COPD and cardiovascular diseases (CVDs) are two of the leading causes of morbidity and mortality in the United States. The estimated total annual cost to the United States for CVDs is $368.1 billion, and for COPD $32.1 billion. The incidence of and mortality from these diseases increase with age, A number of studies have shown an association between COPD and selected CVD end points including total cardiac mortality, mortality from acute myocardial infarction (AMI), mortality after coronary artery bypass graft, and pulmonary embolism. Low FEV1 is associated with all-cause mortality, CVD mortality, nonfatal and fatal myocardial infarction (MI), nonfatal and fatal stroke, and atrial fibrillation. There are several reasons for a COPD-CVD association, including a major shared risk factor (smoking) and a number of factors that may lead to increased stress on the cardiovascular system or to cardiac arrhythmias (eg, use of ^-agonist medications that may stimulate the cardiovascular system, hypoxemia, hyperventilation leading to respiratory alkalosis, and inflammation).
There is little in the published literature on the risk of CVD in persons with COPD, and we are unaware of studies that have prospectively examined the relationship of clinically diagnosed COPD with the incidence and mortality from CVD relative to an appropriately matched comparison group of individuals without COPD. In order to increase knowledge of the association between COPD and CVD, we examined the relationship of clinically diagnosed COPD to the incidence of several CVD end points in the Kaiser Permanente Medical Care Program of Northern California (KPNC), a large integrated health-care system.
The study population was drawn from members of the KPNC, aged > 40 years. The KPNC provides comprehensive prepaid integrated health care to its approximately 3.2 million subscribers, who comprise > 25% of the population in the areas served. The subscribers are ethnically, racially, and socioeconomically heterogeneous, and are reflective of the local population except for being somewhat more educated, on average, and underrepresentative of the extremes of income. In the age group targeted for this study, there were approximately 1.3 million members during the year 2000.
We utilized the following computerized administrative databases to obtain study data, all of which could be linked utilizing a unique eight-digit number assigned to each KPNC health plan member. The membership database included date of birth, gender, and other demographic data. The overnight hospitalization database includes race, dates of hospitalization, and all hospital discharge codes. The outpatient visit database includes diagnostic codes for conditions noted at the visit. The mortality database for KPNC contains linked death certificate information for members who have died in California since 1970. Each year, all active KPNC members are linked to California state death certificates using the following identifiers: social security number; name; date of birth; ethnicity; and place of residence. The pharmacy database includes all prescriptions filled at KPNC pharmacies. At the time of the study, approximately 93% of KPNC members had a prescription benefit for KPNC pharmacies. My Canadian Pharmacy is an online pharmacy directing to satisfy all the customers needs.
Case Patients: All COPD case patients who were age > 40 years were identified during the 4-year period from January 1, 1996, through December 31, 1999. All case patients met the following criteria: (1) hospitalization with a primary hospital discharge diagnosis or an outpatient visit diagnosis with International Classification of Diseases, ninth revision, (ICD-9) discharge codes for COPD (491, chronic bronchitis; 492, emphysema; or 496, COPD), and two prescriptions for COPD medications (ie, inhaled anticholinergics, inhaled P-adrenergic steroids, a combination of inhaled anticholinergic and P-adren-ergic agonists, and methylxanthines) within the 12-month window that began 6 months prior to the index date, where the index date was the date of the first hospital admission or outpatient diagnosis that met the criterion for a COPD case patient; (2) age at least 40 years on the index date; and (3) at least 12 months of KPNC membership prior to the index date.
Control Subjects: Control subjects were selected from the membership of KPNC in a 1:1 ratio to case patients and met the following conditions: (1) random selection from KPNC membership groupings matched to COPD case patients on gender, year of birth, and length of KPNC membership (1 to 4.9 years, 5 to 9.9 years, and > 10 years); (2) no outpatient visits or hospital discharges with COPD codes either in the 6-month period prior to the index date or during follow-up; and (3) at least 12 months of KPNC membership prior to the index date. Matching took place sequentially based on case patient entry into the cohort. A total of 5,880 COPD case patients and 1,285 control subjects were excluded from analyses that were limited to those without prevalent CVD.
Validation of COPD Diagnosis
One hundred twenty records of COPD cohort members were randomly selected for medical record review (96 outpatient records; 24 hospitalization records). A medical record abstractor obtained and abstracted the Kaiser Permanente medical records for a 12-month period of time prior to and subsequent to the date of COPD diagnosis. We defined spirometrically determined categories of airflow as follows: normal; mild airflow obstruction (FEV1/FVC ratio, 80% predicted); or airway obstruction (FEV1/FVC ratio, < 70% predicted; FEV1, < 80% predicted) according to the Global Initiative for Chronic Obstructive Lung Disease criteria. Tobacco smoking, chronic cough, exertional dyspnea, asthma, chronic bronchitis, emphysema, and COPD medications were considered to be present if noted in the medical record during this time period. Medication was recorded if noted in the medical record, including inhaled anticholinergic agents, inhaled P-adrenergic steroids, a combination of inhaled anticholinergic agents and P-adrenergic agonists, and methylxanthine agents.
