Temporal trends in the receipt of aspirin before hospital admission for recurrent acute myocardial infarction: Worcester Heart Attack Study, 1986-1995.
Temporal trends in the receipt of β-blockers before hospital admission for recurrent acute myocardial infarction: Worcester Heart Attack Study, 1986-1995.
Temporal trends in the receipt of lipid-lowering medications before hospital admission for recurrent acute myocardial infarction: Worcester Heart Attack Study, 1986-1995.
McCormick D, Gurwitz JH, Lessard D, Yarzebski J, Gore JM, Goldberg RJ. Use of Aspirin, β-Blockers, and Lipid-Lowering Medications Before Recurrent Acute Myocardial InfarctionMissed Opportunities for Prevention?. Arch Intern Med. 1999;159(6):561-567. doi:10.1001/archinte.159.6.561
For patients who have had a previous myocardial infarction (MI), the use of aspirin, β-blockers, and lipid-lowering agents reduces the risk of recurrent MI and death.
To examine trends in and determinants of receipt of these 3 medications before hospitalization for recurrent acute MI (AMI).
The study population consisted of 1710 patients with a previous history of MI hospitalized with a validated recurrent AMI in all hospitals in Worcester, Mass, during 1986, 1988, 1990, 1991, 1993, and 1995. Logistic regression analyses were used to assess the effect of demographic, clinical, and temporal factors on the receipt of aspirin, β-blockers, and lipid-lowering medications before hospital admission for recurrent AMI.
More than 47% of patients in each study year were not receiving each medication before admission, although significant increases in use were noted over time for aspirin (from 13.5% to 52.6%), β-blockers (from 33.2% to 44.4%), and lipid-lowering medications (from 0.8% to 11.7%). In multivariate analyses, advancing age was associated with not receiving aspirin (odds ratio [OR], 0.67; 95% confidence interval [CI], 0.51-0.89), lipid-lowering medications (OR, 0.14; 95% CI, 0.08-0.25), and β-blockers (OR, 0.75; 95% CI, 0.57-1.00), although this effect was of borderline significance for β-blockers. Being a woman was associated with not receiving aspirin (OR, 0.78; 95% CI, 0.62-0.98) but was positively associated with receiving lipid-lowering medications (OR, 1.59; 95% CI, 1.04-2.43). Coexisting medical conditions and concurrent use of other cardiovascular medications were also associated with receipt of each medication.
Despite encouraging increases over time, the low absolute levels of receipt of medications shown to be efficacious in the long-term treatment of patients after an MI, and their variation by age and sex, suggest that substantial opportunities may exist to prevent recurrent AMIs through the increased use of aspirin, β-blockers, and lipid-lowering medications.
MORTALITY associated with a recurrent acute myocardial infarction (AMI) is appreciably higher than that associated with a first AMI. Several medications exist, however, that have been shown to reduce the likelihood of recurrent AMI and death in patients who have established coronary artery disease, including patients who have had an initial AMI. The effectiveness of therapy with aspirin,1- 3 β-blockers,3- 9 and lipid-lowering medications10- 12 in the secondary prevention of AMI has been well established in large, randomized clinical trials and/or meta-analyses of the published literature. It has been estimated that use of these medications can reduce the risk of cardiovascular death and nonfatal reinfarction, respectively, by 22% and 27% for β-blockers,3,8 13% and 31% for aspirin,2 and 14% and 25% for all lipid-lowering medications combined.13 Based on this evidence, widely publicized guidelines strongly recommend the routine long-term use of aspirin and β-blockers14,15 in patients who recently had an AMI and lipid-lowering medications in patients with elevated cholesterol levels following hospital discharge after AMI.13,16
Despite the positive findings of studies examining the effectiveness of these therapies and the widespread dissemination of these practice guidelines, findings of several previous studies17- 23 suggest that these medications are underprescribed to patients at hospital discharge after an AMI. Results of previous studies also suggest that nonclinical factors such as age,17,21 sex,20 and type of medical insurance20 may affect the likelihood of receiving these medications at hospital discharge after an AMI.
However, among patients experiencing recurrent AMI, little is currently known about the use of these medications before the event. Patients who have had a previous AMI are at particularly high risk for recurrent AMI and death; patients who have recurrent AMI and are not using these medications may thus represent missed opportunities for prevention. Understanding the factors associated with receipt of these medications for patients with recurrent AMI may help to overcome obstacles to optimizing their use.
The objectives of this observational, communitywide study were (1) to examine trends over time in the percentages of patients receiving aspirin, β-blockers, and lipid-lowering medications at hospital admission for recurrent AMI and (2) to identify factors that are associated with the receipt of these agents in patients with previous AMI.
