|Home | About | Journals | Submit | Contact Us | Français|
To develop a more age-appropriate spirometric definition of chronic obstructive pulmonary disease (COPD) among older persons.
Using data from the Third National Health and Nutrition Examination Survey (NHANES III), we developed a two-part spirometric definition of COPD in older persons, aged 65–80 years, that 1) determines a cut-point for the ratio of forced expiratory volume in 1 second to forced vital capacity (FEV1/FVC) based on mortality risk; and 2) among persons below this critical FEV1/FVC threshold, determines cut-points for the FEV1, expressed as a standardized residual percentile (SR-tile) and based on the prevalence of respiratory symptoms and mortality risk. Measurements included spirometry, health questionnaires, and mortality (National Death Index).
There were 2,480 older participants with a mean age of 71.7 years; 1,372 (55.4%) had a smoking history, 1,097 (44.2%) had respiratory symptoms and, over the course of 12-years, 868 (35.0%) had died. Among participants with an FEV1/FVC < .70 and FEV1 < 5th SR-tile, representing 7.7% of the cohort, the risk of death was doubled (adjusted hazard ratio, 2.01; 95% confidence interval [CI], 1.60–2.54). Among participants with an FEV1/FVC < .70 and FEV1 < 10th SR-tile, representing 13.4% of the cohort, the prevalence of respiratory symptoms was elevated (adjusted odds ratio, 2.44; CI, 1.79–3.33).
In a large, nationally representative sample of community-living older persons, defining COPD based on an FEV1/FVC < .70, with FEV1 cut-points at the 10th and 5th SR-tile, identifies individuals with an increased prevalence of respiratory symptoms and an increased risk of death, respectively.
Chronic obstructive pulmonary disease (COPD), a leading cause of disability and death worldwide, is defined by chronic airflow limitation that is not fully reversible [1–9]. The airflow limitation is established spirometrically, based solely on a reduced ratio of forced expiratory volume in 1 second (FEV1) to forced vital capacity (FVC), with severity subsequently categorized according to the FEV1 [2–9]. This strategy for defining COPD has a physiological basis because airflow limitation increases the risk of death and usually precedes the onset of dyspnea or exercise intolerance [10,11]. In contrast, defining COPD on the basis of disease criteria alone (e.g. chronic bronchitis) lacks diagnostic accuracy, because cough and sputum production may occur in the absence of airflow limitation and are not independent predictors of death [10–12].
Current spirometric guidelines are potentially problematic, however, for at least three reasons. First, the threshold that establishes a reduced FEV1/FVC remains controversial [6–8]. For example, the Global Initiative for Obstructive Lung Disease (GOLD) and the American College of Physicians (ACP) advocate an FEV1/FVC threshold based on a fixed-ratio [4,13], but this by itself cannot distinguish clinically-significant pathology from normal age-related increases in airflow limitation . Alternatively, the American Thoracic Society (ATS) and the European Respiratory Society (ERS) advocate an FEV1/FVC threshold based on the lower limit of normal (LLN) , but this is derived from regression equations that have limited explanatory ability when applied to an elderly population (i.e., R2 range of .03 to .08) [8,14,15]. Second, the present method of reporting the FEV1 as percent predicted (%Pred) is fundamentally flawed, because it does not account for differences in the variability of the reference group across the lifespan [16,17]. As a result, a given FEV1 %Pred value is not equivalent for all persons regardless of age, height, sex, and ethnicity [16,17]. Third, current spirometric guidelines, in general, lack clinical validation regarding important measures such as mortality and respiratory symptoms [4,9,13].
These problems with current spirometric guidelines are most evident among older persons. Specifically, conventional cut-points for the FEV1/FVC ratio yield prevalence rates for COPD (in the U.S.) that range widely from 16% to 42%, with most affected individuals subsequently categorized by the FEV1 %Pred as having “mild” COPD [6,7]. Hence, current spirometric guidelines may lead to misclassification (mainly over-diagnosis) of COPD among older persons and, in turn, potentially compromise patient care [7,13,18].
