Difficulties Identifying and Targeting COPD and Population-Attributable Risk of Smoking for COPD

Difficulties Identifying and Targeting COPD and Population-Attributable Risk of Smoking for COPD* A Population Study David Wilson, PhD; Robert Adams, MD; Sarah Appleton, BSc; and Richard Ruffin, MD, FCCP; on behalf of the North West Adelaide (Cohort) Study Team

Study objectives: Respiratory public health interventions depend on accurate identification of the target group, yet this may vary depending on the diagnostic criteria used. We therefore compared the relative performance of various international criteria in identifying COPD cases. The burden of COPD due to smoking can only be determined from population-attributable risk (PAR) studies. These studies, lacking in the COPD literature, are necessary research in support of public health initiatives for COPD. In this representative population study, we also assessed the PAR for current and ex-smokers. Design: A representative biomedical population sample of 2,501 South Australians aged > 18 years (The Northwest Adelaide Health [Cohort] Study). COPD diagnosis and severity were determined according to various FEV1/FVC and FEV1 percentage of predicted criteria recommended by international respiratory authorities. Demographic, health behavior, and quality-oflife data were obtained by telephone interview and self-completed questionnaire. Setting: Northwest Adelaide. Measurements and results: The PAR of smoking (smokers and ex-smokers) for COPD ranged from 51 to 70% depending on the diagnostic respiratory criteria used. COPD prevalence varied depending on the criteria used: American Thoracic Society, 5.4%; British Thoracic Society, 3.5%; European Respiratory Society (ERS), 5.0%; Global Initiative for Chronic Obstructive Lung Disease, 5.4%. There was also substantial disagreement in the cases identified. An alternative approach using ERS reference values one residual SD from the mean produced a COPD prevalence estimate of 6.9%, with improved level of agreement with the established respiratory criteria suggesting their potential as screening criteria. Conclusions: The COPD risks associated with smoking and ex-smoking history were quantifiable using PAR, but PAR also suggests other, yet unquantified, risks. Targeting COPD cases for public health interventions is difficult given the range of spirometry criteria and the associated high level of underdiagnosis. (CHEST 2005; 128:2035–2042) Key words: COPD; epidemiology; public health Abbreviations: 1RSD ⫽ one residual SD; ATS ⫽ American Thoracic Society; BTS ⫽ British Thoracic Society; CI ⫽ confidence interval; ERS ⫽ European Respiratory Society; GOLD ⫽ Global Initiative for Chronic Obstructive Lung Disease; OR ⫽ odds ratio; PAR ⫽ population-attributable risk; PCS ⫽ physical component summary; SA ⫽ South Australian; SF-36 ⫽ Short Form-36

is a leading cause of morbidity and morC OPD tality worldwide and is increasing in prevalence.1–3 Smoking is well established as the main *From the Health Observatory, University of Adelaide, Queen Elizabeth Hospital, Woodville Road, Woodville, Adelaide, South Australia. Ethics approval for the study was obtained from the Queen Elizabeth Hospital Ethics Committee. Financial support for this study was obtained from the Human Services Research & Innovation Program grants of the South Australian Department of Human Services to conduct the research fieldwork. www.chestjournal.org

causal risk factor,4,5 and smoking cessation has been promoted as the only accepted strategy effective in reducing further loss of lung function.6,7 Therefore, the main approaches to reducing the burden of Manuscript received May 18, 2004; revision accepted April 29, 2005. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: David Wilson, PhD, Department of Medicine, Queen Elizabeth Hospital, Woodville, South Australia 5011; e-mail: [email protected] CHEST / 128 / 4 / OCTOBER, 2005

