Between January 1996 and July 1999, a total of 1,971 outpatients were recruited from 24 departments of geriatrics or respiratory medicine within the context of the Salute Respi-ratoria nell’Anziano (Sa.R.A.) [respiratory health in the elderly study]. Details on the Sa.R.A. project are available elsewhere; the Sa.R.A. study is a multicenter Italian project investigating various aspects of chronic airway diseases in the elderly population (age > 65 years) attending pulmonary or geriatric outpatient clinics. Participating centers were geriatric (n = 19) or respiratory medicine departments (n = 5) of university or major teaching hospitals. Researchers had specific and extensive training in respiratory function in the elderly and multidimensional geriatric assessment.
Enrollment was on a consecutive basis up to the achievement of a target number of approximately 200 COPD and 200 asthmatic patients. The study also enrolled as a control group outpatients aged > 65 years and attending the geriatric clinics for nonrespiratory conditions, the most common of which were hypertension (27.3%), arthritis (27%), diabetes mellitus (12.7%), coronary artery disease (11.4%), and cerebrovascular diseases (7%). Data from individual centers were collected by a coordinating center at the Cattedra di Malattie dell’Apparato Respira-torio of the University of Palermo, which was also responsible for the quality control, the retrieval, and the final processing of data.
All the patients underwent a multidimensional geriatric assessment covering several areas: social and environmental status; personal history of smoking habit; disease-specific health status rated by the St. George respiratory questionnaire (SGRQ); quality of sleep (Established Population for Epidemiologic Studies of the Elderly questionnaire); treatment regimen; physical functioning as expressed by the 6-min walking test (6MWT) and the Barthel index; cognitive function rated by the Mini Mental State Examination (MMSE); and mood status assessed by the 15-item geriatric depression scale (GDS). The clinical history related to pulmonary diseases was recorded using a respiratory questionnaire derived with some modifications from the International Union Against Tuberculosis and Lung Disease questionnaire.
Information on drug therapy at study enrolment was collected from all patients and recorded at class category level (eg, long-acting p2-adrenergic drugs, ventolin inhalers, inhaled corticosteroids). If this was not possible, the information was gathered from a structured interview with the patient and/or the caregiver.
Patients underwent a complete physical examination. Co-morbid diseases identified on the basis of history and physical examination were recorded according to the International Classification of Diseases, Ninth Revision, Clinical (ICD9), and the Charlson index was calculated. All centers were provided with an identical fully computerized water-sealed spirometer (Stead-Wells, Baires System; Biomedin; Padua, Italy) that met the standards of the American Thoracic Society for diagnostic spirometry. Pulmonary function tests included baseline spirometry and postbronchodilator reversibility performed using fenoterol, 100 mg, administered throughout a space chamber. Tests were performed with a standardized technique in all centers, and a quality control process was successfully implemented. All the centers achieved high-quality performance in spirometry.
Using an algorithm of classification published elsewhere, we made a diagnosis of asthma if the patient had either:
• FEV1% > 80% of predicted and a history of wheeze in the last year, provided that no other potential cause of wheeze could be recognized (eg, congestive heart failure or goiter); no chronic cough or sputum;
• FEV1 12% after fenoterol and ventolin, with or without history of wheeze in the last year, and chronic cough and sputum absent.
We were aware that these stringent diagnostic criteria might have led to the exclusion of the minority of asthmatic patients with chronic cough and phlegm. However, we judged that it would have been preferable to err on the side of underdiagnosis rather than of overdiagnosis. This view was supported by the uncertainty about whether chronic cough and phlegm identifies a distinctive phenotype of asthma or a mixed asthma-COPD condition.
All the subjects were followed up throughout January 30, 2002, with regard to vital status (and cause of death) by contacting the registry office of the last municipality of residence. Information on vital status was obtained for a total of 1,656 subjects (84.0% of the original series): the sample included 13.4% of asthmatics and 65.4% of control subjects. People unavailable for follow-up were similar to those included in the sample with respect to pulmonary function measures, but they had a higher level of comorbidity (10.7% with a Charlson index > 3, compared to 5.2% of those successfully tracked down), were more frequently disabled (17.9% vs 9%) and cognitively impaired (35.7% vs 12.5%). We excluded 184 participants because they could not be classified for various reasons (n = 90) or because they had nonobstructive chronic bronchitis (ie, cough and sputum present for > 3 mo/yr in 2 consecutive years, but FEV1 > 80% of predicted; n = 94). Among the remaining 1,472 participants, 210 had asthma, 239 had COPD, and 1,023 had no bronchial obstruction. Given our study question, we excluded from our analysis the group with COPD. Follow-up time was calculated from the date of recruitment (first visit) until the date of death or January 30, 2002. This analysis covers a period of 57.9 ± 16.9 months (mean ± SD) [range, 1 to 80 months]. In the asthma group, 126 patients (60% of the initial cohort) were followed up to 60 months; 600 patients (61% of the initial cohort) were followed up to 60 months in the control group. Data on survival and on the causes of death for people who died during the follow-up period were collected by the Osservatorio Epidemiologico of Lazio Region (Rome, Italy). Coding of causes of death was performed by one researcher belonging to the same institution using the ICD9.
The study design was approved by the Ethical Committees of the participating institutions. Patients gave their written consent to participate in the study.
We first compared demographic and clinical characteristics of people with asthma with those of control nonrespiratory subjects using the mean (SD) for continuous variables and proportion for discrete variables. We used the product-limit method to calculate the risk of death in the two groups, and then a multiple Cox regression model to estimate the death rate ratio (hazard ratio [HR]) of people with asthma compared to that of the control group, corrected for potential confounders. Since the control group was free from pulmonary disease, this analysis was not corrected for respiratory functional parameters to avoid collinear-ity with diagnosis of asthma.
We identified the characteristics associated with mortality in both groups by univariable Cox regression, testing the following variables: demography (age, gender); smoking habit (eversmokers vs neversmokers); physical (6MWT) and cognitive (MMSE) performance; mood status (GDS); FEV1 and FVC percentage of predicted); FEV1/FVC; body mass index (BMI) [ratio of weight to squared height]; Charlson index of comorbidity; and, only in people with asthma, SGRQ and drug therapy with inhaled corticosteroids or ventolin inhaler or P2-adrenergic drugs. These variables were chosen because all are potentially related to death. We categorized the variables for which an accepted cutoff value is available: MMSE (cutoff = 24), GDS (cutoff = 5), BMI (cutoff for underweight = 22, cutoff for overweight = 29). To obtain clinically meaningful results, we changed the scale of some of the continuous variables: age was modeled using 5-year increments; FEV1, FVC, and FEVj/FVC ratio using increments of 5%; 6MWT using increments of 10% of walked distance expressed as percentage of predicted; and SGRQ using increments of 4 points (ie, the minimal clinically significant change). The variables found to be associated with the outcome were entered in a multivariable Cox proportional hazard model to estimate the individual contribution of each variable to the mortality rate. All analyses were performed using statistical software (STATA V5; Stata Corporation; College Station, TX; SAS V9; SAS Institute; Cary, NC).