Abstract
It has been suggested that forced expiratory volume in six seconds (FEV6) should be substituted for forced vital capacity (FVC) to measure fractions of timed expired volume for airflow obstruction detection. The present authors hypothesised that this recommendation might be questionable because flow after 6 s of forced expiration from more diseased lung units with the longest time constants was most meaningful and should not be ignored. Furthermore, previous studies comparing FEV6 and FVC included few subjects with mild or no disease.
The present study used spirometric data from the USA Third National Health and Nutrition Evaluation Survey with prior published ethnicity- and sex-specific equations for FEV1/FEV6, FEV1/FVC and FEV3/FVC, and new equations for FEV3/FEV6, all derived from ∼4,000 adult never-smokers aged 20–80 yrs.
At 95% confidence intervals, 21.3% of 3,515 smokers and 41.3% of smokers aged >51 yrs had airway obstruction; when comparing FEV1/FEV6 with FEV1/FVC, 13.5% were concurrently abnormal, 1.5% were false positives and 4.1% were false negatives; and when comparing FEV3/FEV6 with FEV3/FVC, 11.6% were concurrently abnormal, 3.3% were false positives and 5.7% were false negatives.
Substituting forced expiratory volume in six seconds for forced vital capacity to determine the fractional rates of exhaled volumes reduces the sensitivity of spirometry to detect airflow obstruction, especially in older individuals and those with lesser obstruction.
- Airway obstruction
- cigarette smoking
- forced expiratory volume in six seconds
- forced expiratory volume in three seconds
- forced vital capacity
- spirometry
In 1999, using the large National Health and Nutrition Examination Survey (NHANES) III database, a number of spirometric reference equations, including those for forced expiratory volume in six seconds (FEV6) and FEV1/FEV6, were published 1. In 2000, a National Lung Health Education Program consensus statement 2 advocated replacement of forced vital capacity (FVC) and FEV1/FVC with FEV6 and FEV1/FEV6 to detect airways obstruction. Later, Swanney et al. 3 reported high sensitivity and specificity for FEV1/FEV6 compared to gold standard FEV1/FVC in 337 out of 502 patients tested in a tertiary hospital-based university laboratory. In a multicentred lung health study, Enright et al. 4 concluded that FEV1/FEV6 values could be useful for following the course of obstructive airways disease in smokers and for screening smokers for the presence of airway obstruction. Subsequently, Vandevoorde et al. 5 concluded from a large patient study that “the FEV1/FEV6 ratio can be used as a valid alternative for FEV1/FVC in the diagnosis of airway obstruction, especially for screening purposes”. Other investigators recommended using FEV6 rather than FVC for the mean forced expiratory flow between 25 and 75% of FVC (FEF25–75%) and for measuring lung restriction 3, 6–9.
FEV1/FVC, FEV1/FEV6 and FEV3/FVC ratios, derived from the large never-smoking NHANES III database, all decrease in a linear fashion as age increases, indicating an increase in long time-constant lung units or 1-FEV3/FVC 1, 10. Each of these formulae correctly identifies patients with severe airway obstruction. However, subjects with subtler obstruction also commonly exhale an important portion of their FVC after 6 s, i.e. from lung units discharging their gas late in exhalation. Consequently, FEV1/FEV6 and FEV3/FEV6 measurements, with denominators which exclude the FVC–FEV6 volumes, may be less discriminating than FEV1/FVC and FEV3/FVC in detecting milder airway obstruction. In screening for disease, it may be better to focus on detecting lung units with long time constants (after 3 or 6 s) rather than on shaving seconds off expiratory time and centilitres off forced expiratory volumes.
The present authors, therefore, hypothesised that FEV1/FEV6 and FEV3/FEV6 would be less reliable screening parameters than FEV1/FVC and FEV3/FVC in distinguishing changes in lung function due to normal ageing from those due to superimposed airway obstruction from smoking.
METHODS
Data meeting American Thoracic Society (ATS) standards 11 were extracted from the NHANES III database 12 for 13,113 adults, including ex-smokers, 5,943 never-smokers and 3,515 current smokers with no apparent skeletal or neuromuscular disease. Data were obtained nationwide with informed consent. Some data and equations derived from this population were previously published by others 1, 9 and by the present authors 10.
Regression equations 13 for mean and 95% confidence lower limits of normal (95% LLN) for FEV3/FEV6 (table 1⇓) were derived for never-smokers identified ethnically as Black, Latin or White 10. These never-smokers had similar FEV1/FVC values 8 to those of Hankinson et al. 1, derived from the same database.
