Measurement of FEF_25–75% and FEF_75% does not contribute to clinical decision making

Introduction

The forced expiratory vital capacity (FVC) manoeuvre is the fundamental manoeuvre of the test most frequently performed for assessing pulmonary function. Apart from the forced expiratory volume in 1 s (FEV₁) and the FEV₁/FVC ratio, a host of other indices have been derived from the FVC manoeuvre. These include peak expiratory flow, flows at 25%, 50% and 75% of the exhaled FVC and the forced expiratory flow at 25–75% of FVC (FEF_25–75%) (also called maximum mid-expiratory flow (MMEF)). The latter sets of flows, as well as the assessment of closing volume, were described as being more reproducible and more sensitive than FEV₁ to the presence of “small airways disease” [1–3]. It was generally accepted that obstruction in small airways led to reduced flows at low lung volumes, leaving flows at high lung volumes much less affected, resulting in the characteristic concave flow–volume curve [4–6]. The prospect of detecting lung disease at an early stage has led to widespread measurement of FEF_25–75% and forced expiratory flow at 75% of FVC (FEF_75%). However, the greater reproducibility of flows and better sensitivity of FEF_25–75% were also challenged [7, 8]. In addition, flows at a percentage of the FVC are sensitive to measurement errors in the FVC. Furthermore, as FVC and total lung capacity may be affected by disease, forced expiratory flows in patients will be measured at a different lung volume than in healthy subjects, rendering the use of predicted values derived from healthy subjects problematic. For the same reason serial measurements in a subject in whom the FVC has changed due to disease progression and/or therapeutic intervention should only be compared if they have been made at the same lung volume [9]. Finally, the hypothesis that reduced mid-expiratory flows were specific for small airways disease has been shown to be incorrect [10].

These limitations of instantaneous and mid-expiratory flows in clinical practice have led to recommendations to disregard any suspected abnormality in flows if the FEV₁ and FEV₁/FVC ratio are within normal limits [11–13]. However, it is suggested that in the presence of a borderline value of FEV₁/FVC, these tests may suggest the presence of airway obstruction [13]. Reporting of flow is also recommended for quality-control purposes in preschool children [14]. A search on PubMed using the keywords FEF_25–75%, MMEF and FEF_75% revealed that these indices were used in 1143 publications between 1975 and 2012, of which 385 were published since 2000.

The purpose of this study is to investigate the extent to which the use of expiratory flows, namely the FEF_25–75% and FEF_75%, adds information to spirometry over and above that from FEV₁, FVC and FEV₁/FVC ratio.

Results

The numbers of males and females and age ranges in the four datasets are shown in table 1. 68 males and 67 females (0.59% of the total study population) were very short or very tall for age, leading to extreme z-scores for spirometric indices; these data were considered outliers and excluded, leaving 22 632 spirometry results for analysis. The overall prevalence of airways obstruction, defined as FEV₁/FVC ratio less than the lower limit of normal (LLN), was 26.4%; on the basis of FEV₁ % predicted, using the American Thoracic Society (ATS)/European Respiratory Society (ERS) grading system [13], 8.85% were classified as mild, 4.34% as moderate, 4.19% as moderately severe, 5.45% as severe and 3.56% as very severe airways obstruction. A “spirometric restrictive pattern” (where FEV₁/FVC>LLN and FVC<LLN) was found in 15.3% of the total group.

The mean z-scores for males and females for the various spirometric variables in the four datasets are shown in table 2. The LLN for FEF_25–75% and FEF_75% varied between 35–67% and 31–56%, respectively, declining steeply with age. The relationship between the z-scores for the FEV₁/FVC ratio and FEF_25–75% is shown in figure 1a; no differences in this relationship could be discerned between males and females. Of the 22 767 subjects, there were 1862 (8.18%) that showed a FEF_25–75% below the LLN but a FEV₁/FVC ratio in the normal range (quadrant Q3 in fig. 1a). In 1235 (66%) of these individuals this was associated with a FVC below the LLN (figs 1c and 2]). 2.87% of test results revealed a FEF_25–75% in the normal range, but FEV₁/FVC<LLN (suggesting airways obstruction) as indicated in quadrant Q1 in figure 1a. The percentage of subjects in whom FEF_25–75% was less than LLN but FEV₁, FVC and FEV₁/FVC were in the normal range is displayed as a function of age in Figure 2. After adjusting for the z-score for the FEV₁/FVC ratio, the z-score for FEF_25–75% was significantly lower (-0.12) in the Polish than in the other datasets.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Relationship between the z-scores of a) forced expiratory flow at 25–75% of forced vital capacity (FVC) (FEF_25–75%) and b) forced expiratory flow at 75% of FVC (FEF_75%) with that of the forced expiratory volume in 1 s (FEV₁)/FVC ratio. Data in figures c) and d) are subsets of a) and b), respectively, where the FVC is above the lower limit of normal. The dashed lines indicate the lower limits of normal (z= -1.645). The percentage values indicate the percentage of the total 22 767 test results for FEF_25–75% and 15 661 test results for FEF_75% that fall into each quadrant (Q1–Q4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Percentage of subjects, by 2-year age groups, in whom the forced expiratory flow at 25–75% of forced vital capacity (FVC) (FEF_25–75%) was below the lower limit of normal, whereas forced expiratory volume in 1 s (FEV₁), FVC and FEV₁/FVC ratio were within the normal range.