As an alternative spirometric strategy, we propose that COPD be defined by a two-step process that 1) determines a cut-point for the FEV1/FVC based on mortality risk; and 2) among persons below this critical FEV1/FVC threshold, determines cut-points for the FEV1, expressed as a standardized residual percentile (SR-tile) and based on the prevalence of respiratory symptoms and mortality risk. An SR-tile is simply a Z-score that has been converted to a percentile [16,17], and is analogous to what is currently reported for bone mineral density testing . Importantly, the SR-tile method accounts for variability in age, height, sex, and ethnicity [16,17]. In the present study, we have applied our strategy for defining COPD in a large nationally representative sample of community-living older persons, which included a large proportion of women and minorities, and discuss how this approach is more clinically meaningful than those provided in published guidelines [4,9,13].
We used data from the Third National Health and Nutrition Examination Survey (NHANES III), a nationally representative sample of Americans assembled in 1988–1994, with mortality surveillance through December 31, 2000 [20,21]. Our source population included 2,480 community-living NHANES III participants, aged 65–80 years, who were white, African-American, or Mexican-American, had no self-reported asthma, and had completed a health questionnaire, a brief cognitive assessment, and at least two ATS acceptable spirometric maneuvers . As per current ATS recommendations, we did not exclude participants based on spirometric reproducibility criteria .
As described elsewhere , NHANES III recorded the presence of respiratory symptoms in the prior 12 months, including chronic cough or sputum production lasting 3 or more consecutive months, dyspnea on exertion, and wheezing or “whistling in the chest”. Other clinical data included chronic conditions, including self-reported, physician-diagnosed COPD and asthma, as well as smoking history, self-reported health status, body mass index (BMI), and cognitive function. Memory impairment was defined as a score less than 2 on delayed recall of a 3-item word list or a score less than 4 on delayed recall from a 6-item story.
As shown in the appendix, and following current guidelines , we determined regression equations in an NHANES III reference group of age-matched never-smokers, with no chronic conditions and no respiratory symptoms. Also as per guidelines , we used age and height as predictor variables for FEV1 and FVC within each sex and ethnic group. Our subsequent procedures, however, differed from those of prior studies that calculated the FEV1 as %Pred (i.e., measured/predicted × 100) [4,5,6,8]. Instead, in a reference group of age-matched, healthy never-smokers, we calculated the constant that quantified the spread of the reference data, namely the standard deviation of the residuals. (A “residual” is the difference between a measured and predicted value.) Based on the regression equations and the spread of the reference data, we then expressed the measured FEV1 of each participant as a standardized residual (SR), calculated as [(measured-predicted)/(standard deviation of the residuals)]; a percentile based on the SR was subsequently computed (SR-tile) [16,17]. To illustrate, an SR of −1.64 corresponds to the 5th percentile, an SR of 0 corresponds to the 50th percentile, and an SR of 1.64 corresponds to the 95th percentile [16,17]. By standardizing the residuals to the spread of the reference data, the SR method accounts for variability due to age, sex, height, and ethnicity and reports FEV1 SR-tile values on an easy-to-interpret scale of 0–100, whereas %Pred does not [16,17].
We next classified participants based on their SR-tile values for FEV1, using cut-points established for the FEV1 %Pred (80, 50, and 30) by the Global Initiative for Obstructive Lung Disease (GOLD) . We modified these cut-points to 50, 30, 10, and 5 SR-tiles because only a small number of NHANES III participants with FEV1/FVC values below the critical threshold, as defined below, had an FEV1 ≥ 80th SR-tile (n=21) whereas a large number had an FEV1 < 30th SR-tile (n=321).
Lastly, we calculated the LLN for the FVC, to help define the referent group for our mortality risk analysis, i.e., the FVC criterion was used to exclude persons with restrictive lung disease .
We do not report the LLN for the FEV1/FVC because predicted values could not be determined reliably. In contrast to the FEV1 and FVC, regression equations for the FEV1/FVC had very low R2 values, ranging from 0.01–0.07, with only one exception (the value for Mexican-females was 0.15). This lack of explanatory capacity persisted when we analyzed the ratio of the FEV1 to the forced expiratory volume in 6 seconds (FEV6), which yielded R2 values of 0.01–0.08. These results are similar to those published in other cohorts of older persons [14,15].
