Association of Renal Insufficiency with Treatment and Outcomes after Myocardial Infarction in Elderly Patients FREE

–The Editors

Patients with end-stage renal disease who require dialysis have markedly increased mortality after myocardial infarction compared with other patients. One-year mortality in these patients is approximately 60% (1–3). After myocardial infarction, these patients are also unlikely to receive aggressive therapy, such as thrombolytic therapy and primary angioplasty, although these treatments have been associated with improved survival in these patients (4).

Patients with renal insufficiency have a greater risk for cardiovascular disease events (5–6), but the association between mild and moderate renal insufficiency and survival after myocardial infarction has not been evaluated in-depth (1, 7). Two earlier studies have included measures of renal function in prediction models for death after myocardial infarction. Normand and colleagues (8) incorporated both blood urea nitrogen and creatinine levels into a multivariate prediction model for 30-day mortality after myocardial infarction. Jacobs and colleagues included urea nitrogen levels as one of seven categories of predictors for death after hospital admission for acute coronary syndromes (9). However, studies have not compared survival after myocardial infarction among patients with and without renal insufficiency. In addition, the use of medical treatments and procedures after myocardial infarction among patients with and without renal insufficiency, and their association with survival, has not been studied.

We hypothesized that patients with mild and moderate renal insufficiency would have substantially greater 1-year mortality than patients with no renal insufficiency. We also hypothesized that patients with renal insufficiency would be less likely to receive therapeutic interventions known to improve survival after myocardial infarction, such as β-blockers, aspirin, angiotensin-converting enzyme (ACE) inhibitors, thrombolytic therapy, and primary angioplasty. Using data from the Cooperative Cardiovascular Project, we evaluated the treatment of patients with no renal insufficiency and patients with mild and moderate renal insufficiency. We also determined the independent association of renal insufficiency with survival after myocardial infarction.

The Cooperative Cardiovascular Project collected data from all elderly (age ≥ 65 years) Medicare beneficiaries who were admitted between April 1994 and July 1995 to an acute-care hospital and discharged with the diagnosis of acute myocardial infarction (International Classification of Diseases, Ninth Revision, diagnosis code 410) (10). The diagnosis was confirmed by review of the medical records for each patient and required a serum creatine kinase–MB index greater than 5%; an elevated serum lactate dehydrogenase level with lactate dehydrogenase-1 greater than or equal to lactate dehydrogenase-2; or two of the following three criteria: chest pain, serum creatine kinase level more than twice the normal value, or electrocardiographic evidence of acute myocardial infarction (11). A total of 139 567 patients had confirmed myocardial infarction. Only the initial hospitalization for myocardial infarction during the period of evaluation was included. We excluded 6790 patients with severe renal insufficiency (serum creatinine level ≥ 4.0 mg/dL [354 µmol/L]) or estimated creatinine clearance less than 0.17 mL/sec. We also excluded 10 570 patients (8.1%) for whom information on body weight was not available to estimate creatinine clearance.

Within the Cooperative Cardiovascular Project, trained medical record abstracters collected the following data for each patient: date and location of hospitalization, demographic characteristics, comorbid conditions, severity of illness measures, electrocardiogram findings, laboratory values, results from invasive and noninvasive cardiac studies, contraindications to therapy, in-hospital treatments, and discharge medications (9). The reliability of the data abstraction process has been demonstrated (10, 12).

We classified renal function using both the admission serum creatinine level and the estimated creatinine clearance. We defined no renal insufficiency as a serum creatinine level less than 1.5 mg/dL (<132 µmol/L), mild renal insufficiency as a serum creatinine level between 1.5 and 2.4 mg/dL (132 to 212 µmol/L), and moderate renal insufficiency as a serum creatinine level between 2.5 and 3.9 mg/dL (221 to 345 µmol/L). We estimated creatinine clearance by using the Cockroft–Gault equation (13) and divided patients into tertiles. Data on serum creatinine levels were available for all patients. Data on mortality during the first year after hospital admission were obtained from Social Security Administration records.

We compared the characteristics and treatments of patients with no renal insufficiency, mild renal insufficiency, and moderate renal insufficiency using analysis of variance and chi-square tests. We compared the proportions of patients treated with β-blockers among all patients in each category of renal function and in just the patients without relative contraindications to β-blockers (previous heart failure, diabetes, chronic obstructive pulmonary disease, and pulmonary edema at presentation). We compared use of ACE inhibitors among all patients and then restricted the comparison to patients with hypertension or diabetes. For thrombolytic therapy, we compared treatment as a proportion of all patients treated and as a proportion of “ideal” patients treated. Ideal candidates were defined by ST-segment elevation or left bundle-branch block on the electrocardiogram; onset of chest pain within 6 hours of presentation; age of 80 years or younger; and the absence of peptic ulcer disease, chronic liver disease, metastatic cancer, and terminal illness (14).