2035

COPD are through primary prevention smoking programs or early secondary prevention following earliest possible diagnosis.8 –10 Despite this, there are two major gaps in the epidemiologic intelligence of COPD. First, little work has been done to correctly quantify the burden of COPD due to smoking through attributable-risk studies, which is the most precise way to confirm smoking as the major risk factor and identify the burden that remains to be explained by other factors. Risk has been used to compute disability-adjusted life years11 and for estimates of mortality12,13 but not to assess the proportion of all cases that can be ascribed to the smoking risk factor. Second, in identifying COPD cases for earliest possible intervention, we do not know how the various international respiratory criteria perform and how much of the overall target group they identify. The question remains whether early case finding can be improved by alternative or additional strategies. Filling these intelligence gaps are important if a public health approach to COPD is to be effective. Lundback et al8 have shown that COPD developed in 50% of elderly smokers in a cohort study, a higher rate than suggested from previous reports.14,15 This still does not tell us clearly how much of the burden of COPD disease can be explained by smoking and what remains to be explained by other factors. The population-attributable risk (PAR) indicates how much of the community burden of a disease is due to a certain exposure and is especially useful when the cause of the disease is attributed to one major risk factor,16 in this case tobacco. Conversely, the PAR tells us how much of the disease is preventable if we have effective public health and primary care smoking cessation policies. How the data from Lundback et al8 compare with estimates measured using the PAR statistic is unknown. It has also been suggested that the focus on smoking as the sole cause of COPD has hindered a better understanding of other important genetic and environmental factors involved in the development of COPD.17 In this study, we have used the PAR to revisit the importance of smoking in the etiology of COPD using the Northwest Adelaide Health (Cohort) Study population sample. Currently, there is lack of agreement in the diagnostic criteria promoted by major respiratory authorities; the British Thoracic Society (BTS),18 the American Thoracic Society (ATS),19,20 the European Respiratory Society (ERS),21 and the Global Initiative for Chronic Obstructive Lung Disease (GOLD)22,23 have all produced varying guideline criteria for the diagnosis of COPD. This can lead to differing prevalence rates and different people identified as the target for programs, depending on the criteria used.24 It has 2036

also been suggested that fixed respiratory criteria (ATS, BTS, GOLD) may identify too few cases among young adults and too many among the elderly.8 To clarify how well we may be able to target COPD as a public health problem, we compared the prevalence of COPD identified by each set of respiratory criteria in a random population sample assessed clinically by lung function tests. We then used Short Form-36 (SF-36) healthrelated quality-of-life measures to identify the population impact of COPD classified by each of the respiratory criteria.

Materials and Methods The Survey Sample Households in the northwest region of Adelaide comprised the Northwest Adelaide Health (Cohort) Study sample and were selected at random from the electronic “white pages” telephone directory. Within each household, a random selection of adults aged ⱖ 18 years was made without replacement. Telephone interviews were conducted between February and November 2000. A letter introducing the study was sent to each household prior to interview informing participants on the nature of the study. Hotels, motels, nursing homes, and other major institutions were excluded from the survey. An initial sample size of 5,000 households was drawn for this survey. The response rate at preliminary interview was 73.7%, and of these 69% attended the clinic for biomedical testing giving a final sample of 2,501. Data were weighted to Australian census data to address issues of sample nonresponse and provide estimates that represent the population from which the sample was drawn. Trained telephone interviewers assessed self-reported lung health status, smoking status, demographic variables, and quality of life using the SF-36 health-related quality-of-life questionnaire.25 Participants were then invited to be part of the clinical assessment of health status conducted in two teaching hospital clinics. This included spirometry and height and weight. Body mass index was computed, and a body mass index ⱖ 30 kg/m2 was considered to be obese. Spirometry (Microlab 3300; Micro Medical Ltd; Kent, UK) was conducted according to ATS criteria.19 Each subject performed at least three reproducible FVC maneuvers. People with reversibility of airway obstruction were identified by a 15% increase in FEV1 following administration of 400 ␮g of a short-acting bronchodilator (salbutamol) administered via a metered-dose inhaler and large-volume spacer, or those who had at least a 12% increase in FEV1 following short-acting bronchodilator if the absolute difference in FEV1 was ⬎ 200 mL. All identified cases were further classified according to the following: (1) prior physician diagnosis of bronchitis, emphysema, and/or chronic lung disease (self-report); and (2) severity based on FEV1 percentage of predicted as prescribed by each set of international criteria (Table 1). Statistical Analysis Data were analyzed using statistical software (Statistical Package for the Social Sciences, Version 10; SPSS; Chicago, IL; and Epi Info, Version 6; Centers for Disease Control and Prevention; Atlanta, GA). As shown in Table 1, population prevalence and severity rates of airways obstruction were calculated according to Clinical Investigations

