Spirometry interpretation is influenced by the predictive equations defining lower limit of normal (LLN), while ‘distal’ expiratory flows such as forced expiratory flow at 50% FVC (FEF_{50}) are important functional parameters for diagnosing small airway disease (SAD). Area under expiratory flow-volume curve (AEX) or its approximations have been proposed as supplemental spirometric assessment tools. We compare here the performance of AEX in differentiating between normal, obstruction, restriction, mixed defects and SAD, as defined by Global Lung Initiative (GLI) or National Health and Nutrition Examination Survey (NHANES) III reference values, and using various predictive equations for FEF_{50}.

We analysed 15 308 spirometry-lung volume tests. Using GLI versus NHANES III LLNs, and diagnosing SAD by the eight most common equation sets for forced
expiratory flow at 50% of vital capacity lower limits of normal (FEF_{50 LLN}), we assessed the degree of diagnostic concordance and the ability of AEX to differentiate between various definition-dependent patterns.

Concordance rates between NHANES III and GLI-based classifications were 93.7%, 78.6%, 86.8%, 88.0%, 93.8% and 98.8% in those without, with mild, moderate, moderately severe, severe and very severe obstruction, respectively (agreement coefficient 0.81 (0.80–0.82)). The prevalence of SAD was 0.6%–6.9% of the cohort, depending on the definition used. The AEX differentiated well between normal, obstruction, restriction, mixed pattern and SAD, as defined by most equations.

If the SAD diagnosis is established by using mean FEF_{50 LLN} or a set number of predictive equations, AEX is able to differentiate well between various spirometric patterns. Using the most common predictive equations (NHANES III and GLI), the diagnostic concordance for functional type and obstruction severity is high.

Is area under expiratory flow-volume curve (AEX) useful in assessing small airway disease (SAD)?

The AEX can be used as a useful additional tool in diagnosing the spirometric pattern of SAD.

Comparing various predicted equations for normal spirometry, we show that prevalence of SAD varies considerably. Using the most common predictive equations (National Health and Nutrition Examination Survey III and GLI), the diagnostic concordance for functional type and severity is high.

Interpretation of pulmonary function testing (PFT) relies mainly on comparisons between measured flows or volumes and predicted, upper and lower limit of normal values (ULN or LLN, respectively), as determined by specific reference equations, derived from healthy individuals originating from similar populations.

In spirometry, obstruction is defined as forced expiratory volume in 1 s to forced vital capacity (FEV_{1}/FVC) ratio below the fifth percentile of its distribution (LLN). Restriction is _{LLN} or FVC <FVC_{LLN} when FEV_{1}/FVC ratio is ≥FEV_{1}/FVC_{LLN}; in this situation or when mixed ventilatory defects are suspected, full lung volume testing becomes necessary as the diagnostic gold standard. Lastly, small airways disease (SAD) is usually defined when the following criteria are satisfied: normal FVC, normal FEV_{1}/FVC ratio and forced expiratory flow at 50% FVC (FEF_{50}) < forced expiratory flow at 50% of vital capacity lower limits of normal (FEF_{50 LLN}),_{50} has very high variability and wide CIs for its predicted values, but many reference equations have been published and validated over the years, in different populations._{1}, FVC and FEV_{1}/FVC are the National Health and Nutrition Examination Survey (NHANES) III_{50}.

In previous work, we described the diagnostic utility of a spirometric parameter called area under expiratory flow-volume curve (AEX) and its approximations^{–1}, Y axis) by volume (L, X axis) during a forced expiration manoeuvre from TLC to RV. Although AEX can be computed by digital spirometry software, AEX values appear currently to be offered by only a minority of PFT equipment manufacturers.

To address the above shortcomings and to enhance the diagnostic accuracy and value of spirometry, this study evaluates the diagnostic value of AEX in defining the type (normal, obstruction, restriction, mixed defects or SAD) and the severity of impairment compared with the most common reference equations, NHANES III and GLI.

The dataset included 15 308 consecutive tests consisting of spirometry and same-day lung volume determinations, performed in the Cleveland Clinic PFT Laboratory on 9328 distinct adult subjects, over a period of 10 years. Spirometry was performed following the American Thoracic Society (ATS) standards._{50}, we decided to compute the LLN values based on the main published sets of reference equations.^{2-9} Using eight of the most common predictive equations, we classified the presence of SAD as possible, probable, highly probable or definite if the FEF_{50} was <FEF_{50 LLN} from 5, 6, 7 or all 8 of the eight equations. We also computed the average of all eight FEF_{50 LLN} for every subject tested (‘mean FEF_{50 LLN}’).

