|Home | About | Journals | Submit | Contact Us | Français|
Despite evidence from randomized trials and national guidelines that recommend the use of aspirin, β-blockers, and cholesterol-lowering agents for patients with coronary artery disease, many patients are not treated appropriately.1,2,3 Several types of fixed-message physician reminder systems—including checklists, chart tags, and computer-generated reminders—have improved physician compliance with cancer screening guidelines,4 but it is less clear whether they are effective at improving disease management. We examined whether the combination of a computer-generated and written reminder system provided during patient visits could increase patient receipt of aspirin, β-blockers, and cholesterol-lowering agents.
Between March and June 1997, 66 physicians at the Veterans Affairs (VA) Medical Center in San Francisco, the VA Medical Center in Palo Alto, or the VA Northern California Health Care System were randomly assigned to either a control group (n = 33) or an intervention group (n = 33). Physicians assigned to the intervention group received computerized and written reminders for their patients with coronary artery disease, whereas those assigned to the control group were not contacted. The presence of coronary artery disease was assumed if a patient had received an International Classification of Diseases, Ninth Revision (ICD-9) compatible discharge diagnosis or had an active prescription for any form of nitrate therapy.
One year after randomization, we evaluated the proportion of patients who had an active prescription for aspirin; the proportion of patients with myocardial infarction who had an active β-blocker prescription; the proportion of patients receiving a cholesterol-lowering agent; and the proportion of patients with a level of low-density lipoprotein (LDL) cholesterol in the desired range (< 100 mg/dL). Secondary outcome variables included the proportion of patients who were hospitalized for myocardial infarction and patient mortality.
After adjusting the sample size for clustering, χ2 and t tests were used to evaluate differences between patients in the intervention and control groups. In addition, logistic regression analyses that accounted for patient and physician characteristics were performed to evaluate any residual confounding after random sampling.
Of the 66 physicians, 3 physicians (4.5%) left their clinics and were not available for follow-up evaluation. All the reported analyses are based on the remaining patients that were treated by 63 physicians (31 in the intervention group, 32 in the control group). The intervention and control groups were similar with respect to proportion of physician specialists, practice location, age, gender, and number of patients (table 1). Patients in the intervention and control groups were similar with respect to age, gender, race, and history of myocardial infarction.
Outcomes for patients in the intervention and control groups did not differ significantly (table 2). Patient receipt of aspirin, β-blockers, or cholesterol-lowering agents and the proportion of patients who achieved a desired LDL level were similar regardless of treatment group. In addition, patient mortality, number of admissions for myocardial infarction, number of visits to primary care providers, and number of visits to cardiologists were also similar in both groups. After adjusting for baseline differences in patient and physician characteristics, no significant differences in patient outcomes or treatment were found.
Our reminder system may have appeared ineffective in changing physician behavior for a number of reasons. First, patients may have had contraindications to medications that prevented physicians from using these therapies. We found, however, that 64% of patients did not receive aspirin and 82% of patients did not receive β-blockers, proportions that are much larger than expected based on the prevalence of contraindications in similar patient groups.5,6 Second, patients may have obtained aspirin over-the-counter; therefore, the rates of actual aspirin use may have been higher. Third, patients had an average LDL level of 115 mg/dL at the beginning of the study, making it difficult for any intervention to improve on this successful management, especially in the study's short timeframe.
A number of factors related to implementation of the reminder system may have resulted in its failure.7 First, physicians in our study did not participate in the design of the reminder system and, as has been shown in other studies, this “lack of buy-in” may lead to poor compliance. Second, our reminders were fixed rather than dynamic, so physicians may have become desensitized. Finally, study physicians may have felt they were too busy to comply with the guidelines.8 Busy physicians have been shown to be less compliant with preventive services as well as less responsive to reminders because of intense time pressures.
In summary, the results of our study showed that a computerized and written reminder system did not increase the use of aspirin and β-blockers or improve cholesterol management in patients with coronary artery disease. Our findings imply that using fixed reminders to improve disease management may be ineffective. Other types of reminder systems, including those with real-time feedback, are warranted to find new ways of improving the management of patients with coronary artery disease.
Competing interests: None declared