Data on FEF_75%were available in 15 661 subjects and there was a similar relationship with FEV₁/FVC ratio as with FEF_25–75% (fig. 1b). There were 477 cases (3.05% of the total) where FEF_75% was reduced (<LLN) but there was a normal FEV₁/FVC ratio (Q3 in fig. 1b), and the majority of these cases (58%) showed a reduced FVC (fig. 1d). Compared with FEF_25–75% there were significantly more results where FEV₁/FVC ratio was reduced but FEF_75% was not (12.25% of the total). After adjusting for the z-score for FEV₁/FVC ratio, the z-score for FEF_75% was significantly lower (-0.17) in the USA than in the Polish dataset.

Thus, only 2.75% of the total number of test results showed a reduced FEF_25–75% with both FEV₁/FVC and FVC within the normal range. The corresponding value for FEF_75% was lower, at only 1.29% of the total. Of these cases, about half showed a FEV₁/FVC ratio close to the LLN (within 0.25 z-values). When we reviewed the discordances in a random sample of 100 cases, 67% of them were found to have an artefact or submaximal exhalation which could affect results (71% in the 3–10 years age range and 29% in the 10–20 years age range).

A normal FEF_25–75% associated with an abnormal FEV₁/FVC ratio was rare (2.87%, quadrant 1 in fig. 1a). These cases were characterised by a low FEV₁ (mean z-score -0.79), an above-average FVC (mean z-score 0.55) and mostly mild airways obstruction (87.2% using the ATS/ERS grading system [13]). In 12.25% of cases a normal FEF_75% was associated with an abnormally low FEV₁/FVC ratio; characteristically there was more severe airways obstruction (mean z-score for FEV₁ -1.87, for FVC -0.52 and 49.7% cases of mild airways obstruction).

As the flow–volume curve clearly demonstrates, expiratory flows are highly dependent upon the lung volume at which they are determined. Similar to FEV₁ values, they were also expressed as a fraction of the FVC to take this dependency into consideration. The relationship between FEF_25–75%/FVC ratio and FEF_75%/FVC ratio with FEV₁/FVC ratio was curvilinear; however, log-transforming the data linearised the relationship and stabilised the variance, as shown in Figure 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Relationship between a) forced expiratory flow at 25–75% of forced vital capacity (FVC) (FEF_25–75%) and b) forced expiratory flow at 75% of FVC (FEF_75%) and forced expiratory volume in 1 s (FEV₁), both standardised for the FVC. Note that the y-axis is log-transformed.

Forced expiratory flows are commonly presented as a percentage of the mean predicted value. Figure 4 depicts the LLN (5th percentile, at z-score of -1.645) for FEF_25–75% and FEF_75% as per cent predicted, using the GLI-2012 equations and median height for age in males and females.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4–

The lower limit of normal (LLN) as a percentage of the predicted value for forced expiratory flow at 25–75% of forced vital capacity (FEF_25–75%) and forced expiratory flow at 75% of forced vital capacity (FEF_75%) in females and males.

Discussion

To our knowledge this is the first study that has systematically investigated whether the use of maximum mid-expiratory flow and instantaneous flows provides additional information beyond that provided by traditional spirometric indices (FEV₁, FVC and FEV₁/FVC ratio) in detecting lung function impairment. Our findings in a large clinical dataset show that airways obstruction, defined by an abnormally low FEV₁/FVC ratio, goes undetected by the FEF_25–75% and FEF_75% in 2.9% and 12.3% of cases, respectively. We have found that FEF_25–75% is reduced in the presence of a normal FEV₁/FVC ratio in 8.2% of cases. Only 3.0% of cases revealed an abnormally low FEF_75% with a normal FEV₁/FVC ratio. However, a large majority of these discrepant findings occurred when the FVC was abnormally low. As such, in only a very small minority of cases was a low FEF_25–75% or FEF_75% associated with normal FEV₁/FVC ratio and a normal FVC (2.8% and 1.3%, respectively). Of these discordant cases, we identified possible artefact or unsatisfactory FVC efforts in ∼63% of cases.

