Aspirin may provide the best opportunity for secondary prevention after acute myocardial infarction because of its effectiveness, low cost, safety profile, and lack of strong contraindications. More than 18 000 patients have been enrolled in randomized trials to assess the role of antiplatelet therapy for secondary prevention after acute myocardial infarction [3]. Pooled analyses of these patients show that aspirin significantly reduces long-term vascular mortality. Although these trials did not enroll many very elderly patients, other studies [4, 5] have reported that aspirin does not pose a major hazard to elderly patients. The guidelines for treatment of acute myocardial infarction published by the American Heart Association and American College of Cardiology in 1990 strongly recommend the use of aspirin for all survivors of acute myocardial infarction to prevent late recurrence of myocardial infarction and death [6].

Although rates of adherence to the published guidelines for acute myocardial infarction are being evaluated with increasing frequency [7, 8], no indepth analysis has been done on the prescribed use of aspirin at hospital discharge. In an analysis of the short-term use of aspirin in elderly patients [9], we have shown that those who present without chest pain and those who present with high-risk characteristics are most likely to not receive aspirin within the first 2 hospital days. The factors associated with not being prescribed aspirin at discharge are not known and may focus efforts for improving the care of elderly survivors of acute myocardial infarction.

Accordingly, we sought to measure the use of aspirin as a discharge medication among eligible elderly patients hospitalized with an acute myocardial infarction and to identify patient characteristics associated with the decision to prescribe aspirin. We evaluated the associations of age, sex, race, severity of the myocardial infarction, and hospital interventions with the use of aspirin at discharge. Our secondary objective was to compare 6-month mortality rates among patients prescribed aspirin at discharge with rates among patients not prescribed aspirin, after adjustment for baseline differences between the groups. Although observational studies of this type will not supplant the many randomized trials that have established the efficacy of aspirin for secondary prevention, they do allow the rigorous study of patients in actual clinical practice, including many patients in groups not commonly represented in the clinical trials.

We used data from the Cooperative Cardiovascular Project (CCP) pilot, a Health Care Financing Administration initiative to improve care for Medicare beneficiaries who have had acute myocardial infarction and those who undergo invasive cardiovascular procedures [7]. As part of the CCP, the hospital charts of more than 16 000 Medicare patients hospitalized with an acute myocardial infarction were abstracted. The CCP is part of the Health Care Quality Improvement Program, a national effort to help local hospitals and practitioners improve the overall quality of care for Medicare beneficiaries [10].

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample

The study sample consisted of patients in the CCP myocardial infarction cohort. This cohort includes Medicare patients from the 352 nongovernment, acute care hospitals in Alabama, Connecticut, Iowa, and Wisconsin who had a principal discharge diagnosis of myocardial infarction (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] code 410) and were discharged between 1 June 1992 and 28 February 1993. These hospital admissions were identified using the Medicare Provider Analysis Record. We excluded admissions that were not related to the acute care of an acute myocardial infarction (the fifth digit of the ICD-9-CM code, 2) and admissions that resulted in a transfer to another institution.

The study sample was restricted to patients at least 65 years of age who had had a confirmed acute myocardial infarction and who were alive at hospital discharge. Acute myocardial infarction was defined as an elevated creatine kinase-MB level or meeting two of the following three criteria: chest pain, a twofold elevation of the creatine kinase level, or new Q waves on the electrocardiogram. We excluded patients if they had a terminal illness. A terminal illness was considered to be present if the patient's record documented that he or she was terminally ill or had a life expectancy of less than 6 months; if the admission orders indicated that the patient should be given palliative care only; or if a "do not resuscitate" order was written. Patients who had in-hospital resuscitation were not excluded. We also excluded patients who were prescribed warfarin at hospital discharge because fewer than 31% of these patients were taking aspirin, and many physicians consider treating elderly patients with both agents to be improper. In addition, we excluded patients with the following contraindications to long-term aspirin use: history of a hemorrhagic stroke, active bleeding during the hospital stay, history of gastrointestinal bleeding, presence of rectal blood, allergy to aspirin, history of a bleeding disorder, creatinine level greater than 265.2 µmol/L, platelet count less than 100 x 10⁹/L, and a hematocrit less than 0.30.

