- Email: [email protected]
Development and Validation of a Survey-Based COPD Severity Score
clinical investigations Development and Validation of a Survey-Based COPD Severity Score* Mark D. Eisner, MD, MPH, FCCP; Laura Trupin, MPH; Patricia P. Katz, PhD; Edward H. Yelin, PhD; Gillian Earnest, MS; John Balmes, MD, FCCP; and Paul D. Blanc, MD, MSPH, FCCP
Objective: To develop a comprehensive disease-specific COPD severity instrument for survey-based epidemiologic research. Study design and setting: Using a population-based sample of 383 US adults with self-reported physiciandiagnosed COPD, we developed a disease-specific COPD severity instrument. The severity score was based on structured telephone interview responses and included five overall aspects of COPD severity: respiratory symptoms, systemic corticosteroid use, other COPD medication use, previous hospitalization or intubation, and home oxygen use. We evaluated concurrent validity by examining the association between the COPD severity score and three health status domains: pulmonary function, physical health-related quality of life (HRQL), and physical disability. Pulmonary function was available for a subgroup of the sample (FEV1, n ⴝ 49; peak expiratory flow rate [PEFR], n ⴝ 93). Results: The COPD severity score had high internal consistency reliability (Cronbach ␣ ⴝ 0.80). Among the 49 subjects with FEV1 data, higher COPD severity scores were associated with poorer percentage of predicted FEV1 (r ⴝ ⴚ 0.40, p ⴝ 0.005). In the 93 subjects with available PEFR measurements, greater COPD severity was also related to worse percentage of predicted PEFR (r ⴝ ⴚ 0.35, p < 0.001). Higher COPD severity scores were strongly associated with poorer physical HRQL (r ⴝ ⴚ 0.58, p < 0.0001) and greater restricted activity attributed to a respiratory condition (r ⴝ 0.59, p < 0.0001). Higher COPD severity scores were also associated with a greater risk of difficulty with activities of daily living (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.8 to 3.0) and inability to work (OR, 4.2; 95% CI, 3.0 to 5.8). Conclusion: The COPD severity score is a reliable and valid measure of disease severity, making it a useful research tool. The severity score, which does not require pulmonary function measurement, can be used as a study outcome or to adjust for disease severity. (CHEST 2005; 127:1890–1897) Key words: chronic bronchitis; pulmonary disease, chronic obstructive; outcome assessment (health care); pulmonary emphysema; severity of illness index Abbreviations: CI ⫽ confidence interval; HRQL ⫽ health-related quality of life; NHIS ⫽ National Health Interview Survey; OR ⫽ odds ratio; PEFR ⫽ peak expiratory flow rate; SF ⫽ Short-Form
is a major cause of impaired health status, C OPD disability, and mortality. The disease is respon-
sible for ⬎ 2 million emergency department visits and hospitalizations annually in the United States.1 Despite public awareness about the health risks of smoking, mortality from COPD continues to increase.1 Consequently, there is an urgent need for
epidemiologic studies that better delineate the risk factors for adverse health outcomes among adults with COPD. The conduct of such studies has been hampered by the lack of a validated disease-specific COPD severity measure beyond the basic physiologic assessment of pulmonary function. A comprehensive survey-based measure of COPD severity
*From the Department of Medicine (Drs. Eisner, Balmes, and Blanc, Ms. Trupin and Ms. Earnest), University of California, San Francisco; of and Institute for Health Policy Studies (Drs. Katz and Yelin), University of California, San Francisco, San Francisco, CA. Support was provided by grant R01 HL607438 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and Flight Attendants Medical Research Institute grant CoE2001. Dr. Eisner was also supported by grant K23 HL04201 from the National Heart, Lung, and Blood Institute.
Manuscript received August 20, 2004; revision accepted December 21, 2004. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Mark D. Eisner, MD, MPH, FCCP, University of California, San Francisco, 350 Parnassus Ave, Ste 609, San Francisco, CA 94117; e-mail: [email protected]
1890
Clinical Investigations
would be valuable for use in epidemiologic studies as a health outcome or to adjust for disease-specific severity. Pulmonary function measurement, especially FEV1, has traditionally been used to stage the severity of COPD in clinical practice; the recent Global Initiative for Chronic Obstructive Lung Disease has codified this practice in its guidelines.2– 6 Although pulmonary function is an essential clinical tool for characterizing COPD and is a potent predictor of mortality,7 it is not a comprehensive disease-severity measure. FEV1, for instance, correlates weakly with other measures of health-related quality of life (HRQL) and respiratory symptoms.8 –10 Moreover, FEV1 did not capture important treatmentrelated clinical improvements, manifested by fewer COPD exacerbations, in clinical trials of inhaled corticosteroids.11–15 In addition, spirometry can be difficult and expensive to perform in large-scale epidemiologic studies. HRQL instruments, both generic and diseasespecific, have also been used to gauge the severity of COPD.8,16,17 HRQL, however, is conceptually different from disease severity. HRQL is a patient-centered measurement of perceived satisfaction with life, as it is affected by health.18 COPD severity is a distinct construct that reflects the biological impact of disease pathophysiology on diverse aspects of physical functioning. Consequently, COPD severity is proximal to and influential of HRQL. Although HRQL instruments provide an important patientlevel assessment, they cannot substitute for a specific measure of COPD severity. We developed a comprehensive disease-specific COPD severity score for use in epidemiologic and health-outcomes research. The score, which was designed for survey administration, is an integrated measure of disease severity, based on disease status, receipt of clinical treatments, and recent hospitalization for COPD. Using a population-based sample of US adults with COPD, we assessed the reliability and validity of the COPD severity score. Materials and Methods Overview In a population-based sample of 383 US adults with COPD, we developed a disease-specific COPD severity instrument. The severity score was based on responses to a structured telephone interview. Internal consistency reliability was established using standard psychometric techniques. We evaluated concurrent validity by examining the association between the COPD severity score and three aspects of health status: pulmonary function, physical HRQL and health status, and physical disability. Pulmonary function measurements were obtained for a subset of respondents subsequent to the interviews. www.chestjournal.org
Recruitment of Adults With COPD Survey methods have been previously described in detail.19 The study was approved by the University of California, San Francisco Committee on Human Research. Briefly, 383 adults with COPD were selected from a random sample of 2,061 US adults aged 55 to 75 years identified by random-digit dialing telephone interviews. Approximately one half of the overall sample (n ⫽ 1,001) was randomly identified among residents in the 48 contiguous states of the United States. The remainder of the sample (n ⫽ 1,060) was recruited from persons who resided in geographic “hot spots” that had the highest COPD mortality rates based on the National Institute of Occupational Safety and Health Atlas of Respiratory Disease Mortality in the United States: 1982–1993.20 The hot-spot sample was enriched for subjects with COPD. The overall study participation rate was 53% among households with an eligible respondent present. During each telephone contact, one randomly selected adult per household was interviewed. Subjects were asked if they had ever received a physician’s diagnosis of any of several chronic respiratory conditions. Those who reported physician diagnoses of chronic bronchitis or emphysema were considered to have COPD, along with those who specifically reported a diagnosis of COPD. We included respondents with COPD who had concomitant asthma because they clinically resemble persons with COPD alone.21 Development of the COPD Severity Score Participants underwent structured telephone interviews that elicited respiratory symptoms and medications, health status, smoking history, employment history, and sociodemographic characteristics. Survey-item responses were used to construct a disease-specific COPD severity score. The instrument and scoring instructions are available from the authors, on request. We previously developed a disease-specific severity-of-asthma score for use in epidemiologic and outcomes research.22,23 Using a similar process, we created a COPD severity score based on responses to survey items that comprise five overall aspects of COPD severity: respiratory symptoms, systemic corticosteroid use, other COPD medication use, previous hospitalization or intubation for respiratory disease, and home oxygen use (Table 1). Each item was assigned an a priori weight based on clinical aspects of the disease and its expected contribution to overall COPD severity. In order to assess the performance of the a priori weighting scheme, we derived an alternate weighting scheme by factor analysis with orthogonal rotation. Possible total scores range from 0 to 35, with higher scores reflecting more severe COPD. The COPD severity score takes into account both current symptom status and the therapy necessary to achieve this status. For example, a person with few respiratory symptoms who is receiving multiple medications for COPD would have greater COPD severity, and a higher COPD severity score, than another person with comparable symptom status and no medication use. In addition, recent prior hospitalization or intubation for COPD are assigned high weights, as they are indicators of poor future health outcomes.24 Oxygen use, because it is recommended for persons with advanced disease,3 was also heavily weighted. Missing values for COPD severity score components were rare, ranging from 0 to 2% per item. We used previously established methods to impute missing values.23 For dyspnea during the past 14 days or nights, we assigned missing values the maximum of either the day or night response (n ⫽ 9 subjects and n ⫽ 5 subjects, respectively; no subjects were missing both values). Respondents who indicated antibiotic use during the past 12 months but who did not indicate the number of courses were CHEST / 127 / 6 / JUNE, 2005
1891
Table 1—COPD Severity Score Items (n ⴝ 383) Items Respiratory symptoms (maximum 7 points) Dyspnea on exertion, current None Hurrying on level ground or walking uphill Walking with peers on level ground Walking at own pace on level ground Dyspnea during the past 14 d or nights None 1 to 2 d or nights 3 to 6 d or nights 7 to 13 d or nights Every day or night Systemic corticosteroid use (maximum 5 points) Ever used Long-term use in past year* Used in past 2 wk Other medication use (maximum 10 points)† Metered-dose inhaler in past 2 wk‡ Short-acting -agonists Long-acting -agonists Inhaled corticosteroids Ipratropium bromide Nebulizer use, past 2 wk‡ Short-acting -agonists Ipratropium bromide Oral medications Theophylline, past 2 wk -Agonists, past 2 wk Antibiotics for lung condition, past 12 mo One to two courses Three or more courses Hospitalization/intubation/oxygen use (maximum 13 points) Hospitalized for COPD, past 5 yr Intubated for COPD, past 5 yr Home oxygen, current day or nighttime use
Score
No. (%)
0 1 2 3
99 (26) 96 (25) 53 (14) 136 (36)
0 1 2 3 4
128 (33) 51 (13) 50 (13) 29 (8) 125 (33)
1 3 1
115 (30) 42 (11) 20 (5)
1 1 1 1
137 (36) 53 (14) 92 (24) 80 (21)
1 1
43 (11) 13 (3)
1 1
18 (5) 20 (5)
1 2
67 (17) 49 (13)
3 5 5
95 (25) 18 (5) 37 (10)
*At least three times per week for at least 3 months during the past 2 years. †Combination medications were counted under both categories. For example, the combined preparation of ipratropium bromide and albuterol sulfate was counted as both short-acting -agonist and an ipratropium bromide metered-dose inhalers. ‡Missing data: 12 subjects (3%) used an metered-dose inhaler, and 7 subjects (2%) used a nebulizer but did not specify the type of medication.