Table 1—Classification Criteria for COPD Severity Based on Percentage of Predicted FEV1 Criteria

FEV1/FVC

ERS 1RSD*

Men: ⬍ 92.8% of predicted Women: ⬍ 93.5% of predicted19 Men: ⬍ 88.2% of predicted Women: ⬍ 89.3% of predicted ⬍ 70% ⬍ 70% ⬍ 70% 19

ERS† ATS BTS GOLD

Mild

Moderate

Severe

ⱖ 70

50–69

ⱕ 49

ⱖ 70

50–69

⬍ 50

ⱖ 50 80–60 ⱖ 80

35–49 40–59 30–79

⬍ 35 ⱕ 39 ⱕ 29

*Mean minus 1 SD. †Mean minus 1.64 SDs.

the FEV1/FVC and FEV1 percentage of predicted criteria promoted by the ERS,21 the ATS,19 the BTS,18 and the GOLD.23 With the exception of the ERS, each set of criteria uses fixed FEV1/FVC ratios to determine airflow limitation across age and gender (Fig 1). ERS criteria are based on mean population reference values determined by Quanjer,26 which acknowledges that airflow limitation may vary according to age and gender, among other things. Failure to account for these variations could lead us to make false assumptions regarding the respiratory health of individuals who have the same predicted ratio but vary in age. Figure 1 graphically illustrates the theoretical FEV1/FVC ratios by age group for each of the respiratory criteria determining airway obstruction in men. In identifying airway obstruction only, the ERS provide age- and gender-specific criteria that take into account the fact that lung function deteriorates with age. This, as Figure 1 shows, results in very different population prevalence rates for COPD at a given age group, according to ERS and the other criteria. This will be discussed further. The ERS criteria for airflow limitation are determined by a cutoff level of 1.64 residual SDs below mean predicted values for age and gender.26 The ERS criteria also provide reference values based on one residual SD (1RSD) below mean predicted values.26 This is a more inclusive definition of airways limitation and may allow us to improve early detection of COPD. As no population studies have produced prevalence estimates of COPD

based on ERS 1RSD criteria, they were used in this study to assess the additional case finding value they provided in comparison with the other international criteria. Cases identified using ERS 1RSD are considered to at least have airways limitation and constitute an important time to establish diagnosis. For cases identified by ERS 1RSD, severity of airways limitation was determined using ERS criteria. The PAR statistic was used to estimate the proportion of COPD disease that can be attributed to smoking (current and ex-smokers, and current smokers only) in the population. This was calculated on the basis of the formula PAR ⫽ (p[r ⫺ 1]/ (p[r ⫺ 1] ⫹ 1) ⫻ 100, where p is the proportion of the population who smoke and r is the relative risk in the population.27 The results were expressed as the percentage of COPD attributable to smoking. The SF-36 physical component summary (PCS) was used to compare the physical health status of those with COPD identified by each of the respiratory criteria with the PCS score for the South Australian (SA) population. Firstly, mean PCS scores were calculated for those categorized with mild and moderate-tosevere COPD according to each of the respiratory criteria by multivariate analysis of variance controlling for the main effects of age and gender. The interaction term for age and gender was not statistically significant. The method of Garrat et al28 was used to calculate standardized PCS scores according to COPD classi-

Figure 1. Theoretical values of airways obstruction, as determined by FEV1/FVC percentage of predicted according to major respiratory guidelines, at each age group, for men. www.chestjournal.org

CHEST / 128 / 4 / OCTOBER, 2005

2037

Figure 2. The SF-36 physical health scale (PCS) according to COPD criteria and severity. mod-severe ⫽ moderate to severe.

