The Role of Pulmonary Function Testing in the Diagnosis and Management of COPD

Abstract

Pulmonary function testing (PFT) has a long and rich history in the definition, diagnosis, and management of COPD. For decades, spirometry has been regarded as the standard for diagnosing COPD; however, numerous studies have shown that COPD symptoms, pathology, and associated poor outcomes can occur, despite normal spirometry. Diffusing capacity and imaging studies have called into question the need for spirometry to put the “O” (obstruction) in COPD. The role of exercise testing and the ability of PFTs to phenotype COPD are reviewed. Although PFTs play an important role in diagnosis, treatment decisions are primarily determined by symptom intensity and exacerbation history. Although a seminal study positioned FEV₁ as the primary predictor of survival, numerous studies have shown that tests other than spirometry are superior predictors of mortality. In years past, using spirometry to screen for COPD was promulgated; however, this only seems appropriate for individuals who are symptomatic and at risk for developing COPD.

Introduction

Pulmonary function testing (PFT) has a long and rich history in the definition, diagnosis, and management of COPD. Indeed, our early insights into the complex pathophysiology of COPD would not have occurred without data derived from PFTs. As our understanding of COPD has evolved, so has our appreciation of the strengths and many weaknesses of PFTs to diagnose and manage COPD. Moreover, disagreements among experts on many issues with regard to PFTs and COPD, including the very definition of air-flow obstruction, have yet to be resolved. This paper reviews the role of PFTs in the diagnosis and management of COPD, with an emphasis on appreciating important limitations of PFTs. We also offer a sensible, balanced approach to the controversial topic of how air-flow obstruction should be defined.

Spirometry as the Standard: Putting the “O” in COPD

Early PFT was restricted to measuring vital capacity (VC) with volume displacement spirometers.¹ However, in 1925, Alfred Fleisch introduced a pneumotach device that was capable of measuring both volume and flows.² By using this technology, Barach³ observed that subjects with asthma and emphysema had reduced expiratory flows during forced exhalation. In 1958, Hyatt⁴ described the utility of the flow-volume loop for the diagnosis of air-flow obstruction. These findings and innovations spawned a new age in the understanding of COPD pathophysiology and produced new models in which physiologic measurements both defined COPD and became the accepted standard for diagnosis and assessing the response to therapy both in clinical practice and clinical trials.

Early versions of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) strategy reports⁵ stated that a COPD diagnosis is confirmed by spirometry with a post-bronchodilator FEV₁/FVC < 0.7. Disease severity classification was defined largely by FEV₁.⁵ With a standard in place, much effort was focused on addressing the underutilization of spirometry to diagnose COPD. Today spirometers are widely available, even as an attachment on a smartphone. However, spirometry remains underutilized, and, even when performed, spirometry is affected by poor test quality and the misinterpretation of results. Han et al⁶ performed a retrospective study of 5,039 subjects with a new COPD diagnosis who were enrolled in 5 insurance plans. The primary outcome was the percentage of subjects who underwent spirometry testing 2 years before or 6 months after their COPD diagnosis was made.⁶ In this cohort, only 32% of subjects had undergone spirometry testing. The underutilization of spirometry to detect COPD, most importantly in its early stages, led to the creation of the National Lung Health Education Program, led by Thomas Petty MD.⁷ A consensus statement recommended that primary care providers perform office spirometry in patients who are symptomatic to detect asthma and COPD.⁸ Obstructive airway disease was defined as an FEV₁/FEV₆ (forced expiratory volume in the first 6 s) and FEV₁ < lower limit of normal (LLN) according to the National Health and Nutrition Examination Survey III reference equations.^8,9 This initiative led to slogans such as “a spirometer in every doctor’s office” and “putting the “O” in COPD.” The paradigm seemed to be clear: an obstructive defect on a spirometry test confirms a COPD diagnosis, whereas a normal spirometry test rules out COPD. However, an intense debate would ensue with regard to the definition of obstruction: the LLN versus FEV₁/FVC < 0.7.^10,11

The Great Debate: Lower Limit of Normal or FEV₁/FVC < 0.7

There are 2 approaches to interpreting FEV₁/FVC, the statistical LLN and a fixed ratio of 0.7, which has long been endorsed by GOLD.⁵ It is interesting to note that, in 2004, the American Thoracic Society (ATS)/European Respiratory Society (ERS) Task Force for the diagnosis and treatment of COPD¹² endorsed an FEV₁/FVC of 0.7 to diagnose COPD, and, in 2005, the ATS/ERS Task Force for PFT interpretation¹³ endorsed the LLN. The statistical lower limit of normal is defined as the fifth percentile of data derived from subjects classified as normal on the basis of being asymptomatic never-smokers. The primary determinates of expected or predicted lung function in health are age, born sex, and height, although height is an imperfect proxy for the size of the thorax. Although differences in lung size among ethnic groups have long been recognized, the etiology of these differences is poorly understood and the use of ethnic-specific reference equations is controversial. If the pulmonary function data collected from presumably healthy subjects are normally distributed, then the average value is at the center of a bell curve. This value is referred to as the predicted value or 100% of predicted. The term “100% of predicted” may be misleading because the moniker “100%” implies perfect health, but “100% of predicted” is actually the 50th percentile, average.

According to the empirical rule, 90% of values recorded from patients who are disease-free should fall within ±1.645 standard deviations or z-scores from the center of the bell curve. A z-score of ±1.645 excludes 5% of values on the low and 5% of values on the high side of the bell curve (2-tailed curve). However, because FEV₁/FVC can only be abnormally low, a 1-tailed approach is used and the LLN is defined as a z-score of –1.645 from the mean value. This means that only 5% of patients with normal lungs can be expected to have values below this threshold. In other words, a value at or just below the LLN would only be expected to be recorded in 1 in 20 patients with normal lungs. Therefore, an FEV₁/FVC value below this threshold is an unusual finding in a patient without obstructive lung disease, so the post-test probability of COPD is increased and, in many cases, a diagnosis is made. Using sophisticated statistical techniques allows modern reference equations to account for age-related declines and variability in FEV₁/FVC.^14,15 This seems to be a logical, perhaps an ideal approach because FEV₁/FVC clearly changes as a function of age, and the LLN in older patients is < 0.7.¹⁵ Results of studies have suggested that using the 80% of predicted and FEV₁/FVC < 0.7 fixed thresholds instead of the statistical LLN results in false-negative results in the young and false-positive results in the aged, and may misclassify up to 20% of patients.^16,17

Using the fixed FEV₁/FVC of 0.7 is not as statistically robust as the fifth percentile LLN; however, the fixed ratio does have some advantages. The fixed ratio is generalizable, regardless of the reference equation used. It can be applied to all demographics, is easy for clinicians to remember, and does not require the spirometer to have sophisticated software to calculate z-scores and the LLN. The detractors of using the 70% fixed ratio contend that these proposed benefits are outweighed by the fact that relying on the fixed ratio will result in the underdiagnosis of obstruction in the young (the statistical LLN is > 0.7) and the overdiagnosis of obstruction in older patients (the statistical LLN is < 0.7) (Fig. 1).^11,16-18 If these were the only factors to consider, then it would be reasonable to conclude that the answer to this debate is simple: FEV₁/FVC should be judged on the basis of the statistical LLN because the risk of misdiagnosis outweighs the ease, simplicity, and convenience of a fixed ratio.

Fig. 1.

FEV₁/FVC as a function of age in never-smoking women. The ∼20° angle line represents the statistical lower limit of normal, and the horizontal line represents the 70% fixed threshold of normality. The light gray triangle depicts potential false-negative results in younger adults, whereas the darker triangle depicts potential false-positive results in older adults when the fixed 70% threshold is applied. From Reference 18, with permission.

However, Mannino¹⁹ argued that asthma and COPD should not be defined solely on the basis of statistics but that other factors, including patient outcomes, should be taken into account. In his argument, Mannino et al²⁰ pointed to data that showed subjects in GOLD stage 0 (respiratory symptoms, no obstruction) experienced higher mortality despite having statistically normal spirometry, although most of the mortality was due to cardiac not respiratory disease. Mannino et al¹⁸ went on to show that elderly subjects with an FEV₁/FVC < 0.7 but > LLN had a higher risk of death (hazard ratio 1.3) and hospitalization for COPD (hazard ratio 2.6) (Fig. 2). More recently, Bhatt et al²¹ studied the relationship between FEV₁/FVC and clinical outcomes in 24,207 subjects from 4 population-based studies that recorded spirometry and COPD-related events. The prevalence of air-flow obstruction was 15% when using the LLN and was 26% when the 0.7 fixed threshold was applied. The C statistic (equivalent to the area under the curve derived from receiver operating characteristic analysis) was highest for FEV₁/FVC of 0.71, and the difference between the fixed ratio and the LLN was significant. However, the methods used in this study were criticized by Miller and Stanojevic²² because international classification of diseases codes were used with the assumption that they represented an independent verification of a COPD diagnosis. Adding to the confusion is the fact that patients can have evidence of COPD on computed tomography (CT) and normal spirometry by either threshold. Lutchmedial et al²³ performed a retrospective analysis of subjects with emphysema on CT. FEV₁/FVC of 0.7 identified 83% of subjects and the LLN identified 75% of subjects with radiographic evidence of emphysema. When averaged, spirometry failed to detect emphysema in 10.4% of cases.

Fig. 2.

Kaplan-Meier plots of first hospitalization related to COPD according to Global Initiative for Obstructive Lung Disease (GOLD) severity classification according to FEV₁. GOLD severity classifications 1–3 are divided based on whether FEV₁/FVC is below the statistical lower limit of normal. Subjects classified as GOLD 0 and GOLD 1–3 with FEV₁ < 0.7 but > lower limit of normal had a higher risk of a first COPD hospitalization than did normal subjects. LLN = lower limit of normal. From Reference 18, with permission.

What is the answer to the fixed threshold versus the LLN debate when patients with FEV₁/FVC < 0.7 but > LLN may not have COPD and patients with a ratio > 0.7 and the LLN can have COPD? Perhaps there is no answer. Maybe the most logical conclusion is to accept the fact that, in many cases, FEV₁/FVC, regardless of how normality is defined, is not reliable enough to confidently to exclude COPD. Perhaps the best approach is to use the more statistically valid FEV₁/FVC LLN to interpret spirometric obstruction, with the understanding that a normal ratio cannot exclude COPD. In other words, FEV₁/FVC interpretation and COPD diagnosis should be viewed as separate processes. Our answer to the question of Mannino,¹⁹ “should we be using statistics to define disease?” is no. However, in our opinion, the answer to the question, “should we be using statistics to interpret spirometry?” is yes.