Data Collection

Trained nurses and medical record technicians abstracted the medical records using software developed as part of the Uniform Clinical Data Set System [11]. This is an automated system designed to provide standardized abstraction of data to be used for evaluating the quality of care for Medicare beneficiaries. Data were entered directly into the computer using this menu-driven database software. Strategies to decrease abstraction errors and variability included standardized training sessions, online data definitions, and range checks. We abstracted 594 charts a second time to test the reliability of ascertaining the use of aspirin in the hospital and found very high agreement, with a value of 0.966.

Outcome Variables

The outcome variable for our first objective was the frequency with which aspirin was prescribed at discharge. The principal end point for the second phase of the study was 6-month mortality; the 6-month period started on the day of admission. We obtained this information, which is derived from Social Security data, from Medicare's Enrollment Database. Follow-up was 100%.

Independent Variables

The independent variables included age; sex; race; medical history; preadmission, hospital, and discharge medications; clinical status at discharge; discharge disposition; hospital complications; and hospital procedures. Patients were placed into three strata according to age: 65 to 74 years, 75 to 84 years, and 85 years and older. Comorbid conditions included a chart-documented history of hypertension, diabetes mellitus, chronic obstructive pulmonary disease, renal dysfunction (defined as admission creatinine level more than 177 µmol/L), history of myocardial infarction before the index admission, malnutrition (albumin level less than 30 g/L), and dementia. Initial hospital treatment variables included the use of aspirin, ß-blockers, and thrombolytic therapy in the emergency department or in the hospital on day 1. Hospital treatment variables included the use of aspirin, ß-blockers, or thrombolytic therapy. Hospital procedure variables included cardiac catheterization, percutaneous coronary revascularization, and coronary artery bypass surgery. Variables that describe clinical events or patient characteristics at any time during the hospitalization included congestive heart failure, atrial fibrillation or flutter, stroke, recurrent chest pain, ventricular tachycardia, intubation, and creatine kinase levels more than four times the normal level. Left ventricular systolic dysfunction was indicated if the patient had a documented ejection fraction of less than 40%. Because the left ventricular ejection fraction was not available for 38% of the study sample, we created an indicator variable for the missing variable. Hospital length of stay was coded as longer than 12 days (yes or no), which was the 75th percentile for length of stay. Discharge medication variables included ß-adrenergic blocking agents. Discharge disposition was coded as discharge to the patient's home or to a care facility.

Statistical Analysis

For the first part of the study, we determined how frequently elderly patients who had had an acute myocardial infarction were being prescribed aspirin at hospital discharge. In a restricted cohort of patients whose medical charts did not list any contraindications to the long-term use of aspirin, we evaluated the bivariate associations of demographic and clinical variables with the prescription of aspirin at discharge. Then, using the variables from the bivariate analysis, we developed a multivariable logistic-regression model by backward stepwise selection, with the prescription of aspirin as the dependent variable. We dropped variables from the model at a significance level of P > 0.01.

For the second part of the study, we determined whether the prescribed use of aspirin at discharge was associated with better 6-month survival in an unadjusted model and after adjustment for potential confounders in a multiple logistic regression model. Covariates, which were selected on the basis of the previous analyses and clinical relevance, were forced into this model. Covariates included demographic characteristics (age, sex, and race), clinical history (preadmission use of aspirin, history of diabetes, dementia, chronic obstructive pulmonary disease, myocardial infarction, hypertension, renal dysfunction [creatinine level more than 177 µmol/L], peptic ulcer disease, stroke, and left ventricular ejection fraction less than 40%), a variable indicating whether a left ventricular ejection fraction had been measured, variables that described the clinical course in the hospital (use of aspirin, use of thrombolytic therapy, use of ß-blockers, congestive heart failure, ventricular tachycardia, endotracheal intubation, length of hospital stay more than 12 days, creatine kinase level more than four times normal, cardiac catheterization, percutaneous transluminal coronary angioplasty, and coronary artery bypass surgery), and the prescription of ß-adrenergic blocking agents at hospital discharge.

We constructed and examined partial residual plots to evaluate potential problematic areas of model fit [12]. Goodness-of-fit was evaluated by comparing fitted probabilities of 30-day mortality with observed 30-day mortality within deciles of risk and calculating the corresponding observed chi-square statistics [13]. To evaluate the discriminating power of the fitted model, we also calculated an area under the receiver-operator characteristic curve for each model [14]. All calculations were done using the STATA software system (STATA Corp., College Station, Texas).

Results

Top Methods Results Discussion Author & Article Info References

Study Sample

The CCP database consists of 16 189 patients (Table 1). After we excluded patients younger than age 65 years, patients in whom an acute myocardial infarction had not been confirmed, patients transferred to another institution, patients with a terminal illness or recent trauma, patients who had died during the hospitalization, and patients with contraindications to aspirin, 5490 patients remained in the study sample for subsequent analyses.

