Are patient characteristics helpful in recognizing mild COPD (GOLD I) in daily practice?

Abstract

Objective. To determine whether in a high-risk group of middle-aged male current smokers, patient characteristics are useful to recognize mild COPD (GOLD stage I). Design. In a cross-sectional study spirometry was performed according to the American Thoracic Society criteria. COPD was defined according to the GOLD criteria for COPD. Setting. Primary care. Subjects. Male smokers, aged 40–65 years, without documented lung disease in the population at large. Main outcome measures. Medical records were scrutinized to collect patient characteristics. Multiple logistic regression analysis was used to identify independent determinants of mild COPD. Results. A total of 567 subjects participated. COPD, defined by a FEV1/FVC ratio < 0.7, was detected in 170 subjects (30.0%, 95% CI 26.2–33.9%). In 149 subjects (26.3%; 22.7–30.1%) COPD was mild (GOLD stage I) and in 21 subjects (3.7%; 2.3–5.6%) moderate (GOLD stage II). Only age and cough were independently associated with the presence of mild COPD. The ability of these determinants to discriminate between subjects with or without mild COPD was relatively poor (ROC area 0.65). Therefore no prediction rule was developed. Conclusion. Our results indicate that patient characteristics are not helpful to recognize mild COPD (GOLD stage I) in middle-aged male smokers.

• COPD
• family practice
• middle-aged
• patient characteristics
• screening
• smokers

Chronic obstructive pulmonary disease (COPD) is a major health problem. The estimated prevalence of COPD, defined as a FEV1/FVC ratio < 0.7, in the population at large in the US varies from about 3% in never smokers to 14% in both male and female current smokers [1]. In only a small proportion of patients is COPD recognized as such by their primary care physician. For example, the prevalence of physician-diagnosed COPD in the UK is 1.4% among women and 1.7% among men [2]. The diagnostic criteria for COPD according to the GOLD guidelines are currently widely accepted, although the one-dimensional severity grading – i.e. based only on lung function – is discussed [3–5]. Relatively novel in the GOLD Guidelines is GOLD stage I (mild COPD) defined by an abnormal FEV1/FVC (<0.7) ratio but with an FEV1 above 80% predicted. Identifying subjects with early stages of COPD is important since treatment of early COPD by intensive smoking-cessation programs has proven to be effective [6]. There is no consensus regarding an optimal screening strategy. Screening strategies involving all elderly smokers (>65 years) has been criticized because of the lack of evidence of a significant number of undiagnosed severe disease cases, the risk of false positive diagnosis, and the smaller benefit of smoking cessation [7–9]. Screening in the younger age group (<40 years) is not recommended because of the low prevalence of COPD. Various authors advocate targeted screening, e.g. spirometry in smokers 45 years or older who report smoking during a clinical encounter [10], [11]. However, screening of all middle-aged smokers seems not to be feasible because of shortage of resources. Screening of a subgroup of smokers with a high prevalence of COPD, identified for example by means of easily obtainable patient characteristics, would be more efficient. Therefore, the aim of the present study was to determine whether in a high-risk group of middle-aged male smokers patient characteristics can be used to recognize mild COPD (GOLD stage I).

Chronic obstructive pulmonary disease (COPD) is a major health problem with an estimated prevalence of 10–15% among smokers.

• Patient characteristics are not helpful to recognize mild COPD (GOLD stage I) in middle-aged male smokers.

Material and methods

This study was part of the IJsselstein Study, a cohort study, aimed to assess the prevalence and determinants of undetected COPD in middle-aged smokers. Subjects with COPD were offered smoking cessation interventions. The study was conducted between 1998 and 2003.

Study population

In 1998 all men (n = 3985) aged 40 to 65 years and listed with a local GP in IJsselstein, a small city in the centre of the Netherlands, were asked by letter if they had smoked one or more cigarettes a day during the previous 12 months and if so whether they were willing to participate in a study to identify determinants of COPD. Subjects with documented lung disease, i.e. a diagnosis of asthma, COPD, or any other chronic pulmonary condition reported in the GP medical records (222/3985, 5.6%), were excluded. A total of 2596 of the 3763 men without previously documented lung disease responded (69%). Among the 2596 respondents 978 (37.7%) reported beeing current smokers. At the start of the 1998 survey 60 (6.1%) ‘current’ smokers were excluded because of smoking cessation more than 12 months previously (33) or because they smoked only a pipe or cigars (17); 10 subjects were excluded since they reported a chronic lung disease not recorded in the GP medical records (10). Thus the eligible population consisted of 918 current cigarette smokers without known chronic lung disease (35.4%). Eventually, 805 of these 918 eligible subjects (87.7%) attended the 1998 survey. During the examinations spirometry was performed, a short questionnaire was filled in by the participants, and smoking history, height, and weight were assessed by a nurse practitioner.