assigned to the most common “one to two courses” category (n ⫽ 9 subjects). Missing values for medication use and other questions were defined as zero (ie, not used or not present) [range, 0 to 2% missing per item]. COPD Severity Score Reliability Internal consistency reliability was evaluated by principal component analysis to examine whether the COPD severity score appeared to measure a single construct. Cronbach ␣ was also calculated as a measure of internal consistency.25 As noted above, we compared the a priori weighting scheme with an alternate weighting scheme derived from factor analysis. Concurrent Validity We assessed concurrent validity of the COPD severity score by evaluating its association with three aspects of health status 1892
defined a priori: pulmonary function, physical HRQL and health status, and physical disability. The variables selected to represent these aspects of health status were based on our theoretical model of COPD severity. We anticipated that greater COPD severity would correlate with impairment in these three dimensions of health status. We used two methods for obtaining pulmonary function measurements among a subset of 264 subjects (70%) who we recontacted in follow-up to the initial interview. Of the 264 subjects, 170 respondents indicated that they had undergone pulmonary function testing during the past 5 years. These subjects were contacted by mail to request permission to retrieve their pulmonary function testing results from their outpatient medical record. The health-care providers for 71 participating subjects were contacted, and 49 pulmonary function reports were obtained. From these reports, we ascertained FEV1 for 49 subjects and peak expiratory flow rate (PEFR) for 44 subjects. In the remaining 24 instances, no record of pulmonary function testing was located by the health-care provider. The mean time difference between the pulmonary function test and the initial interview date, which generated the survey responses for calculation of the COPD severity score, was 0.4 ⫾1.6 years [⫾ SD]). We used the predicted values provided by the pulmonary function testing laboratory where the tests were conducted. Respondents who indicated no recent pulmonary function testing at follow-up were contacted by letter to recruit them for a telephone-based PEFR measurement. Of the 94 eligible subjects, 68 subjects were sent peak flowmeters by mail, and 49 participated in the peak expiratory flowmeter program. Participating subjects were mailed a peak flowmeter (Astech; Dey Laboratories; Napa, CA). One research assistant, who was trained by an investigator with expertise in asthma care, carried out a structured telephone-based program with subjects that lasted approximately 10 min per session. Each subject performed three peak expiratory flow maneuvers according to American Thoracic Society guidelines and similar to other telephone-based peak flow monitoring programs.26,27 The mean time difference between the initial interview and the peak flow measurement was 1.4 ⫾ 0.2 years. Predicted values were based on reference standards developed for the third National Health and Nutrition Examination Survey.28 In total, we had peak expiratory flow data for 93 subjects (44 subjects who underwent pulmonary function testing and 49 subjects who completed the home peak flowmeter program). Physical HRQL was assessed using the Short-Form (SF)-12 physical component summary score. The SF-12 is derived from the Medical Outcomes Study SF-36 instrument, which is the most widely used measure of generic HRQL. The SF-36 has been extensively validated in the general population29 and among adults with COPD.30 Defined from the eight SF-36 subscales by factor analysis, the physical component summary score reflects an underlying physical dimension of physical HRQL.31 Higher scores reflect more favorable health states. As another measure of physical health status, daily activity restriction was ascertained using a question adapted from the National Health Interview Survey (NHIS).32 Specifically, respondents were asked to indicate how many days their activity was limited due to a respiratory condition during the past month. Respondents were also asked to rate the severity of their fatigue or tiredness on a scale of 0 to 10, in which 0 reflected no fatigue and 10 indicated very severe fatigue. Concurrent validity was also assessed using three measures of physical disability and health impacts. Self-rated general health was assessed with a question developed for the NHIS and also used in the SF-36 questionnaire.29 Based on this item, we defined adverse physical health status as self-reported fair or poor general health (as opposed to good, very good, or excellent). Although Clinical Investigations
this question comprises one aspect of the SF-12 physical component summary score, it has independent value as a measure of health status.21 Current employment status was ascertained by questions that were patterned after the Bureau of Labor Statistics Current Population Survey.33 Using these items, we ascertained whether respondents were unable to work secondary to a respiratory condition. Activities of daily living were assessed with the “body care” section of the Functional Performance Inventory SF.34 These activities included dressing or undressing, showering or bathing, caring for feet, washing hair, shaving or applying makeup, and preparing meals. We defined difficulty with activities of daily living as being unable to do the activity or able to do the activity with much difficulty. Statistical Analysis The analysis was conducted using statistical software (Version 8.1; SAS Institute; Cary, NC). Internal consistency reliability was evaluated using principal component analysis and Cronbach ␣. We evaluated concurrent validity by examining the association between the COPD severity score and pulmonary function, physical HRQL/health status, and physical disability. The Spearman rank correlation was used to evaluate the association between the COPD severity score and pulmonary function measures (percentage of predicted FEV1 and percentage of predicted PEFR). Using the Spearman rank correlation, we examined the association between COPD severity score and physical health status in three separate analyses: among subjects with available FEV1 data, among subjects with PEFR data, and among the total cohort. In the first two analyses, we controlled for the level of physiologic impairment using percentage of predicted FEV1 or percentage of predicted PEFR, respectively. This was accomplished by calculating the partial correlation between COPD severity score and each physical health status measure, after adjusting for pulmonary function.35 Using an analogous strategy, we used logistic regression analysis to examine the association between COPD severity score and physical disability. Because sociodemographic factors may be on the causal pathway between COPD severity and health status, we did not control for these variables in multivariate analysis. As a sensitivity analysis, we also repeated the analysis using a narrower definition of COPD that only included subjects reporting emphysema or COPD, excluding those with chronic bronchitis alone.
Table 2—Personal Characteristics of 383 Adults With COPD* Characteristics
Data
Age at interview, yr Age at COPD diagnosis, yr† Male gender White, non-Hispanic race/ethnicity Education‡ High school or less Some college College or graduate school Cigarette smoking Never smoked Current smoker Former smoker
64 ⫾ 6 46 ⫾ 18 138 (36) 336 (88) 202 (53) 105 (28) 73 (19) 74 (19) 126 (33) 183 (48)
*Data are presented as mean ⫾ SD or No. (%). †Twenty-nine individuals did not report age at diagnosis. ‡Two subjects did not report education level.