Forced expiratory volume in three seconds(FEV3)/forced expiratory volume in six seconds (FEV6) per cent formula for never-smoking adults
The 3,515 current smokers were then divided into four similar-sized groups according to age: 20–29.3 yrs; 29.4–38.1 yrs; 38.2–50.7 yrs; and 50.8–80 yrs. Using sex- and ethnicity-specific equations for FEV1/FVC and FEV1/FEV6 1, FEV3/FVC 10 and the newly derived FEV3/FEV6, each value of the current smokers was categorised as “normal” or “abnormal”, depending on whether it was above or below 95% LLN values, as all ratios had normal distributions in never-smokers.
Given x = FEV1 or FEV3, deviations of discordant x/FEV6 values from x/FVC values, i.e. false positive or negative, were calculated as follows: 
When the paired ratios (x/FVC and x/FEV6) were both above their LLN, they were concordant normal; when both were below their LLN, they were concordant abnormal. However, when an individual FEV1/FVC was normal and the FEV1/FEV6 was below LLN, for example, the FEV1/FEV6 value was considered discordant and false positive. When an FEV3/FVC was below LLN and FEV3/FEV6 was normal, the FEV3/FEV6 was considered discordant and false negative.
Deviations of false positive or negative x/FEV6 values from x/FVC values were calculated as follows: 
Ratios of false positive and false negative to concordant abnormal values were calculated for groups of differing age and severities of obstruction. In those with abnormal FEV1/FVC and/or abnormal FEV3/FVC, severity of obstruction was based on FEV1 % predicted: severe <50%; moderate 50–65%; mild 65–80%; and minimal >80% and <120%. Two-by-two tables were created for the calculation of sensitivity, specificity, and positive and negative predictive values for each age group. To counter the potential criticism that a 95% LLN (mean-1.645×se) might be spurious or too strict, all analyses were repeated (but not necessarily reported) using a confidence limit of 99% (99% LLN; mean–2.33×se).
RESULTS
Figure 1⇓ shows the FVC manoeuvre durations for 13,113 subjects in the NHANES III survey aged ≥20 yrs who had optimal tests. In individuals with longer forced expirations (late emptying of long time-constant units), the mean and variability of the volume differences from FEV6 values increased markedly (almost similar in mL to the square of the duration of FVC in s). Table 2⇓ shows that the volume differences between FVC and FEV6 were higher for smokers and increased with age, especially in current smokers.
Increase in mean±sd of forced vital capacity (FVC)-forced expiratory volume in six seconds (FEV6) volumes in 13,113 adults as durations of FVC increase.
Changes in forced vital capacity minus forced expiratory volume in six seconds with age and smoking status
Using ethnicity- and sex-specific formulae, percentages of NHANES III current smokers found to have abnormal FEV1/FVC or FEV3/FVC are displayed in figure 2⇓ for each age group.
Percentage of incidence of airway obstruction using 95% confidence lower limit of normal (LLN) for forced expiratory volume in one second (FEV1)/forced vital capacity (FVC) and FEV3/FVC in 3,515 current smokers divided into equal-sized age groups. □: 20.0–29.3 yrs; ░: 29.4–38.1 yrs; ▓: 38.2–50.7 yrs; ▒: 50.8–80.0 yrs.
As expected, abnormalities increased in older age groups. Using 95% LLN values, >41% of 50.8–80-yr-old smokers had evidence of airway obstruction. Most commonly, in all age groups, both FEV1/FVC and FEV3/FVC were abnormal. If only one was abnormal, it was more likely to be FEV1/FVC in younger smokers and FEV3/FVC in older smokers. As noted in table 3⇓, using FEV1/FVC and FEV3/FVC as standards, the overall incidence of airway obstruction exceeded 20% in smokers. Severe airway obstruction was rare except in the oldest age group.
Severity and incidence of airway obstruction in 3,515 current smokers
Table 4⇓ displays, across age groups, both 95% and 99% LLN values. Using 95% LLN for FEV1/FEV6 and FEV1/FVC, a total of 473 concordant abnormal pairs and 194 discordant pairs (51 false positives and 143 false negatives) were found with a discordant/concordant abnormal ratio of 194/473 = 41%. For FEV3/FEV6 and FEV3/FVC, 408 concordant abnormal pairs and 315 discordant (115 false positives and 200 false negatives) pairs were found, with a discordant/concordant abnormal ratio of 315/408 = 77%. The number of false negatives increased strikingly with age. Using 95% LLN for both FEV1 and FEV3 comparisons, total specificities were relatively high, negative and positive predictive values were intermediate, while sensitivities were low.
Concordant/discordant spirometric measurements in 3,515 current smokers at 95% and 99% confidence limits
Using 99% LLN for FEV1/FEV6 and FEV1/FVC pairs and FEV3/FEV6 and FEV3/FVC pairs, (table 4⇑), proportions of discordant to concordant abnormal pairs actually increased while sensitivities declined. The high incidence of discordant values, false-negative values, low sensitivity and even FEV3/FEV6 false positives confirm the low reliability of the FEV1/FEV6 and FEV3/FEV6 to detect airway obstruction in this population.