Prescribed Use of Aspirin

Of the 5490 hospitalized patients in the study sample, 4149 (76%) were discharged with instructions to take aspirin. Among these, aspirin therapy was newly started in 2644 patients (48%). Table 2 lists the use of aspirin at discharge and various demographic and clinical characteristics. Patients who were older; who were female; who had more comorbid conditions; and who did not receive thrombolytic therapy, ß-adrenergic blocking agents, or aspirin at admission were less likely to be prescribed aspirin at hospital discharge. In addition, patients who did not have cardiac catheterization, percutaneous transluminal coronary angioplasty, and coronary artery bypass surgery were also less likely to receive aspirin. Patients with congestive heart failure, low left ventricular ejection fraction, atrial fibrillation, stroke, the need for intubation during the hospitalization, and long lengths of hospital stay were also less likely to receive aspirin at discharge. The use of ß-blocker therapy at discharge was positively associated with the prescribed use of aspirin.

In the multivariable model, 10 variables were strongly associated with the prescription of aspirin at discharge (Table 3). The variables associated with an increased prescription of aspirin included several indicators of better overall health status (higher albumin level, no documented depressed ejection fraction, no extended length of stay, and discharge to the patient's home). The prescription of aspirin was also associated with several specific patterns of care, including the use of cardiac procedures, ß-blockers at discharge, and aspirin during the hospitalization. Age, sex, and race were not independently associated with a lower likelihood of being prescribed aspirin at discharge. The chi-square goodness-of-fit statistic for the model had a P value of 0.23 (P > 0.05 indicates that the data did not significantly depart from the model).

Aspirin and Mortality

The 6-month mortality rate in the study sample was 10.4% (571 of 5490 patients). Patients prescribed aspirin at discharge had a 6-month mortality rate of 8.4% (348 of 4149 patients) compared with 17% (223 of 1341 patients) for patients not prescribed aspirin (P < 0.001). The survival curves for the two groups are shown in Figure 1. We developed a series of multivariable logistic regression models to test the association between the prescribed use of aspirin at discharge and 6-month survival (Table 4). An "aspirin only" model showed an association between aspirin at discharge and 6-month mortality (odds ratio, 0.46; 95% CI, 0.38 to 0.55). In a model that adjusted for demographic variables, clinical variables, use of procedures, and use of ß-blockers at discharge, aspirin remained significantly associated with lower mortality (odds ratio, 0.77; CI, 0.61 to 0.98).

View larger version (12K): [in this window] [in a new window]   Figure 1. Cumulative probability of death for patients who were prescribed aspirin at discharge and those who were not prescribed aspirin (P < 0.001).

Graphic displays of partial residual plots from these models did not show any problems with our model assumptions. The chi-square goodness-of-fit statistic for the final model with demographic, clinical, and treatment variables had a P value of 0.39. The area under the receiver-operating characteristic curves for the final model was 0.79, indicating good model discrimination.

Discussion

Top Methods Results Discussion Author & Article Info References

Our principal finding is that 24% of elderly patients who survived an acute myocardial infarction and did not have a contraindication to long-term aspirin therapy were not prescribed aspirin at discharge. We focused on the care of patients who were the best candidates for aspirin. The use of aspirin in randomized trials is associated with an odds ratio of 1.5 to 2.0 for all categories of gastrointestinal bleeding, although fatal bleeding is rare [15]. These data, established by pooling the findings of major trials, are based on the use of large aspirin dosages, commonly in excess of 1000 mg/d. Although the risk with current recommended dosages should be much lower, we excluded patients with any increased risk for bleeding. Thus, our findings represent the prescribed use of aspirin among the best candidates for aspirin therapy.

The rate of prescribing aspirin at discharge in our study is similar to the 1990 rate reported by investigators from the Survival and Ventricular Enlargement (SAVE) study [16]. The SAVE cohort, however, represented a select group of patients at premier medical centers participating in a multicenter randomized trial. In addition, the patients had a mean age of approximately 59 years, and more than 80% were men; in our study, the mean patient age was 76 years, and 54% of patients were men. Other than that provided by the SAVE study, remarkably little information is available on the contemporary use of aspirin for secondary prevention after myocardial infarction.