Spirometry

In all participants spirometry was performed using a hand-held spirometer (Vitalograph 2170). Each subject had to perform at least three acceptable forced vital capacity manoeuvres while sitting. The results were shown on a computer screen and the procedure was supported by computer software (Spirotrack). When FEV1 predicted was less than 85%, the bronchodilator response was tested by inhalation of two puffs of terbutaline 250 µg through an inhalation chamber in order to exclude previous unrecognized asthma. When FEV1 predicted, after an interval of 15 minutes, was still less than 85% a second bronchodilator test (after an interval of at least 30 minutes after two puffs of ipratropium bromide 20 µg), was performed on another day within a month. All measurements were performed by one experienced and specially trained nurse practitioner. The spirometer was calibrated with a one-litre syringe at least once a week. Two investigators (RMMG, APES) independently assessed the quality of the flow–volume curves and time–volume curves according to the criteria of the American Thoracic Society [12]. In the case of disagreement, a final assessment was made by a lung physiologist. The tracings with the largest sum of FEV1 and FVC were used in this analysis. In total 702 participants performed an adequate lung function test.

COPD was classified according to the 2003 update of the WHO/GOLD criteria [3], [4]. According to these guidelines COPD is defined by a FEV1/FVC ratio of < 0.7. The severity of COPD is distinguished in four stages:

• mild (GOLD stage I): FEV1 predicted ≥ 80%;

• moderate (GOLD stage II): 50% ≥ FEV1 predicted < 80%;

• severe (GOLD stage III): 30% ≥ FEV1 predicted < 50%;

• very severe (GOLD stage IV): FEV1 predicted < 30% (or < 50% with signs of chronic respiratory failure).

All cut-off values refer to post-bronchodilator measurements.

Medical records

All GPs were using computer-based medical records. From these medical records the following items were extracted retrospectively:

• number of clinical encounters because of lower respiratory tract complaints or because of any other reason in the previous 12 months;

• number of prescription of antibiotics because of lower respiratory tract complaints in the previous 12 months.

A clinical encounter because of lower respiratory tract complaints was defined as the presence of one of the following items in the medical records:

• a diagnosis of acute or chronic bronchitis, pneumonia or lower respiratory tract infection;

• symptoms such as coughing, dyspnoea, phlegm production, or wheezing, without specific diagnosis.

A diagnosis recorded by a primary or secondary care physician in the patient records or in the medical correspondence was classified as present. In total 135 of 702 (19.2%) medical records were missing, 60 (8.5%) records because one GP refused to participate and 75 (10.7%) records because the participants had moved to another city by the time we inspected the medical records.

Questionnaire

The Airways Questionnaire (AQ20), a short questionnaire to measure health-related quality of life among patients with chronic obstructive pulmonary disease, was filled in by the participants before the pulmonary function test [13]. The item concerning coughing (do you have coughing spells in the daytime, yes/no) was included in the current analysis.

Data analysis

The data analysis was performed in the 567 subjects with complete medical records. The association between all aforementioned determinants and the presence of COPD was assessed using univariate logistic regression analysis. Odds ratios (OR) and 95% confidence interval (95% CI) were calculated in subjects with mild COPD (GOLD I) using subjects with normal spirometry as the reference group. Determinants with a univariate p-value of < 0.15 were entered together in a multivariable logistic regression model to assess which determinant was independently associated with the presence of mild COPD. In order to retain a more reduced model only the strongest determinants of the presence of COPD were selected, excluding determinants with a p-value > 0.05 from the multivariate model. The reliability (goodness of fit) of the model was assessed by means of the Hosmer & Lemeshow test [14]. The ability of the model to discriminate between subjects with or without mild COPD was quantified using the area under the Receiver Operator Characteristic curve (ROC area) [15]. The ROC area is a parameter to summarize discriminative ability of a prediction model and can range from 0.5 (no discrimination) to 1.0 (perfect discrimination). A value ≥ 0.7 is considered to be reasonable and > 0.8 good.