Reliability Internal consistency was high (Cronbach ␣ ⫽ 0.80). In addition, the principal component analysis revealed loading on a single factor (Eigenvalue 3.3) that contributed 71% of the variance of the model. There was a rapid fall-off in the scree plot, with a second component Eigenvalue of 0.55. Concurrent Validity: Pulmonary Function Measurement of pulmonary function was available for 93 subjects. In the subset of 49 subjects with available FEV1 data, higher COPD severity scores were associated with poorer FEV1 percentage of predicted (r ⫽ ⫺0.40, p ⫽ 0.005) [Table 3]. Among the 93 persons with PEFR measurements, greater COPD severity was also related to worse PEFR percentage of predicted (r ⫽ ⫺ 0.35, p ⬍ 0.001). Concurrent Validity: Physical HRQL and Health Status, Disability, and Health Impacts
Results Subject Characteristics The mean age of subjects with COPD was 64 ⫾ 6 years (Table 2). The majority of subjects had a lifetime history of cigarette smoking (81%). Other sociodemographic characteristics are shown in Table 2. COPD Severity Score The COPD severity score ranged from 0 to 28 (mean, 7.3 ⫾ 6.5; median, 6.0; 25th to 75th interquartile range, 2 to 25). The COPD severity score, which was based on an a priori weighting system, correlated very closely with the score whose alternative weighting was derived from factor analysis with orthogonal rotation (r ⫽ 0.97, p ⬍ 0.0001). www.chestjournal.org
Among the 49 subjects with available FEV1 data, higher COPD severity scores were associated with Table 3—Correlation Between COPD Severity Score and Pulmonary Function
Variables
No.
Mean ⫾ SD
Correlation With COPD Severity Score
FEV1 % predicted* PEFR % predicted†
49 93
54.6 ⫾ 25.4 63.4 ⫾ 26.8
⫺ 0.40 ⫺ 0.35
p Value 0.005 0.001
*Obtained from pulmonary function testing records (see Materials and Methods). †Obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home peak flowmeter measurements only were r ⫽ ⫺ 0.29 (p ⫽ 0.04); results for PEFR obtained using spirometry only were r ⫽ ⫺ 0.27 (p ⫽ 0.07). CHEST / 127 / 6 / JUNE, 2005
1893
poorer physical HRQL (r ⫽ ⫺ 0.37, p ⫽ 0.01) and greater restricted activity due to a respiratory condition (r ⫽ 0.59, p ⬍ 0.001) after controlling for FEV1 (Table 4). In the subgroup with PEFR measurements, higher COPD severity scores were also related to poorer physical HRQL (r ⫽ ⫺ 0.55, p ⬍ 0.001) and greater restricted activity (r ⫽ 0.61, p ⬍ 0.001) after controlling for percentage of predicted PEFR. Among the entire cohort, COPD severity scores were also strongly related to physical HRQL (r ⫽ ⫺ 0.58, p ⬍ 0.001) and restricted activity (r ⫽ 0.59, p ⬍ 0.001). Higher COPD severity scores were also associated with greater fatigue (r ⫽ 0.61, p ⬍ 0.0001). In the subgroup with available FEV1 measurements, the risk of impaired self-perceived general health increased with greater COPD severity scores, after controlling for FEV1 (odds ratio [OR], 2.4 per SD-sized score increment; 95% confidence interval [CI], 1.1 to 5.1) [Table 5]. Higher COPD severity scores were also associated with a greater risk of inability to work secondary to a respiratory condition (OR, 6.7; 95% CI, 2.2 to 20.3) and difficulty performing activities of daily living (OR, 1.7; 95% CI, 0.8 to 3.4). Among those with available PEFR measurements, the COPD severity score was also related to each measure of physical disability. In the overall cohort, higher COPD severity scores were associated with a greater risk of impaired general health (OR, 3.1; 95% CI, 2.3 to 4.1), difficulty performing activities of daily living (OR, 2.3; 95% CI, 1.8 to 3.0), and inability to work (OR, 4.2; 95% CI, 3.0 to 5.8). COPD or Emphysema Subgroup We evaluated the reliability and validity of the COPD severity score among a more restricted subgroup limited to subjects who reported physician-
diagnosed emphysema or COPD (n ⫽ 190), excluding those who indicated chronic bronchitis only. This group had greater COPD severity, as indicated by higher COPD severity scores (mean, 10.0 ⫾ 6.8). As in the overall analysis, internal consistency reliability was high, with a Cronbach ␣ of 0.79. The concurrent validity analyses also provided similar results to the group overall. Higher COPD severity scores correlated with poorer FEV1 percentage of predicted (r ⫽ ⫺ 0.35, p ⫽ 0.04), PEFR percentage of predicted (r ⫽ ⫺ 0.37, p ⫽ 0.008), physical HRQL (r ⫽ ⫺0.47, p ⬍ 0.0001), and restricted activity days due to a respiratory condition (r ⫽ 0.60, p ⬍ 0.0001). Discussion In this population-based study, we developed a survey-based COPD severity score that incorporates clinical aspects of the disease, including respiratory symptoms, oral corticosteroid use, other COPD medication use, previous hospitalization and intubation, and home oxygen therapy. The COPD severity score is internally consistent, reliable, and appears to capture a broad range of disease severity. We have demonstrated the concurrent validity of the COPD severity score using diverse measures of physical health status, disability, and pulmonary function. This severity instrument could prove to be a valuable tool for survey-based research investigating the epidemiology and health outcomes of COPD. The COPD severity score can function as a study outcome variable in studies of risk factors for adverse health impacts, such as environmental exposures or health-care processes. The severity score can also be used to adjust for differences in baseline disease severity in cohort studies. This may be particularly valuable in effectiveness studies of therapeutic inter-
Table 4 —COPD Severity Score as a Predictor of Physical HRQL and Health Status Physical HRQL/Health Status Measure Among subjects with FEV1 data, controlling for FEV1 % predicted SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month Among subjects with PEFR data, controlling for PEFR % predicted† SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month Among entire group SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month
Subjects, No.
Mean ⫾ SD
Relationship With Severity Score (r)
p Value
49 49
32.5 ⫾ 11.7 12.7 ⫾ 14.6
⫺ 0.37 0.59
0.01 0.001
93 93
34.9 ⫾ 12.1 9.5 ⫾ 13.5
⫺ 0.55 0.61
0.001 0.001
383 383
36.2 ⫾ 13.0 7.0 ⫾ 12.1
⫺ 0.58 0.59
0.001 0.001
*Higher scores indicate more favorable health status. †PEFR obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home peak flowmeter measurement only were SF-12 physical component summary score (r ⫽ ⫺ 0.65, p 0.0001) and restricted activity days (r ⫽ 0.59, p 0.0001). Results for PEFR obtained using spirometry were SF-12 physical component summary score (r ⫽ ⫺ 0.43, p ⫽ 0.003) and restricted activity days (r ⫽ 0.65, p 0.0001). 1894
Clinical Investigations
Table 5—COPD Severity Score as a Predictor of Physical Disability Disability Measure Among subjects with FEV1, controlling for FEV1 % predicted General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡ Among subjects with PEFR, controlling for PEFR % predicted§ General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡ Among entire group (n ⫽ 383) General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡
Subjects, No.
Prevalence, No. (%)
Risk of Disability, OR (95% CI)*
49 49 46
27 (55) 18 (37) 15 (33)
2.4 (1.1–5.1) 1.7 (0.8–3.4) 6.7 (2.2–20.3)
93 93 85
49 (53) 24 (26) 24 (28)
2.6 (1.5–4.6) 2.5 (1.4–4.2) 4.0 (2.1–7.6)
383 383 349
193 (50) 89 (23) 82 (23)
3.1 (2.3–4.1) 2.3 (1.8–3.0) 4.2 (3.0–5.8)
*OR expressed per SD-sized increment in COPD severity score. †Unable to do one or more activity of daily living or much difficulty doing the activity. ‡Among subjects with history of any labor force participation. §PEFR obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home monitoring only were general health fair or poor (OR, 4.0; 95% CI, 1.5 to 10.3), difficulty performing activities of daily living (OR, 2.8; 95% CI, 1.3 to 6.2), and unable to work (OR, 3.5; 95% CI, 1.5 to 8.3). Results for PEFR obtained from spirometry were general health fair or poor (OR, 2.3; 95% CI, 1.1 to 5.0), difficulty performing activities of daily living (OR, 1.8; 95% CI, 0.91 to 3.7), and unable to work (OR, 5.5; 95% CI, 1.9 to 16.2).