Is There More to Spirometry Than FVC and FEV₁?

FEV₁/VC

In some patients, the VC from a non-forced expiratory (so-called slow VC) or inspiratory VC is significantly larger than the FVC. This disparity may be related to air-trapping that occurs from airway closure associated with the high intrathoracic pressures applied to the airways during a forced expiratory maneuver. By itself this finding may be significant. Yuan et al²⁴ showed that an VC-FVC gap is inversely correlated with peak oxygen consumption during exercise. In addition, in patients with a significant VC-FVC gap, the FEV₁/FVC will be higher than the FEV₁/VC, which potentially creates a scenario of a normal FEV₁/FVC coupled with an abnormal FEV₁/VC. Torén et al²⁵ compared FEV₁/FVC with FEV₁/VC in 1,050 subjects ages 50–64 years after bronchodilator administration. When FEV₁/VC was used, the prevalence of air-flow obstruction increased from 10% to 16.4% when using the fixed threshold of 0.7 and from 9.5% to 15.6% when using the LLN. The additional subjects identified as having air-flow obstruction according to FEV₁/VC had a lower FEV₁, higher residual volume (RV), and more wheezing. Fortis et al²⁶ studied 854 current and former smokers with a post-bronchodilator FEV₁/FVC > 0.7 and FEV₁ > 80% of predicted. At the time of enrollment, 120 subjects had an FEV₁/VC < 0.7. Subjects with an FEV₁/VC < 0.7 were older (mean age 65 vs 59 years), had more pack-years of smoking, a higher body mass index, and a lower before and after bronchodilator FEV₁. In addition, subjects with an FEV₁/VC < 0.7 had more emphysema on CT and more-severe exacerbations, and were more likely to progress to having an abnormal FEV₁/FVC (Fig. 3).²⁶ Multiple studies have shown that obesity results in a higher FEV₁/FVC, which potentially results in the underdiagnosis of air-flow obstruction.^27-29 O’Donnell et al²⁷ found that the impact of obesity on FEV₁/FVC in subjects with COPD was greatest in GOLD stages III/IV. Substituting FEV₁/VC for FEV₁/FVC should be done with caution because, currently, there are no robust reference equations for FEV₁/VC, so there is a risk of overdiagnosing air-flow obstruction in elderly patients.^13,30 Interestingly, the 2005 ATS/ERS interpretation standard¹³ recommended the use of FEV₁/VC and the 2022 ERS/ATS interpretation standards³¹ do not. The gap between FVC and VC can be reduced by the use of a modified spirometry technique in which the patient is instructed to exhale in a more-relaxed fashion after the FEV₁ has been recorded.³²

Fig. 3.

COPD-free survival in smokers with normal spirometry. Subjects with a post-bronchodilator FEV₁/SVC < 0.7 were more likely to progress to spirometry-defined COPD. SVC = slow vital capacity. From Reference 26, with permission.

FEV₃/FEV₆

COPD pathology likely begins in the small airways, an area named “the silent zone” by Woolcock et al³³ in 1969, because the small airways contribute very little to the overall airway resistance of the tracheobronchial tree.^34,35 It is not surprising, therefore, that pathology that predominately affects the silent zone might not be detected by measuring the volume of air forcefully exhaled over 1 s (ie, FEV₁). Forced expiratory volume in the first 3 s (FEV₃)/FEV₆, or FEV₃/FVC have been proposed as more sensitive measures of small airways disease than FEV₁. Dilektasli et al³⁶ examined 7,853 current and former smokers from the COPDGene phase I study with regard to FEV₃/FEV₆. They found that 15.4% of the subjects with an FEV₁/FVC ≥ the LLN had an FEV₃/FVC₆ < the lower limit of normal.³⁶ Despite a similar incidence of emphysema, the subjects with an FEV₃/FEV₆ < LLN had more air trapping and segmental wall area on CT.³⁶ These subjects also had functional impairments, including a lower FEV₁ and 6-min walk distance (6MWD), more dyspnea, and reduced quality of life.³⁶ More recently, Yee et al³⁷ studied 832 current and former smokers with an FEV₁/FVC > 0.7. Among this cohort, 17.2% had an FEV₃/FEV₆ < the LLN. These subjects had more CT evidence of small airways disease and, unlike the cohort in Dilektasli et al,³⁶ had more emphysema than subjects with a normal FEV₃/FEV₆. These subjects were found to have a lower FEV₁ and quality of life, and were at a higher risk for a severe exacerbation and a shorter time to a first exacerbation. The subjects with an FEV₃/FEV₆ < the lower limit of normal were also more likely to progress to COPD as defined by FEV₁/FVC, even though they had similar longitudinal changes in FEV₁.³⁶ The investigators also found that FEV₃/FEV₆ was as reproducible as FEV₁.³⁶

Expiratory Flow Indices

Expiratory flow measures derived from spirometry have fallen out of favor in recent years. These indices are easily affected by spirometry technique and tend to have a very wide range of normality, particularly in older individuals. Older adults can have a forced expiratory flow between 25% and 75% of the FVC (FEF_25–75%) and forced expiratory flow at 75% of the FVC (FEF_75%) values < 50% of predicted and still be > the LLN.¹⁵ Quanjer et al³⁸ examined the additional information added by FEF_25–75% and FEF_75% when FVC and FEV₁ was > the LLN in a very large cohort of children and adults. They found that the FEF_25–75% identified obstruction in 2.9% of cases and that FEF_75% identified obstruction in 12.3% of cases with otherwise normal spirometry. Based on these findings and others, the ATS Committee on Proficiency Standards for Pulmonary Function Laboratories³⁹ recommended only reporting FVC, FEV₁, and FEV₁/FVC on spirometry reports because other values, such as FEF_25–75%, have not shown clinical value. However, an important consideration is whether these findings are applicable to symptomatic smokers at risk for COPD. Gelb et al⁴⁰ studied the expiratory flow indices in a small cohort of current or former smokers (range 30–75 pack-years) with respiratory symptoms and a COPD Assessment Test⁴¹ score > 10. All the subjects had an FEV₁/FVC > 70% and FEV₁ > the lower limit of normal. Despite normal-appearing spirometry, 62.5% had a forced expiratory flow at 50% of the FVC < the LLN, only 25% had FEF_25–75% < the lower limit of normal, but 100% had an FEF_75% < the LLN according to 3 different reference equations.^9,15,42 In addition, 50% of the subjects had an abnormal diffusion coefficient (K_CO) and 69% had emphysema on CT.

Can Anything Be Gleaned From Unacceptable Spirometry?

Although multiple studies have shown that most subjects, even elderly subjects and those with severe lung disease, are capable of performing high-quality spirometry, some patients will still be unable to perform spirometry correctly.^43-45 For these patients, pulmonary function technologists must be careful not to declare the testing session a failure and report no data because there may be usable information gleaned from lower-quality spirometry tests. A common error is failure to reach end-of-forced-expiration criteria.⁴⁶ Even though the FVC is underestimated, the FEV₁/FVC may still be below the LLN. An acceptable FEV₁ can be compared with an unforced VC, including an inspiratory VC.^25,26 The 2019 ATS/ERS standardization of spirometry technical statement⁴⁶ removed a forced expiratory time of 6 seconds as an acceptable end-of-test criterion. Therefore, spirometry tests with a valid FEV₁/FEV₆ might be deemed unacceptable and potentially go unreported. Bhatt et al⁴⁷ investigated the ability of FEV₁/FEV₆ to diagnose air-flow obstruction in 10,018 subjects from the COPDGene cohort. When using a diagnostic cutoff of FEV₁/FVC < 0.7, receiver operating characteristic analysis indicated that FEV₁/FEV₆ could identify air-flow obstruction with a high level of accuracy (area under the curve 0.99).⁴⁷ The best cutoff value for FEV₁/FEV₆ was 0.73, which produced 92% sensitivity and 97% specificity.⁴⁷ The subjects with an abnormal FEV₁/FEV₆ and FEV₁/FVC had greater segmental bronchial wall thickness, more dyspnea, shorter 6MWD, and worse quality of life compared with the subjects who had a normal FEV₁/FEV₆ and abnormal FEV₁/FVC.⁴⁷ As stated previously, the FEV₁ can be compared with VC if the patient cannot produce a valid FVC or FEV₆. Another approach is to evaluate FEV₁/FEV₃. Allen et al⁴⁸ found that 25% of elderly subjects who could not perform a full FVC were able to achieve FEV₃ and that FEV₁/FEV₃ < 0.8 showed good agreement with FEV₁/FVC < 0.7. Bhattarai et al⁴⁹ found that FEV₁/FEV₃ correlated well with FEV₁/FEV₆ and FEV₁/FVC, r² = 0.91 and 0.84, respectively. They also showed that an FEV₁/FEV₃ cutoff of 80% accurately identified air-flow obstruction with 94% sensitivity and 96% specificity.

As Hyatt et al⁴ showed many years ago, the flow-volume loop can be helpful in detecting air-flow obstruction, and this may apply even to patients who can only forcefully exhale for 3 s or cannot forcefully exhale at all. Li et al⁵⁰ used a sophisticated approach to characterize the morphology of the expiratory limb of the flow-volume curve. They drew a line from the peak expiratory flow to FEV₃, and identified the triangular area under this line as AT₃. The area under the expiratory flow-volume tracing was termed AUC₃ (area under the curve before 3 s). They reported that AUC₃/AT₃ correlated very well with FEV₁/FVC (r² = 0.88) and that there was agreement between AUC₃/AT₃ and both FEV₁/FVC < 0.7 and the LLN criteria for the detection of air-flow obstruction. Other measures of flow-volume loop curvilinearity, such as the classic β angle, angle of collapse, and concavity indices, have shown good correlation with obstructive lung disease.^51-53 In patients unable to perform forced expiratory maneuvers, concavity in the flow-volume loops of spontaneous breaths have shown good correlation with FEV₁.^54,55

Alternatives to FEV₁/FVC to Detect COPD: A Reason for Optimism or Pessimism?