fication and severity criteria. Standardized scores were calculated by dividing the difference between the mean PCS score of the overall SA population29 and the PCS score for the COPD severity classification by the SD of the mean SA population score: mean PCS (ERS mild COPD) – mean PCS (SA population)/SD (SA population mean PCS). In Figure 2, the mean SA population PCS score is set at zero, allowing comparisons of the PCS score for each respiratory classification severity criterion (mild, moderate, and severe) to be made with the SA population in terms of SDs. Effect sizes were interpreted according to the criteria of Kazis et al30: an effect size of 0.2 or one fifth of an SD is described as small or mild, an effect size of 0.4 is described as moderate, and an effect size of ⱖ 0.6 is described as large or severe. The study data were weighted to Australian census data to provide estimates for the northwest population. Bivariate and logistic regression analyses were conducted to examine the relationship between a number of demographic and risk factor variables and those with or without airways obstruction according to the ERS 1RSD (the group providing the largest sample size and a potential target group). The ␹2 test was used to examine the statistical significance of any differences in proportions, and relative risk ratios were produced to show the relative difference between groups with and without airways limitation. Logistic regression analyses in this study were used only for COPD patients determined to have nonreversible airways obstruction and were conducted only for patients with previously undiagnosed COPD using the ERS 1RSD criterion. This was done to examine the relationship of smoking with COPD in cases that would not have been detected using the standard international criteria. Variables entered into the logistic regression had p values ⬍ 0.2531 in the bivariate analyses and were used in the logistic regression analyses to determine the effects of smoking, age, and gender on previously undiagnosed COPD cases using ERS 1RSD. Once an appropriate model was obtained, the presence of multicollinearity was assessed using SPSS software. There was no significant collinearity between the explanatory variables. In addition, the data were examined for interactions for smoking by age and gender. No interaction terms included in the analyses were significant. 2038

Results Table 2 shows that estimates of COPD varied from 1.4% (95% confidence interval [CI], 0.9 to 1.7%) for the BTS criteria to a high of 6.9% (95% CI, 6.0 to 7.8%) for ERS 1RSD when people with reversibility are excluded, and a low of 3.5% (95% CI, 2.8 to 4.2%) for the BTS criteria to a high of 10.9% for ERS 1RSD when people with reversibility were included. It can be seen from Table 2 that for all criteria, the prevalence of COPD was higher in men than in women, with the closest gender prevalence rates being observed for the ERS criteria that compute age- and sex-adjusted rates. The PAR of smoking (current smokers and exsmokers) in COPD also varied across respiratory criteria from 61% for ATS and GOLD criteria to a high of 78% for the ERS criteria (Table 3). Table 3 shows that for current smokers alone, the highest

Table 2—Prevalence of COPD by Type of Respiratory Criteria, Gender, and Reversibility Status* Nonreversible Obstruction Variables

Men

Women

Total Nonreversible Obstruction

ERS 1RSD† ERS ATS BTS GOLD

99 (8.0) 31 (2.5) 55 (4.4) 26 (2.1) 55 (4.4)

74 (5.7) 28 (2.2) 19 (1.5) 8 (0.6) 19 (1.5)

173 (6.9) 59 (2.4) 74 (3.0) 34 (1.4) 74 (3.0)

All Obstruction 275 (10.9) 126 (5.0) 136 (5.4) 87 (3.5) 136 (5.4)

*Data are presented as No. (%). †Reference values based on 1RSD below mean predicted values. Clinical Investigations

Table 3—PAR for COPD According to Respiratory Criteria Current Smokers

Current and Ex-smokers

Criteria

%

Relative Risk†

PAR†

%

Relative Risk†

PAR†

ERS 1RSD* ERS ATS BTS GOLD

52.3 55.9 37.0 42.4 37.8

3.03 3.50 1.62 2.03 1.68

35.4 40.3 14.3 21.8 15.5

79.2 89.8 82.2 84.8 82.2

3.11 7.54 3.93 4.78 3.93

53.3 77.9 61.3 67.1 61.3

*Reference values based on 1RSD below mean predicted values. †The sample size used to compute relative risk and PAR for each of the respiratory criteria was 2,501.

PAR was 40% for the ERS criteria. Variation in the PAR is due to different relative risk estimates produced for each of the respiratory criteria. For current smokers and ex-smokers, the relative risk varied from 3.93 for the GOLD/ATS criteria to 7.5 for the ERS criteria. Further analysis showed that this was due to differential classification of cases by each of the criteria. For example, of those classified with COPD by the GOLD criteria, only 58% were also classified by the ERS criteria. Of those classified with COPD in this study by international criteria (with no reversibility), most were previously undiagnosed by a physician (Table 4). This occurred for 93.3% of those classified by ERS, 90.2% of those classified by ATS, 84.4% of those classified by BTS, and 93.0% of those classified by GOLD. With the exception of the ATS and GOLD criteria, which use the same diagnostic cutoff points, there was substantial disagreement between each of the other criteria: ERS and BTS, 39%; ERS and ATS, 72%; and BTS and ATS, 45%. The one exception to this was the ERS 1RSD, which captured most of the cases in this study for each of the other standard respiratory criteria: ERS, 100%; BTS, 100%; GOLD, 92%; and ATS, 92%. A logistic regression analysis was conducted for smoking controlling for age and gender with undiag-