Our primary outcome was all-cause mortality, ascertained from a public-use linked mortality file that contains information based on the National Death Index, with follow-up through December 31, 2000 . Vital status was available on all but one participant .
SUDAAN version 9.0.1 (Research Triangle Park) was used to estimate hazard ratios (Cox proportional hazards analysis) and odds ratios (logistic regression), with a p-value < 0.05 (two-sided) denoting statistical significance . Proportional hazards assumptions for survival models were tested using interaction terms crossing the time-to-event outcome with each variable in the multivariable model. If significant at the 0.05 level, these interaction terms were retained in the final model. The incidence rate and 95% confidence interval (CI) for death were expressed in person-years. Participants who had not died were censored at the end of the follow-up period.
As a first step, we varied the cut-point for the FEV1/FVC in 0.05 decrements below a ratio of 0.80, and fit separate Cox regression models for all-cause mortality, each adjusted for age, height, sex, ethnicity, smoking history, BMI, chronic conditions, self-reported health status, and cognition. Higher order effects were tested for the continuous covariates and were included in the final models if they met the forward selection criterion of p < 0.20 . The first FEV1/FVC cut-point below 0.80 that yielded a statistically significant increase in the risk of death was termed the “critical” threshold.
Next, among participants with an FEV1/FVC below the critical threshold, we evaluated time-to-death using the Kaplan-Meier method with strata defined according to FEV1 cut-points, expressed as SR-tile. We then determined the independent association between the FEV1 SR-tile cut-points and death using Cox proportional hazards, adjusted for the same covariates as described earlier. The FEV1 SR-tile cut-points were treated as nominal categories, with the referent group including participants with normal pulmonary function, defined by an FEV1/FVC ≥ critical threshold and an FVC ≥ LLN. Similarly, we also evaluated the associations between FEV1 SR-tile cut-points and the presence of respiratory symptoms, by calculating odds ratios using logistic regression.
To enhance the interpretation of our findings, we evaluated the prevalence of smoking exposure and reduced health status according to the FEV1 SR-tile cut-points among participants with an FEV1/FVC below the critical threshold, and we calculated the FEV1 %Pred values that corresponded to these SR-tile cut-points.
As shown in Table 1, our study population had a mean age of approximately 72 years, with a majority being current or former smokers. The five most common self-reported chronic conditions were hypertension, arthritis, diabetes mellitus, myocardial infarction, and chronic obstructive pulmonary disease. About a third of the participants had fair-to-poor health status and memory impairment, respectively, and 44.4% endorsed respiratory symptoms. Over the 12-year follow-up period, 868 (35.0%) participants died, yielding a mortality rate of 4.6 per 100 person-years (95% confidence interval [CI] 4.3, 4.9).
Table 2 provides adjusted hazard ratios for all-cause mortality according to the FEV1/FVC. An increased mortality risk was first observed at a cut-point below .70, with an adjusted hazard ratio of 1.23 (95%CI, 1.03–1.47). Figure 1 displays Kaplan-Meier survival curves among participants with an FEV1/FVC < .70, stratified by FEV1 SR-tile. Survival was notably lower for participants with an FEV1 < 5th SR-tile. As shown in Table 3, a threshold effect for mortality was observed at an FEV1 < 5th SR-tile, with an adjusted hazard ratio of 2.01 (95%CI, 1.60–2.54), representing 7.7% of the cohort.
Table 4 reports adjusted odds ratios for respiratory symptoms, based on FEV1 SR-tile among participants with an FEV1/FVC < .70 versus those with normal pulmonary function. A graded relationship was observed, with FEV1 stages at < 5th SR-tile and 5th–9th SR-tile, respectively, demonstrating statistically significant elevations in the odds ratio. When these two stages were combined, i.e., FEV1 < 10th SR-tile, the adjusted odds ratio for respiratory symptoms was 2.44 (95%CI, 1.79–3.33), representing 13.4% of the cohort.
Table 5 provides the mean values for the FEV1/FVC and the prevalence of ever-smokers and reduced health status according to FEV1 SR-tile stages among participants with an FEV1/FVC < .70. Graded relationships were observed, with an FEV1 < 5th SR-tile having the lowest mean value for the FEV1/FVC and the highest prevalence of ever-smokers and reduced health status, respectively.