This investigation was conducted as part of the Worcester Heart Attack Study, a multihospital, population-based investigation of time trends in the attack and survival rates associated with AMI.24- 28
The population studied consisted of patients hospitalized with a primary or secondary diagnosis of AMI (International Classification of Diseases, Ninth Revision, code 410)29 in all acute care general hospitals in the Worcester (Mass) Standard Metropolitan Statistical Area (1990 census estimate=437,000) during 1986, 1988, 1990, 1991, 1993, and 1995, who had a history of myocardial infarction (MI). Sixteen university-affiliated and community hospitals were originally included in this study, with fewer hospitals included in recent study years because of hospital closures or conversions to long-term care facilities. The medical records of greater Worcester residents with a discharge diagnosis of AMI from these hospitals were individually reviewed and validated according to pre-established diagnostic criteria that have been described previously.30- 32 In brief, these criteria included a clinical history of prolonged chest pain not relieved by nitrate therapy or rest; increased total and isoenzyme subfractions of creatine kinase or lactate dehydrogenase; and serial electrocardiographic findings of ST segment changes or Q waves typical of AMI. At least 2 of these 3 criteria needed to be satisfied for study inclusion. Presence of a previous history of MI was assessed through information provided by the patient that was documented in the medical record at hospital admission and confirmed through review of the medical record of previous hospitalization for AMI at areawide hospitals.
The hospital records of patients with validated AMI were abstracted for demographic (age, sex, and race) and clinical data (medical history of previous MI, angina, hypertension, diabetes, congestive heart failure, or stroke), type of medical insurance, total serum cholesterol level observed during hospitalization, and preadmission medication use (aspirin, β-blockers, lipid-lowering medications, calcium channel blockers, diuretics, warfarin sodium, antiarrhythmic medications, and digoxin). Patients were considered to be taking a medication before admission if it was listed as a current outpatient medication in the medical record on the day of hospital admission for recurrent AMI. This information was provided by patients themselves, their usual outpatient physicians, or other referring institutions, such as long-term care facilities. Outpatient medical records were not used for verification of medical regimens.
Time trends in the use of aspirin, β-blockers, and lipid-lowering medications before hospitalization for recurrent AMI were analyzed by determining the percentages of patients who received each of these medications by study year. A 2-sided Cochran-Armatage test for trend was used to determine statistical significance.
Demographic and clinical correlates of receiving each medication were evaluated for the study sample (all study years combined), and relative risks and 95% confidence intervals (CIs) were calculated for each variable. These variables were used to develop separate stepwise multivariate logistic regression models with use of aspirin, β-blockers, and lipid-lowering medication as the outcome variables. Candidate variables in these analyses included race (white vs all other races), medical history (angina, hypertension, diabetes, congestive heart failure, and stroke), concurrent (preadmission) use of other medications (aspirin, β-blockers, lipid-lowering medications, calcium channel blockers, diuretics, warfarin, antiarrhythmic medications, and digoxin), medical insurance (private vs Medicare, Medicaid, and uninsured), and total serum cholesterol level observed in the hospital (≥5.17 vs <5.17 mmol/L [≥200 vs <200 mg/dL]). Complete lipid profile laboratory data were not available for study participants. Our definition of a high serum cholesterol level (total cholesterol level ≥5.17 mmol/L [≥200 mg/dL]) was used to serve as a proxy for a low-density lipoprotein cholesterol level of approximately 3.36 mmol/L or greater (≥130 mg/dL),16 the level at which medical treatment of hypercholesterolemia in patients with established coronary heart disease is recommended by the National Cholesterol Education Project guidelines, published in 1994.13 Variables were dropped from each model at a significance level of P<.05. Because of a priori importance, age (<65, 65-74, and ≥75 years), sex, and study year (with 1995 as the referent category) were forced into all models. For each model, we reported adjusted odds ratios (ORs) and 95% CIs for all variables.
The total study population comprised 1710 patients, most of whom were older than 65 years, male, and white, with no private medical insurance (Table 1). Coexisting medical conditions were common: 43.0% had angina, 58.1% had hypertension, 34.3% had diabetes, 31.2% had a history of congestive heart failure, and 36.2% had a total cholesterol level greater than 5.17 mmol/L [>200 mg/dL]. The percentages of patients receiving aspirin, β-blockers, and lipid-lowering medications at hospital admission were 37.1%, 37.3%, and 7.2%, respectively. Concurrent use of additional cardiovascular medications varied, ranging from 43.9% of patients taking a calcium channel blocker to 7.8% of patients taking warfarin.