ventions for COPD. Importantly, the COPD severity score does not require measurement of pulmonary function, which may be major logistical advantage for population-based studies. The methods that we used to recruit adults with COPD had both strengths and limitations. The population-based methods used to identify subjects should ensure the generalizability of the test characteristics of the COPD severity score. The use of self-reported physician diagnosis, however, may result in some misclassification of disease status. This methodology is a standard epidemiologic approach for identifying COPD using survey-based techniques.1 The high prevalence of lifetime smoking, which was ⬎ 80%, supports the diagnosis of COPD. Moreover, the prevalence of COPD in our general population sample (13.5%) was similar to that reported in two other population-based studies conducted in the United States, The third National Health and Examination Survey and the NHIS.1 In addition, the majority of subjects indicated recent respiratory symptoms or COPD medication use (89%), consistent with a diagnosis of COPD. Finally, reanalysis using a more restrictive definition of COPD that excludes chronic bronchitis did not appreciably affect the results. Although the enrolled sample appears similar to COPD patients in general, we cannot exclude some degree of misdiagnosis and misclassification of disease status. Another limitation is the availability of pulmonary function testing on a subset of the overall sample, the use of such data from several pulmonary function laboratories, and the time lag between interview and www.chestjournal.org
pulmonary function measurement. When we examined the subgroups with available pulmonary function measurements, we found strong evidence of reliability and validity. This indicates that estimates of reliability and validity are robust and not subject to selection bias. If the time lag between interview and pulmonary function measurement introduced bias, it would be expected to be a conservative one (ie, toward the null). The consistency of findings across different study outcomes, ranging from pulmonary function to physical health status and disability, also supports the psychometric soundness of the COPD severity score. The COPD severity score was a reliable and valid tool in our cohort study. It is based on survey data, which could be affected by subject misreporting, but has the significant advantage of deployment in largescale epidemiologic studies in which pulmonary function measurement and physical examination are not logistically feasible. The score, however, should be validated in a separate, larger cohort of COPD patients, which we will carry out in a different ongoing cohort study of COPD. In addition, we did not have additional objective measures, such as the 6-min walk test, which will also provide additional validation in our ongoing COPD cohort study. Although the COPD severity score was designed for epidemiologic and health-outcomes research, it could have applications in clinical settings. Before clinical use can be recommended, further validation and comparison with the recently published multidimensional BODE score need to be undertaken,36 which we are planning in an ongoing COPD cohort CHEST / 127 / 6 / JUNE, 2005
1895
study. It will be instructive to compare our COPD severity score, which takes into account both current symptom status and the therapy necessary to achieve this status, with the BODE score, which grades patients based on dyspnea, airflow obstruction, exercise capacity, and body mass index. One key difference is that the BODE score requires in-person physical testing, which could reduce its applicability in large-scale population-based epidemiologic studies based on feasibility concerns. The COPD severity score is a simple, reliable, and valid survey-based instrument that can be used in epidemiologic and health-outcomes studies. Developed for telephone interview administration, the COPD severity score may be adapted to other research settings, such as personal interview or written questionnaire. Future studies will evaluate other psychometric properties, such as longitudinal responsiveness to change in COPD status.
References 1 Mannino DM, Homa DM, Akinbami LJ, et al. Chronic obstructive pulmonary disease surveillance–United States, 1971–2000. MMWR Surveill Summ 2002; 51:1–16 2 BTS guidelines for the management of chronic obstructive pulmonary disease. The COPD Guidelines Group of the Standards of Care Committee of the BTS. Thorax 1997; 52(suppl):S1–S28 3 Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. American Thoracic Society. Am J Respir Crit Care Med 1995; 152:S77–S121 4 Siafakas NM, Vermeire P, Pride NB, et al. Optimal assessment and management of chronic obstructive pulmonary disease (COPD). The European Respiratory Society Task Force. Eur Respir J 1995; 8:1398 –1420 5 Pauwels RA, Buist AS, Calverley PM, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary. Am J Respir Crit Care Med 2001; 163:1256 –1276 6 Fabbri LM, Hurd SS. Global strategy for the diagnosis, management and prevention of COPD: 2003 update. Eur Respir J 2003; 22:1–2 7 Speizer FE, Fay ME, Dockery DW, et al. Chronic obstructive pulmonary disease mortality in six U.S. cities. Am Rev Respir Dis 1989; 140:S49 –55 8 Wijnhoven HA, Kriegsman DM, Hesselink AE, et al. Determinants of different dimensions of disease severity in asthma and COPD: pulmonary function and health-related quality of life. Chest 2001; 119:1034 –1042 9 Prieto L, Alonso J, Ferrer M, et al. Are results of the SF-36 health survey and the Nottingham Health Profile similar? A comparison in COPD patients. Quality of Life in COPD Study Group. J Clin Epidemiol 1997; 50:463– 473 10 Nishimura K, Izumi T, Tsukino M, et al. Dyspnea is a better predictor of 5-year survival than airway obstruction in patients with COPD. Chest 2002; 121:1434 –1440 11 Alsaeedi A, Sin DD, McAlister FA. The effects of inhaled corticosteroids in chronic obstructive pulmonary disease: a systematic review of randomized placebo-controlled trials. Am J Med 2002; 113:59 – 65 1896
12 Burge PS, Calverley PM, Jones PW, et al. Randomised, double blind, placebo controlled study of fluticasone propionate in patients with moderate to severe chronic obstructive pulmonary disease: the ISOLDE trial. BMJ 2000; 320:1297– 1303 13 Lung Health Study Research Group. Effect of inhaled triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease. N Engl J Med 2000; 343: 1902–1909 14 Pauwels R. Inhaled glucocorticosteroids and chronic obstructive pulmonary disease: how full is the glass? Am J Respir Crit Care Med 2002; 165:1579 –1580 15 Verhoeven GT, Hegmans JP, Mulder PG, et al. Effects of fluticasone propionate in COPD patients with bronchial hyperresponsiveness. Thorax 2002; 57:694 –700 16 Hajiro T, Nishimura K, Tsukino M, et al. A comparison of the level of dyspnea vs disease severity in indicating the healthrelated quality of life of patients with COPD. Chest 1999; 116:1632–1637 17 Oga T, Nishimura K, Tsukino M, et al. Analysis of the factors related to mortality in chronic obstructive pulmonary disease: role of exercise capacity and health status. Am J Respir Crit Care Med 2003; 167:544 –549 18 Curtis JR, Martin DP, Martin TR. Patient-assessed health outcomes in chronic lung disease. Am J Respir Crit Care Med 1997; 156:1032–1039 19 Trupin L, Earnest G, San Pedro M, et al. The occupational burden of chronic obstructive pulmonary disease. Eur Respir J 2003; 22:462– 469 20 Kim J. Atlas of respiratory disease mortality, United States: 1982–1993. Cincinnati, OH: Department of Health and Human Services, National Institute for Occupational Safety and Health, 1998 21 Eisner MD, Yelin EH, Trupin L, et al. The influence of chronic respiratory conditions on health status and work disability. Am J Public Health 2002; 92:1506 –1513 22 Blanc PD, Cisternas M, Smith S, et al. Asthma, employment status, and disability among adults treated by pulmonary and allergy specialists. Chest 1996; 109:688 – 696; erratum 2000; 118:564 23 Eisner MD, Katz PP, Yelin EH, et al. Assessment of asthma severity in adults with asthma treated by family practitioners, allergists, and pulmonologists. Medical Care 1998; 36:1567– 1577 24 Fan VS, Curtis JR, Tu SP, et al. Using quality of life to predict hospitalization and mortality in patients with obstructive lung diseases. Chest 2002; 122:429 – 436 25 Nunally JC, Berstein IH. Psychometric theory. 2nd ed. New York, NY: McGraw-Hill, 1994 26 American Thoracic Society. Standardization of spirometry– 1987 update. Am Rev Respir Dis 1987; 136:1285–1298 27 Weinberger M, Murray MD, Marrero DG, et al. Effectiveness of pharmacist care for patients with reactive airways disease: a randomized controlled trial. JAMA 2002; 288: 1594 –1602 28 Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med 1999; 159:179 –187 29 Ware JE Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36): I. Conceptual framework and item selection. Med Care 1992; 30:473– 483 30 Benzo R, Flume PA, Turner D, et al. Effect of pulmonary rehabilitation on quality of life in patients with COPD: the use of SF-36 summary scores as outcomes measures. J Cardiopulm Rehabil 2000; 20:231–234 31 Ware J Jr., Kosinski M, Keller SD. A 12-item short-form Clinical Investigations
health survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220 –233 32 LaPlante M. Data on disability from the National Health Interview Survey 1983–1985: an InfoUse report. Washington, DC: U.S. National Institute on Disability and Rehabilitation Research, 1988 33 Bureau of Labor Statistics. Labor force statistics from the current population survey. Available at: http://www.bls.gov/ bls/proghome.htm; accessed March 9, 2005
www.chestjournal.org
34 Leidy NK. Psychometric properties of the functional performance inventory in patients with chronic obstructive pulmonary disease. Nurs Res 1999; 48:20 –28 35 Selvin S. Statistical analysis of epidemiologic data. New York, NY: Oxford University Press, 1996 36 Celli BR, Cote CG, Marin JM, et al. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004; 350:1005–1012
CHEST / 127 / 6 / JUNE, 2005
1897