Figure 3⇓ shows that at 95% LLN, discord increased markedly as the severity of airway obstruction decreased. Table 5⇓ shows that such false-negative discords also tended to be larger than false-positive discords; mean absolute mismatch of these ratios was 2.64%.
Ratios of a) discordant forced expiratory volume in one second (FEV1)/FEV6 (false positive or false negative) to concordant abnormal FEV1/forced vital capacity (FVC) and b) discordant FEV3/FEV6 (false positive or false negative) to concordant abnormal FEV3/FVC in current smokers with severe, moderate, mild and minimal obstruction, using 95% confidence limits. FEV1/FEV6 and FEV3/FEV6 usually correctly identify severe obstruction but are progressively more unreliable in identifying lesser degrees of airway obstruction. ░: false positive; □: false negative.
Differences of discordant ratios at 95% confidence limits
DISCUSSION
The present study found low sensitivities, and a high incidence of false negative FEV1/FEV6 and FEV3/FEV6 and a moderate incidence of false positive FEV3/FEV6 in this NHANES III population, supporting the present authors' hypothesis that the use of FEV6 in place of FVC reduces the sensitivity of spirometry in detecting airway disease. Prior findings 3–7 that promote FEV6 as an acceptable surrogate for FVC are therefore further detailed in table 6⇓. The study from a university hospital-based laboratory 3, which used spirometry from 310 patients, found high sensitivities and specificities when comparing FEV1/FEV6 to FEV1/FVC. However, 53% of their patients had severe (35%) or moderate (18%) obstruction. Their conclusion, stated in table 6⇓, might not be valid in populations with a lower severity of airway obstruction. Enright et al. 4 followed over 2,800 smokers and concluded that the FEV1/FEV6 was nearly as strong a predictor of decline in function in smokers as FEV1/FVC. Without giving statistical evidence, they stated that “use of the FEV1/FEV6 is a good substitute for the FEV1/FVC when screening smokers for the presence of airways obstruction” 4. A large study from another academic hospital laboratory 5 used patients with an overall incidence of 12.9% for severe obstruction, 12.1% for moderate obstruction and 13.3% for mild obstruction, and 95% confidence limits to define abnormality. They concluded that “the FEV1/FEV6 ratio can be used a valid alternative for FEV1/FVC in the diagnosis of airway obstruction, especially for screening purposes in high-risk populations for COPD [chronic obstructive pulmonary disease] in primary care”. Their re-analysis of the same data 6, using a fixed ratio of FEV1/FVC <70% versus a selected FEV1/FEV6 of <73%, was remarkably similar. The last study 7, which used excellent equipment and technicians in industrial settings, presented similar findings. However, the higher incidence of false positives than false negatives is surprising, since eliminating flow after 6 s would favour a finding of false-negative FEV1/ FEV6 ratios, as in the present study.
Studies comparing forced expiratory volume in six seconds(FEV6) to forced vital capacity (FVC)
First, the differences in the populations studied will be considered. In the NHANES III database, adult never-smokers outnumbered current smokers. Furthermore, <20% of current smokers had an FEV1/FVC below 95% LLN and only 3% had obstruction considered moderate or severe. This incidence is much lower than that of 53% and 25% of moderate-to-severe obstruction found in the two university hospital patient studies (table 6⇑). The present authors believe that NHANES III smokers and nonsmokers in their study better represent the USA (or other) general populations likely to request or receive spirometric screening by primary care physicians or other providers. The NHANES III analyses disclose that discord increases as severity of obstructive airways declines (fig. 3⇑) and average sensitivities (table 4⇑) fall to 77 and 67% at 95% LLN and to 66 and 54% at 99% LLN for FEV1/FEV6 and FEV3/FEV6, respectively, for all NHANES III smokers, with even lower sensitivities for older smokers.
Secondly, as late flow occurs when longer time-constant lung units play a more prominent role in expiratory airflow, it is not surprising that exclusion of late flow by terminating flow, volume and ratio measurements at 6 s, causes low sensitivities and false negatives. As has long been recognised and recently re-emphasised 14, patients with airway obstruction frequently have slow or unforced vital capacity volumes exceeding those of forced manoeuvres. The fact that nearly one-quarter to one-third of smokers with airway obstruction as discerned using FEV1/FVC and FEV3/FVC at 95% LLN and one-third to one-half at 99% LLN are excluded by substituting FEV1/FEV6 and FEV3/FEV6 should curb enthusiasm for use of the latter measures to detect obstructive disease in a general population, including smokers.