To determine the association of renal function with survival after myocardial infarction, we constructed Kaplan–Meier curves for the three groups of serum creatinine levels (and the tertiles of creatinine clearance); statistical significance was assessed by using the log-rank test. We repeated these analyses within subgroups of patients who presented without pulmonary edema or cardiogenic shock (Killip class I and class II) to determine whether the association of renal function with 1-year survival persisted among patients with less severe myocardial infarctions.

We used Cox proportional-hazards models to determine whether mild and moderate renal insufficiency were independent predictors of 1-year mortality (15). These models were adjusted for patient demographic characteristics (age, sex, race, rural or urban setting, and region of the United States); comorbid conditions (history of diabetes mellitus, hypertension, hypercholesterolemia, tobacco use, congestive heart failure, stroke, peripheral vascular disease, angina, myocardial infarction, percutaneous transluminal coronary angioplasty, coronary artery bypass graft surgery, chronic obstructive pulmonary disease, dementia, inability to ambulate, depression, and incontinence); severity of clinical presentation (Killip class, electrocardiogram findings, heart rate, mean arterial blood pressure, alertness and orientation according to the Glasgow coma scale, duration of chest pain, and blood urea nitrogen level [< 30 mg/dL; <10.7 mmol/L as urea or ≥ 30 mg/dL; ≥ 10.7 mmol/L as urea]); hospital characteristics (capability to do coronary angiography and revascularization; volume of myocardial infarction admissions); in-hospital treatments (aspirin, β-blockers, ACE inhibitors, thrombolytic therapy, intravenous nitroglycerin, coronary angiography, percutaneous transluminal coronary angioplasty, and coronary artery bypass graft surgery); and discharge medications (aspirin, β-blockers, calcium-channel blockers, and ACE inhibitors). We evaluated the inclusion of a categorical variable for each hospital (n = 4200) in a 10% sample of our data set, but it had little effect on the association of renal function with survival after myocardial infarction. We also checked for violations of linearity by examining augmented models, which included quadratic terms for continuous predictors.

We tested the validity of the proportional hazards assumption by determining whether the risk estimates for the renal function categories varied significantly over time. Because we found a significant interaction of the risk for renal function over time, we elected to explore stratified models over time. We separately evaluated the association of our renal function categories with survival after myocardial infarction for the following time intervals: months 1, 2 to 3, 4 to 6, 7 to 9, and 10 to 12. Because the association of renal insufficiency with mortality was the same for months 7 to 9 and 10 to 12, we combined these follow-up intervals. We confirmed that there were no residual violations of the proportional hazards assumption within each time strata by again testing for the presence of an interaction of each renal function predictor with time in predicting mortality during that time strata. We used Martingale and deviance residuals to check the overall fit of each model and found no evidence for a lack of fit.

We evaluated the association of medication use with aspirin, β-blockers, and ACE inhibitors with survival during the first month after myocardial infarction—overall and stratified by baseline renal function. Because of the large sample size, we found statistically significant interactions for renal function with aspirin and ACE inhibitors, although the point estimates were very similar. Because we did not believe that these statistical interactions had clinical relevance, we did not include these interactions in the models estimating the effect of renal function on survival after myocardial infarction.

Because mortality follow-up was complete, informative censoring could be ruled out for this analysis. Information on discharge medications was available for all patients. Hot-deck imputation, which imputed values by cluster, was used for other missing variables (16–17). We repeated our analyses when missing variables were not imputed and verified that the results were essentially unchanged. Statistical analysis was performed by using the SAS statistical package, version 6.12 (SAS Institute, Inc., Cary, North Carolina). Statistical significance was set at a two-sided P value of less than 0.05.

This study was supported by a research grant from the Centers for Medicare & Medical Services (formerly the Health Care Financing Administration) and the National Institute on Aging. The funding source had no role in the collection, analysis, and interpretation of the data or in the decision to submit the paper for publication.