Table 6 provides the FEV1 %Pred values that corresponded to the 10th and 5th SR-tiles, respectively, for a person of average height. For each of these SR-tile cut-points, the %Pred values varied considerably based on age, ethnicity, and sex. For example, a 65-year old male of average height had an FEV1 value at the 5th SR-tile that ranged from 51%Pred in African-Americans to 69%Pred in white Americans.
In a large, nationally representative sample of community-living persons aged 65–80 years, we found that a two-part spirometric definition of COPD that includes an FEV1/FVC < .70, with FEV1 cut-points at the 10th and 5th SR-tile, identifies individuals with an increased prevalence of respiratory symptoms and an increased risk of death, respectively.
Our strategy for defining COPD has a strong clinical and physiological basis. First, we have established spirometric cut-points that are associated with important clinical measures . All-cause mortality is considered the most definitive health outcome, and was the primary endpoint in landmark studies of oxygen therapy in COPD . In addition, respiratory symptoms are the most common and distressing feature of COPD . Second, we have evaluated the FEV1 as an SR-tile rather than %Pred [4,9,13]. Because it does not consider the spread of the reference data, reporting the FEV1 as %Pred incorrectly assumes that a given value is equivalent for persons of different age, height, sex, and ethnicity [16,17]. The SR-tile method yields, however, a value that is applicable to all persons because it considers the spread of the reference data [16,17]. To illustrate, for an NHANES III participant of average height, we found that FEV1 values at the 5th SR-tile varied from 51%Pred to 77%Pred, depending on the ethnicity, sex, and age of the individual; see Table 6. In contrast to our proposed strategy, current guidelines for COPD recommend spirometric cut-points that have not been clinically validated among older persons, and report the FEV1 as %Pred [4,9,13].
Our results also help to clarify the best approach for establishing a critical threshold for the FEV1/FVC ratio, as part of a spirometric definition of COPD. We found that a fixed-ratio at a threshold of .70 is associated with an increased risk of death, which is consistent with GOLD guidelines . Among persons with an FEV1/FVC ratio below this critical threshold, we established COPD—in contrast to GOLD and other spirometric guidelines [4,9,13]—on the basis of FEV1 SR-tile cut-points that were defined by mortality risk and the prevalence of respiratory symptoms. Using this strategy, 7.7% of our study population had a severe form of COPD, defined by an FEV1/FVC < .70 and an FEV1 < 5th SR-tile, which conferred an increased risk of death and an increased prevalence of respiratory symptoms. This subgroup also had the highest prevalence of smoking exposure and reduced health status, and the lowest mean value for the FEV1/FVC. Participants with an FEV1/FVC < .70 and an FEV1 at the 5th to 9th SR-tile, representing 5.7% of the study population, had a milder form of COPD, which conferred an increased prevalence of respiratory symptoms but not an increased risk of death. This subgroup had the second highest prevalence of smoking exposure and reduced health status, and the second lowest mean value for the FEV1/FVC.
Because neither the risk of death nor prevalence of respiratory symptoms was elevated, we would argue that participants with an FEV1/FVC < .70 but an FEV1 ≥ 10th SR-tile have airflow limitation, but not COPD. Although longitudinal studies are needed, this latter group may be heterogeneous, including persons who simply have normal age-related increases in airflow limitation and those who will experience declines in pulmonary function over time (i.e., transition to COPD) .
As compared with current guidelines,4,9,13 our two-part spirometric definition posits that a reduced FEV1/FVC, although necessary for defining airflow limitation, is insufficient to establish a diagnosis of COPD. Developmentally, after achieving peak pulmonary function at about 20 years of age, airflow limitation increases with age , principally due to increasing rigidity of the chest wall and decreasing elastic recoil of the lung . Although COPD is also characterized by airflow limitation, this effect is due to small airways’ disease and parenchymal destruction [4,36]. Consequently, a reduced FEV1/FVC as a measure of airflow limitation can be caused by normal aging and/or clinically-significant pathology [7,8]. To make this distinction, it is necessary to consider the FEV1 because it is associated with COPD-related airway inflammation in persons with established airflow limitation . Consequently, among older persons, our spirometric strategy likely reduces potential misclassification of COPD, as evidenced by an overall prevalence (13.4%) that is more clinically realistic than that generated by GOLD guidelines (34.5%).