The numerous alternatives to FEV₁/FVC that may allow earlier detection of COPD and diagnostic alternatives for patients unable to perform high-quality spirometry might be expected to engender a feeling of optimism with regard to COPD detection. However, the clinical utility of less traditional means of assessing air-flow obstruction faces many formidable challenges. Some of the alternative methods (eg, geometric analysis of flow-volume curves) are not available in standard spirometry software packages. Although many of the non-traditional values are available in spirometry software (eg, FEV₁/FEV₃, FEF_75%) and have reliable reference equations, very few clinicians who perform or interpret spirometry are aware of their clinical usefulness. The ATS Committee on Proficiency Standards for Pulmonary Function Laboratories³⁹ recommend only reporting FVC, FEV₁, and FEV₁/FVC on spirometry reports, and the 2022 ERS/ATS technical standard for PFT interpretation³¹ discourages the reporting of FEV₁/VC. Perhaps most importantly, even the value of standard PFTs are affected by poor testing and interpretation quality. Poor quality spirometry performed outside of pulmonary function laboratories has made many proponents of office spirometry reconsider their position.⁵⁶ Schermer et al⁵⁷ found that only 39% of spirometry tests performed in an office setting met acceptability and repeatability standards.

More recently, Hegewald et al⁵⁸ tested the accuracy of 17 spirometers used in a primary care setting. Only 1 of 17 of the devices was deemed to be accurate when tested with an ATS waveform generator. When the percent error was applied to patient test results, 28% of the tests were found to be miscategorized. In addition, only 60% of the recorded spirometry tests were of acceptable quality. It is problematic that PFT laboratory accreditation does not exist in the United States; however, Borg et al⁵⁹ found similar rates of spirometry quality (∼60% acceptable and repeatable) in 2 PFT laboratories based at tertiary hospitals in Australia where laboratory accreditation is required. When one of the laboratories initiated a program of technologist proficiency monitoring and feedback, the percentage of high-quality spirometry tests rose to 92%. Technologist proficiency monitoring and feedback programs have repeatedly been shown to be the most powerful driver of high-quality spirometry testing both in clinical practice and epidemiologic studies of COPD.^59-64 The value of spirometry in detecting air-flow obstruction in many settings is also hindered by its reliance on human test interpretation.

Topalovic et al⁶⁵ compared PFT interpretations made by artificial intelligence versus pulmonologists. Artificial intelligence identified PFT patterns and made the correct diagnosis 100% and 82% of the time, respectively, in which pulmonologists identified PFT patterns and made the correct diagnosis only 73% and 45% of the time, respectively.⁶⁵ With such poor performance interpreting traditional PFT data, it is unlikely that human interpreters will have a more nuanced understanding of the potential usefulness of non-traditional spirometry indices, for example, FEF_75%. Although artificial intelligence interpretation software has been available on spirometers for many years, the algorithms may not be uniform and up to date across all manufacturers and may not encompass non-traditional indices. Ideally uniform artificial intelligence interpretation algorithms endorsed by major pulmonary and COPD organizations, which analyze non-traditional indices, would be used by all manufacturers.

Spirometric Restriction: Hiding the “O” in COPD?

Preserved ratio impaired spirometry (PRISm) is a spirometric pattern defined as an FEV₁ < 80% of predicted coupled with an FEV₁/FVC > 0.7 (which typically requires a reduced FVC).⁶⁶ This pattern was first described by Hyatt et al⁶⁷ as the “non-specific pattern,” with a prevalence of 9.5% in a large PFT database. However, the definition by Hyatt et al⁶⁷ required a normal total lung capacity (TLC) and diffusion capacity of the lung for carbon monoxide ( ${\text{D}}_{\text{LCO}}$). In the COPDGene⁶⁶ cohort, the prevalence of PRISm was between 10.4 and 11.3%, higher than the prevalence in the Rotterdam Study cohort⁶⁸ (7.1%). However, this is not unexpected given that the latter cohort included subjects who never smoked.⁶⁸ Both studies showed that a significant number of subjects transition in and out of the PRISm patten.^66-68 By using the Rotterdam Study cohort, Wijnant et al⁶⁸ showed that, among PRISm subjects re-examined after 4 years, 16% had transitioned to normal spirometry and 49% had transitioned to COPD. Subjects with inconsistent spirometry patterns had a higher annual decline in FEV₁ and FVC compared with those who retained a PRISm pattern. Subjects with PRISm had a similar all-cause mortality and a higher risk of cardiovascular mortality when compared with subjects with COPD and with GOLD 2–4 severity. In clinical practice, many of these patients would not be diagnosed with COPD, despite risk factors (eg, smoking), based on a single spirometry test that did not meet the definition of air-flow obstruction. Fortis et al⁶⁹ showed that 7% of subjects retained a COPD diagnosis after a non-obstructive spirometry, but 24% continued to be prescribed bronchodilator and/or inhaled corticosteroids. These patients may have received continuing therapy because they experienced symptomatic improvement despite a normal FEV₁/FVC.

Moving Beyond Putting the “O” in COPD

Because there is not a perfect answer to the LLN versus 0.7 debate and other non-traditional indices are unlikely to gain widespread utilization, perhaps the paradigm should shift to a new strategy for COPD diagnosis. Data from COPDGene led Lowe et al⁷⁰ to propose a new approach to a COPD diagnosis that includes multiple factors, including environmental exposures, symptoms, CT images, and spirometry (Fig. 4). This approach increased the COPD diagnosis rate from 46% when using GOLD criteria to 82% in a cohort of current and former symptomatic smokers.⁷⁰ When using this strategy, subjects were classified as having no, possible, probable, and definite COPD. The hazard ratio for all-cause mortality for the COPD groups were 1.28, 1.89, and 5.2, respectively.

Fig. 4.

Four characteristics used to make a COPD diagnosis from the COPDGene 2019 redefinition of COPD. From Reference 70, with permission.

The Role of Diffusing Capacity for COPD Diagnosis

${\text{D}}_{\text{LCO}}$ is a technically and physiologically complex test that reflects multiple components of lung function, including alveolar-capillary membrane diffusivity, lung volume, and pulmonary vascular volume, and the distribution of ventilation and perfusion.^71,72 The pathophysiology of COPD can affect all of these components of lung function, which makes ${\text{D}}_{\text{LCO}}$ a potentially important diagnostic tool for patients suspected of having COPD.⁷³ Indeed a ${\text{D}}_{\text{LCO}}$ < 80% of predicted is associated with emphysema in smokers with normal spirometry.⁷⁴ This finding is of consequence; as mentioned earlier, Lutchmedial et al²³ found that, on average, spirometry failed to detect emphysema in 10.4% of cases. In addition, an abnormal ${\text{D}}_{\text{LCO}}$ in ex-smokers with normal spirometry and CT of the chest is associated with more dyspnea, a shorter 6MWD, and evidence of early emphysema as detected by hyperpolarized helium 3 magnetic resonance imaging apparent diffusion coefficients (Fig. 5).⁷⁵ ${\text{D}}_{\text{LCO}}$ may also be able to predict the development of air-flow obstruction. Harvey et al⁷⁶ compared the risk of developing GOLD-defined COPD in active smokers. In <4 years of follow-up, 22% of the smokers with normal spirometry but abnormal ${\text{D}}_{\text{LCO}}$ developed GOLD-defined COPD compared with just 3% of the smokers with both normal spirometry and ${\text{D}}_{\text{LCO}}$.⁷⁶ Due in part to its lack of availability outside of formal pulmonary function laboratories, the 2022 GOLD report⁷⁷ did not assign ${\text{D}}_{\text{LCO}}$ an important role in the diagnosis of COPD. In addition to the lack of availability outside of pulmonary function laboratories, another limitation of ${\text{D}}_{\text{LCO}}$ for the diagnosis of COPD is its lack of specificity. For example, a normal spirometry and abnormal ${\text{D}}_{\text{LCO}}$ pattern may be present in numerous disorders, including pulmonary vascular disease, interstitial lung disease, and anemia.

Fig. 5.

Hyperpolarized helium 3 (³He) static ventilation and apparent diffusion coefficient images from former smokers with normal spirometry and normal diffusion capacity, normal spirometry and abnormal diffusion capacity, and subjects with GOLD I COPD. The subjects with abnormal diffusion capacity had worse ³He images than did the subjects with normal diffusion, which suggests early emphysema, despite normal thoracic computed tomography. GOLD = Global Initiative for Obstructive Lung Disease. From Reference 75, with permission.

In health, alveolar volume derived from ${\text{D}}_{\text{LCO}}$ testing is > 80% of the TLC. This finding is representative of a homogenous distribution of ventilation as the non-soluble ${\text{D}}_{\text{LCO}}$ tracer gas (eg, methane, helium) ventilates most alveolar units. Ventilation heterogeneity, an important feature of COPD results in a low alveolar volume/TLC. The poor communicating fraction has been used to describe this relationship and is easily calculated: 1 – (alveolar volume/TLC) ×100. A poor communicating fraction as a pathophysiologic construct is supported by imaging studies that show moderate correlation between a poor communicating fraction and ventilation defects derived from magnetic resonance images of ventilation with hyperpolarized helium.⁷⁸ A higher poor communicating fraction is associated with worsening key physiologic measures, including FEV₁, maximum oxygen uptake ( ${\dot{\text{V}}}_{{\text{O}}_2}$), and inspiratory capacity (Fig. 6).⁷⁹ A poor communicating fraction has been shown to increase in a stepwise fashion according to GOLD COPD stages and was a better predictor of exercise intolerance than FEV₁, ${\text{D}}_{\text{LCO}}$, or RV in subjects with mild-to-severe COPD.⁷⁹

Fig. 6.

The mean percent of predicted FEV₁, maximum O₂ uptake (V̇_O2), and inspiratory capacity based on poor communicating fraction tertiles. Adapted from Reference 79.