Objective: To develop a comprehensive disease-specific COPD severity instrument for survey-based epidemiologic research. Study design and setting: Using a population-based sample of 383 US adults with self-reported physiciandiagnosed COPD, we developed a disease-specific COPD severity instrument. The severity score was based on structured telephone interview responses and included five overall aspects of COPD severity: respiratory symptoms, systemic corticosteroid use, other COPD medication use, previous hospitalization or intubation, and home oxygen use. We evaluated concurrent validity by examining the association between the COPD severity score and three health status domains: pulmonary function, physical health-related quality of life (HRQL), and physical disability. Pulmonary function was available for a subgroup of the sample (FEV1, n ⴝ 49; peak expiratory flow rate [PEFR], n ⴝ 93). Results: The COPD severity score had high internal consistency reliability (Cronbach ␣ ⴝ 0.80). Among the 49 subjects with FEV1 data, higher COPD severity scores were associated with poorer percentage of predicted FEV1 (r ⴝ ⴚ 0.40, p ⴝ 0.005). In the 93 subjects with available PEFR measurements, greater COPD severity was also related to worse percentage of predicted PEFR (r ⴝ ⴚ 0.35, p < 0.001). Higher COPD severity scores were strongly associated with poorer physical HRQL (r ⴝ ⴚ 0.58, p < 0.0001) and greater restricted activity attributed to a respiratory condition (r ⴝ 0.59, p < 0.0001). Higher COPD severity scores were also associated with a greater risk of difficulty with activities of daily living (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.8 to 3.0) and inability to work (OR, 4.2; 95% CI, 3.0 to 5.8). Conclusion: The COPD severity score is a reliable and valid measure of disease severity, making it a useful research tool. The severity score, which does not require pulmonary function measurement, can be used as a study outcome or to adjust for disease severity. (CHEST 2005; 127:1890–1897) Key words: chronic bronchitis; pulmonary disease, chronic obstructive; outcome assessment (health care); pulmonary emphysema; severity of illness index Abbreviations: CI ⫽ confidence interval; HRQL ⫽ health-related quality of life; NHIS ⫽ National Health Interview Survey; OR ⫽ odds ratio; PEFR ⫽ peak expiratory flow rate; SF ⫽ Short-Form
is a major cause of impaired health status, C OPD disability, and mortality. The disease is respon-
sible for ⬎ 2 million emergency department visits and hospitalizations annually in the United States.1 Despite public awareness about the health risks of smoking, mortality from COPD continues to increase.1 Consequently, there is an urgent need for
epidemiologic studies that better delineate the risk factors for adverse health outcomes among adults with COPD. The conduct of such studies has been hampered by the lack of a validated disease-specific COPD severity measure beyond the basic physiologic assessment of pulmonary function. A comprehensive survey-based measure of COPD severity
*From the Department of Medicine (Drs. Eisner, Balmes, and Blanc, Ms. Trupin and Ms. Earnest), University of California, San Francisco; of and Institute for Health Policy Studies (Drs. Katz and Yelin), University of California, San Francisco, San Francisco, CA. Support was provided by grant R01 HL607438 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and Flight Attendants Medical Research Institute grant CoE2001. Dr. Eisner was also supported by grant K23 HL04201 from the National Heart, Lung, and Blood Institute.
Manuscript received August 20, 2004; revision accepted December 21, 2004. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Mark D. Eisner, MD, MPH, FCCP, University of California, San Francisco, 350 Parnassus Ave, Ste 609, San Francisco, CA 94117; e-mail: [email protected]
1890
Clinical Investigations
would be valuable for use in epidemiologic studies as a health outcome or to adjust for disease-specific severity. Pulmonary function measurement, especially FEV1, has traditionally been used to stage the severity of COPD in clinical practice; the recent Global Initiative for Chronic Obstructive Lung Disease has codified this practice in its guidelines.2– 6 Although pulmonary function is an essential clinical tool for characterizing COPD and is a potent predictor of mortality,7 it is not a comprehensive disease-severity measure. FEV1, for instance, correlates weakly with other measures of health-related quality of life (HRQL) and respiratory symptoms.8 –10 Moreover, FEV1 did not capture important treatmentrelated clinical improvements, manifested by fewer COPD exacerbations, in clinical trials of inhaled corticosteroids.11–15 In addition, spirometry can be difficult and expensive to perform in large-scale epidemiologic studies. HRQL instruments, both generic and diseasespecific, have also been used to gauge the severity of COPD.8,16,17 HRQL, however, is conceptually different from disease severity. HRQL is a patient-centered measurement of perceived satisfaction with life, as it is affected by health.18 COPD severity is a distinct construct that reflects the biological impact of disease pathophysiology on diverse aspects of physical functioning. Consequently, COPD severity is proximal to and influential of HRQL. Although HRQL instruments provide an important patientlevel assessment, they cannot substitute for a specific measure of COPD severity. We developed a comprehensive disease-specific COPD severity score for use in epidemiologic and health-outcomes research. The score, which was designed for survey administration, is an integrated measure of disease severity, based on disease status, receipt of clinical treatments, and recent hospitalization for COPD. Using a population-based sample of US adults with COPD, we assessed the reliability and validity of the COPD severity score. Materials and Methods Overview In a population-based sample of 383 US adults with COPD, we developed a disease-specific COPD severity instrument. The severity score was based on responses to a structured telephone interview. Internal consistency reliability was established using standard psychometric techniques. We evaluated concurrent validity by examining the association between the COPD severity score and three aspects of health status: pulmonary function, physical HRQL and health status, and physical disability. Pulmonary function measurements were obtained for a subset of respondents subsequent to the interviews. www.chestjournal.org
Recruitment of Adults With COPD Survey methods have been previously described in detail.19 The study was approved by the University of California, San Francisco Committee on Human Research. Briefly, 383 adults with COPD were selected from a random sample of 2,061 US adults aged 55 to 75 years identified by random-digit dialing telephone interviews. Approximately one half of the overall sample (n ⫽ 1,001) was randomly identified among residents in the 48 contiguous states of the United States. The remainder of the sample (n ⫽ 1,060) was recruited from persons who resided in geographic “hot spots” that had the highest COPD mortality rates based on the National Institute of Occupational Safety and Health Atlas of Respiratory Disease Mortality in the United States: 1982–1993.20 The hot-spot sample was enriched for subjects with COPD. The overall study participation rate was 53% among households with an eligible respondent present. During each telephone contact, one randomly selected adult per household was interviewed. Subjects were asked if they had ever received a physician’s diagnosis of any of several chronic respiratory conditions. Those who reported physician diagnoses of chronic bronchitis or emphysema were considered to have COPD, along with those who specifically reported a diagnosis of COPD. We included respondents with COPD who had concomitant asthma because they clinically resemble persons with COPD alone.21 Development of the COPD Severity Score Participants underwent structured telephone interviews that elicited respiratory symptoms and medications, health status, smoking history, employment history, and sociodemographic characteristics. Survey-item responses were used to construct a disease-specific COPD severity score. The instrument and scoring instructions are available from the authors, on request. We previously developed a disease-specific severity-of-asthma score for use in epidemiologic and outcomes research.22,23 Using a similar process, we created a COPD severity score based on responses to survey items that comprise five overall aspects of COPD severity: respiratory symptoms, systemic corticosteroid use, other COPD medication use, previous hospitalization or intubation for respiratory disease, and home oxygen use (Table 1). Each item was assigned an a priori weight based on clinical aspects of the disease and its expected contribution to overall COPD severity. In order to assess the performance of the a priori weighting scheme, we derived an alternate weighting scheme by factor analysis with orthogonal rotation. Possible total scores range from 0 to 35, with higher scores reflecting more severe COPD. The COPD severity score takes into account both current symptom status and the therapy necessary to achieve this status. For example, a person with few respiratory symptoms who is receiving multiple medications for COPD would have greater COPD severity, and a higher COPD severity score, than another person with comparable symptom status and no medication use. In addition, recent prior hospitalization or intubation for COPD are assigned high weights, as they are indicators of poor future health outcomes.24 Oxygen use, because it is recommended for persons with advanced disease,3 was also heavily weighted. Missing values for COPD severity score components were rare, ranging from 0 to 2% per item. We used previously established methods to impute missing values.23 For dyspnea during the past 14 days or nights, we assigned missing values the maximum of either the day or night response (n ⫽ 9 subjects and n ⫽ 5 subjects, respectively; no subjects were missing both values). Respondents who indicated antibiotic use during the past 12 months but who did not indicate the number of courses were CHEST / 127 / 6 / JUNE, 2005
1891
Table 1—COPD Severity Score Items (n ⴝ 383) Items Respiratory symptoms (maximum 7 points) Dyspnea on exertion, current None Hurrying on level ground or walking uphill Walking with peers on level ground Walking at own pace on level ground Dyspnea during the past 14 d or nights None 1 to 2 d or nights 3 to 6 d or nights 7 to 13 d or nights Every day or night Systemic corticosteroid use (maximum 5 points) Ever used Long-term use in past year* Used in past 2 wk Other medication use (maximum 10 points)† Metered-dose inhaler in past 2 wk‡ Short-acting -agonists Long-acting -agonists Inhaled corticosteroids Ipratropium bromide Nebulizer use, past 2 wk‡ Short-acting -agonists Ipratropium bromide Oral medications Theophylline, past 2 wk -Agonists, past 2 wk Antibiotics for lung condition, past 12 mo One to two courses Three or more courses Hospitalization/intubation/oxygen use (maximum 13 points) Hospitalized for COPD, past 5 yr Intubated for COPD, past 5 yr Home oxygen, current day or nighttime use
Score
No. (%)
0 1 2 3
99 (26) 96 (25) 53 (14) 136 (36)
0 1 2 3 4
128 (33) 51 (13) 50 (13) 29 (8) 125 (33)
1 3 1
115 (30) 42 (11) 20 (5)
1 1 1 1
137 (36) 53 (14) 92 (24) 80 (21)
1 1
43 (11) 13 (3)
1 1
18 (5) 20 (5)
1 2
67 (17) 49 (13)
3 5 5
95 (25) 18 (5) 37 (10)
*At least three times per week for at least 3 months during the past 2 years. †Combination medications were counted under both categories. For example, the combined preparation of ipratropium bromide and albuterol sulfate was counted as both short-acting -agonist and an ipratropium bromide metered-dose inhalers. ‡Missing data: 12 subjects (3%) used an metered-dose inhaler, and 7 subjects (2%) used a nebulizer but did not specify the type of medication.