nosed COPD cases determined by ERS 1RSD as the dependent variable. Those classified by ERS 1RSD were significantly more likely to be ex-smokers (adjusted odds ratio [OR], 2.2; 95% CI, 1.4 to 3.3) or current smokers (adjusted OR, 5.3; 95% CI, 3.4 to 8.0). It also showed that there was no difference in age group for these undiagnosed cases according to ERS 1RSD indicating that younger people are also at risk for airways obstruction according to ERS 1RSD. Figure 2 shows the physical summary component (PCS) of the SF-36 quality-of-life questionnaire. This shows that by any of the criteria used, the physical health of those with moderate-to-severe COPD is significantly impaired compared to the normal population. The effect size is large for both the BTS and ERS, and the effect size for the moderate-to-severe group determined by ERS 1RSD is comparable to the effect size for both ATS and GOLD criteria. Even the mild group of ERS 1RSD are below the population norm for quality of life. Discussion If ex-smokers were excluded from the PAR statistic, the highest estimate of PAR for smokers alone would be 40% using the ERS criteria. This theoretically means that the PAR for ex-smokers is almost 38%, emphasizing the importance of former smokers in COPD programs in which continued monitoring and prevention of exacerbations with the use of inhaled corticosteroids, annual influenza, and pneumococcal vaccinations can achieve important public health benefits.32 The PAR statistic also identifies that, depending on the respiratory criteria used, between 20% and 40% of the PAR of COPD still needs to be explained by other genetic and environmental risk factors. It should be emphasized, however, that this study found a high prevalence of smoking history in those with COPD, with 90% of the ERS group being a current or ex-smoker (Table 3).

Table 4 —Diagnosed and Undiagnosed COPD Cases According to Standard International Criteria and Severity Status* Previously Undiagnosed

Previously Physician Diagnosed

Criteria

Mild

Moderate

Severe

Total

Mild

Moderate

Severe

Total

ERS ATS BTS GOLD ERS 1RSD†

44 (78.6) 62 (95.4) 20 (74.0) 39 (58.2) 144 (90.0)

9 (16.1) 3 (4.6) 7 (26.0) 28 (41.8) 13 (8.1)

3 (5.4)

56 65 27 67 160

2 (28.9) 1 (20.0) 1 (16.6) 7 (58.3)

4 (57.1) 2 (20.0) 4 (66.7) 1 (8.3)

4 (100) 1 (14.3) 2 (40.0) 1 (16.7) 4 (33.3)

4 7 5 6 12

3 (1.9)

*Data are presented as No. (%) or No. †Reference values based on 1RSD below mean predicted values. www.chestjournal.org