All statistical analyses were performed using the statistical package SPSS (SPSS for Windows, version 11.0, SPSS Inc.).

Results

General characteristics of the participants are given in Table I. COPD (GOLD stage I or up) was established in 30.0% of the participants (95% CI 26.2–33.9). In 26.3% of the subjects (22.7–30.1) the airflow obstruction was mild (GOLD stage I) and in 3.7% of the individuals (2.3–5.6%) moderate (GOLD stage II). No one met the criteria for severe COPD (GOLD stage III or IV). In Table II unadjusted and adjusted odds ratios for potential determinants of mild COPD are presented. In the univariate analysis age, pack-years of smoking, and cough were associated with previously unrecognized mild airflow obstruction (GOLD I) while in the multivariate analysis only age and cough correlated independently with the presence of GOLD I. Entering these two variables in a prediction model resulted in an area under curve (ROC area) of 0.65 (95% CI 0.59–0.70) (Table II). The number of pack-years of smoking was strongly correlated with age and did not increase the ROC area. Although the model showed a good fit, indicated by a non-significant Hosmer & Lemeshow test (p > 0.5), a prediction rule was not developed because of the relatively poor performance of the combination of the determinants (ROC area 0.65) to predict the presence of unrecognized mild COPD.

Table I.  General characteristics of a cohort male smokers aged 40–65 years.

	All participants (n = 567)	Normal spirometry (n = 397)	GOLD I (n = 149)	GOLD II (n = 21)
Age (years)	50.2 (6.6)	49.2 (6.2)	52.3 (6.8)	53.5 (6.6)
Pack-years of smoking	24.8 (9.6)	23.4 (8.7)	28.2 (11.2)	27.3 (5.3)
BMI (kg/m^2)	25.6 (3.4)	25.8 (3.4)	25.1 (3.4)	25.6 (3.2)
FVC (% predicted)	110 (14)	108 (14)	117 (13)	94 (12)
FEV1 (% predicted)	100 (14)	103 (13)	95 (10)	70 (7)
FEV1/FVC	0.73 (0.07)	0.77 (0.04)	0.65 (0.04)	0.60 (0.05)
Reversibility (FEV% predicted)^1	5.3 (4.3)	–	5.8 (4.1)	4.0 (4.7)
GP contact (≥1) for all reasons, previous 12 months	60%	61%	55%	71%
GP contact (≥1) for LRT complaints, previous 12 months	18%	16%	22%	33%
Antibiotics (≥1 prescription) for LRT complaints, previous 12 months	11%	9%	14%	14%
Cough	19%	14%	29%	48%

Figures denote means (standard deviation), unless else stated; BMI = body mass index; FVC = Forced Expiratory Vital Capacity; FEV1 = Forced Expiratory Volume in 1 second; LRT: lower respiratory tract; ¹Reversibility was tested in 45/149 subjects with GOLD I and 20/21 of subjects with GOLD II.

Table II.  Association between patient characteristics and mild COPD (GOLD I) in smokers, aged 40–65 years, and ability (ROC area) of characteristics to discriminate between subjects with and without GOLD 1.

Patient characteristic		n	Mild COPD
		(546)	(GOLD I; n = 149)
			OR (95% CI) unadjusted^1	OR (95% CI) adjusted^1
Age (years)	40–44	138	1.0^2	1.4 (1.2–1.7)
	45–49	137	1.4 (0.8–2.5)	(per 5 years)
	50–54	130	1.5 (0.9–2.8)
	≥55	141	3.2 (1.9–5.6)
Pack-years	<20	159	1.0^2
	20–29	252	1.2 (0.7–2.0)
	≥30	135	3.0 (1.8–5.0)
GP contact (≥1) for all reasons, previous 12 months		323	0.8 (0.5–1.1)
GP contact (≥1) for LRT complaints, previous 12 months		96	1.5 (0.9–2.4)
Antibiotics (≥1 prescription) for LRT complaints previous 12 months		57	1.5 (0.8–2.7)
Cough		100	2.4 (1.5–3.8)	2.3 (1.5–3.7)
ROC area			0.65 (0.59–0.70)

¹Reference group: subjects without COPD (n = 397); ²reference category; n = number of subjects positive for the determinant; OR = odds ratio; CI = confidence interval; LRT + lower respiratory tract; ROC + Receiver Operating Characteristic curve.

