Blood Pressure. 2010; 19: 67–74
Chronic obstructive pulmonary disease and cardiovascular mortality in elderly subjects from general population
ALBERTO MAZZA1,3, SERGIO ZAMBONI1, DOMENICO RUBELLO2, LAURA SCHIAVON1, SARA ZORZAN1 & EDOARDO CASIGLIA3 1Department
of Internal Medicine – General Hospital of Rovigo, Italy, 2Unit of Nuclear Medicine, General Hospital of Rovigo, Italy, 3Department of Clinical and Experimental Medicine, University of Padova, Italy
Abstract Aims. To ascertain whether chronic obstructive pulmonary disease (COPD) is an independent risk factor for cardiovascular (CV) mortality in the elderly subjects from general population. Methods. 3282 subjects of the Northern Italy aged 65 years were followed up for 12 years in the frame of the CArdiovascular STtudy in the ELderly. Multivariate stepwise proportional hazard Cox regression was therefore used to identify the prognostic role of COPD on CV mortality in hypertensive (HT) and normotensive (NT) subjects. The hazard ratio (HR) of COPD with 95% confidence interval (CI) for mortality was adjusted for confounders in both genders. Results. COPD resulted to be an independent predictor of CV mortality (HR 1.34, CI 1.13–1.61) in HT but not in NT subjects. This was evident both in men (HR 1.44, 1.25–1.95) and women (HR 1.32, CI 1.14–1.53); pulse pressure (PP) was directly related and anti-hypertensive therapy inversely related to risk of CV mortality, an association that was greater in subjects with than without COPD. Conclusion. COPD should be included in the computation of global risk in HT subjects. PP is the main BP component in increasing CV risk in subjects with COPD. Controlled trials should be performed to evaluate the pressor targets to be reached in HT subjects with COPD, with the aim of decreasing their CV risk. Key Words: Arterial hypertension, chronic obstructive pulmonary disease, elderly, mortality
Introduction Chronic obstructive pulmonary disease (COPD) represents an increasing burden worldwide, the fourth cause of death (1) projected to reach the third place by 2020 (2). High mortality observed in subjects with COPD is in part attributable to comorbidities, also representing the main cause of hospitalization (3,4). Diabetes mellitus, congestive heart failure and chronic renal failure are known to be comorbidities increasing cardiovascular (CV) risk in subjects with COPD (5). Nevertheless, very little is known about the relative impact of COPD and arterial hypertension on mortality. This is particularly true in the elderly, where hypertension represents the main comorbid condition (4). Moreover, data concerning the prognostic role of COPD in hypertensive subjects mainly derive from data collected in middle-aged hospitalized patients (6) examined in retrospective
analysis (7), while few evidences are present in literature about the role of COPD in elderly hypertensives from general population (8,9). Although the increased risk of mortality in subjects with COPD is not surprising, the coexistence of hypertension and old age increase per se the risk of CV mortality. Furthermore, subjects with COPD are frequently treated with drugs – such as corticosteroids and β2-stimulating agents – that, as mentioned in the last ESH/ESC guidelines, limit the effect of some anti-hypertensive drugs. The rationale of the present epidemiological study was to stimulate the research to optimize the management of hypertensive subjects with COPD, also in light of the progressive population aging. For this aim, the predictive role of COPD on CV mortality was investigated at a population level in a cohort of elderly subjects taking part in the Cardiovascular Study in the Elderly (CASTEL). The 12-year results are described herein.
Correspondence: Alberto Mazza, Department of Internal Medicine, General Hospital of Rovigo, Viale Tre Martiri, 140, I–45100 Rovigo, Italy. Tel: +39-0425-394567. Fax: +39-0425-394157. E-mail: [email protected]
(Received 20 July 2009; accepted 3 November 2009) ISSN 0803-7051 print/ISSN 1651-1999 online © 2010 Informa UK Ltd. (Informa Healthcare, Taylor & Francis AS) DOI: 10.3109/08037050903464642