Reasons for false positives in the present study using FEV1/FEV6 and FEV3/FEV6 are less obvious. On review of the present data, the correlations with age are inferior to those for FEV1/FVC and FEV3/FVC ratios for each ethnic and sex group. Perhaps more importantly, se values for FEV6 ratios are invariably, but minimally, lower than se values for FVC ratios for every ethnic and sex group. This results in defining narrower “windows” of abnormality, so that ratios using FEV6 may “find” airway obstruction outside those windows when it is not present.
It also appears (fig. 2⇑) that FEV1/FVC identifies airway obstruction slightly less often than FEV3/FVC, especially in older smokers. Confirming that both ratios detect deterioration with smoking, the present authors previously found that, by middle age, both FEV1/FVC and FEV3/FVC values of current smokers are similar to those of never-smokers who are 20 yrs older 10. Excluding FEV3/FVC from spirometric analyses misses some airway obstruction.
Possible limitations
As in all prior studies comparing values and ratios of FEV6 to FVC, sharp cut-off lines were used in order to distinguish the actual differences between the equations. Although it could be argued that sharp cut-off lines are inappropriate, differences between equations cannot be detected and statistically analysed without using such limits. As populations have a greater variability of FVC or FEV1 than their ratios, subjects with abnormal ratios and FEV1 within normal limits (i.e. 80–120% pred) could be normal or be minimally obstructed, but have a higher morbidity 14.
Perspective
A recent publication 15 reviewed, discarded, selected and analysed a large number of past studies based on clinical evaluation, spirometry and questionnaires to evaluate the effect of multiple therapies on patients with or suspected of having obstructive lung disease. It concluded that: “80 percent of adults reporting a clinical diagnosis of chronic bronchitis or emphysema did not have current airflow obstruction”, spirometry increased 1-yr smoking cessation quit rates by only 1% and “COPD treatment trials including inhaled medications, pulmonary rehabilitation, disease management, or surgery, improved […] functional status […] less than considered clinically significant”. After stating that spirometry was a useful diagnostic tool in evaluating individuals with symptoms suggestive of COPD, Wilt et al. 15 concluded that: “spirometric testing is likely to label a large number of individuals (many who do not report respiratory symptoms) with disease and result in considerable testing and treatment costs and healthcare resource utilization”. In reaching their conclusions, which might not find agreement from other pulmonologists, it must be noted that many of their referenced studies inappropriately used fixed ratios of FEV1/FVC as criteria for obstruction, rather than ratios dependent on age.
Eaton et al. 16 placed quality spirometers in 30 primary care practices and assessed the results. They found: 1) 2 h of physician and nurse training, and further experience were important in improving quality of tracings; 2) spirometric manoeuvres were commonly terminated prematurely; 3) even with training it was rare to get two (33%) or three (19%) blows meeting ATS criteria; 4) primary care physician interpretations were deemed to be correct only 53% of the time; and 5) only an average of 2.3 tests were performed weekly at each site. Nevertheless, the practitioners believed that 13% of the tests helped in counselling smokers. To place these findings in perspective, one might ask if 13% of radiographs, electrocardiograms, or mammograms performed and interpreted in a primary care practice are helpful in counselling patients. The present authors wonder whether physicians should rely on tests from equipment and personnel used so infrequently.
Despite the recommendation of the Global Initiative for Chronic Obstructive Lung Disease Committee 17, the present authors believe that it is unwise to ignore age and use fixed ratios of FEV1/FVC such as 70% to identify airway obstruction. Rather than identifying 5% of normal individuals at the 95% confidence limits as abnormal, such fixed limits are certain to underdiagnose airway obstruction in younger individuals and overdiagnose airway obstruction in older individuals.
Therefore, there are diverse approaches and recommendations regarding spirometry. On the one hand, there is the desire to reduce costs by having minimally trained personnel use simpler and cheaper equipment, with emphasis on measurement of FEV1, FEV6 and FEF25–75% and subsequent interpretation at the primary care level. On the other hand, there are concerns that spirometry is costly and of limited value in detecting early lung disease, reducing the incidence of smoking, or following the effect of therapy in those with known lung disease. The present authors favour a third approach: referral of patients with pulmonary symptoms or a significant smoking history to sites where well-trained personnel with excellent equipment test a large number of patients per day, measuring FEV1, FEV3, FVC, and their ratios, adding, when indicated, measurements of the slow vital capacity, inspiratory capacity, expiratory reserve volume, inspiratory flow and total lung capacity, with interpretation of the values and tracings by experienced pulmonologists or other similarly well-trained physicians. In the present authors' opinion, this latter approach would be cost-effective, result in more accurate diagnoses and be in everyone's best interests.
It is concluded that quality spirograph measures, with proper reference standards, must be used to accurately identify airway obstruction. The perceived benefit of terminating forced expiratory manoeuvres at 6 s discards data from the most obstructed lung units and reduces the sensitivity of detection of obstructive lung disease.
- Received November 22, 2005.
- Accepted January 25, 2006.
- © ERS Journals Ltd
References
