < .01) and older patients (P < .0001) and also missed airflow obstruction and restrictive defects in younger patients (P < .0001).

Conclusions

Using lung function tests on their own with 80% predicted and fixed cut points to determine whether a test is abnormal could misdiagnose > 20% of patients referred for pulmonary function tests. The GOLD/PP method introduces clinically important biases in assessing disease status that could affect allocation to treatment groups. This misclassification is avoided by using the LLN based on the fifth-percentile values.

Section snippets

Materials and Methods

PFT results were obtained retrospectively from the patient databases of relatively large PFT laboratories from three countries in order to improve the chance that the results can be used to generalize to other populations. Full PFT results from consecutively tested white adults were obtained from the University Hospital Trust (Birmingham, England, United Kingdom) (n = 5,702; men, 51.2%; mean age, 58.3 years; age range, 20-92 years); Saint Louis University Hospital (St Louis, Missouri) (n =

Results

The mean results expressed as standardized residuals10 for FEV1, FVC, TLC, and Dlco for the patients from the three centers are shown in Table 1. The mean results for each index were not significantly different for each laboratory from the other two except for the FEV1 and TLC results for the United States and New Zealand compared with the UK data (P < .005, two-sample t test), which relates to differences in the prediction equations and methodology used for measuring TLC.

When fixed thresholds

Discussion

This study demonstrates that using a percent predicted and fixed thresholds method (GOLD/PP) instead of fifth-percentile LLNs when interpreting lung function results can mean that up to 24% of patients are misclassified with regard to possible lung disease. These errors may vary with the prediction equations applied, sex, ethnic group, and, especially, age.19, 20 We also show that the GOLD/PP method fails to identify properly about 10% of patients with normal lung function and falsely

Acknowledgments

Author contributions: Dr Miller: contributed data and ideas for the study and to the analysis and writing. He assembled the data set and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Dr Quanjer: contributed ideas for the study and to the analysis and writing.

Dr Swanney: contributed data and ideas for the study and to the writing.

Dr Ruppel: contributed data and ideas for the study and to the writing.

Dr Enright: contributed ideas for the study and to the

References (33)

  • A Miller

    Prediction equations and ‘normal values.’

  • MR Miller et al.

    Predicted values: how should we use them?

    Thorax

    (1988)
  • S Stanojevic et al.

    Reference ranges for spirometry across all ages: a new approach

    Am J Respir Crit Care Med

    (2008)
  • American Thoracic Society

    Lung function testing: selection of reference values and interpretative strategies

    Am Rev Respir Dis

    (1991)
  • PH Quanjer et al.

    Lung volumes and forced ventilatory flows. Report Working Party Standardization of Lung Function Tests, European Community for Steel and Coal. Official Statement of the European Respiratory Society

    Eur Respir J

    (1993)
  • R Pellegrino et al.

    Interpretative strategies for lung function tests

    Eur Respir J

    (2005)
  • Cited by (0)

    Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/site/misc/reprints.xhtml).

    View full text