The Role of Lung Volume Testing for COPD Diagnosis

Although lung volumes are not required to make a COPD diagnosis, lung volumes can provide additional information to assess the underlying pathophysiology.³¹ Lung volumes are commonly measured by one of several methods, including nitrogen washout, helium dilution, and body plethysmography. The gas dilution methods can be either a single breath or multi-breath technique. In the context of COPD, the multi-breath methodologies are superior to single-breath techniques because the maldistribution of ventilation causes an underestimation of lung volume, more so in the single-breath techniques.⁸⁰ The accepted standard for measuring lung volumes in COPD is whole-body plethysmography, which is relatively impervious to gas maldistribution and trapped gas within the lung. However, O’Donnell et al⁸¹ called this paradigm into question when they showed that TLC via helium dilution agreed better with TLC via CT than plethysmography. The researchers’ theory was that plethysmography overestimated lung volume in subjects with COPD for a number of possible reasons, including discordance between alveolar pressure and its proxy, mouth pressure. However, this idea was refuted by Tantucci et al⁸² who showed that, when postural reductions in lung volume that occur during supine CT imaging were accounted for, plethysmography and CT were in agreement and helium dilution underestimated TLC. Body plethysmography also provides the opportunity to repeat measurements in a relatively short interval, whereas the gas dilution techniques are more time consuming. Measurement of functional residual capacity, when combined with VC and its subdivisions, provides the necessary information to calculate TLC and RV, along with RV/TLC.

The main physiologic information provided by measuring functional residual capacity, TLC, RV, and RV/TLC is the presence or absence of air trapping and/or hyperinflation. Increases in any of these parameters above the expected values suggests pathology that is often associated with COPD. Although there are no widely accepted thresholds for air trapping or hyperinflation, a TLC or RV > 120% of predicted has been used in the past. However, using thresholds based on a fixed percentage of predicted introduces an age and sex bias into the assessment. Lung volume data for adults from the Global Lung Function Initiative (GLI) project⁸³ suggest that an upper limit of normal is a more appropriate measure of air trapping and hyperinflation. For example, a 65-year-old man, 175 cm tall, would have a predicted TLC of 6.93 L, with an upper limit of normal of 8.35 L, which equates to 120% of predicted. But at the same time, his RV predicted would be 2.29 L, with a upper limit of normal of 3.36 L, equivalent to 147% of predicted. Analysis of the GLI lung volume data suggests that RV along with RV/TLC are more variable in healthy subjects than previously thought. The ERS/ATS guidelines³¹ for interpretation now recommend using an upper limit of normal for all measurements of lung volumes.

Two main patterns of increased lung volumes are found in COPD. RV > upper limit of normal while the TLC is within normal limits is a common pattern, which indicates that air trapping is occurring at the expense of the VC. In the second pattern, RV is increased while VC is preserved, which results in an overall increase in the TLC. In both cases, the functional residual capacity is typically increased. Results of studies suggest that the increased functional residual capacity, with a concomitant loss of inspiratory capacity, is related to the severity of dyspnea and exercise intolerance in COPD.^84,85 Inspiratory capacity/TLC or just inspiratory capacity alone has been proposed to monitor changes in air trapping or hyperinflation in patients with COPD.^86,87 In a study of smokers with normal spirometry, an RV/TLC > upper limit of normal was associated with increased respiratory medication use, health-care utilization, all-cause mortality, and a greater likelihood to progress to spirometry-defined COPD.⁸⁸

Can PFTs Identify COPD Phenotypes?

The 2 classic COPD phenotypes, made famous by cartoon illustrations by Frank Netter MD, are the pink puffer and the blue bloater. The characteristics of the pink puffer are cachexia, marked dyspnea, and a predominance of emphysema pathology, whereas the characteristics of the blue bloater are obesity, hypercapnia, hypoxemia, and a predominance of chronic bronchitis pathology.⁸⁹ However, as our understanding of COPD has evolved, it has become clear that there are considerably more than two COPD phenotypes that may be clinically important. The hope is that identifying COPD phenotypes will lead to personalized therapy and improved outcomes.⁹⁰ Some COPD phenotypes may be based on clinical events (eg, frequent exacerbations),⁹¹ physiologic characteristics (eg, rapid decline in lung function),⁹² and characteristics derived from imaging studies.⁹³ The identification of preserved ratio impaired spirometry is a well-known example of PFT-based phenotyping that has an impact on patient outcomes.^66-68 Another example of PFT-aided phenotyping is the presence of abnormal ${\text{D}}_{\text{LCO}}$ to identify patients with emphysema, with or without abnormal spirometry^74-76 However, an isolated reduction in ${\text{D}}_{\text{LCO}}$ in a patient at risk for COPD is not specific for centrilobular emphysema because this pattern can also be seen in patients with combined pulmonary fibrosis and emphysema syndrome,⁹⁴ and the proposed pulmonary vascular COPD phenotype.⁹⁵ A lack of specificity clearly limits the usefulness of PFTs to phenotype COPD. Indeed spirometry cannot identify markedly different forms of emphysema.^96,97

Vaz Fragoso et al⁹⁸ analyzed spirometry data from the COPDGene cohort and compared GLI spirometry categories to respiratory-related phenotypes: dyspnea, quality of life, exercise capacity, bronchodilator responsiveness, emphysema, and gas trapping. Graded associations were found between spirometric impairment and respiratory-related phenotypes. As expected, subjects with severe COPD according to the GLI categories scored worse on all phenotypes. However, there was no statistical difference in dyspnea, quality of life, and 6MWD between subjects categorized as normal and subjects with mild COPD. These data agree with the study published by Regan et al,⁹⁹ which demonstrated that 54% of smokers with normal spirometry experienced ≥1 forms of respiratory impairment. When compared with never-smokers, smokers with normal spirometry had worse quality of life and shorter 6MWD, and more commonly had emphysema and/or airway thickening on CT. These findings call into question the concept of the healthy smoker based solely on spirometry and supports the concept of pre-COPD as a clinical entity.^100,101

A proposed phenotype that has generated a lot of interest and debate is the asthma-COPD overlap syndrome. Much of the controversy surrounding asthma-COPD overlap syndrome as a clinical entity is the lack of a clear definition. The lack of a clear definition limits the utility of asthma-COPD overlap syndrome in clinical practice and in research because many studies use different definitions of asthma-COPD overlap syndrome.¹⁰² Most definitions of asthma-COPD overlap syndrome describe patients who have features of both asthma and COPD, including co-diagnosis, eosinophilia, and responsiveness to bronchodilator.¹⁰² Many definitions include a ≥15% and ≥400 mL increase in FEV₁ as a major criteria and a ≥12% and ≥200 mL increase in FEV₁ as a minor criteria after bronchodilator for asthma-COPD overlap syndrome.^102-107 Some definitions include an elevated fraction of expired nitric oxide in their criteria.^102,104,105 The 2020 GOLD report¹⁰⁸ removed asthma-COPD overlap syndrome as a recognized entity, stating that “asthma and COPD are different disorders, although they may share some common traits and clinical features.”

Assessing Exercise Capacity

Exercise limitation is a major source of morbidity in patients with COPD.^85,109 In addition, maximum ${\dot{\text{V}}}_{{\text{O}}_2}$ in subjects with COPD has been shown to be a stronger predictor of mortality than FEV₁.¹⁰⁹ Exercise testing is an objective way to assess a patient’s functional capacity, coalescing the negative impact of both the pulmonary and systemic consequences of COPD.^85,110-113 There are 2 basic forms of exercise testing used for patients with COPD, field walking tests and cardiopulmonary exercise testing (CPET). Field walking tests are sometimes referred to as submaximal exercise tests; however, in a small cohort of subjects with severe COPD, Troosters et al¹¹⁴ found no difference in maximum ${\dot{\text{V}}}_{{\text{O}}_2}$, peak heart rate, or blood gases measured during a 6-min walk test (6MWT) and CPET.

Because the 6MWT is markedly affected by how the test is administered, it is important that laboratories follow the protocol recommendations of the ERS/ATS field walking test standards.¹¹⁵ The 6MWT should be performed on a 30-m course with each end marked with cones. Using a course shorter than 30 m will reduce the 6MWD because the patient will spend more time turning around the cones at each end of the course.^115,116 Beekman et al¹¹⁶ compared the 6MWD in 45 subjects with COPD on a 10-and a 30-m course. The 6MWD was 49.5 m shorter and the percent of predicted was 8% lower when using a 10-m course. The 49.5-m difference in 6MWD exceeds the minimum clinically importance difference of 30 m.¹¹⁵ Patients should perform the 6MWT unaccompanied by the testing personnel. A randomized crossover study found that subjects with COPD had a shorter 6MWD when the testing personnel walked with the subject.¹¹⁷ It is recommended, particularly for a first visit, that the 6MWT be performed twice because there is a learning effect and the second 6MWD may surpass the minimum clinically importance difference when compared with the first 6MWD.¹¹⁵ The measured 6MWD can be assessed with the use of a reference equation; however, the results may vary greatly in patients with COPD.

Andrianopoulos et al¹¹⁸ compared the 6MWD percent of predicted in 2,757 subjects with COPD between the reference equations published by Troosters et al¹¹⁹ with 21 other reference equations. Only 4 of the 21 reference equations agreed with the reference equations by Troosters et al.¹¹⁹ Sixteen of the equations resulted in a significantly higher and one equation resulted in a significantly lower 6MWD percent of predicted value. This result may be somewhat unexpected because the equations by Troosters et al¹¹⁹ were derived by using a 50-m course.¹¹⁶ The absolute 6MWD without comparison to a reference equation can shed light on the severity of the condition and prognosis of the patient with COPD. A study of data from subjects with COPD in the ECLIPSE study¹²⁰ found that a 6MWD ≤ 357 m was associated with an increased frequency of hospitalization and that 6MWD ≤ 334 m increased the risk of death. In a 2-year follow-up study of subjects with severe COPD, Pinto-Plata et al¹²¹ found that 6MWD was an independent predictor of survival and that the change in 6MWD did not correlate well with changes in FEV₁ (Fig. 7). Specifically, non-survivors had larger declines in 6MWD (–40 m/year) than did survivors (–22 m/year), despite similar yearly declines in FEV₁.¹²¹ The 6MWD is also 1 of the 4 variables in the BODE (body mass index, air-flow obstruction, dyspnea, exercise capacity) index, which has been shown to have a higher C statistic than FEV₁ (0.74 vs 0.65, respectively) for predicting the risk of respiratory causes of death in subjects with COPD.¹²² The ability of the 6MWD to accurately assess exercise capacity in COPD is validated by the modified BODE index, which replaced 6MWD with maximum ${\dot{\text{V}}}_{{\text{O}}_2}$. Cardoso et al¹²³ found excellent correlation between the original and modified BODE index (r = 0.92), whereas Cote et al¹²⁴ found no difference in the ability of either BODE index to predict mortality.