Of the 130 099 elderly patients with myocardial infarction in our sample, 36 756 (28.3%) met the definition of mild renal insufficiency based on serum creatinine level and 10 888 (8.4%) met the criteria for moderate renal insufficiency. Patients with renal insufficiency were older and more likely to be male and black compared with patients who had relatively preserved renal function (Table 1). Several chronic disease conditions, including diabetes, hypertension, chronic heart failure, peripheral vascular disease, and dementia, were more prevalent among patients with renal insufficiency. Smoking, however, was less common.

Myocardial infarction was more severe for patients with renal insufficiency (Table 1). More than half of patients with mild and moderate renal insufficiency presented with Killip class III or IV compared with 30% of patients with preserved renal function. Patients with renal insufficiency also had higher heart rates and lower systolic blood pressures; however, presenting electrocardiogram findings were similar.

In-hospital treatment differed substantially among the groups of patients (Table 2). Patients with no renal insufficiency were treated with aspirin and β-blockers 20% more often (on an absolute scale) than patients with moderate renal insufficiency. In addition, patients with no renal insufficiency were more than twice as likely to receive thrombolytic therapy, coronary angiography, and angioplasty than patients with moderate renal insufficiency but were only slightly more likely to have coronary artery bypass graft surgery. Use of ACE inhibitors, in contrast, was greatest in patients with mild renal insufficiency, followed by patients with moderate renal insufficiency, and lowest in patients with no renal insufficiency. A similar pattern was observed when analyses were restricted to patients with hypertension or diabetes.

Among patients surviving to hospital discharge, those with no renal insufficiency were most likely to be prescribed aspirin and β-blockers, those with mild renal insufficiency most often received ACE inhibitors, and those with moderate renal insufficiency most often received calcium-channel blockers. Indeed, calcium-channel blockers were prescribed more often for patients with moderate renal insufficiency than either ACE inhibitors or β-blockers.

In unadjusted analyses, renal insufficiency was strongly associated with survival after myocardial infarction (Figure 1). At 1 month after hospital admission, mortality for patients with moderate renal insufficiency was 44% compared with 13% for patients who had relatively normal renal function. At 1 year, almost three times as many patients with moderate renal insufficiency had died compared with patients without renal insufficiency (Figure 1). Among patients who presented without pulmonary edema (Killip class I and class II), the unadjusted relative hazard was 1.60 (95% CI, 1.56 to 1.66) for patients with mild renal insufficiency and 2.56 (CI, 2.45 to 2.67) for patients with moderate renal insufficiency. We also repeated these analyses among patients who underwent echocardiography with measurement of left ventricular ejection fraction (LVEF) during hospitalization and survived to hospital discharge (n = 84 948); 34% (n = 28 545) of these patients had an LVEF less than 0.40. Left ventricular ejection fraction and renal insufficiency had no statistical interaction with 1-year survival, and the relative risk associated with renal insufficiency was similar in both LVEF subgroups.

Because of the time interaction of renal insufficiency and survival after myocardial infarction, we conducted separate multivariate proportional hazards models for months 1, 2 to 3, 4 to 6, and 7 to 12. After adjustment for both patient characteristics and treatments, moderate renal insufficiency was associated with a more than twofold risk for death in month 1, a 60% increased risk for death in months 2 to 3, and a 25% increased risk for death in months 3 to 6 (Table 3) . Mild renal insufficiency was also associated with a significantly elevated mortality risk in each of these time intervals. However, during months 7 to 12 of follow-up, the mortality risks were similar and low in all three groups of patients.

We found an inverse association between renal function as measured by tertiles of estimated creatinine clearance and 1-year mortality (Figure 2). Compared with patients in the highest tertile (creatinine clearance > 0.92 mL/sec), those in the lowest tertile (<0.55 mL/sec) had a 2.6-fold higher mortality and those in the middle tertile (0.55 to 0.92 mL/sec) had a 1.5-fold higher mortality for the entire year of follow-up.

After multivariate adjustment, patients in the lowest tertile of creatinine clearance had a high mortality risk during the first month of follow-up (Table 3). However, after 6 months of follow-up, this high risk disappeared. Patients in the middle tertile of creatinine clearance were primarily at increased mortality risk during the first month after myocardial infarction compared with patients in the highest tertile.

Because aspirin, β-blockers, and ACE inhibitor use differed among patients according to their renal function, we evaluated the association of these medications with 1-month survival after stratification by renal function (Table 4). All three medications seemed to be associated with a substantial survival benefit in patients with no renal insufficiency as well as those with mild and moderate renal insufficiency. Although these associations seemed similar across renal function subgroups, the relative hazard for aspirin was significantly lower in patients with no renal insufficiency than in patients with renal insufficiency; the relative hazard for ACE inhibitor use was significantly greater.