Discussion

In our study among 567 male middle-aged smokers only age and cough were associated with the presence of previously unrecognized mild COPD (GOLD I). Combining these determinants did not satisfactorily predict the presence of mild COPD (ROC area 0.65).

The difficulty in predicting GOLD I by means of patient characteristics is at least partly attributable to the fact that unrecognized mild COPD is often asymptomatic (see Table I). In a second analysis of 409 subjects who underwent spirometry after five years’ follow-up and completed the MRC questionnaire, GOLD stage I was found in 123 individuals (30.1%). Approximately two-thirds were asymptomatic, i.e. not reporting cough, phlegm, dyspnoea, or wheezing. In addition neither a family history of COPD nor blue-collar employment was reported more frequently among participants with GOLD I compared with those without airflow obstruction. These findings are in line with previous reports indicating that mild and even moderate COPD remains to a large extent asymptomatic and that an FEV1 decline up to 50% of predicted does not correlate well with symptoms or other patient characteristics [16–18].

Some limitations of our study should be considered. First, only men were included. Because of the increased smoking rate of women in recent decades, the prevalence of (undetected) COPD in female smokers is rising [19]. The proportion of middle-aged female patients visiting the GP for respiratory complaints compared with middle-aged male patients is about 40% higher [20]. In addition, the prevalence of reported respiratory complaints is higher in female smokers than in male smokers [21]. Therefore, patient characteristics such as respiratory complaints could possibly be more helpful in recognizing mild COPD in female than in male smokers. Second, 805 subjects participated in our study and 702 of them (87.2%) produced an acceptable spirometric curve according to ATS criteria, which is comparable with other studies [10]. There were no relevant differences in the characteristics, listed in Table I, of subjects with and without acceptable curves. Third, of the 702 subjects with acceptable spirometric curves, 567 medical records (80%) were able to be collected. For obvious reasons our analysis was restricted to those smokers with complete medical records. Importantly, however, age, pack-years, and lung function parameters were comparable in subjects with and without retrievable medical records. Finally, age, cough, and ≥ 1 GP contact for lower respiratory tract complaints in the previous 12 months were independently associated with the presence of previously undetected moderate COPD (GOLD II) (data not shown). In contrast to our finding regarding GOLD stage I, it is likely that patient characteristics are useful to identify smokers with moderate COPD (GOLD II). The prevalence of GOLD II in our sample, however, was too low to develop such a prediction rule.

The major strength of our study is that a large proportion of the eligible male smokers, in a small town with a population representative of the Dutch population, participated. In addition, we collected patient characteristics in general practices without any special instruction to record medical history, in order to make the results of our study applicable to general practices at large in the Netherlands. Finally we used the currently widely accepted GOLD criteria to define COPD including post-bronchodilator values.

As far as we know, no other studies have been performed in the population at large aimed to develop a prediction rule based on patient characteristics in order to select middle-aged smokers for screening spirometry. In the literature several screening strategies are advocated, e.g. screening spirometry by primary care physicians among smokers aged > 45 years old, with or without lower respiratory tract complaints [10], [11]. According to our study, patient characteristics are of little use to select smokers for screening spirometry in order to identify mild COPD (GOLD stage I). Identifying subjects with mild COPD is important since further deterioration of the COPD to GOLD stage II was assessed in 20% of the subjects with GOLD stage I after five years in our study and treatment of early COPD by intensive smoking-cessation programs has proven to be effective [6]. The National Lung Health Education Program (United States) has recommended using office spirometry to screen for sub-clinical lung disease in smokers 45 years or older during a clinical encounter [22]. No published studies have evaluated the costs and benefits of the latter recommendation. Many of the available data on the effectiveness of screening are from studies that included smoking-cessation interventions, making it difficult to determine the effects of screening alone. There is no convincing evidence that spirometry, as an isolated intervention, aids smoking cessation [23], [24]. The balance of costs and benefits of screening for COPD among middle-aged smokers is probably less favourable than case findings among smokers visiting their GP for respiratory complaints due to the lower yields of a mass screening program and its higher additional costs. More studies are needed to assess cost-effectiveness and feasibility of various screening approaches in smokers to arrive at the optimal screening strategy.

Our results indicate that patient characteristics are not helpful to recognize mild COPD (GOLD stage I) in middle-aged male smokers.

The authors would like to thank the local GPs in IJsselstein for participating in the study and the Dutch Asthma Foundation for providing the financial support to fund the study. The study was approved by the ethics committee of the University Medical Centre Utrecht.