The percentages of patients with AMI who were receiving aspirin at hospital admission increased significantly (P<.001) during the 6 study years, from 13.5% in 1986 to 52.6% in 1995 (Figure 1). Older patients and women were significantly less likely to be receiving aspirin, whereas patients who had a history of angina or stroke or who were concurrently receiving β-blockers, lipid-lowering medications, or calcium channel blockers were more likely to be receiving aspirin (Table 1).
In the multivariate regression model, 10 demographic and clinical variables were identified as having an independent association with receiving aspirin at hospital admission (Table 2). Enrollment in earlier study years (compared with more recent study years) was associated with not receiving aspirin (OR, 0.13; 95% CI, 0.09-0.21; comparing 1986 with 1995). Advancing patient age (OR, 0.67; 95% CI, 0.51-0.89; for age >75 years compared with age <65 years) and female sex (OR, 0.78; 95% CI, 0.62-0.98) were also associated with not receiving aspirin. As in the bivariate analysis, history of angina or stroke and concurrent use of several other cardiovascular medications, including β-blockers and lipid-lowering medications, were associated with receiving aspirin.
The percentages of patients with AMI who were receiving β-blockers at hospital admission increased modestly (P<.003) during the 6 study years, from 33.2% in 1986 to 44.4% in 1995 (Figure 2). Advancing age, white race, a history of congestive heart failure, and use of diuretics or digoxin were associated with not receiving a β-blocker (Table 1). Patients with a history of angina or hypertension or who were concurrently receiving aspirin, lipid-lowering medications, warfarin, or antiarrhythmic medications were more likely to receive a β-blocker.
In the multivariate regression model, 12 demographic and clinical variables were identified as having an independent association with receiving a β-blocker (Table 3). Although there was a statistically significant increase in the odds of receiving a β-blocker from 1990 (OR, 0.62; 95% CI, 0.42-0.90) and 1991 (OR, 0.63; 95% CI, 0.43-0.91) to 1995, the odds of receiving this medication in earlier years (1986 or 1988) were no lower than in the most recent study year (1995). As in the bivariate analysis, advancing patient age (OR, 0.75; 95% CI, 0.57-1.00) and white race (OR, 0.53; 95% CI, 0.35-0.72) were associated with not receiving a β-blocker. History of angina or hypertension and concurrent use of other cardiovascular medications, including aspirin and lipid-lowering medications, were again positively associated with receiving a β-blocker.
The percentages of patients with AMI who were receiving lipid-lowering medications increased significantly (P<.001) during the 6 study years, from 0.8% in 1986 to 11.7% in 1995 (Figure 3). Older patients, women, and those not covered by private medical insurance were significantly less likely to be receiving a lipid-lowering medication, whereas patients who had an elevated cholesterol level (≥5.17 mmol/L [≥200 mg/dL]) or who were concurrently receiving β-blockers or aspirin were more likely to be receiving a lipid-lowering medication (Table 1). This low level of receipt of lipid-lowering medications was despite the finding that more than 36.2% of patients had a total cholesterol level greater than 5.17 mmol/L (>200 mg/dL) (Table 1).
In the multivariate regression model, 6 demographic and clinical variables were identified as having an independent association with receiving a lipid-lowering medication (Table 4). Enrollment in earlier study years was associated with not receiving a lipid-lowering medication. Advancing patient age (OR, 0.14; 95% CI, 0.08-0.25) also remained associated with not receiving a lipid-lowering medication. An elevated serum cholesterol level and concurrent use of other cardiovascular medications were, as in the bivariate analysis, also associated with receipt of a lipid-lowering medication. In contrast to the bivariate analysis, women were significantly more likely to be receiving a lipid-lowering medication after controlling for demographic and clinical factors.
Evidence-based clinical guidelines strongly endorse the use of aspirin and β-blockers in nearly all patients who have experienced an MI and do not have specific contraindications14,15 and the use of lipid-lowering medications in those with elevated serum cholesterol levels.13,16 Studies establishing the efficacy of using these medications for secondary prevention in coronary heart disease, on which these practice guidelines are based, were published before the first year of the present investigation. Despite widespread dissemination of this information, we found that, even by 1995, more than half of all patients who were first seen with recurrent AMI were not receiving aspirin or a β-blocker and that most patients (>90%; two thirds of those with an elevated cholesterol level) were not receiving a lipid-lowering medication. Although receipt of aspirin and lipid-lowering medications increased substantially during the approximately 10-year study, only modest changes were noted for β-blocker use. In addition to comorbidities and concurrent cardiovascular medication use, receipt of these medications was significantly affected by nonclinical factors such as age and sex. These findings suggest substantial missed opportunities for the prevention of recurrent AMI with the use of these effective therapies.