Fig. 7.

The inverse relationship between mortality and 6-min walk distance (6MWD) in subjects with severe COPD. From Reference 121, with permission.

Continuous pulse oximetry should be recorded during a 6MWT.¹¹⁵ Exercise-induced desaturation during a 6MWT is not uncommon in patients with COPD, but there is significant variability in the prevalence of exercise-induced desaturation in different cohorts.^110,125-129 Exercise-induced desaturation during a 6MWT is associated with worse air-flow obstruction,^110,126-128 higher GOLD stage and BODE score¹¹⁰ and ${\text{D}}_{\text{LCO}}$ < 50% of predicted.¹²⁸ Similar to exercise-induced desaturation prevalence, there is significant variability in the factors associated with exercise-induced desaturation during 6MWT.^125-130 Exercise-induced desaturation during 6MWT in subjects with COPD has been shown to be associated with poorer outcomes and survival.^125,127 Multiplying the 6MWD by the nadir ${\text{S}}_{{\text{pO}}_{\text{2}}}$ during the 6MWT produces the distance-desaturation product. Data from subjects with COPD in the ECLIPSE trial¹²⁰ indicated that the distance-desaturation product was a better prognostic value for mortality than the 6MWD alone.

There are alternatives to the 6MWT that may be attractive to clinics that do not have access to a low-traffic, 30-m course to perform the 6MWT. In 1992, Singh et al¹³⁰ introduced the incremental shuttle walk test for patients with COPD. The incremental shuttle walk test only requires a 10-m course with the center of 2 cones spaced 9 m apart. Walking from one cone to the other is considered a single shuttle. The incremental shuttle walk test resembles a progressive multi-stage CPET protocol because the pace of walking is progressively increased with an audio recording (beeps) to a maximum of twelve 1-min stages.^115,130,131 The patient continues walking at the pace dictated by the audio recording until he or she has to stop due to dyspnea, fatigue, or orthopedic-related discomfort, if he or she fails to complete 2 consecutive shuttle distances within the allotted time, or if the ${\text{S}}_{{\text{pO}}_{\text{2}}}$ falls < 80%.¹¹⁵ In the original description by Singh et al,¹³⁰ the test would be terminated if the heart rate exceeds a specific threshold of 0.85 (210 – [0.65 × age]). However, this provision was not included in the 2014 ERS/ATS technical standard for field walking tests.¹¹⁵ The measured outcome of the incremental shuttle walk test is the number of meters walked for completed shuttles.¹¹⁵ Singh et al¹³⁰ and other studies^132-134 found a good correlation between the incremental shuttle walk test distance and the 6MWD, and that both were related to the maximum ${\dot{\text{V}}}_{{\text{O}}_2}$ measured during a cycle ergometer test.^134,135

Similar to the 6MWT, there is evidence of a learning effect, and it is recommended that 2 incremental shuttle walk tests be performed.^116,131-133 In terms of predicting outcomes, Emtner et al¹³⁵ found that the subjects who required rehospitalization for exacerbation of COPD had a shorter walked distanced during the incremental shuttle walk test versus those who did not require rehospitalization (174 m vs 358 m). Ringbaek et al¹³⁶ found that a walked distance < 170 m during an incremental shuttle walk test was associated with a marked increase in mortality. Moreover, replacing the 6MWD with the distance walked during the incremental shuttle walk test in the BODE formula was shown to be a good predictor of the need for hospitalization and mortality in the subjects with COPD.¹³⁷ Based on outcome data from the subjects who participated in a pulmonary rehabilitation program, the suggested minimum clinically importance difference (improvement) in the incremental shuttle walk test is 48 m (∼5 shuttles).¹³⁸

The endurance shuttle walk test uses the same 10-m course as the incremental shuttle walk test; however, during the endurance shuttle walk test, the patient walks at a constant pace set at a percentage of the patient’s incremental shuttle walk test distance (eg, 75–95%) for up to 20 min.¹³⁹ Unlike the 6MWT and incremental shuttle walk test, the endurance shuttle walk test is measured in seconds walked; however, meters walked can also be recorded. The endurance shuttle walk test has been shown to be a reliable indicator of functional gains after pulmonary rehabilitation.^139,140 A proposed minimum clinically importance difference (improvement) after pulmonary rehabilitation is 192 (186–199) s or 159 (154–164) m.¹⁴⁰

An alternative to field walking tests in the sit-to-stand test. There are numerous iterations of the sit-to-stand test, ranging from 30-s to 3-min time-based protocols as well as repetition-based protocols.¹⁴¹ Ozalevli et al¹⁴² compared a 1-min sit-to-stand test with a 6MWT in subjects with severe COPD (mean FEV₁ 47% of predicted). They found excellent correlation between 6MWD and sit-to-stand test repetitions as well as age, quality of life, dyspnea, and quadricep strength.¹⁴² Quadricep weakness is common in patients with COPD, even in patients with mild disease, and is associated with poor outcomes.^143,144 The cardiovascular responses to the sit-to-stand test are significantly lower than the 6MWT, which might make the sit-to-stand test more attractive in patients with COPD and with a significant cardiac comorbidity.^142,145 Puhan et al¹⁴⁶ showed that subjects with COPD who died within 2 years had a significantly lower 1-min sit-to-stand test repetition count when compared with survivors (11.8 vs 19.5 repetitions, respectively).

As stated above, field walking tests can produce maximum ${\dot{\text{V}}}_{{\text{O}}_2}$ similar to CPET in patients with COPD.^114,133,134 However, CPET can provide more-detailed information about the source and mechanism of exercise limitation in COPD.¹⁴⁷ The typical exercise performance pattern for patients with COPD is a ventilatory limitation to exercise. In health, exercise is limited by exhaustion of the cardiovascular capacity, and, at peak exercise, some ventilatory reserve remains.¹⁴⁸ A common method to detect a ventilatory limitation to exercise is to divide the peak exercise minute ventilation ( ${\dot{V}}_E$) by either the maximum voluntary ventilation (MVV) or the product of the FEV₁ multiplied by a factor of 40 (MVV proxy):

$$\frac{{\dot{V}}_{E}peak}{MVV}$$

Where:

${\dot{V}}_E$ peak = peak minute ventilation achieved during exercise.

A ${\dot{V}}_E$ peak/MVV > 0.85 indicates a ventilatory limitation to exercise.^148,149 This relationship can also be expressed as the ventilatory (or breathing) reserve:

$$\text{Ventilatory reserve}=\frac{MVV-{\dot{V}}_{E}peak}{MVV}\times 100$$

A ventilatory reserve ≤ 15% indicates a ventilatory limitation to exercise.^148,149 However, Neder et al¹⁵⁰ showed a limitation of relying solely on ventilatory reserve to determine a ventilatory limitation to exercise in a study of 288 subjects with mild-to-severe COPD. Ventilatory reserve < 20% was compared with critical inspiratory constraint (end-inspiratory lung volume/TLC > 90%) and ventilatory inefficiency (nadir ${\dot{V}}_E$/ $\dot{\text{V}}\text{C}\text{O}2$ > 34) during CPET. Among subjects without a compromised ventilatory reserve, 50% showed evidence of critical inspiratory constraint. Ventilatory inefficiency was associated with disproportionately high dyspnea scores regardless of critical inspiratory constraint or ventilatory reserve.¹⁵⁰ Attaining critical inspiratory constraint was more strongly associated with dyspnea and low maximum ${\dot{\text{V}}}_{{\text{O}}_2}$ than ventilatory reserve. Subjects with a low ventilatory reserve who did not reach critical inspiratory constraint had similar levels of dyspnea and peak maximum ${\dot{\text{V}}}_{{\text{O}}_2}$ as did subjects to those with a preserved ventilatory reserve. Moreover, critical inspiratory constraint and ventilatory inefficiency were more strongly associated with poor exercise performance.¹⁵⁰ Ventilatory inefficiency is associated with dyspnea and poor exercise tolerance in smokers and patients with early COPD who have yet to develop significant air-flow obstruction.^151,152

Exercise limitation in COPD is multifactorial, including impaired gas exchange, respiratory and peripheral muscle dysfunction, increased airway resistance, dynamic hyperinflation, and increased respiratory drive.^147,153,154 In patients with both COPD and congestive heart failure, those with increased ventilatory drive leading to hypocapnia have worse exercise tolerance and dyspnea than individuals without hypocapnia, despite similar pulmonary function and ejection fraction.¹⁵⁵ Dynamic hyperinflation plays a central role in exercise limitation in patients with COPD by reducing inspiratory reserve volume, reducing chest wall compliance, and moving the respiratory muscles into a weak length-tension position.^154,156,157 Measurement of inspiratory capacity before and during a CPET can estimate the degree of dynamic hyperinflation as TLC minus inspiratory capacity equals end-expiratory lung volume. As ventilation increases to keep up with metabolic demands during an incremental CPET, the end-expiratory lung volume increases to a point to which the tidal volume becomes fixed and inadequate, which results in the cessation of exercise (Fig. 8).^147,154,157 In patients with very-severe COPD, exercise may terminate due to dynamic hyperinflation before reaching the lactate threshold. However, many patients with COPD will reach the lactate threshold but at a low ventilatory reserve, which may help to distinguish between a pulmonary and cardiac limitation to exercise. Medoff et al¹⁵⁸ showed that subjects with a pulmonary limitation to exercise had a ventilatory reserve of 27% at the lactate threshold compared with 75% and 73% in normal subjects and in those with a cardiac limitation to exercise, respectively.

Fig. 8.

Progressive reduction of inspiratory reserve volume (IRV) and increase in end-expiratory lung volume (EELV) due to dynamic hyperinflation (DH) during exercise in a subject with COPD. TLC = total lung capacity; EILV = end-inspiratory lung volume. From Reference 154, with permission.

Maximal Voluntary Ventilation and Respiratory Muscle Strength

In some patients with COPD, tests other than standard spirometry, ${\text{D}}_{\text{LCO}}$, and lung volumes may be helpful. MVV was described earlier because it is used to estimate the ventilatory ceiling during maximum exercise testing and thus a way to judge ventilatory limitation to exercise.^148,149 MVV correlates well with FEV₁ in normal subjects and in those with COPD.¹⁵⁹ Pitta et al¹⁶⁰ found a better correlation between MVV and activities of daily living, including total energy expenditure, steps taken, and moderate-to-vigorous activities than inspiratory capacity and FEV₁. Recently, Andrello et al¹⁶¹ found MVV to be a better predictor of numerous COPD outcomes, including dyspnea and exercise capacity than FEV₁. The downside of routinely performing MVV in patients with COPD is that it adds to testing time and is often poorly tolerated due to dyspnea, cough, fatigue, and dizziness.