CHEST / 128 / 4 / OCTOBER, 2005

2039

The literature identifies that the relationship between smoking and deterioration of lung function levels off to that demonstrated by nonsmokers following smoking cessation.33 Smoking cessation is therefore an important goal of COPD management, as ex-smokers have been shown to have improved survival rates.34 It is possible that ex-smokers in this study quit smoking because of smoking-related symptoms, and it has been shown previously that people who perceive their symptoms to be smoking related are more likely to want to quit.35 In terms of public health programs, it is important that exsmokers are included in assessing the overall burden of disease and development of appropriate policy and intervention responses. Furthermore, if large numbers of smokers and ex-smokers with COPD are not being detected, as was the case in this study, then other important aspects of management that could improve survival and quality of life will be foregone. Depending on which respiratory criteria are used, many COPD cases will be missed, given the variability in classification by each set of respiratory criteria and the large proportions of people with airways obstruction who were not detected by any of the standard international criteria. Although these are largely mild cases, they are the group with potentially the greatest health gain. The study also shows that the PAR must include ex-smokers in considering the total burden of COPD due to this risk factor. Taking both pieces of information together, we are failing to adequately target COPD at a population level following which more aggressive public health smoking cessation programs and improved management of cases to minimize exacerbations must occur. Calverley32 has emphasized the difficulty of early diagnosis and targeting, and notes that, in an ideal world, we would be able to identify our target population clearly. If, however, the major respiratory criteria are failing to identify cases, especially early cases, then an alternative approach is required if health services are to be more effective in intervention. Our results suggest that the ERS 1RSD criterion is a candidate criterion for screening in general practice, given its basis as population reference values. The question that arises from using this more liberal criterion is concerned with the number of COPD false-positive results that this would yield. As no “gold standard” is yet available to guide us, the value of the method must be imputed from other information. First, the ERS 1RSD criterion clearly identified people with airways limitation at a younger age when preventive activity is likely to have better health outcomes and clinical monitoring can begin. Second, the criteria captured most or all of the cases detected by all of the standard criteria. ERS 1RSD also identified twice as many previous physician2040

diagnosed COPD cases as any of the other standard criteria. This raises the question as to why the criteria would be less precise for previously undiagnosed cases. Third, the ERS 1RSD definition identified undiagnosed cases with airways limitation that were five times more likely to be smokers than those without airways limitation and were as likely to include younger as well as older smokers. This relative risk for smokers was higher than the next highest OR produced by the standard ERS criteria for smokers and ex-smokers (OR, 3.5). Fourth, the moderate-to-severe cases identified by ERS 1RSD are already moderately physically impaired, as was apparent from the quality of life data, and this impairment is comparable to those classified by both ATS and GOLD. The mild cases are also below the population norm with regard to the PCS. On the basis of this information, the ERS 1RSD criteria would seem to be potentially useful screening criteria for earlier identification of people with airways limitation who will then benefit from ongoing monitoring in primary care together with appropriate interventions and treatment to minimize future impact.36 It should be stated, however, that further work needs to be done to identify the predictive value of ERS1 RSD criteria in a prospective study, which would allow us to more precisely determine sensitivity and specificity. Further support for the use of the ERS 1RSD screening criterion to identify early COPD cases comes from the smoking data in this study. Supplementary analyses showed that 15%, or 1 in 7 current or previous smokers, were classified with some airways limitation by ERS 1RSD, compared with only 6%, or 1 in 16 subjects, with no smoking history. This high yield of potential COPD cases among smokers and ex-smokers is supported by a recent Swedish study8 that identified a higher COPD rate among smokers than reported in current literature and again showed ex-smokers to be at increased risk of COPD. This would argue for an approach in general practice that asks about smoking history followed by spirometric screening using the ERS 1RSD criterion. However, not all primary care physicians are equipped with spirometers. Peak expiratory flow is potentially an alternative to spirometry, but further research is needed to establish the efficacy of this approach.37 Our findings also support those of Lundback et al8 that a fixed ratio (Fig 1) will produce more falsenegative results among younger adults and more false-positive results among older adults, again confounding target group identification and secondary prevention initiatives. It is also of note that the ERS criteria identify a much closer gender balance of COPD than the other criteria, and raises the quesClinical Investigations