Results of well-designed clinical trials have been shown to affect the prescription of cardiovascular medications by physicians.28,33 However, results of several previous studies conducted in the 1990s found significant underuse of aspirin17- 19 and β-blockers18,20- 23 at hospital discharge after an initial AMI despite the demonstrated efficacy of these medications in published clinical trials. Undertreatment of hypercholesterolemia among patients with established coronary heart disease also has been previously described.34 The low rates of prescribing these medications have been attributed to deficits in physician knowledge of medication effectiveness because of long delays in the dissemination of the results of clinical research to practicing clinicians.22,34- 38 For example, results of a recent study38 show that the time lag between published meta-analyses that established the efficacy of using aspirin and β-blockers in the secondary prevention of MI and the recommended use of these medications by more than half the authors of review articles and textbook chapters on the subject was 6 and 2 years, respectively. In addition, despite the publication of evidence-based practice guidelines recommending treatment of hypercholesterolemia for patients with established coronary heart disease in 1988,16 the use of lipid-lowering medications was not yet recommended by more than half the authors of review articles and textbook chapters on the subject 2 years later.38 This considerable time lag between the publication of results of clinical trials and the acquisition of this knowledge by physicians may be largely responsible for the low percentages of patients receiving aspirin, β-blockers, and lipid-lowering medications that we observed and the dramatic increases in receipt of aspirin and lipid-lowering medications years after their effectiveness was first demonstrated.
We also found that, although aspirin, β-blockers, and lipid-lowering medications were received by relatively small percentages of patients after an MI, these medications were received at even lower levels by several clinically and demographically defined patient subgroups. As expected, patients with cardiovascular risk factors and comorbid conditions tended to be more likely to be receiving any one of the 3 medications of interest, and patients who were currently using any 1 of these medications were more likely to be using another of them. However, nonclinical factors such as age and sex also affected patterns of receipt of these medications. Patients who were older (≥75 years) were significantly less likely to use any of the 3 medications, a finding consistent with previous studies that show underuse of β-blockers20- 22 and aspirin17 in elderly patients at hospital discharge after AMI. Yet, the survival benefit from β-blocker21,39 and aspirin1 therapy in patients with established coronary heart disease appears to be at least as great for elderly as nonelderly patients. The benefits of therapy with lipid-lowering medications in elderly patients with established coronary heart disease have not been clearly demonstrated because most large clinical trials do not include adequate numbers of elderly patients. Nonetheless, because there is no evidence that the basic pathophysiological processes underlying coronary atherogenesis are different for elderly and nonelderly patients, National Cholesterol Education Program guidelines recommend that age alone should not be a reason to treat hypercholesterolemia less aggressively.13
Women in the present population-based study were significantly less likely to be receiving aspirin at the time of reinfarction, a finding consistent with that of a previous study40 showing that aspirin is used less often for women than for men after an initial MI. As with increasing age, results of previous research do not suggest a less beneficial effect of aspirin use in women than in men with previous MI.1 However, in our multivariate models, being a woman was associated with greater odds of receiving a lipid-lowering agent. Thus, patient sex seems to have a variable effect on the odds of receiving effective therapies at the time of recurrent AMI.
Lack of physician awareness of the results of clinical trials demonstrating the effectiveness of these medications or of practice guidelines recommending the use of these medications irrespective of age or sex may contribute to the low percentages of patients receiving these medications that we observed. It is also likely that use of these 3 medications was affected by the presence of clinical contraindications. β-blocker use is contraindicated in patients with heart block, bradycardia, congestive heart failure, reactive airways disease, diabetes mellitus, and depression, and aspirin use may be contraindicated in patients with bleeding disorders, peptic ulcer disease, thrombocytopenia, and aspirin allergy. Use of lipid-lowering medications is contraindicated in few patients. Because data were only available for a few coexisting illnesses that could constitute a contraindication to medication use, we could not assess the magnitude of the impact of clinical contraindications on the rates of use of these medications. However, of patients screened for possible inclusion in the largest β-blocker trials, the proportion who had contraindications to β-blocker use did not exceed 18%.7,41 In a previous aspirin trial,42 the percentage of screened patients who were excluded because of contraindications to aspirin use was less than 4%. Although it is possible that patients with recurrent AMI may have more contraindications to medication use than patients in these clinical trials, it seems unlikely that such differences could completely account for the low medication use rate that we observed.