Respiratory muscle strength is routinely measured in patients with known or suspected neuromuscular disease; however, respiratory muscle weakness is also a consequence of COPD.^85,112 As with neuromuscular disease, respiratory muscle strength in patients with COPD can be assessed by measuring maximum inspiratory pressure ( $\text{PImax}$) and maximum expiratory pressure (PEmax). Terzano et al¹⁶² studied $\text{PImax}$ and PEmax in subjects with COPD in comparison with a healthy control group. Subjects with COPD were divided into 3 disease severity groupings based on FEV₁ percent predicted: mild, >80%; moderate, 50%-80%; severe, <50%. PEmax was only significantly lower than that measured in healthy controls in the severe COPD group. However, $\text{PImax}$ was significantly lower in comparison with healthy controls in all COPD severity groupings. Although hyperinflation is known to place the diaphragm in a shortened and weak position, poor correlation was found between respiratory muscle strength and RV or RV/TLC. Inspiratory muscle training has been shown to increase $\text{PImax}$; however, in most studies, the improvement in $\text{PImax}$ failed to exceed the minimum clinically importance difference.¹⁶³ Formiga et al¹⁶⁴ hypothesized that sustained $\text{PImax}$ might better reflect diaphragm function during normal activities than did $\text{PImax}$ in subjects with COPD. In contrast to $\text{PImax}$, sustained $\text{PImax}$ measures sustained inspiratory pressure over several seconds from RV to TLC. Sustained $\text{PImax}$ correlated better with FEV₁, inspiratory capacity, inspiratory capacity/TLC. A reduced sustained PImax was associated with poor outcomes and quality of life.

Should Lung Function Guide COPD Therapy?

Although air-flow obstruction is defined by a low FEV₁/FVC, lung function is not used to guide pharmacologic therapy of COPD.^77,165-169 According to the latest GOLD guidelines,⁷⁷ therapy should be guided by symptoms and exacerbations. Spirometry is used to diagnose air-flow obstruction, classify the functional severity of disease, and monitor progression of disease. However, there is poor correlation between FEV₁ and symptoms, quality of life, or exacerbations, and some patients may have normal spirometry despite symptoms or radiographic signs of disease.^23,74,170 The latest GOLD guidelines⁷⁷ recommend a short-acting bronchodilator as needed for group A patients, characterized by 0–1 moderate exacerbations and a low level of dyspnea or symptoms. A long-acting bronchodilator should be used if patients are in group B, characterized by more dyspnea or symptoms but still few exacerbations. Patients in group C have a low level of symptoms and dyspnea but have > 1 exacerbation per year, which leads to hospitalization and should be treated with dual bronchodilator therapy that consists of long-acting β agonists and long-acting muscarinic antagonists. Patients with frequent exacerbations and more-severe symptoms or dyspnea should be treated with long-acting β agonists and/or long-acting muscarinic antagonists and/or inhaled corticosteroids, and consideration of additional therapy. Therapy should be modified based on continued assessment of dyspnea and exacerbations. If dyspnea is the primary problem, then long-acting β agonists and/or long-acting muscarinic antagonists should initially be used, with the addition of inhaled corticosteroids for persistent symptoms. If exacerbations are the main issue, then, in addition to escalating up to long-acting β agonists and/or long-acting muscarinic antagonists and/or inhaled corticosteroids, further therapy with roflumilast (frequent bronchitis) or azithromycin should be considered. Pulmonary function does not impact any of these steps, other than to determine eligibility for roflumilast (FEV₁ < 50% of predicted) and rapidity of lung function loss over time (minimum clinically importance difference for FEV₁ in COPD is estimated to be 100 mL).¹⁶⁹

When FEV₁ < 50% predicted, additional lung function testing with lung volumes, ${\text{D}}_{\text{LCO}}$, and 6MWT, in addition to CT imaging, may be useful to determine eligibility for lung volume reduction or lung transplantation,¹⁶⁶ as discussed next. Although lung function testing does not guide pharmacologic therapy of COPD, some lung function parameters are important in guiding eligibility for lung-volume-reduction surgery and lung transplantation for COPD. In the late 20th century and up to the present, there has been great interest in lung-volume-reduction surgery for end-stage COPD. The National Emphysema Treatment Trial evaluated different surgical approaches to treating hyperinflation in COPD.¹⁷⁰ More recently, the use of endobronchial valves, coils, or related devices has become commonplace in efforts to reduce the effects of air trapping and hyperinflation in patients with COPD. Evaluation of the effectiveness surgical or endobronchial approaches typically include FEV₁ and 6MWD but frequently use lung volume measurements, notably the RV, as well.¹⁷¹ A recent meta-analysis by van Geffen et al¹⁷² of surgical and bronchoscopic lung volume reduction found an overall mean reduction in RV of 0.58 L, 95% CI –0.80 to –0.37 L. Guidelines for eligibility for lung-volume-reduction surgery include FEV₁ < 45% predicted, TLC > 100% post-bronchodilator, and RV ≥ 150% post-bronchodilator.¹⁷³ In addition, arterial blood gases must demonstrate ${\text{P}}_{{\text{aCO}}_{\text{2}}}$ ≤ 60 mm Hg and ${\text{P}}_{{\text{aO}}_{\text{2}}}$ > 45 mm Hg., both at sea level.¹⁷³ Finally, the post-rehabilitation 6MWD must be ≥ 140 m. Lung-volume-reduction surgery is contraindicated if FEV₁ < 20% predicted and ${\text{D}}_{\text{LCO}}$ < 20% predicted or if there is homogeneous emphysema seen on a CT. Lung-volume reduction may also be accomplished by placement of endobronchial valves, for which the pulmonary function requirements are similar.¹⁷⁴ For lung transplantation, criteria based on lung function differ depending on the condition. For COPD, FEV₁ < 25% is a typical requirement; for interstitial lung disease (ILD), FVC < 80% predicted and ${\text{D}}_{\text{LCO}}$ < 40% predicted are considered appropriate.¹⁷⁵

Lung function may also guide the use of noninvasive ventilation at home. A recent study demonstrated the benefit of noninvasive ventilation plus oxygen versus oxygen alone in subjects with persistent hypercapnia ( ${\text{P}}_{{\text{aCO}}_{\text{2}}}$ > 53 mm Hg, 2–4 weeks after hospital discharge for an exacerbation of COPD.)¹⁷⁶ Lung function also guides eligibility for Medicare coverage of pulmonary rehabilitation because such coverage is provided for moderate-to-severe COPD defined as GOLD grades 2–4, which are defined by spirometry (FEV₁ ≤ 79% predicted when post-bronchodilator FEV₁/FVC < 0.7)⁷⁷ Although bronchodilator responsiveness testing is commonly performed during PFT, the response to bronchodilator in the pulmonary function laboratory should not guide therapy.¹⁷⁷ All patients with symptomatic COPD should be offered bronchodilator therapy, with short-acting β₂ agonists used for mild disease and only intermittent mild symptoms, and long-acting bronchodilators (long-acting β agonists and/or long-acting muscarinic antagonists) used for more-severe disease or persistent symptoms. The response to bronchodilator in the PFT laboratory does not guarantee a clinical response, and the lack of a response in the laboratory likewise does not mean a patient will not respond clinically to bronchodilator therapy. In fact, some patients respond with a change in lung volumes but not FEV₁.^178,179 In addition, the presence of a bronchodilator response does not indicate asthma as opposed to COPD because up to 32% of patients with COPD may have a bronchodilator response¹⁸⁰ and many patients with asthma will not have a bronchodilator response.¹⁶⁹ The presence of a bronchodilator response in COPD has been associated with improved exercise tolerance¹⁸⁰ and more phenotypic features of asthma,¹⁸¹ with a recent study that suggests less dyspnea and improved quality of life (based on FEV₁ response)¹⁸² but, again, that does not impact therapy.

Can PFTs Predict COPD Survival?

The ability of measures of pulmonary function to predict an early death was recognized by Hutchinson¹⁸³ himself not long after he invented the spirometer. He observed that reduced VC (measured in cubic inches) preceded an early death, in many cases due to tuberculosis.^1,183 Approximately 130 years later, the seminal paper by Fletcher and Peto¹⁸⁴ established that subjects with COPD experienced a rapid decline in FEV₁, which was associated with disability and premature death. The iconic Fletcher and Peto¹⁸⁴ graph of FEV₁ plotted against age was enormously important in advancing the concept that measuring and monitoring the trajectory of lung function in patients with COPD could help predict outcomes and, more importantly, serve as a therapeutic target to alter the course of the patient’s disease. However, our contemporary knowledge with regard to the trajectory of FEV₁ in COPD has rendered the iconic figure in Fletcher and Peto¹⁸⁴ historically interesting but clinically antiquated. In fact, we now know that there are different FEV₁ trajectories among patients with COPD. Lange et al¹⁸⁵ described 2 distinct FEV₁ trajectories among subjects who developed air-flow obstruction. Half of the subjects who developed COPD had a normal FEV₁ in young adulthood, followed by a rapid decline in FEV₁, similar to the “smoked regularly and susceptible to its effects” curve in the Fletcher and Peto¹⁸⁴ model, whereas the other half had a low FEV₁ in young adulthood but did not have a rapid decline in FEV₁.¹⁸⁵ Although a low FEV₁ is associated with COPD mortality,¹⁸⁶ the relationship between FEV₁ and survival is not as strong as one might think.^187,188 An important concept to appreciate is that COPD is associated with many serious comorbidities. Many patients with COPD die with their disease, not from it,¹⁸⁹ which may limit the ability of spirometry alone to predict survival. Alternative ways to express FEV₁ have been proposed but have yet to gain widespread utilization. The FEV₁ quotient expresses FEV₁ as a function of the sex-specific first percentile FEV₁ of subjects with pulmonary disease instead of expressing FEV₁ as a percentage of a population mean (ie, percent of predicted).^31,190 For male patients, the first percentile is ∼0.5 L; for female patients, 0.4 L. An example calculation of the FEV₁ quotient in a male subject with COPD and an FEV₁ of 0.8 L is

$$\frac{0.8}{0.5}=1.6$$

As the FEV₁ declines toward the first percentile, the FEV₁ quotient becomes smaller and the risk of death increases. FEV₁ quotient was found to be vastly superior to FEV₁ percent of predicted to predict mortality (Cox regression analysis hazard ratio 18.8 vs 6.1, respectively).¹⁹⁰