tion as to whether or not the fixed ratio criteria are also missing more female patients than male patients. The National Institute for Clinical Excellence has recently produced guidelines for the diagnosis and management of COPD.38 These suggest that an absolute change in FEV1 of at least 400 mL is required as the threshold for the diagnosis of asthma. While acknowledging these guidelines, were this cut-off to be used in the present study it would only serve to increase the numbers diagnosed with COPD and strengthen the arguments made in this discussion. We believe that this has been accounted for in Table 2 by identifying all those in the population sample with airways obstruction, regardless of reversibility. Other factors could be involved in the development of COPD, which were not assessed in this study including passive smoking, low socioeconomic status, diet, and early environmental exposure.39 Passive smoking has been identified as among the more important of these, but studies33,39 have identified that the effect is small when considered alongside direct smoking effects. In the review by Anto et al,40 the effect of passive smoking was concluded to account for approximately 5% reduction in the maximal attainable FEV1. Finally, there are some limitations in a survey of this type. Some sample loss occurred between telephone interview and recruitment to the clinic. There were higher nonparticipation rates in 18- to 24-yearold men, 24- to 35-year-old women, and two-person households, but no major differences were observed in recruitment rates by area. Given, however, the data were weighted by age, gender, household size, and probability of selection in the household, and that good cell sizes were available even where some small sample differences were observed, it is believed that weighting may have overcome some of the potential for bias. By way of example, the sample obtained at the clinic was very close to census proportions by gender: women, 52.4% (census, 51.6%); men, 47.6% (census, 48.5%). In addition, the data were examined for collinearity and interactions for smoking by age and gender, and none were found. We would therefore conclude that these results are reliable and generalizable to this regional population. Self-reported smoking status may have underestimated smoking history in the survey given that lower rates are more likely in telephone surveys compared to household interview surveys in which evidence of smoking can be seen. Underestimation of smoking history in this survey would, however, have served to downplay the risk of COPD attributable to smoking. The survey was not able to assess the effects of other possible contributing influences www.chestjournal.org

to COPD, such as environmental pollution or, as previously mentioned, passive smoking. If passive smoking and other sources of exposure influence the occurrence of COPD, it is assumed that these additional effects are distributed across smokers and nonsmokers. COPD status in both smokers and nonsmokers is included in the PAR equation. In summary, we have argued that this study shows a large proportion of all COPD cases go undetected irrespective of criteria used. On the basis of the data presented, we have also argued that ERS 1RSD provides a potential screening criteria and that ERS 1RSD is more appropriate for the early detection of COPD cases when intervention would be most effective. If the ERS 1RSD criteria are used together with questions on smoking history, then detection rates for COPD will increase, given that at least one in every seven smokers has this level of airway limitation. A randomized controlled trial should be used to assess the cost benefit of using ERS 1RSD together with smoking history to assess the cost benefit of this approach. In detecting COPD, physicians should not only ask about current smoking but also about smoking history.

References 1 Manton KG. The global impact of non-communicable diseases: estimates and projections. World Health Stat Q 1988; 41:255–266 2 Mannino DM. COPD: epidemiology, prevalence, morbidity and mortality, and disease heterogeneity. Chest 2002; 121(suppl):121S–126S 3 van Schayk CP, Loozen JMC, Wagena E, et al. Detecting patients at high risk of developing chronic obstructive pulmonary disease in general practice: cross sectional case finding. BMJ 2002; 324:1–5 4 Warner KE, Hodgson TA, Carroll CE. Medical costs of smoking in the Unites States: estimates, their validity, and their implications. Tob Control 1999; 8:290 –300 5 Centers for Disease Control and Prevention. Cigarette smoking-attributable morbidity–United States, 2000. MMWR Morb Mortal Wkly Rep 2003; 52:842– 844 6 Sutherland ER, Allmers H, Ayas NT, et al. Inhaled corticosteroids reduce the progress of airflow limitation in chronic obstructive pulmonary disease: a meta-analysis. Thorax 2003; 58:937–941 7 Highland KB, Strange C, Heffner JE. Long term effect of inhaled corticosteroids on FEV1 in patients with chronic COPD. Ann Intern Med 2003; 138:969 –973 8 Lundback B, Lindberg A, Lindstrom, et al. Not 15 but 50% of smokers develop COPD?: report from the Obstructive Lung Disease in Northern Sweden Studies. Respir Med 2003; 97:115–122 9 Petty LT. Definition, epidemiology, course and prognosis of COPD. Clin Cornerstone 2003; 5:1–10 10 Chen JC, Manino DM. Worldwide epidemiology of chronic obstructive pulmonary disease. Curr Opin Pulm Med 1999; 5:93–99 11 Nilunger L, Diderichsen F, Burstrom B, et al. Using risk analysis on health impact assessment: the impact of different CHEST / 128 / 4 / OCTOBER, 2005