Beyond physician prescribing practices, the patient's inability to comply with physician recommendations may contribute to lower use rates. It is possible that the cost of long-term therapy may discourage some patients from continuing to use these medications. We found that patients with private insurance were more likely to be using lipid-lowering medications, most classes of which are relatively expensive (eg, 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors), at the time of their recurrent MI. This relationship did not persist in the multivariate model, but it is possible that this was caused by either low statistical power because of the small number of patients using lipid-lowering medications or residual confounding by age because patients with nonprivate insurance (including Medicare) were more likely to be older. The relationship between insurance status and use of lipid-lowering medications in patients after an MI, therefore, deserves further study. For use of β-blockers and aspirin, 2 relatively inexpensive medications, we found no association with medical insurance status.
It is also likely that patients' inability to tolerate the adverse effects of these medications contributed to the low usage rates we observed. In a previous large study41 of patients who have experienced an MI, the Beta-blocker Heart Attack Trial, withdrawal of β-blocker therapy because of adverse effects was 13% for 2 years of follow-up. In the Coronary Drug Project Research Group trial43 of aspirin for secondary prevention, only 4.5% of patients taking aspirin were compliant with therapy less than 20% of the time during an average of 22 months of follow-up. However, results of a previous study44 show that rates of discontinuation of lipid-lowering medication use in the primary care setting (primary and secondary prevention patients combined) were variable, depending on the particular agent prescribed, and were higher, in general, than the rates reported in clinical trials. Unfortunately, no data were collected as part of our study that would allow us to assess the impact of this problem on receipt of medications examined in this study.
Although it is likely that each factor discussed has some effect on the rate of use of these medications, physician prescribing behavior is likely to have the greatest impact. Results of previous studies on receipt of aspirin and β-blockers by patients after an MI indicate that physicians do not prescribe these medications to 15% to 25% of eligible patients at hospital discharge. Furthermore, the rapid rise in the use of aspirin and lipid-lowering medications over time in the present study more likely reflects substantial changes in physician prescribing behavior because of the acquisition of new knowledge about health benefits associated with these agents rather than dramatic changes in the prevalence of contraindications to using these medicines or changes in patients' ability to comply with recommended therapy.
Residents of the Worcester metropolitan area are similar to those of the overall United States with respect to characteristics such as age, sex, and socioeconomic status but not for race.24,30 By including all patients hospitalized with recurrent AMI from a defined geographic area, this study minimized the likelihood of selection biases that may be present in studies of patients hospitalized in single or referral hospitals. Several limitations of this study should also be noted. First, to assess medication use just before admission, we relied on documentation of the patient's outpatient medical regimen in the medical record at hospital admission for recurrent AMI. To the extent that this information in many cases came from patient self-report, inaccuracies in patient recall could have led to some underestimates or overestimates of the rates of use of these medications.
Second, although detailed information about cardiovascular comorbidities and concurrent medication use was available for study patients, information on the complete range of additional comorbidities that could represent absolute or relative contraindications to use of the 3 medications examined was not available. In addition, information about discontinuation of medication use because of adverse effects was not known. Thus, we could not determine the "right" percentage of patients who should have been receiving each medication we examined. Third, although this study highlights the small percentages of patients receiving these medications in actual practice, we were not able to assess the reasons underlying these patterns of care. Future studies will need to address the relative impact of physicians' failure to prescribe these medications and patients' inability or choice not to comply with recommended therapies.
In summary, this study documents the extent of underuse of aspirin, β-blockers, and lipid-lowering medications by patients with a previous history of MI who later experience a recurrent AMI and identifies clinical and nonclinical factors associated with this underuse. Although the rates of use of the cardiovascular medications we studied may be higher for patients after an MI who did not have a recurrent event (to the extent that these medications are effective), our findings confirm that there remain substantial missed opportunities to treat patients after an MI with medications that are shown to reduce the risk of recurrent MI and cardiovascular death. Given the high prevalence of MI, concerted efforts should be undertaken to facilitate more rapid transmission of the results of clinical trials of cardiovascular medications to practicing physicians and to reduce substantial variation in treatment practices that seems to be related to patient age and sex.
Accepted for publication July 14, 1998.
Supported by grant HL R01 54534 from the National Institutes of Health, Bethesda, Md.
This research was made possible by the cooperation of the medical records, administration, and cardiology departments of participating metropolitan Worcester, Mass, hospitals.
Reprints: Danny McCormick, MD, MPH, Division of General Medicine, University of Massachusetts Medical Center, 55 Lake Ave N, Worcester, MA 01655.