Categorical variables were displayed as counts and percentages and compared using Pearson χ^{2} tests. The goodness of fit for the continuous variables’ distributions was assessed using Shapiro-Wilk W (n<2000) or Kolmogorov-Smirnov-Lillefors (n>2000) tests (normal distributions), Kolmogorov’s D test (exponential or logarithmic distributions) or Cramer-von Mises W (Weibull or gamma distributions), as appropriate. Continuous variables were reported as mean±SD or median (25th–75th IQR), and were compared using Student’s t-test or Tukey-Kramer HSD (parametric), Welch’s analysis of variance or Wilcoxon/Kruskal-Wallis rank sum (non-parametric) tests, as appropriate.

Statistical significance was defined as p<0.05. Analyses were performed using JMP Pro V.14.

None (as this involves secondary analyses of data).

Among the subjects tested, 51% were men and 49% women; 86% were White and 13% African-American. Median (IQR) for age, height, weight and body mass index were 57 (47–67) years, 168 (161–175) cm, 79 (66–94) kg and 28 (24–32) kg/m^{2}, respectively. All subjects had spirometry. The helium dilution method was used in 40%, and body plethysmography in 60% of volume determinations. The median (IQR) for FEV_{1}, FVC and FEV_{1}/FVC were 1.77 (1.09–2.49) L, 2.81 (2.14–3.59) L and 0.62 (0.47–0.78), respectively. Mean±SD and median (IQR) for AEX and Sqrt AEX were 6.64±6.13, 4.88 (1.95–9.48) L^{2}·s^{–1}; and 2.32±1.13, 2.21 (1.40–3.08) L·s^{–0.5}, respectively.

Based on FEV_{1} percent predicted by NHANES III references, obstruction was present in 39.4% of the tests; among them, the degree of severity (using standard ATS/ERS stratification criteria)

(A) Mosaic plot illustrating concordance and discordance rates between NHANES III and GLI—defined obstruction. P<0.0001 (Pearson test). Colour codes: green: no obstruction; red: obstruction. (B) Mosaic plot illustrating concordance and discordance rates between NHANES III and GLI—defined obstructive ventilatory defects, using standard American Thoracic Society severity stratification. P<0.0001 (Pearson test). Colour codes: green: normal; light pink: mild obstruction; darker pink: moderate obstruction; light red: moderately severe; red: severe; dark red: very severe. GLI, Global Lung Function Initiative; NHANES III, National Health and Nutrition Examination Survey 3.

When using both spirometry and lung volume determinations (per the ATS/ERS diagnostic algorithm),_{LLN} on lung volume testing.

Ventilatory pattern distribution versus classification schema and predictive equations used. Classification 1 included spirometry only (FEV_{1}/FVC vs FEV_{1}/FVC_{LLN}). Classification 2 partitioned ventilatory impairments based on FEV_{1}/FVC versus FEV_{1}/FVC_{LLN}, FVC versus FVC_{LLN} and TLC versus TLC_{LLN} (observed vs lower limit of normal, LLN)

Classification | Reference equations | Normal | Obstruction | Restriction | Mixed | Small airway disease (SAD) |

Classification 1 | NHANES III | 39.4 | ||||

GLI | 42.6 | |||||

Classification 2 | NHANES III | 27.6 | 48.4 | 19.7 | 4.2 | – |

GLI | 28.2 | 49.7 | 17.2 | 4.9 | – |

NHANES III, National Health and Nutrition Examination Survey 3.

Using the most commonly used predictive equations for FEF_{50 LLN} (_{50}<FEF_{50 LLN} using at least 5, 6, 7 or all 8 of the eight equations used for comparison. By comparing FEF with the mean FEF_{50 LLN} (ie, the average of all eight FEF_{50 LLN} from the respective predictive equations), the prevalence of SAD was 6.9%.

Box-and-whisker and histogram plots of FEF_{50 LLN} by eight different predictive equations_{50} measurements. FEF_{50}, forced expiratory flow at 50% FVC; FEF_{50 LLN}, forced expiratory flow at 50% of vital capacity lower limits of normal.

By using either NHANES III or GLI reference equations (^{2}·s^{–1} in possible, probable, highly probable and in definite SAD, respectively. When Sqrt AEX was assessed by functional patterns, all in-between differences were statistically significant when SAD was defined by FEF_{50} <mean FEF_{50 LLN} (p<0.0001, _{1}/FVC ratio).

Sqrt AEX differentiates well between various functional patterns in both NHANES III ((A), top) and GLI ((B), bottom) classifications. P<0.0001 (for all in-between group comparisons, Welch’s ANOVA and Kruskal-Wallis tests). Colour codes: green: normal; red: obstruction; blue: restriction; purple: mixed ventilatory defects (as defined by GLI equations). ANOVA, analysis of variance; GLI, Global Lung Function Initiative; NHANES III, National Health and Nutrition Examination Survey 3; Sqrt, square root.