assigned to the most common “one to two courses” category (n ⫽ 9 subjects). Missing values for medication use and other questions were defined as zero (ie, not used or not present) [range, 0 to 2% missing per item]. COPD Severity Score Reliability Internal consistency reliability was evaluated by principal component analysis to examine whether the COPD severity score appeared to measure a single construct. Cronbach ␣ was also calculated as a measure of internal consistency.25 As noted above, we compared the a priori weighting scheme with an alternate weighting scheme derived from factor analysis. Concurrent Validity We assessed concurrent validity of the COPD severity score by evaluating its association with three aspects of health status 1892
defined a priori: pulmonary function, physical HRQL and health status, and physical disability. The variables selected to represent these aspects of health status were based on our theoretical model of COPD severity. We anticipated that greater COPD severity would correlate with impairment in these three dimensions of health status. We used two methods for obtaining pulmonary function measurements among a subset of 264 subjects (70%) who we recontacted in follow-up to the initial interview. Of the 264 subjects, 170 respondents indicated that they had undergone pulmonary function testing during the past 5 years. These subjects were contacted by mail to request permission to retrieve their pulmonary function testing results from their outpatient medical record. The health-care providers for 71 participating subjects were contacted, and 49 pulmonary function reports were obtained. From these reports, we ascertained FEV1 for 49 subjects and peak expiratory flow rate (PEFR) for 44 subjects. In the remaining 24 instances, no record of pulmonary function testing was located by the health-care provider. The mean time difference between the pulmonary function test and the initial interview date, which generated the survey responses for calculation of the COPD severity score, was 0.4 ⫾1.6 years [⫾ SD]). We used the predicted values provided by the pulmonary function testing laboratory where the tests were conducted. Respondents who indicated no recent pulmonary function testing at follow-up were contacted by letter to recruit them for a telephone-based PEFR measurement. Of the 94 eligible subjects, 68 subjects were sent peak flowmeters by mail, and 49 participated in the peak expiratory flowmeter program. Participating subjects were mailed a peak flowmeter (Astech; Dey Laboratories; Napa, CA). One research assistant, who was trained by an investigator with expertise in asthma care, carried out a structured telephone-based program with subjects that lasted approximately 10 min per session. Each subject performed three peak expiratory flow maneuvers according to American Thoracic Society guidelines and similar to other telephone-based peak flow monitoring programs.26,27 The mean time difference between the initial interview and the peak flow measurement was 1.4 ⫾ 0.2 years. Predicted values were based on reference standards developed for the third National Health and Nutrition Examination Survey.28 In total, we had peak expiratory flow data for 93 subjects (44 subjects who underwent pulmonary function testing and 49 subjects who completed the home peak flowmeter program). Physical HRQL was assessed using the Short-Form (SF)-12 physical component summary score. The SF-12 is derived from the Medical Outcomes Study SF-36 instrument, which is the most widely used measure of generic HRQL. The SF-36 has been extensively validated in the general population29 and among adults with COPD.30 Defined from the eight SF-36 subscales by factor analysis, the physical component summary score reflects an underlying physical dimension of physical HRQL.31 Higher scores reflect more favorable health states. As another measure of physical health status, daily activity restriction was ascertained using a question adapted from the National Health Interview Survey (NHIS).32 Specifically, respondents were asked to indicate how many days their activity was limited due to a respiratory condition during the past month. Respondents were also asked to rate the severity of their fatigue or tiredness on a scale of 0 to 10, in which 0 reflected no fatigue and 10 indicated very severe fatigue. Concurrent validity was also assessed using three measures of physical disability and health impacts. Self-rated general health was assessed with a question developed for the NHIS and also used in the SF-36 questionnaire.29 Based on this item, we defined adverse physical health status as self-reported fair or poor general health (as opposed to good, very good, or excellent). Although Clinical Investigations
this question comprises one aspect of the SF-12 physical component summary score, it has independent value as a measure of health status.21 Current employment status was ascertained by questions that were patterned after the Bureau of Labor Statistics Current Population Survey.33 Using these items, we ascertained whether respondents were unable to work secondary to a respiratory condition. Activities of daily living were assessed with the “body care” section of the Functional Performance Inventory SF.34 These activities included dressing or undressing, showering or bathing, caring for feet, washing hair, shaving or applying makeup, and preparing meals. We defined difficulty with activities of daily living as being unable to do the activity or able to do the activity with much difficulty. Statistical Analysis The analysis was conducted using statistical software (Version 8.1; SAS Institute; Cary, NC). Internal consistency reliability was evaluated using principal component analysis and Cronbach ␣. We evaluated concurrent validity by examining the association between the COPD severity score and pulmonary function, physical HRQL/health status, and physical disability. The Spearman rank correlation was used to evaluate the association between the COPD severity score and pulmonary function measures (percentage of predicted FEV1 and percentage of predicted PEFR). Using the Spearman rank correlation, we examined the association between COPD severity score and physical health status in three separate analyses: among subjects with available FEV1 data, among subjects with PEFR data, and among the total cohort. In the first two analyses, we controlled for the level of physiologic impairment using percentage of predicted FEV1 or percentage of predicted PEFR, respectively. This was accomplished by calculating the partial correlation between COPD severity score and each physical health status measure, after adjusting for pulmonary function.35 Using an analogous strategy, we used logistic regression analysis to examine the association between COPD severity score and physical disability. Because sociodemographic factors may be on the causal pathway between COPD severity and health status, we did not control for these variables in multivariate analysis. As a sensitivity analysis, we also repeated the analysis using a narrower definition of COPD that only included subjects reporting emphysema or COPD, excluding those with chronic bronchitis alone.
Table 2—Personal Characteristics of 383 Adults With COPD* Characteristics
Data
Age at interview, yr Age at COPD diagnosis, yr† Male gender White, non-Hispanic race/ethnicity Education‡ High school or less Some college College or graduate school Cigarette smoking Never smoked Current smoker Former smoker
64 ⫾ 6 46 ⫾ 18 138 (36) 336 (88) 202 (53) 105 (28) 73 (19) 74 (19) 126 (33) 183 (48)
*Data are presented as mean ⫾ SD or No. (%). †Twenty-nine individuals did not report age at diagnosis. ‡Two subjects did not report education level.
