Design, setting and study population
Study design has been extensively described elsewhere [14]. In brief, this is a closed population-based prospective cohort study including 27,204 individuals 60 years or older assigned to nine primary care centers in the Health Region of Tarragona (a mixed residential-industrial urban area in the Mediterranean coast of Catalonia, Spain). The study was approved by the ethical committee of the Catalonian Health Institute (P09/49) and was conducted in accordance with the general principles for observational studies. In this first-time report, we analyse the primary end-points resulting from the first year of survey. Cohort members were followed since the start of the study (December 1, 2008) until the occurrence of any event, change in pneumococcal vaccination status, disenrollment from the primary care center, death, or until the end of first 12-month follow-up (November 30, 2009).
Data sources
All participating primary care centers have a computerized clinical record system that includes administrative data, medical conditions, prescriptions, laboratory results and diagnosis associated with hospital and outpatient visits. This electronic clinical record system was used to classify cohort members by their pneumococcal vaccination status as well as to identify comorbidities or underlying conditions and establish baseline characteristics of the cohort at study entry. The hospital diagnosis discharge databases of the two reference hospitals in the study area (Joan XXIII and Santa Tecla), coded according to the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9) were used to identify study events.
Outcomes
The primary outcomes were hospitalisation for community acquired pneumonia, acute myocardial infarction, ischaemic stroke and death from any cause. Outcomes were initially identified on the basis of listed ICD-9 diagnosis codes for pneumonia (480 to 487.0), myocardial infarction (410) and ischaemic stroke (433, 434, 436 and 437). Cases were only definitively included if, on conclusion of the medical record review, the physician reviewer (two specialist family physicians previously trained) verified the diagnosis according to criteria mentioned below.
Pneumonia was defined when a new radiological infiltrate was identified with one major criteria (cough, expectoration or fever) or two minor criteria (dyspnea, pleuritic pain, altered mental status, pulmonary consolidation on auscultation and leukocytosis) [14].
Acute Myocardial Infarction was defined as a detection of rise and/or fall in cardiac biomarkers together with at least one of the following: symptoms of ischaemia, ECG changes indicative of new ischaemia (new ST-T changes, new left bundle branch block and/or development of pathological Q waves) and/or imaging evidence of new loss of viable myocardium or new regional wall motion abnormality [15].
Ischaemic stroke was considered when a patient rapidly developed signs of focal or global disturbance of cerebral function lasting more than 24 h (unless interrupted by surgery or death), with no apparent nonvascular cause and a neuro image showing an ischaemic brain lesion [16].
Vaccination history
Pneumococcal vaccination status was determined by a review of the primary care centers' electronic clinical records, which contain specially designated fields for pneumococcal and influenza vaccinations. We assumed that information in the computerized clinical records system (working since 1999) was complete, so a subject was considered as unvaccinated when a vaccination was not recorded. Cohort members were classified as vaccinated against pneumococcus if they had received at least one dose of PPV23 in the last 60 months before study start.
Statistical analysis
Incidence rates were calculated as person-years, considering in the denominator the sum of the persons-time contributed to each individual during the study period. Baseline characteristics according to pneumococcal vaccination status were compared using Chi-squared test.
Cox proportional hazards models were used to assess the association between having received the pneumococcal vaccine and the time to the first outcome [17]. The final models were adjusted for significant and confounder variables.
The following variables were considered in all the initial models: age, sex, number of outpatient visits to family physician in 12-months before study start (< 3,3-5,6-9,≥10), influenza vaccination in prior autumn, history of coronary artery disease (myocardial infarction or angina), history of stroke, history of chronic heart disease (congestive heart failure, hypertensive heart disease, cardiomyopathy, valvulopathy, cardiac dilatation or ventricular hypertrophy), chronic pulmonary disease (chronic bronchitis, emphysema and asthma), hypertension, hypercholesterolemia, obesity, diabetes mellitus, smoking status (non-smoker, quit, current), alcoholism, chronic severe liver disease(chronic viral hepatitis, alcoholic hepatitis and cirrhosis), chronic severe nephropathy (nephrotic syndrome, renal failure, dialysis or transplantation), cancer (solid organ or haematological neoplasia), dementia and nursing-home residence. Age, sex and influenza vaccine status were judged epidemiologically relevant variables, being included in all the final models. The variables were time-invariant and defined at study entry.
The authors checked for confounders and multicolinearity among the independent variables. The proportional hazard assumptions were assessed by adding the covariate by time interactions to the model and plotting the scaled and smoothed Schoenfeld residuals obtained from main effects model, where possible. All results were expressed with 95% confidence intervals (CIs). Statistical significance was set at p < 0.05 (two-tailed). The analyses were performed using Stata/SE version 11.1 for Windows (StataCorp. LP).