Our findings are surprising given the strong evidence for the benefit of aspirin [3]. The Antiplatelet Trialists' Collaboration combined data to determine the effects of antiplatelet therapy on vascular events (myocardial infarction, stroke, or vascular death) in many settings. Among 11 randomized trials of antiplatelet therapy for secondary prevention after myocardial infarction, including 19 791 patients, the therapy was associated with a 25% reduction in the odds of having a vascular event. Moreover, on the basis of these studies, the official American Heart Association and American College of Cardiology guidelines for the care of patients who have had an acute myocardial infarction recommend that aspirin be used as a discharge medication [6].

In our sample of elderly patients who were good candidates for aspirin, age was not an independent predictor of the recommended use of aspirin at discharge. The effect of age in the bivariate analysis was mitigated by adjustment for other factors. This finding may reflect clinicians' understanding that although older patients may be more susceptible than younger patients to the adverse effects of aspirin, they may also be eligible for more benefit because of their higher rate of events. For example, the GISSI-2 investigators [1] reported that octogenarians who survive an acute myocardial infarction face a 12% mortality over the following 6 months. If aspirin could reduce that mortality by only 10% within 6 months, we would expect 12 patients to be saved for every 1000 patients treated with aspirin. It is unlikely that the complication rate of aspirin in the elderly, even among those with relative contraindications, could overshadow the magnitude of this benefit.

Our study cannot determine the reasons some patients do not receive aspirin. However, our findings show, that among patients without a contraindication to aspirin, certain categories of patients were much more likely to be discharged without a prescription for aspirin. Interestingly, in multivariable analyses, the single cardiovascular characteristic that was consistently associated with a lower use of aspirin was a documented left ventricular ejection fraction of less than 40%. This association was seen even though we had excluded patients receiving warfarin. Noncardiovascular variables were also associated with the use of aspirin. Patients with diabetes and poor nutritional status were significantly associated with a lower likelihood of being prescribed aspirin at hospital discharge.

Aspirin was most strongly associated with a constellation of care patterns. Patients who had clinical interventions such as cardiac catheterization, percutaneous transluminal coronary angioplasty, or coronary artery bypass surgery were more likely to be prescribed aspirin at discharge. The SAVE investigators also reported this association [16]. In addition, patients treated with aspirin during the hospitalization and patients prescribed ß-blockers as a discharge medication were much more likely to be treated with aspirin at discharge. Although the survival advantage after discharge associated with the prescription of aspirin at discharge persists despite adjustment for these differences, these observations suggest that the apparent protective effect of aspirin after discharge may be due to residual confounding. Nevertheless, our results are consistent with those of randomized trials of aspirin for secondary prevention [3]. A particular strength of our study is that it provides information on aspirin therapy in a broad spectrum of elderly patients spanning four states, a group that is poorly represented in randomized trials.

Our study does, however, have some potential limitations. First, we determined the use of aspirin at discharge through retrospective chart review. Aspirin use may not have been properly recorded, and treatment may have been continued after hospital discharge. Also, we could not determine patterns of aspirin use after hospital discharge. These problems, however, would have tended to diminish the association between aspirin and improved survival. Finally, our ability to ascertain contraindications was limited by our retrospective methods. Nevertheless, we sought to ascertain all contraindications by the use of laboratory results and medications in addition to written notes.

In conclusion, as part of a national effort to improve care to Medicare beneficiaries, our study has identified an opportunity to improve the delivery of health care to elderly patients who have had a myocardial infarction. About one fourth of Medicare patients who survive acute myocardial infarction are not prescribed aspirin at discharge, and these patients have a higher adjusted risk for death at 6 months that is consistent with what would be expected from randomized trials. Thus, increasing the use of aspirin at discharge appears to be an excellent opportunity to improve the care of elderly patients who have survived an acute myocardial infarction.

Presented at the 68th Scientific Session of the American Heart Association, 13 November 1995, Anaheim, California.

Dr. Radford: University of Connecticut Health Center, Cardiology, 283 Farmington Avenue, Farmington, CT 06030.

Dr. Ellerbeck: Health Care Financing Administration, Regional Office, 601 East 12th Avenue, Room 235, Kansas City, MO 64106.

Drs. Hennen and Meehan, Ms. Petrillo, and Mr. Wang: Connecticut Peer Review Organization, 100 Roscommon Drive, Middletown, CT 06457.

Dr. Jencks: Health Care Financing Administration, 2-D-2 Meadows Easy, 6325 Security Boulevard, Baltimore, MD 21207.