Reliability Internal consistency was high (Cronbach ␣ ⫽ 0.80). In addition, the principal component analysis revealed loading on a single factor (Eigenvalue 3.3) that contributed 71% of the variance of the model. There was a rapid fall-off in the scree plot, with a second component Eigenvalue of 0.55. Concurrent Validity: Pulmonary Function Measurement of pulmonary function was available for 93 subjects. In the subset of 49 subjects with available FEV1 data, higher COPD severity scores were associated with poorer FEV1 percentage of predicted (r ⫽ ⫺0.40, p ⫽ 0.005) [Table 3]. Among the 93 persons with PEFR measurements, greater COPD severity was also related to worse PEFR percentage of predicted (r ⫽ ⫺ 0.35, p ⬍ 0.001). Concurrent Validity: Physical HRQL and Health Status, Disability, and Health Impacts
Results Subject Characteristics The mean age of subjects with COPD was 64 ⫾ 6 years (Table 2). The majority of subjects had a lifetime history of cigarette smoking (81%). Other sociodemographic characteristics are shown in Table 2. COPD Severity Score The COPD severity score ranged from 0 to 28 (mean, 7.3 ⫾ 6.5; median, 6.0; 25th to 75th interquartile range, 2 to 25). The COPD severity score, which was based on an a priori weighting system, correlated very closely with the score whose alternative weighting was derived from factor analysis with orthogonal rotation (r ⫽ 0.97, p ⬍ 0.0001). www.chestjournal.org
Among the 49 subjects with available FEV1 data, higher COPD severity scores were associated with Table 3—Correlation Between COPD Severity Score and Pulmonary Function
Variables
No.
Mean ⫾ SD
Correlation With COPD Severity Score
FEV1 % predicted* PEFR % predicted†
49 93
54.6 ⫾ 25.4 63.4 ⫾ 26.8
⫺ 0.40 ⫺ 0.35
p Value 0.005 0.001
*Obtained from pulmonary function testing records (see Materials and Methods). †Obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home peak flowmeter measurements only were r ⫽ ⫺ 0.29 (p ⫽ 0.04); results for PEFR obtained using spirometry only were r ⫽ ⫺ 0.27 (p ⫽ 0.07). CHEST / 127 / 6 / JUNE, 2005
1893
poorer physical HRQL (r ⫽ ⫺ 0.37, p ⫽ 0.01) and greater restricted activity due to a respiratory condition (r ⫽ 0.59, p ⬍ 0.001) after controlling for FEV1 (Table 4). In the subgroup with PEFR measurements, higher COPD severity scores were also related to poorer physical HRQL (r ⫽ ⫺ 0.55, p ⬍ 0.001) and greater restricted activity (r ⫽ 0.61, p ⬍ 0.001) after controlling for percentage of predicted PEFR. Among the entire cohort, COPD severity scores were also strongly related to physical HRQL (r ⫽ ⫺ 0.58, p ⬍ 0.001) and restricted activity (r ⫽ 0.59, p ⬍ 0.001). Higher COPD severity scores were also associated with greater fatigue (r ⫽ 0.61, p ⬍ 0.0001). In the subgroup with available FEV1 measurements, the risk of impaired self-perceived general health increased with greater COPD severity scores, after controlling for FEV1 (odds ratio [OR], 2.4 per SD-sized score increment; 95% confidence interval [CI], 1.1 to 5.1) [Table 5]. Higher COPD severity scores were also associated with a greater risk of inability to work secondary to a respiratory condition (OR, 6.7; 95% CI, 2.2 to 20.3) and difficulty performing activities of daily living (OR, 1.7; 95% CI, 0.8 to 3.4). Among those with available PEFR measurements, the COPD severity score was also related to each measure of physical disability. In the overall cohort, higher COPD severity scores were associated with a greater risk of impaired general health (OR, 3.1; 95% CI, 2.3 to 4.1), difficulty performing activities of daily living (OR, 2.3; 95% CI, 1.8 to 3.0), and inability to work (OR, 4.2; 95% CI, 3.0 to 5.8). COPD or Emphysema Subgroup We evaluated the reliability and validity of the COPD severity score among a more restricted subgroup limited to subjects who reported physician-
diagnosed emphysema or COPD (n ⫽ 190), excluding those who indicated chronic bronchitis only. This group had greater COPD severity, as indicated by higher COPD severity scores (mean, 10.0 ⫾ 6.8). As in the overall analysis, internal consistency reliability was high, with a Cronbach ␣ of 0.79. The concurrent validity analyses also provided similar results to the group overall. Higher COPD severity scores correlated with poorer FEV1 percentage of predicted (r ⫽ ⫺ 0.35, p ⫽ 0.04), PEFR percentage of predicted (r ⫽ ⫺ 0.37, p ⫽ 0.008), physical HRQL (r ⫽ ⫺0.47, p ⬍ 0.0001), and restricted activity days due to a respiratory condition (r ⫽ 0.60, p ⬍ 0.0001). Discussion In this population-based study, we developed a survey-based COPD severity score that incorporates clinical aspects of the disease, including respiratory symptoms, oral corticosteroid use, other COPD medication use, previous hospitalization and intubation, and home oxygen therapy. The COPD severity score is internally consistent, reliable, and appears to capture a broad range of disease severity. We have demonstrated the concurrent validity of the COPD severity score using diverse measures of physical health status, disability, and pulmonary function. This severity instrument could prove to be a valuable tool for survey-based research investigating the epidemiology and health outcomes of COPD. The COPD severity score can function as a study outcome variable in studies of risk factors for adverse health impacts, such as environmental exposures or health-care processes. The severity score can also be used to adjust for differences in baseline disease severity in cohort studies. This may be particularly valuable in effectiveness studies of therapeutic inter-
Table 4 —COPD Severity Score as a Predictor of Physical HRQL and Health Status Physical HRQL/Health Status Measure Among subjects with FEV1 data, controlling for FEV1 % predicted SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month Among subjects with PEFR data, controlling for PEFR % predicted† SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month Among entire group SF-12 physical component summary score* Restricted activity days due to a respiratory condition, past month
Subjects, No.
Mean ⫾ SD
Relationship With Severity Score (r)
p Value
49 49
32.5 ⫾ 11.7 12.7 ⫾ 14.6
⫺ 0.37 0.59
0.01 0.001
93 93
34.9 ⫾ 12.1 9.5 ⫾ 13.5
⫺ 0.55 0.61
0.001 0.001
383 383
36.2 ⫾ 13.0 7.0 ⫾ 12.1
⫺ 0.58 0.59
0.001 0.001
*Higher scores indicate more favorable health status. †PEFR obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home peak flowmeter measurement only were SF-12 physical component summary score (r ⫽ ⫺ 0.65, p 0.0001) and restricted activity days (r ⫽ 0.59, p 0.0001). Results for PEFR obtained using spirometry were SF-12 physical component summary score (r ⫽ ⫺ 0.43, p ⫽ 0.003) and restricted activity days (r ⫽ 0.65, p 0.0001). 1894
Clinical Investigations
Table 5—COPD Severity Score as a Predictor of Physical Disability Disability Measure Among subjects with FEV1, controlling for FEV1 % predicted General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡ Among subjects with PEFR, controlling for PEFR % predicted§ General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡ Among entire group (n ⫽ 383) General health, fair or poor Difficulty performing activities of daily living† Unable to work due to a respiratory condition‡
Subjects, No.
Prevalence, No. (%)
Risk of Disability, OR (95% CI)*
49 49 46
27 (55) 18 (37) 15 (33)
2.4 (1.1–5.1) 1.7 (0.8–3.4) 6.7 (2.2–20.3)
93 93 85
49 (53) 24 (26) 24 (28)
2.6 (1.5–4.6) 2.5 (1.4–4.2) 4.0 (2.1–7.6)
383 383 349
193 (50) 89 (23) 82 (23)
3.1 (2.3–4.1) 2.3 (1.8–3.0) 4.2 (3.0–5.8)
*OR expressed per SD-sized increment in COPD severity score. †Unable to do one or more activity of daily living or much difficulty doing the activity. ‡Among subjects with history of any labor force participation. §PEFR obtained from a pulmonary function testing records (n ⫽ 44) or peak flowmeter used by the subject at home (n ⫽ 49). Results for PEFR obtained from home monitoring only were general health fair or poor (OR, 4.0; 95% CI, 1.5 to 10.3), difficulty performing activities of daily living (OR, 2.8; 95% CI, 1.3 to 6.2), and unable to work (OR, 3.5; 95% CI, 1.5 to 8.3). Results for PEFR obtained from spirometry were general health fair or poor (OR, 2.3; 95% CI, 1.1 to 5.0), difficulty performing activities of daily living (OR, 1.8; 95% CI, 0.91 to 3.7), and unable to work (OR, 5.5; 95% CI, 1.9 to 16.2).