2041

12 13 14 15 16 17 18 19 20 21 22

23

24 25

relative risks for men and women in different socio-economic groups. Health Policy 2004; 67:215–224 Davis RM, Novotny TE. The epidemiology of cigarette smoking and its impact on chronic obstructive disease. Am Rev Respir Dis 1989; 140:S82–S84 Ezzati M, Lopez AD. Estimates of global mortality attributable to smoking in COPD. Lancet 2003; 362:847– 852 American Thoracic Society. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1995; 152:77s–120s Rijcken B, Britton J. Epidemiology of chronic obstructive pulmonary disease. Eur Respir Mon 1998; 7:41–73 Northridge ME. Public health methods: attributable risk as a link between causality and public health action. Am J Public Health 1995; 85:1202–1204 Weiss ST, DeMeo DL, Postma DS. Eur Respir J 2003; 21(suppl):4s–12s British Thoracic Society. BTS guidelines for the management of chronic obstructive airways disease. Thorax 1997; 52(suppl):1S–28S American Thoracic Society. Standardization of spirometry: 1987 update. Am Rev Respir Dis 1987; 136:1285–1298 American Thoracic Society. Lung function testing: selection of reference values and interpretive strategies. Am Rev Respir Dis 1991; 144:1202–1218 Siafakas NM, Vermiere P, Pride NB, et al. Optimal assessment and management of chronic obstructive pulmonary disease (COPD). Eur Respir J 1995; 8:1398 –1420 Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Pulmonary Disease. NHLBI/ WHO Workshop Report. Bethesda, MD: National Institutes of Health, National Heart, Lung, and Blood Institute, 2001; publication No. 2701 NHLBI/WHO Workshop Report. Global Initiative for Chronic Obstructive Lung Disease (GOLD): global strategy for the diagnosis, management and prevention of chronic obstructive pulmonary disease. Bethesda, MD: National Institutes of Health, National Heart, Lung, and Blood Institute, April 2001 Vieggi G, Pedreschi M, Pistelli F, et al. Prevalence of airways obstruction in a general population. Chest 2000; 117:339S– 345S Ware JE, Kosinski M, Bayliss MS, et al. Comparison of

2042

26 27 28 29 30 31 32 33 34 35 36

37 38 39 40

methods for the scoring and statistical analysis of the SF-36 health profile and summary measures: summary of results from the Medical Outcomes Study. Med Care 1995; 33: AS264 –AS272 Quanjer P. Lung volumes and forced ventilatory flows. Eur Respir J 1993; 6(suppl):5– 40 Gordis L. Epidemiology. Philadelphia, PA: WB Saunders, 1996 Garrat AM, Ruta DA, Abdalla MI, et al. The SF-36 health survey questionnaire: an outcome measure suitable for routine use within the NHS. BMJ 1993; 306:1440 –1444 Wilson D, Wakefield M, Taylor A. The South Australian Health Omnibus Survey. Prom J Aust 1992; 2:47– 49 Kazis LE, Anderson JJ, Mennan RF. Effect sizes for interpreting changes in health status. Med Care 1989; 27:S178 – S189 Hosmer D, Lemeshow S. Applied logistic regression. New York, NY: John Wiley, 1989 Calverley P. COPD: early detection and intervention. Chest 2000; 117(suppl):365S–371S Anto JM, Vermeire P, Vestbo J, et al. Epidemiology of chronic obstructive pulmonary disease. Eur Respir J 2001; 17:982–994 Fletcher CM, Peto R, Tinker CM, et al. The natural history of chronic bronchitis and emphysema. Oxford, UK: Oxford University Press, 1976 Gorecka D, Bednarek M, Nowinski A, et al. Diagnosis of airflow limitation combined with smoking cessation advice increases stop-smoking rate. Chest 2003; 123:1916 –1923 Van den Boom G, van Schayck CP, Rutten MPMH, et al. Active detection of chronic obstructive pulmonary disease and asthma in the general population. Am J Respir Crit Care Med 1998; 158:1730 –1738 Jackson H, Hubbard R. Detecting chronic obstructive pulmonary disease using peak flow rate: cross sectional survey. BMJ 2003; 327:653– 654 Halpin D. NICE guidance for COPD [editorial]. Thorax 2004; 59:181–182 Coultas DB. Passive smoking and the risk of adult asthma and COPD: an update. Thorax 1998; 53:381–387 Anto JM, Vermiere P, Vestbo J, et al. Epidemiology of chronic obstructive pulmonary disease. Eur Respir J 2001; 17:982–984

Clinical Investigations