${\text{D}}_{\text{LCO}}$ has also emerged as a strong predictor of all-cause mortality in subjects with COPD. Balasubramanian et al¹⁹¹ followed up the outcomes of 2,329 subjects from the COPDGene cohort over a median of 4.9 years. Every 10% decline in the ${\text{D}}_{\text{LCO}}$ percent of predicted was associated with a 29% increase in mortality. There was no difference in the ability of ${\text{D}}_{\text{LCO}}$ and the multi-variable BODE index¹²² to predict mortality.¹⁹¹ Importantly, in subjects with COPD, female sex is associated with a more-rapid annual decline in the ${\text{D}}_{\text{LCO}}$ percent of predicted despite having a higher FEV₁ percent of predicted than male subjects.¹⁹² As discussed above, ${\text{D}}_{\text{LCO}}$ may help to overcome the shortcomings of spirometry to diagnose COPD. For example, ${\text{D}}_{\text{LCO}}$ < 80% of predicted is associated with emphysema in smokers with normal spirometry.⁷⁴ ${\text{D}}_{\text{LCO}}$ also seems to be superior to spirometry to predict clinical outcomes and survival in subjects with COPD. In a study of subjects at GOLD I (FEV₁/FVC < 0.7, FEV₁ ≥ 80% predicted), ${\text{D}}_{\text{LCO}}$ < 60% of predicted was associated with more dyspnea, air trapping, shorter 6MWD and higher mortality (23% vs 9%, mortality) when compared with subjects at GOLD I with ${\text{D}}_{\text{LCO}}$ ≥ 60% of predicted, respectively (Fig. 9).¹⁹³

Fig. 9.

Cumulative survival over months of follow-up of subjects with GOLD I COPD and different levels of diffusion capacity ( ${\mathrm{D}}_{\text{LCO}}$) impairment. pred = predicted. From Reference 193, with permission.

In recent years, the clinical importance of performing lung volume testing has been questioned,¹⁹⁴ and, in most cases, lung volumes are not required to make a diagnose of COPD⁷⁷ However, lung volumes may help to predict survival in patients with COPD. Casanova et al⁸⁶ examined lung volume data from 689 subjects with varying degrees of COPD severity based on spirometry. Reduced inspiratory capacity/TLC, a measure of lung hyperinflation, was found to be an independent predictor of mortality. An inspiratory capacity/TLC cutoff value of 25% had a C statistic of 0.74 according to receiver operating characteristic analysis.⁸⁶ Zeng et al⁸⁸ studied lung volumes in 7,479 subjects at risk for COPD but with normal spirometry. They found that subjects with air trapping defined as an RV/TLC > the upper limit of normal were more likely to develop COPD, be hospitalized for a COPD-related illness, and had a higher risk of mortality.⁸⁸

The results of exercise testing may also be predictive of survival in patients with COPD. As mentioned above, data from the ECLIPSE study¹²⁰ found that a 6MWD ≤ 334 m was associated with an increased the risk of death. Pinto-Plata et al¹²¹ found that 6MWD was an independent predictor of survival and that non-survivors had larger declines in 6MWD (–40 m/year) than did survivors (–22 m/year) despite similar yearly declines in FEV₁. In a 3-year observational study of subjects with COPD, Casanova et al¹²⁵ found the 6MWD to be highly predictive of mortality in subjects with an FEV₁ ≤ 50% of predicted. The median 6MWD was 420 m in survivors and 339 m in non-survivors. Lower resting ${\text{P}}_{{\text{aO}}_{\text{2}}}$ and developing oxygen desaturation during the 6MWT increased the risk of death.¹²⁵ Kim et al¹²⁷ applied 2 definitions of exercise-induced desaturation during a 6MWT ( ${\text{S}}_{{\text{pO}}_{\text{2}}}$ ≤ 88%; ${\text{S}}_{{\text{pO}}_{\text{2}}}$ < 90% or a decrease of ≥4%) in a cohort of subjects with COPD. Meeting both definitions of exercise-induced desaturation during 6MWT was associated with higher mortality; however, desaturating to ≤88% was a stronger predictor of mortality when compared with subjects who did not desaturate (50% vs 11%, respectively).¹²⁷ Golpe et al¹⁹⁵ also showed that 6MWD, baseline oxygenation, and desaturation during a 6WMT was predictive of mortality. When accounting for the impact of body weight on 6MWD, specifically 6MWT work (body weight × 6MWD), was not superior to 6MWD in predicting survival.

PFTs for COPD Screening and Case Finding

Although spirometry is the standard for detecting air-flow obstruction as the primary physiologic feature of COPD, there is less consensus with concern to who should be screened for the disease. The United States Preventive Services Task Force has examined the evidence for and against screening for COPD in 2008,¹⁹⁶ 2016,¹⁹⁷ and most recently in 2022.¹⁹⁸ In each instance, their recommendation has been that screening persons who are asymptomatic for COPD has little benefit for improving quality of life, morbidity, or mortality. This recommendation does not apply to individuals who present with cough, sputum production, or wheezing, or who otherwise have symptoms suggestive of air-flow limitation.

The use of spirometry to assess patients with symptoms of respiratory disease is widely accepted. Targeting of current or former smokers, usually ages > 40 years, is often performed in the clinic or by primary care practitioners.¹⁹⁹ Patients often do not self-report symptoms (eg, dyspnea), until their COPD has advanced. They may modify their lifestyle and avoid performing activities that cause shortness of breath. They may not indicate symptoms of dyspnea because their activities of daily living are at a low level. Screening for COPD in susceptible subjects has frequently used the GOLD recommendation of an FEV₁/FVC < 0.7 after bronchodilator.^5,77 The fixed 0.7 threshold has been shown to introduce an age-related bias, so ATS/ERS guidelines recommend using the LNN from equations appropriate for the patient being tested.¹⁷ Some case-finding studies have used pre- and post-bronchodilator spirometry, whereas others relied on only pre-bronchodilator measurements. However, there is no clear evidence that post-bronchodilator testing is better at detecting actual COPD.²⁰⁰ Studies of case findings for COPD conclude that the disease is either over- or under-diagnosed, largely due to how COPD is defined.²⁰¹ Misdiagnosis of COPD often occurs when spirometry is not performed.

Spirometry is not the only tool that can be used to assist in making a diagnosis of COPD. Other modalities include questionnaires and peak expiratory flow measurements, which have shown demonstrated sensitivity but with reduced specificity. Questionnaires designed to detect COPD have been shown to perform well when used to select patients for diagnostic spirometry. The COPD Diagnostic Questionnaire,²⁰² the Lung Function Questionnaire,²⁰³ and the COPD Assessment Test⁴¹ are examples in which cut points, based on questionnaire score, identified patients whose subsequent spirometry confirmed a diagnosis of COPD. Combining questionnaires and peak expiratory flow measurements to select patients for diagnostic spirometry has been shown to be an efficient and cost-effective approach for COPD case finding.²⁰⁴ The Burden of Obstructive Lung Disease study used peak expiratory flow to reduce the number of subjects at risk as determined by questionnaire alone.²⁰⁵ In patients with severe COPD, instability in daily peak expiratory flow measurements predicts poor outcomes, including reduced exercise capacity, the need for hospitalization, and all-cause mortality.²⁰⁶

Spirometry pre- and post-bronchodilator has historically been used to differentiate between fixed airway obstruction (ie, COPD) and reversible obstruction (ie, asthma). There is significant overlap in the physiology between these 2 entities. There is also significant variability in the spirometric findings of patients at risk of developing COPD. Aaron et al²⁰⁷ describe how measurement of FEV₁/FVC in individuals with mild or moderate airway obstruction can change from within normal limits to obstruction and vice versa. Buhr et al²⁰⁸ looked at subjects in a large clinical cohort whose air flow reverted from obstructed to within normal limits. They found that there was significant risk for future development of COPD in those with pre- but not post-bronchodilator obstruction. All of these factors need to be considered when screening for potential COPD.

Summary

PFTs have and will continue to play an important role in the diagnosis and management of COPD in clinical medicine. PFTs will also be pivotal to expand our evolving understanding of the complex pathophysiology of COPD. What has become crystal clear is that the paradigm that a COPD diagnosis hinges on putting the “O” in COPD with abnormal spirometry is false. Patients who are symptomatic may have early or pre-COPD, or even extensive emphysema despite normal spirometry.^100,101 Future studies are needed to expand our understanding of abnormalities in the silent zone of the lung not detected by conventional spirometry, which may contribute to the pathophysiology of COPD. Clinicians and researchers require a deep understanding of the strengths and the many limitations of PFTs in the realm of COPD to properly use PFT data to solve clinical problems and research challenges.

Discussion

MacIntyre: The ATS/ERS says to use the largest FVC of all the tests, but it doesn’t have to be from the same maneuver. They do say you can report the FEV₁/slow vital capacity but it’s a separate line item on the report.

Haynes: Yes, and I was saying earlier that some of these things might be better if they were reported with the concept of documenting by exception. So, if it was abnormal, then the software would flag it for you. The less traditional indices like FEV₁/slow vital capacity and FEF₇₅ wouldn’t necessarily be on the standard report unless they were abnormal, and the computerized interpretation could inform the human interpreter that these findings are consistent with early COPD.

MacIntyre: Well, I don’t think a report should ever put a diagnosis in.

Haynes: No, but at least point out if it’s abnormal.

MacIntyre: So, a major tool in assessing COPD. Thoughts?

Orr: Due to this issue of making a diagnosis based on a diagnostic test, many radiologists on their reports will write ‘clinical correlation advised,’ which really bothers some people. But it sounds like you’re talking about the importance of Bayesian or pre-test probability when we have an abnormal value. So, should we be reporting abnormal values with the warning that “clinical coordination is advised”? I worry that people will be assuming a diagnosis if I report LLN in a 35-year-old, when the rate of false positives is likely quite high.