We observed a strikingly high mortality after myocardial infarction among elderly patients with renal insufficiency. Patients with renal insufficiency were also less likely to receive beneficial therapies, such as aspirin, β-blockers, and thrombolytic therapy, during hospitalization and at discharge. This undertreatment may contribute to the very high mortality.

Although 1-year mortality is known to be greater than 60% among patients receiving renal dialysis, the effects of mild and moderate renal insufficiency on mortality after myocardial infarction have not been well described (2). Earlier investigations have incorporated the serum creatinine or blood urea nitrogen levels into predictive models for mortality (8–9). Increased serum creatinine levels have also predicted cardiac complications and mortality after noncardiac and cardiac surgery (18–21). We found that serum creatinine level and estimated creatinine clearance were strongly associated with early mortality after myocardial infarction in elderly patients.

The association of renal insufficiency with mortality after myocardial infarction was strongest in the initial few months and disappeared after 6 months. However, by 6 months after myocardial infarction, half of the patients with moderate renal insufficiency had already died; relatively few patients died during months 7 to 12 of follow-up in all of the patient subgroups. Appropriate secondary prevention for patients with renal insufficiency should therefore be initiated immediately after myocardial infarction. Treatment with aspirin, β-blockers, and ACE inhibitors seemed to be associated with substantial benefit in the first month after myocardial infarction for patients with or without renal insufficiency.

In light of the substantially increased mortality risk for patients with renal insufficiency, their undertreatment is of concern. The low use of certain treatments, such as aspirin and ACE inhibitors, in patients with renal insufficiency may result from fear of adverse effects. However, two recent observational studies of elderly patients with reduced LVEF after myocardial infarction found that ACE inhibitors and β-blockers were associated with greater benefit in patients with renal insufficiency than in patients with preserved renal function (22–23). Similarly, aspirin therapy has been recommended for the treatment and secondary prevention of myocardial infarction in patients with renal insufficiency (24).

It is not clear why patients with renal insufficiency—even those considered ideal candidates—received thrombolytic therapy less often. Similarly, we noted a lower rate of β-blocker use in patients with renal insufficiency, even among those without chronic obstructive pulmonary disease, heart failure, and diabetes. In some of these patients, the treating physician may have been aware of potential contraindications not contained in the medical record and, thus, unknown to us. More likely, patients with renal insufficiency receive less aspirin, β-blockers, and thrombolytic therapy because they are thought to be frail and less likely to benefit or more likely to experience side effects. Patients with the highest mortality risk, however, such as those with renal insufficiency, should gain the greatest survival benefit from medications that reduce their risk (25). Unfortunately, clinicians who evaluate this risk–benefit tradeoff may emphasize the tangible risk for short-term side effects rather than the long-term benefits of reduced mortality.

One strength of our study is that we analyzed a national sample of elderly patients with myocardial infarction; thus, our findings should be generalizable (10). Although our data set relied on chart abstraction, the data collection process has been validated extensively. The primary predictors that we evaluated, serum creatinine level and estimated creatinine clearance, are routinely obtained in patients with myocardial infarction who are admitted to hospitals. These measures are much less precise than a true glomerular filtration rate for assessing renal function, but this imprecision would be expected to reduce the magnitude of the associations that we saw between renal insufficiency and mortality after myocardial infarction.

We cannot be certain that the serum creatinine levels measured at presentation indicated baseline renal function, because the severity of the myocardial infarction could have caused an acute elevation in the serum creatinine level. We did observe, however, similar relative risks associated with renal insufficiency among the patients with the least severe infarctions (Killip class I and class II); these patients would have been unlikely to present with acute renal failure. Another limitation is that the events evaluated in our study occurred in 1994 and 1995. Because the indications for the use of β-blockers and ACE inhibitors have expanded since that time, treatment of persons with and without renal insufficiency may have changed (26–29). In addition, the treatment disparities that we observed may be less pronounced in younger patients with renal insufficiency; we cannot extend our findings to persons younger than 65 years of age.

In conclusion, we found mortality for elderly patients after myocardial infarction to be substantially greater in those with mild and moderate renal insufficiency than in patients who have relatively normal renal function. Undertreatment of these patients with proven medical therapies could contribute to their excess mortality. Interventions such as clinical practice guidelines should address the quality of care for this high-risk subgroup of patients with myocardial infarction.