It is commonly perceived that the FEF_25–75% is diagnostically more helpful in a paediatric population than in adults. Indeed, we did find a trend of a higher prevalence rate of an abnormally low FEF_25–75% when FEV₁, FVC, and FEV₁/FVC ratio are in the normal range in youngsters than in adults (fig. 2), but even then the rate is very low at only 3%. It is difficult to always obtain high quality measurements in children. Indeed, in a random sample of 100 children with discordant findings in the 3–10-year and 10–20-year age ranges, 71% and 29%, respectively, were found to be associated with flow–volume curves with an artefact or submaximal exhalation or inhalation, which could affect results. The very low incidence of discordant findings argues against the notion that forced expiratory flows are more sensitive to small airways obstruction than other spirometric indices. We suggest that these infrequent occurrences where flow indices result in discordant findings from conventional spirometric indices probably represent statistical noise and biological variability associated with spirometry, and in particular with the high reliance of expiratory flow indices on the valid measurement of FVC. We would also suggest that such findings should prompt a critical review of the quality of the FVC manoeuvre. The practical implication is that the flow indices contribute little or no additional information over and above that provided by the FEV₁ and FVC and, therefore, do not add to the diagnostic arsenal in detecting airways obstruction.

The frequent use of these flow indices is based on the notion that airways obstruction affects flows towards the end of a forced expiratory manoeuvre much more than at the start of the FVC manoeuvre, because at low lung volumes lung elastic recoil is low and airway dimensions are small. Hence, such flows are thought to be more sensitive to small airways obstruction than the FEV₁. This misconception arose at least in part because McFadden and Linden [1] based their conclusions on the erroneous, yet still popular, idea that 80% pred denotes the LLN for any marker of lung function. This ignores the considerable age related difference in variability of lung function with age [17] and has led to considerable overdiagnosis of abnormality. The LLN for FEF_25–75% is 67% of predicted mean in childhood and 35% in those aged ≥80 years; corresponding values for FEF_75% are 56% and 31%. Indeed, as early as 1988, Flenley [10] contested the view that mid-expiratory flows were specific for small airways disease. By inference, partitioning into large and small airways obstruction in chronic lung diseases on the basis of spirometric test results is not warranted.

The dependency of forced expiratory flow on lung volume is immediately clear from flow-–volume curves, unlike the mechanisms responsible for determining such flows. During a forced expiration the high pleural pressure leads to dynamic compression of intrathoracic airways. Flow through compressed airways is then determined by wave speed limitation, i.e. when local gas velocity is equal to the speed of propagation of pressure waves in the airways; the location of the flow-limiting segment (choke point) is determined by the interplay between lung elastic recoil and airway compliance [19, 20]. As more volume is expelled, lung elastic recoil and pleural pressure diminish, which allows the choke point to settle upstream in airway segments with pressure–area characteristics that allow lower flow [21, 22]. It follows that forced expiratory flows are highly dependent on lung volume so that, like the FEV₁ (the average flow over the first second of the forced expiration), they should be standardised for true lung volume, which is the total lung capacity. This is usually not feasible, and hence the vital capacity is used as a proxy for lung size. The FEV₁/FVC ratio is dimensionally s⁻¹ and represents the lung emptying rate. As our analysis has shown, standardising the FEF_25–75% and FEF_75% in a similar fashion by expressing them as a ratio of the FVC, reveals high correlations with the FEV₁/FVC ratio (fig. 3), signifying that the information content of these indices is very similar.

Obstructive respiratory disease with increasing residual volume causes flow at a fixed percentage of the FVC to be measured closer to the total lung capacity (which may be increased due to disease) than in a reference population; this partly masks a fall in forced expiratory flows due to airways obstruction and will diminish the sensitivity in diagnosing obstructive lung disease. In addition, normal reference ranges for forced expiratory flows reflect between-subject variability both in flows and in the FVC. This leads to much higher coefficients of variation than for FEV₁, FVC or their ratio and, therefore, to larger reference intervals around predicted values (fig. 4). Variability might have been somewhat smaller if flows had been taken from the envelope of flow–volume curves [12, 23], but in this study it was taken from the FVC manoeuvre with the highest sum of FEV₁ and FVC in accordance with recommended guidelines [15]. Such wide intervals render these indices essentially worthless for diagnostic purposes [12, 17, 24]. This also highlights that using 80% pred as the LLN for lung function indices lacks any scientific basis and leads to highly biased and erroneous clinical decisions. The use of a lower limit defined as the lower 5th or other percentile of the distribution in a healthy population of nonsmokers is, therefore, the recommended procedure [12, 14, 17, 25–28]. Z-scores indicate the number of age-specific standard deviations that a measured value differs from the predicted value and, unlike using a fixed percentage of predicted as the basis for interpretation of lung function, they are free of bias due to age, height, sex and ethnic group [12, 14, 17, 25–28] and were, therefore, used in this study.

This study is based on a large number of data from patients referred to tertiary hospitals for suspected or known lung disease. The relationship between the z-score for FEV₁ and that for flows differed only slightly, albeit significantly, between centres, indicating that our findings are robust and applicable to children and adults with a wide range of normal and abnormal lung function test results. However, the very large number of data precluded rigorous post hoc quality control of all original spirograms. In routinely collected data there will be a proportion where data are clinically useful and therefore accepted, but where the FVC manoeuvres do not meet stringent quality criteria. As shown by this study, a low FEF_25–75% or FEF_75% when the FVC, FEV₁ and FEV₁/FVC ratio are within normal limits is rare and should lead to reviewing whether the FVC manoeuvre was performed correctly.