Box-and-whisker plots and distributions of Sqrt AEX by functional patterns, as defined by GLI equations for normal, obstruction, restriction, mixed ventilatory defects and SAD; the latter, a subgroup of normal spirometry if no distal flows are analysed, is defined as FEF_{50}<mean FEF_{50 LLN} (ie, the mean of all eight FEF_{50 LLN}, as per the respective predictive equations, L·s^{–1})._{50 LLN}, forced expiratory flow at 50% of vital capacity lower limits of normal; GLI, Global Lung Function Initiative; NHANES III, National Health and Nutrition Examination Survey 3; SAD, small airway disease; Sqrt, square root.

In spirometry, obstruction is optimally defined by FEV_{1}/FVC<FEV_{1}/FVC_{LLN}, while restriction is _{LLN} or FVC <FVC_{LLN} when FEV_{1}/FVC≥FEV_{1}/FVC_{LLN}. In the latter situation or when mixed ventilatory defects are suspected, confirmatory lung volume measurements are necessary. When comparisons are performed between total lung capacity (TLC), TLC_{LLN} and TLC_{ULN,} the following patterns can be identified: normal (TLC_{LLN}_{ULN}), thoracic overdistension (TLC >TLC_{ULN}) or restriction (TLC <TLC_{LLN}). Similarly, airway hyperinflation can be defined as functional residual capacity (FRC) >FRC_{ULN}, residual (RV) >RV_{ULN} or RV/TLC >RV/TLC_{ULN}. Lastly, SAD is usually defined as normal FVC, normal FEV_{1}/FVC ratio and FEF_{50}<FEF_{50 LLN}._{50} has very high variability and wide CIs for its predicted values but, over the years, many reference equations have been published and validated in different populations._{50}.

The appeal of AEX relates to the existing limitations of spirometry and its current interpretive strategies. For example, an obstructive defect is currently defined per ATS/ERS recommendations_{1}/FVC ratio below the fifth percentile. However, it is well-known that sometimes, the earliest changes of airflow limitation occur in the small airways, and can be seen in the last portion of the flow-volume diagram (with a downward ‘concave’ appearance of the curve), even when the initial portion is still normal. Quantitatively, this can be described as reduced isovolumic instantaneous flows, such as forced expiratory flows at 50% or 75% of FVC (FEF_{50} or FEF_{75}, respectively) or the flow between 25% and 75% of FVC (FEF_{25–75})._{50 pred} and FEF_{50 LLN} have not been developed in neither the NHANES III nor in the GLI equation sets. The use of AEX is intended to address these limitations, being able to better separate qualitatively (the type) and quantitatively (the severity) of these ventilatory impairments.

Indeed, this analysis suggests that AEX can also quantitatively assess flow changes that impact terminal or distal segment of the flow-volume curve. The terminal segment of the flow-volume loop is considered effort-independent and due to a complex interplay between resistance to airflow in small airways or units with long emptying time and the respiratory system’s elastic recoil.

To date, several studies have addressed the possible utility of AEX._{1}._{1}._{1}; unfortunately, their study lacked a validation group. They wrote, ‘we speculate that when dealing with lungs of normal capacity, this new parameter might be sensitive enough to detect small airways involvement when the Raw is normal’.

Extending the scope and the scale of the studies which examined AEX in narrow clinical settings or in small populations, in this investigation and in our prior communication,

One possible limitation of the current investigation is related to the fact that our subjects’ demographics were different from the general US population. For example, in our dataset, 86% were White, 13% were African-American, while other ethnic or racial groups were poorly represented. Further, any specific working definition used for SAD will inevitably affect the performance of the new measurement, inducing ‘imprecision’. As shown in _{50 LLN}. We found that even when the FEF_{50 LLN} ‘outliers’ were excluded from the set of predictive equations (eg, Forche and Miller equations),

The AEX is a useful tool for assessing respiratory dysfunction, including that of SAD. In this setting, the formulas used for FEF_{50 LLN} can influence greatly the accuracy of the categorisation. This study also shows that, using the two most common predictive equations (NHANES III vs GLI), the diagnostic concordance for the type of defect and severity of obstruction is relatively high.

Kevin McCarthy RCPT (data extraction).

OCL: concept, data analysis and interpretation, article writing, submission. JKS: concept, interpretation, article writing.

The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

None declared.

Not required.

The study received Institutional Review Board approvals (Cleveland Clinic IRB EX#0504 and EX#19–1129; Emory IRB #00049576).

Not commissioned; externally peer reviewed.

No data are available.