A. Mazza et al.
Materials and methods Data collection The CASTEL enrolled 3282 subjects (1281 men and 2001 women) aged 65 years or over, representing 73% of elderly subjects from the Northern Italian towns of Castelfranco Veneto and Chioggia. The study was approved by the CASTEL ethics committee and all the procedures were in accordance with the Helsinki declaration and with institutional guidelines. Each subject gave informed consent to the study. Details on recruitment and procedures for baseline examination and follow-up have been described elsewhere (10). Briefly, at the initial screening historical data, medical examination, including anthropometric and laboratory measurements, resting heart rate, blood pressure (BP) and electrocardiogram, were recorded. Forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) were measured with an automated wedge spirometer; a printed record was available immediately after the test, and the measurement was repeated after 5 min if the result was unreliable or the patient did not cooperate. FEV1 (in per cent of theoretical value) was corrected by gender using the algorithm proposed by Cherniack & Raber (11): FEV1 = 0.091 × heightinches − 0.023 × ageyears − 1.507 for men FEV1 = 0.060 × heightinches − 0.194 × ageyears − 1.87 for women. At the initial screening, blood pressure was measured eight times in a 3-month period (three times at the first impact with the patient, three times 1 month later and two times another month later); in order to avoid any white coat/alert reaction, only the average of the last two measurements was considered for data analysis. Periodical investigations in individual patients were scheduled annually to determine how many subjects developed hypertension during the follow-up, and to ascertain whether later antihypertensive treatment was necessary. No drop-out was observed during the follow-up. Subjects were labelled as hypertensive when systolic BP was 140 mmHg or diastolic BP was 90 mmHg or when treated. Pulse pressure was the difference between systolic and diastolic. Anti-hypertensive drugs used in the CASTEL were clonidine (15 µg/ day), nifedipine (20 mg/day from a slow-release preparation), verapamil (240 mg/day), atenolol (100 mg/day) or hydrochlorothiazide (25 mg/day), sometimes combined. There was no significant difference about the number of anti-hypertensive drugs taken in subjects with and without COPD. Diagnosis of COPD has completely been modified during the last decade. In the past, antiobstructive medications – like bronchodilators,
steroids or anti-cholinergic drugs – were included in the diagnosis of COPD. However, bronchodilators and steroids are not specific medications for COPD, as they may be used for other respiratory diseases, such as asthma and bronchiectasia. For this reason, the guidelines of the American Thoracic Society and European Respiratory Society requires FEV1 and FVC measurements for the diagnosis of COPD by spirometry. According to these joints guidelines, subjects having a FEV1/FVC ratio 0.7 were considered as having COPD (12,13). Body mass index (BMI) was calculated in kg/m2 from weight/squared height ratio. Subjects were considered diabetic if taking anti-diabetic drugs or having fasting blood glucose level 7.0 mmol/l in at least two separate samples without therapy (14). Fasting lipid concentrations were enzymatically. Subjects were classified according to whether they were current smokers (1 cigarette daily) or never smokers. Creatinine clearance (CrCl, in ml/min) was calculated from serum creatinine (Scr, in mg/dl) using the formula of Cockcroft & Gault (15); this equation measured CrCl accurately also in the elderly (16) and was corrected for the female gender by multiplying CrCl by 0.85. All subjects were screened for albuminuria with a dipstick using a random morning urine sample, as 24-h samples are impractical for population screenings because of inconvenience and frequent collection errors (17); urine infection was previously excluded. Subjects with proteinuria 300 mg/l were defined as proteinuric. Electrocardiogram was analysed on the basis of the Minnesota code (18) by an expert who did not know the aim of the study. Subjects with at least one of the following items at baseline: Minnesota code equal to 1.1 or 1.2 or 1.3 (if absent 6.4.1) or 4.1 or 4.4 (if absent 6.4.1, 7.1.1 and 7.2.1) or 5.1 or 5.2 or 5.3 or 5.4 (if absent 6.4.1, 7.1.1, 7.2.1 and 7.4), positive myocardial scintigraphy, a positive stress test, a history of myocardial infarction confirmed by hospital files, a history of angina pectoris confirmed by hospital of physician’s files or appropriate anti-anginal long-term treatment, were labelled as having coronary heart disease (CHD). Subjects having a Minnesota code equal to 3.1 or 3.3 were considered to have left ventricular hypertrophy (LVH).