Haynes: The other potential problem with radiology results is chest x-ray reports where big lungs are categorized as “hyperinflated compatible with COPD”. Many years ago, I witnessed a long run of young people coming in for PFTs for emphysema because a particular radiologist often interpreted large lungs as “hyperinflated lungs compatible with COPD”. But yes, I think spirometry interpretation and diagnosis should be separate, they’re not the same thing. There is a gray zone. We accept a 5% error that if your z-score is below –1.64, 1/20 patients like that won’t have lung disease they’re just on the lower end of the normal distribution curve. So it’s important to separate making a diagnosis and interpreting spirometry or any other test, for that matter.

Carlin: How do you see artificial intelligence (AI) coming into play here? And when can we expect that to happen?

Haynes: AI is already here, it’s whether or not people want to use it. I think that the interpretation software is very good, we use it for pediatrics at our hospital because we don’t have a pediatric pulmonologist right now. It’s always been there and it can outperform doctors as I showed. I think sometimes the resistance might be that physicians are reimbursed for reading tests. I like a combined approach where you use the AI software and there would be some over-reading by a pulmonologist – I wouldn’t take the humans out of interpretation completely, but the AI software outperforms human beings.¹ Even in the same practice, you can see different pulmonologists interpret the same patterns differently.

MacIntyre: AI I think refers to two different concepts in my brain. True AI to me is when you have a computer analyze all the raw data and develop an interpretive strategy to match some sort of accepted standard. In contrast, what you are describing is a computerized interpretive strategy that follows a clear-cut algorithm. If you fall below LLN you go this way and if you’re above you go that way. I’m not sure I’d call that AI, that’s just following an algorithm.

Haynes: And that’s what we teach people to do.

MacIntyre: That’s fine and I don’t have a problem with that, but that’s not real AI. AI to me is going beyond that and actually taking, not just these two values (FEV₁ and FVC), but actually looking at the entire flow-volume loop, analyzing the whole curve, and then trying to match that to some kind of clinical, imaging, or other type of standard. Determining a standard gets tough. Maybe it’s imaging, but imaging has its own silent zone. You can image lungs down to about a mm diameter airway but there are a lot of airways between that 1-mm airway and the alveolus. These are silent even in the imaging world. So that’s where AI is going to come in, it’s going to try to get at those issues and I’m afraid we still have a way to go.

Carlin: I liken this to current discussion in major league sports – particularly baseball and in particular baseball umpiring. Making a correct call of balls and strikes is a cornerstone of baseball. The strike zone is well defined yet interpreted differently by the umpires. They are now using a form of AI to help with those determinations in the minor leagues and it is proving to be successful. Right now in the major leagues, the home plate umpire will call around 180 pitches per game. Following the game, every pitch is reviewed and correlated with the umpire’s calls. Usually a 95% or greater concordance is attained by the umpire yet we all remember the one that was a bad call. That umpire is then judged to be a lousy umpire. As the umpires continue to use AI to help with called balls and strikes that should increase their accuracy and make the game outcomes less controversial. We can likely do this with our COPD diagnosis and just think if we are correct in our diagnosis in over 95% of cases. Wouldn’t that be great? I don’t see why we can’t do that in medicine. As Neil mentioned, looking at ways to integrate all the information collected to make an accurate diagnosis.

MacIntyre: The ERS/ATS interpretation guidelines have just come out.² There’s a tremendous amount of discussion in that document about this uncertainty zone. This is an attempt to get away from absolute cutpoints like normal vs abnormal. And they’re encouraging people to put in the notion that this patient is borderline and we’re not sure on which side of the fence they’re falling. Clinicians will have to live with that, that’s reality. And in our interpretation codes we have back home, I am now putting in codes that say exactly that. If you are sitting on the fence, I don’t say if you’re obstructed or not obstructed, I don’t say if you’re restricted or not restricted – only that you may be. Future testing will be needed to see which way you’re going to fall on this thing. Otherwise, you end up with these crazy statements where you’re normal one week and then abnormal the next week, and then you’re back to normal again. And clinicians go nuts with that.

Haynes: I’m glad you mentioned that Neil, because I didn’t show it here but a study by Aaron and colleagues showed that 20% of data from COPD subjects in the Lung Health Study went above and below LLN over time and this can be affected by how long you have the patient exhale.³ The FEV₁/FVC is affected by expiratory time, so if a technologist has the patient exhale longer than the technologist did on a previous test the FVC will likely be larger and the ratio will be lower.

MacIntyre: So automatic interpretation algorithms are going to work, I would guess, in somewhere in the 75% range. That’s just a wild guess on my part. But there’s that 25% there where an eyeball and a brain and experience and intelligence has to come into play as to what I think this might actually be and then I hedge my bets with statements that it might be. This is an area of uncertainty that we can’t be more definitive on.

Haynes: This supports the idea of testing patients more than once. If you only test a patient once, and their values are normal the clinician might declare that the patient doesn’t have COPD, and it might be assumed that the patient is not susceptible to developing COPD from smoking. But if the patient is at risk, they should be tested multiple times because you may find different results.

Carlin: We do that in sleep medicine all the time in regards to the diagnosis of obstructive sleep apnea. If I know my apnea hypopnea index is 5.9 events per hour, am I normal or abnormal? While I am outside the normal range (less than 5 apnea/hypopnea events per hour), if I am asymptomatic should I receive treatment or not? Using AI to help integrate all the clinical and laboratory findings should ultimately help to determine the correct approach to therapy.

MacIntyre: You’re not normal, Brian.

Carlin: You might certainly be correct there, Neil.

MacIntyre: Uncertainty is a reality. I have to ask one spirometry question just for completeness’ sake. I know you can’t answer it in 2 minutes or less, probably 2 hours or less. Should we be race adjusting?

Haynes: When you look at different ethnic groups there’s no questions that there are differences in lung size as a function of height, age, and sex. That’s really beyond question. However, this is a very complicated and controversial topic. There may be many reasons for differences in lung size, maybe it’s not completely genetic, and socioeconomic and environmental influence may play a role. But the differences are still there, and the risk you run of saying we should not use reference equations from specific ethnic groups is that some people are going to be misclassified. For example, a study of Southeast Asian never smokers found that when Caucasian-derived reference equations were applied 30% had an abnormal FEV₁, but <2% had an abnormal FEV₁ when an ethnically-derived equation was used.⁴ But this problem is very complicated because the world is more diverse than ever and it is difficult, and perhaps inappropriate, to try to put patients exclusively in one ethnic group.

MacIntyre: It’s safe to say that trying to use self-identified race/ethnicity as factors to determine normal is open to all kinds of variability and misinterpretation. But there is tons of controversy on this. I go back to the interpretation guidelines which, like I said, just came out online.² The ATS/ERS hedged their bets. They agreed there were race/ethnicity issues but they stopped short of saying to use (or not use) race-adjusted formulas, they just said use GLI as you see fit. Also, there is the other/mixed category with GLI that some people are espousing.

Haynes: The problem with the ‘other’ category is it’s 80% weighted towards white. So each group is equally represented but 80% of it is Caucasian.

MacIntyre: I understand that, but what I’m saying is if it’s not 100% white it’s going to get dropped a bit by including these other ethnicities. I’m not espousing using race corrections but it’s a fascinating topic.

Haynes: I do not support using race corrections in reference equations because those are just using a fixed correction factor for all indices, for both sexes over all ages and heights and the differences vary with those factors.

MacIntyre: Fair point.

Haynes: Using correction factors is different than using reference equations that were derived from a certain ethnic population. And I think something that hasn’t been asked is what do the patients want? If you are from Thailand and your parents and grandparents are from Thailand and you had a lung function test to qualify for employment that was normal according to GLI Southeast Asian and the same data is abnormal using a Caucasian-derived equation, which one would you want to be used? I think we’re leaving the patients out of the discussion.

Mike Hess: I’d also throw out there that calling them corrections implies that one of them is right.

Haynes: GLI does not use correction factors, the GLI equations are derived from sampling from those specific populations. The problem is that the available equations do not sample from every population on earth which you might assume from the word global in GLI. Available reference equations are not going to apply to every group of people and when I’ve been in discussions about on this topic, I always push the point that you’re never, ever, ever, going to have a reference equation that will fit every demographic. It’s impossible. I’ve also raised the question that maybe global isn’t the way to go, maybe local is better. Applying the statistical approach that GLI uses, which is better than the old polynomial equations, to locally-derived pulmonary function data might be better. It’s important to keep in mind that if a patient’s lung function is really bad the equation that is used probably doesn’t matter, it’s really an issue for patients who are at risk of being miscategorized because their data is near the LLN and that’s why I think recognizing that there is a gray zone and resisting the tendency to put patients in either the healthy box or the disease box can reduce the incidence of misclassification. In medicine everyone wants the “yes/no”, “positive/negative” test result for which I say pregnancy tests and autopsies, everything else has a gray zone and if you don’t appreciate the gray zone, you’re going to make a lot of mistakes.

Orr: I’d like to point out as far as the mixed category, it’s one that’s growing when you look at demographics. I believe in California it’s now 10%. My children are included in that. Mixed is not a real category because the reality is the parents may be of different racial/genetic backgrounds.

Haynes: And how do you quantify that as well? I serve as a member of the GLI Executive Task Force, we have developed a new equation that has been named “GLI Global”.⁵ GLI Global is similar to the ‘other’ GLI category, but it more equally represents all of the ethnic groups. I don’t think GLI Global should be used for all patients, rather I think it should be used as a replacement of the current GLI “other” equation which is heavily weighted towards Caucasian.

†Branson: Jeff, the data you showed on office spirometry were pretty dismal. But we often hear that’s a place for respiratory therapists to work, in a pulmonary physician’s office doing disease management, including spirometry. Do you have any hope for that or ideas as to where we’re going in the future?

Haynes I think that would be wonderful, I just don’t know that it will ever happen.

†Branson: Because of finances or logistics?

Haynes Yes. You can see that a lot of primary care offices aren’t even using many registered nurses. I see spirometry reports coming from offices and the technicians a lot of times are medical assistants. I don’t believe spirometry is part of their training, maybe it should be. The other problem with office spirometry is if you don’t perform tests frequently you’ll never become proficient. Once I asked a patient who supposedly had an abnormal test in the office whether the person doing the test appeared confident, and they said, “well actually there were two people, one pushing the buttons and the other one was reading the manual”. That’s all I needed to know.