If routinely implemented, our spirometric strategy for defining COPD could lead to improvements in patient care. For example, with GOLD guidelines, exercise intolerance in an older patient who has coexisting cardiopulmonary risk factors, anemia, and a history of physical inactivity might be attributed primarily to COPD instead of to one or more of the patient’s comorbid conditions. Because cut-points are based on clinical measures relevant to COPD, our spirometric strategy could lead to a more targeted use of COPD-specific pharmacotherapy and, as a result, reduce the frequency of medication-related adverse events .
We recognize potential limitations to our study. First, a threshold based on the LLN for the FEV1/FVC could not be reliably determined. Unlike the FEV1 and FVC, regression equations for the FEV1/FVC lack explanatory ability in older persons. This may be attributable to age-related increases in lung function variability, leading to normal values for the FEV1/FVC that range widely and are significantly skewed — in comparison to the FEV1 and FVC alone . Moreover, because prior work has shown that elderly persons with an FEV1/FVC < .70 but > LLN have an increased risk of death , the clinical validity of the LLN is uncertain. Our inability to calculate the LLN of the FEV1/FVC likely had little effect on our estimate of COPD prevalence, since our strategy relies ultimately on the FEV1 SR-tile to establish a diagnosis.
Second, the cut-points that we identified for the FEV1/FVC and FEV1 SR-tile were based on a single cohort of older persons. Further validation of these cut-points, with an expanded array of clinically relevant outcomes (including health care utilization), will be required in other cohorts of older persons, as well as in other age groups.
Third, evaluating an outcome based on all-cause mortality rather than COPD-specific mortality might be considered a limitation. Prior work has shown, however, that COPD is commonly underreported as a cause of death, even among patients with symptomatic COPD . Furthermore, COPD increases the risk of death from cardiovascular disease and lung cancer, and the number of deaths from these causes is much greater than those from respiratory disease among patients with COPD [28–30].
Fourth, our spirometric definition of mild COPD is based on the presence of respiratory symptoms, which are not necessarily specific to COPD . Nonetheless, respiratory symptoms are the most distressing feature of COPD and can lead to disability and increased healthcare utilization [25,37]. In addition, the graded relationship observed for the presence of respiratory symptoms across the FEV1 SR-tile stages (Table 4) enhances the validity of our spirometric definition of COPD.
Finally, because spirometry in NHANES III was not obtained after a bronchodilator, we could not assess reversibility, a recommended criterion for defining COPD . It is unlikely, however, that the absence of information on “reversibility” had a meaningful effect on our results, because 1) persons with self-reported asthma were excluded, 2) prior work has shown that only 3% of abnormal prebronchodilator FEV1/FVC ratios normalize in response to a bronchodilator [31,32], and 3) bronchodilator reversibility, as defined by the FEV1 response, is neither a sufficient criterion to exclude COPD nor an independent predictor of mortality [33,34]. Nevertheless, we acknowledge that our spirometric analyses may have included a small number of participants with asthma, either as a sole form of obstructive airways disease or comorbidity.
In conclusion, among older persons, we are proposing a revised spirometric definition of COPD that identifies individuals with an increased prevalence of respiratory symptoms and an increased risk of death. If confirmed in other populations of older persons, this definition of COPD could lead to improvements in patient care.
The work for this report was supported by the Atlantic Philanthropies, and the ASP/CHEST and Hartford Foundations. The study was conducted at the Yale Claude D. Pepper Older Americans Independence Center (P30AG21342). Dr. Fragoso is currently a recipient of an ASP/CHEST Foundation Geriatric Research Development Award and of a Career Development Award from the Department of Veterans Affairs. Dr. Concato is supported by the Department of Veterans Affairs Cooperative Studies Program. Dr. Van Ness received support from the Claude D. Pepper Older Americans Independence Center at Yale (2P30AG021342-06). Dr. Yaggi is supported by a career development transition award from the Department of Veterans Affairs Clinical Science Research and Development Service. Dr. Gill is the recipient of an NIA Midcareer Investigator Award in Patient-Oriented Research (K24AG021507). The investigators retained full independence in the conduct of this research and report no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.