Mortality Annual mortality was monitored over a period of 12 years according to the Register’s Office that is responsible for recording of all death and for maintaining a complete central archive. Death certificates were verified, and causes of death were doublechecked by analysing hospital files, retirement home files and physicians’ files. All Italian hospitals have an archive, and no deaths can be missed in their database. All records were coded according to the Inter-
COPD in arterial hypertension national Classification of Disease, ninth revision, by a trained research physician supervised by another expert, to determine the cause of death accurately. If necessary, a third physician was contacted to resolve any problem with uncertain data. Mortality was defined as CV when deriving from coronary disease (code 410–414), stroke (436), heart failure (398.91, 402.11, 404.11, 404.13, 404.91, 404.93, 428.0– 428.11), arrhythmias (427) or pulmonary embolism (415.1). No cause of death was missed.
and time and the effect of the covariate on survival was dependent on time. COPD, gender, age, pulse pressure, LVH, smoking, CHD, congestive heart failure, CrCl, serum total cholesterol (TC), lowdensity-lipoprotein serum cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), proteinuria and anti-hypertensive treatment were included as covariates in the analysis of mortality. As gender was accepted in the Cox models for mortality (see below), the analysis was performed separately in each gender.
Statistical analysis Only the 3257 subjects having complete data were included in the statistical analysis. Continuous variables were averaged, expressed as mean and standard deviation, and compared with analysis of covariance and the Bonferroni’s post-hoc test. Categorical variables were expressed as percentage rates and compared with the Pearson’s χ2 test. Annual mortality rates were compared with the Kaplan–Mayer approach after generating cumulative survival curves. Multivariate stepwise proportional hazard Cox regression was used to identify the variables having a prognostic role on CV mortality (19). Hazard ratio (HR) of mortality with 95% confidence intervals (CI) was calculated for each item and adjusted for confounders. Covariates were selected because of their potential relation with COPD or mortality. The proportionality assumption of the Cox model was previously tested by adding a time-dependent covariate for each variable, so that the conditional hazard at each point in time was a function of the covariate
Results The general characteristics of normotensive and hypertensive subjects are summarized in Table I. The latter had in comparison with the former significantly higher BMI, systolic and diastolic BP, pulse pressure, heart rate and serum lipids, higher prevalence of smoking, diabetes, historical CHD and LVH. The normotensives also had higher levels of CrCl than the hypertensives. General characteristics of the hypertensive subjects with and without COPD are shown in Table II, also showing gender stratification. Among subjects with COPD, systolic and pulse BPs were significantly higher in treated than in untreated ones (166.1 vs 163.1 mmHg, p 0.0001; and 75.8 20.6 vs 73.2 18.1, p 0.03). This was evident in both men and women (data by gender not shown). During the 12 years of follow-up, there were 1599 deaths (713 in men and 886 in women); 862 of them were CV (336 in men and 526 in women). In
Table I. General characteristics of the population.
Age (years) BMI (kg/m2) SBP (mmHg) DBP (mmHg) PP (mmHg) HR (beats/min) CrCl (ml/min) TC (mmol/l) HDL-C (mmol/l) LDL-C (mmol/l) TG (mmol/l) FEV1 (%) COPD (%) Smoking (%) Diabetes (%) CHD (%) CHF (%) LVH (%) Proteinuria (%)
Entire population (n = 3257)
Normotensive (n = 451)
Hypertensive (n = 2806)
p-value, BP status
73.7 ± 5.2 26.3 ± 4.3 159.8 ± 24.7 88.7 ± 11.5 71.1 ± 19.3 76.0 ± 11.5 64.3 ± 23.6 5.6 ± 1.1 1.5 ± 0.4 3.6 ± 1.0 1.4 ± 0.8 78.1 ± 24.6 42.1 12.8 17.0 8.6 10.2 6.8 11.0
73.8 ± 5.8 24.7 ± 4.1 126.8 ± 8.7 76.4 ± 7.21 50.4 ± 8.9 73.4 ± 11.0 64.8 ± 23.9 5.3 ± 1.1 1.4 ± 0.4 3.2 ± 1.0 1.3 ± 0.7 81.9 ± 26.8 48.5 10.9 11.9 6.2 9.1 6.3 10.4
73.7 ± 5.1 26.6 ± 4.3 165.1 ± 22.2 90.7 ± 10.8 74.4 ± 18.5 76.4 ± 11.6 61.1 ± 21.7 5.7 ± 1.1 1.5 ± 0.4 3.5 ± 1.0 1.4 ± 0.8 74.6 ± 24.7 41.1 24.6 17.8 9.0 11.2 14.2 11.2