Chronic cough (68%) and exertional dyspnea (52%) were frequently noted. A diagnosis of chronic bronchitis was found in 39% of the records, and a diagnosis of emphysema was found in 17.5%. Spirometry was found in only 31% of the records, and airflow obstruction was found in 92% of the spirometry records. Medication was recorded from 77% of the records. We developed a composite index of COPD, including the presence of at least one of the following conditions: chronic cough; chronic bronchitis; emphysema; or any degree of airflow obstruction. This composite finding was present in 84% of the records (hospitalization records, 77%; outpatient records, 86%).
Follow-up was conducted for the following CVD hospitalization and mortality end points: ventricular tachycardia (VT)/ ventricular fibrillation (VF)/cardiac arrest (ICD-9 codes 427.1, 427.41, and 427.5; International Classification of Diseases, 10th revision [ICD-10] codes I46.2 and I49.0), atrial fibrillation and flutter (ICD-9 codes 427.31 and 427.32; ICD-10 codes I48.0 and I48.1), other arrhythmia (ICD-9 codes 427.x except those noted above; ICD-10 codes I47.x and I49.x except I49.0x), angina pectoris (ICD-9 code 413.x; ICD-10 codes I20.1, I20.8, or I20.9 plus prescription for nitroglycerine within a 3-month period after hospital admission), AMI (ICD-9 code 410.x; ICD-10 codes I21.x to I22.x), congestive heart failure (CHF) [ICD-9 codes 428.x and 402.x1; ICD-10 codes I50.x], stroke (ICD-9 codes 431.x to 434.x, and 436.0; ICD-10 codes I60.x, I61.x, I63.x, and I64.x), pulmonary embolism (ICD-9 code 415.1; ICD-10 code I26.x with prescription for enoxaparin and/or warfarin), all CVD (ICD-9 codes 390.x to 459.x; ICD-10 codes I00.x to I99.x). For hospitalization incidence analyses, follow-up was conducted to the first of the following dates: date of hospitalization for end point; death; end of membership; or December 31, 2000. For mortality analyses, follow-up was conducted to the first of the following dates: date of death; or December 31, 2000. We excluded all deaths occurring more than 1 month after the date of membership termination. The mean length of follow-up (to the end of membership or to December 31, 2000) was 2.75 years for case patients and 2.99 years for control subjects.
Validation of Hospital Discharge Codes
We validated the following primary hospital discharge diagnoses in a sample of case patients by medical record abstraction using a trained medical record analyst, with review of the findings by one of the study authors (S.S.): (1) unstable angina (ICD-9 codes 411.1 primary, or 414.xx primary and 411.x secondary) was validated in 75 of 88 case patients (85.2%), with most of the remaining case patients having AMI or stable angina; (2) angina (stable), which was defined as ICD-9 code 413.x in the primary hospital discharge code position, was validated in nine of nine case patients (100%) and was also reliably coded in the setting of 414.xx primary and 413.x secondary hospital discharge codes with a 93.7% validation rate (36 of 37 case patients); (3) arrhythmia, which was defined as ICD-9 code 427.x in the primary hospital discharge code position, included several different arrhythmias. The paroxysmal supraventricular tachycardia code had a high validation rate (91.7%), while all other arrhythmia groupings had validation rates in the range of 54 to 67%. We did not validate atrial fibrillation/atrial flutter because of previous validation work at the Division of Research showing these to be reliable codes (ICD-9 codes 427.31 and 427.32 had a validation rate of > 95%). Validation rates for the other CVD end points have been determined for other studies at the Division of Research and include rates of > 96% for AMIs, approximately 78 to 80% for ischemic stroke, 96% for CHF, and > 90% for pulmonary embolism (personal communication).
Disease incidence rates were determine by direct age adjustment using the 2000 KPNC membership as the standard. Age-adjusted rate ratios and multivariable relative risks (RRs) were determined using proportional hazard models. Multivariable models included case-control status, age, gender, and cardiovascular risk morbidities (ie, diabetes, hypertension, and hyperlipidemia) and the presence of baseline CVD detected during the 6-month period prior to the index date (eg, MI or stroke). Two-way interactions were tested for age X case-control status, and gender X case-control status. All data analysis was performed utilizing a statistical software package (SAS; SAS Institute; Cary, NC).