ventions for COPD. Importantly, the COPD severity score does not require measurement of pulmonary function, which may be major logistical advantage for population-based studies. The methods that we used to recruit adults with COPD had both strengths and limitations. The population-based methods used to identify subjects should ensure the generalizability of the test characteristics of the COPD severity score. The use of self-reported physician diagnosis, however, may result in some misclassification of disease status. This methodology is a standard epidemiologic approach for identifying COPD using survey-based techniques.1 The high prevalence of lifetime smoking, which was ⬎ 80%, supports the diagnosis of COPD. Moreover, the prevalence of COPD in our general population sample (13.5%) was similar to that reported in two other population-based studies conducted in the United States, The third National Health and Examination Survey and the NHIS.1 In addition, the majority of subjects indicated recent respiratory symptoms or COPD medication use (89%), consistent with a diagnosis of COPD. Finally, reanalysis using a more restrictive definition of COPD that excludes chronic bronchitis did not appreciably affect the results. Although the enrolled sample appears similar to COPD patients in general, we cannot exclude some degree of misdiagnosis and misclassification of disease status. Another limitation is the availability of pulmonary function testing on a subset of the overall sample, the use of such data from several pulmonary function laboratories, and the time lag between interview and www.chestjournal.org
pulmonary function measurement. When we examined the subgroups with available pulmonary function measurements, we found strong evidence of reliability and validity. This indicates that estimates of reliability and validity are robust and not subject to selection bias. If the time lag between interview and pulmonary function measurement introduced bias, it would be expected to be a conservative one (ie, toward the null). The consistency of findings across different study outcomes, ranging from pulmonary function to physical health status and disability, also supports the psychometric soundness of the COPD severity score. The COPD severity score was a reliable and valid tool in our cohort study. It is based on survey data, which could be affected by subject misreporting, but has the significant advantage of deployment in largescale epidemiologic studies in which pulmonary function measurement and physical examination are not logistically feasible. The score, however, should be validated in a separate, larger cohort of COPD patients, which we will carry out in a different ongoing cohort study of COPD. In addition, we did not have additional objective measures, such as the 6-min walk test, which will also provide additional validation in our ongoing COPD cohort study. Although the COPD severity score was designed for epidemiologic and health-outcomes research, it could have applications in clinical settings. Before clinical use can be recommended, further validation and comparison with the recently published multidimensional BODE score need to be undertaken,36 which we are planning in an ongoing COPD cohort CHEST / 127 / 6 / JUNE, 2005
1895
study. It will be instructive to compare our COPD severity score, which takes into account both current symptom status and the therapy necessary to achieve this status, with the BODE score, which grades patients based on dyspnea, airflow obstruction, exercise capacity, and body mass index. One key difference is that the BODE score requires in-person physical testing, which could reduce its applicability in large-scale population-based epidemiologic studies based on feasibility concerns. The COPD severity score is a simple, reliable, and valid survey-based instrument that can be used in epidemiologic and health-outcomes studies. Developed for telephone interview administration, the COPD severity score may be adapted to other research settings, such as personal interview or written questionnaire. Future studies will evaluate other psychometric properties, such as longitudinal responsiveness to change in COPD status.
References 1 Mannino DM, Homa DM, Akinbami LJ, et al. Chronic obstructive pulmonary disease surveillance–United States, 1971–2000. MMWR Surveill Summ 2002; 51:1–16 2 BTS guidelines for the management of chronic obstructive pulmonary disease. The COPD Guidelines Group of the Standards of Care Committee of the BTS. Thorax 1997; 52(suppl):S1–S28 3 Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. American Thoracic Society. Am J Respir Crit Care Med 1995; 152:S77–S121 4 Siafakas NM, Vermeire P, Pride NB, et al. Optimal assessment and management of chronic obstructive pulmonary disease (COPD). The European Respiratory Society Task Force. Eur Respir J 1995; 8:1398 –1420 5 Pauwels RA, Buist AS, Calverley PM, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease (GOLD) Workshop summary. Am J Respir Crit Care Med 2001; 163:1256 –1276 6 Fabbri LM, Hurd SS. Global strategy for the diagnosis, management and prevention of COPD: 2003 update. Eur Respir J 2003; 22:1–2 7 Speizer FE, Fay ME, Dockery DW, et al. Chronic obstructive pulmonary disease mortality in six U.S. cities. Am Rev Respir Dis 1989; 140:S49 –55 8 Wijnhoven HA, Kriegsman DM, Hesselink AE, et al. Determinants of different dimensions of disease severity in asthma and COPD: pulmonary function and health-related quality of life. Chest 2001; 119:1034 –1042 9 Prieto L, Alonso J, Ferrer M, et al. Are results of the SF-36 health survey and the Nottingham Health Profile similar? A comparison in COPD patients. Quality of Life in COPD Study Group. J Clin Epidemiol 1997; 50:463– 473 10 Nishimura K, Izumi T, Tsukino M, et al. Dyspnea is a better predictor of 5-year survival than airway obstruction in patients with COPD. Chest 2002; 121:1434 –1440 11 Alsaeedi A, Sin DD, McAlister FA. The effects of inhaled corticosteroids in chronic obstructive pulmonary disease: a systematic review of randomized placebo-controlled trials. Am J Med 2002; 113:59 – 65 1896
12 Burge PS, Calverley PM, Jones PW, et al. Randomised, double blind, placebo controlled study of fluticasone propionate in patients with moderate to severe chronic obstructive pulmonary disease: the ISOLDE trial. BMJ 2000; 320:1297– 1303 13 Lung Health Study Research Group. Effect of inhaled triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease. N Engl J Med 2000; 343: 1902–1909 14 Pauwels R. Inhaled glucocorticosteroids and chronic obstructive pulmonary disease: how full is the glass? Am J Respir Crit Care Med 2002; 165:1579 –1580 15 Verhoeven GT, Hegmans JP, Mulder PG, et al. Effects of fluticasone propionate in COPD patients with bronchial hyperresponsiveness. Thorax 2002; 57:694 –700 16 Hajiro T, Nishimura K, Tsukino M, et al. A comparison of the level of dyspnea vs disease severity in indicating the healthrelated quality of life of patients with COPD. Chest 1999; 116:1632–1637 17 Oga T, Nishimura K, Tsukino M, et al. Analysis of the factors related to mortality in chronic obstructive pulmonary disease: role of exercise capacity and health status. Am J Respir Crit Care Med 2003; 167:544 –549 18 Curtis JR, Martin DP, Martin TR. Patient-assessed health outcomes in chronic lung disease. Am J Respir Crit Care Med 1997; 156:1032–1039 19 Trupin L, Earnest G, San Pedro M, et al. The occupational burden of chronic obstructive pulmonary disease. Eur Respir J 2003; 22:462– 469 20 Kim J. Atlas of respiratory disease mortality, United States: 1982–1993. Cincinnati, OH: Department of Health and Human Services, National Institute for Occupational Safety and Health, 1998 21 Eisner MD, Yelin EH, Trupin L, et al. The influence of chronic respiratory conditions on health status and work disability. Am J Public Health 2002; 92:1506 –1513 22 Blanc PD, Cisternas M, Smith S, et al. Asthma, employment status, and disability among adults treated by pulmonary and allergy specialists. Chest 1996; 109:688 – 696; erratum 2000; 118:564 23 Eisner MD, Katz PP, Yelin EH, et al. Assessment of asthma severity in adults with asthma treated by family practitioners, allergists, and pulmonologists. Medical Care 1998; 36:1567– 1577 24 Fan VS, Curtis JR, Tu SP, et al. Using quality of life to predict hospitalization and mortality in patients with obstructive lung diseases. Chest 2002; 122:429 – 436 25 Nunally JC, Berstein IH. Psychometric theory. 2nd ed. New York, NY: McGraw-Hill, 1994 26 American Thoracic Society. Standardization of spirometry– 1987 update. Am Rev Respir Dis 1987; 136:1285–1298 27 Weinberger M, Murray MD, Marrero DG, et al. Effectiveness of pharmacist care for patients with reactive airways disease: a randomized controlled trial. JAMA 2002; 288: 1594 –1602 28 Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general U.S. population. Am J Respir Crit Care Med 1999; 159:179 –187 29 Ware JE Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36): I. Conceptual framework and item selection. Med Care 1992; 30:473– 483 30 Benzo R, Flume PA, Turner D, et al. Effect of pulmonary rehabilitation on quality of life in patients with COPD: the use of SF-36 summary scores as outcomes measures. J Cardiopulm Rehabil 2000; 20:231–234 31 Ware J Jr., Kosinski M, Keller SD. A 12-item short-form Clinical Investigations
health survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220 –233 32 LaPlante M. Data on disability from the National Health Interview Survey 1983–1985: an InfoUse report. Washington, DC: U.S. National Institute on Disability and Rehabilitation Research, 1988 33 Bureau of Labor Statistics. Labor force statistics from the current population survey. Available at: http://www.bls.gov/ bls/proghome.htm; accessed March 9, 2005
www.chestjournal.org
34 Leidy NK. Psychometric properties of the functional performance inventory in patients with chronic obstructive pulmonary disease. Nurs Res 1999; 48:20 –28 35 Selvin S. Statistical analysis of epidemiologic data. New York, NY: Oxford University Press, 1996 36 Celli BR, Cote CG, Marin JM, et al. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004; 350:1005–1012
CHEST / 127 / 6 / JUNE, 2005
1897