Lack of Benefit for Intravenous Thrombolysis in Patients With Myocardial Infarction Who Are Older Than 75 Years
Background—The benefit of intravenous thrombolytic therapy in elderly patients with myocardial infarction is uncertain. There are no randomized trials of thrombolytic efficacy or observational studies of clinical effectiveness that focus specifically on the elderly.
Methods and Results—To determine whether thrombolytic therapy for elderly patients is associated with a survival advantage in a large observational database, we conducted a retrospective cohort study of 7864 Medicare fee-for-service patients aged 65 to 86 years with the primary discharge diagnosis of acute myocardial infarction who were admitted with clinical and ECG indications for thrombolytic therapy and no absolute contraindications. The study included all US acute care nongovernment hospitals without on-site angioplasty capability. Using proportional-hazards methods, we found that in a comprehensive multivariate model, there was a significant interaction (P<0.001) between age and the effect of thrombolytic therapy on 30-day mortality rates. For patients 65 to 75 years old, thrombolytic therapy was associated with a survival benefit, consistent with randomized trials. Among patients aged 76 to 86 years, thrombolytic therapy was associated with a survival disadvantage, with a 30-day mortality hazard ratio of 1.38 (95% CI 1.12 to 1.71, P=0.003). For these patients, there was no benefit from thrombolytic therapy in any clinical subgroup.
Conclusions—In nationwide clinical practice, thrombolytic therapy for patients >75 years old is unlikely to confer survival benefit and may have a significant survival disadvantage. Reperfusion research that is focused on elderly patients is urgently needed.
Although patients >75 years old compose ≈30% of patients with myocardial infarction (MI),1 the role of intravenous thrombolysis in this age group is uncertain. Published subgroup analyses2 and meta-analyses3 suggest a benefit for thrombolytic therapy in the elderly without statistically significant conclusions. There are no randomized trials or observational studies that focus specifically on elderly patients with MI, and the clinical effectiveness for any therapy may diverge from efficacy in randomized trials.4
We conducted a retrospective cohort study of thrombolytic therapy using a database compiled by the Health Care Financing Administration (HCFA) for its Cooperative Cardiovascular Project (CCP). This cohort is uniquely suited to study actual clinical practice because it contains a nearly 100% nationwide sample of elderly fee-for-service patients with MI, with extensive clinical data, blinded data abstraction, and reliable long-term follow-up.
The CCP methods are described elsewhere.5 6 7 Briefly, for each US acute care hospital for a continuous 8-month sampling period between February 1994 and July 1995, HCFA identified from administrative data all Medicare fee-for-service beneficiaries with the principal discharge diagnosis of MI. Using photocopied medical records, trained abstracters entered ≈170 historical, clinical, and demographic data elements into an electronic database.
To reduce potential selection biases, we limited the analysis to patients aged ≥65 years, who were admitted from home, a clinic, or a physician’s office with a verified MI. We excluded duplicate admissions, interhospital transfers, admissions from nursing and retirement homes, and patients whose MIs occurred after admission. To eliminate selection biases associated with primary angioplasty, for which intention-to-treat data are not available, we excluded patients at hospitals with on-site angioplasty capability. An exploratory analysis was used to study same-day angioplasty (or next-day angioplasty within 6 hours of arrival) without antecedent thrombolytic agents because the CCP database does not include times for many angioplasty procedures.
We limited analysis to patients who presented within 12 hours of symptom onset and who had ECG ST-segment elevation in ≥2 contiguous leads, with ≥1.5 mm in limb leads (to exclude borderline ECGs) and ≥2 mm in precordial leads. We excluded patients with coma on arrival, preexisting dementia, terminal illness, serum urea nitrogen level of >25 mmol/L (70 mg/dL), or inability to walk independently. We also excluded patients with left bundle-branch block, because CCP data cannot distinguish between old and new left bundle-branch block, and those with absolute contraindications to thrombolysis.8 Because the benefit of thrombolytic therapy in cardiogenic shock is uncertain,9 we excluded patients with a systolic blood pressure of <90 mm Hg on arrival. We excluded patients who did not receive aspirin and heparin, which are typically administered regardless of thrombolytic treatment. Survival information was obtained from hospital records and the Social Security Administration’s Enrollment Database.
We defined thrombolytic patients as those receiving an intravenous thrombolytic agent within 4 hours of admission. MI was deemed present if patients had a creatine kinase–MB fraction of >0.05 or any 2 of these 3 criteria: chest pain, doubling of the total serum creatine kinase, or ECG evidence of acute MI.
Because thrombolytic benefit occurs entirely within 30 days after MI,10 our primary end point was 30-day mortality. Our published multivariate model,7 which combined 33 clinical, historical, and health system variables previously linked to 30-day mortality after MI,11 12 was modified in 2 respects. We added a variable for the maximum amplitude of ST-segment elevation and omitted physician specialty (because cardiology specialty often is required for thrombolytic agents). All independent variables had Spearman correlations of <0.5. We performed alternative analyses without clinical exclusion criteria, without missing categorical variables, and limited to variables significant at the P<0.001 level in multivariate analysis within the study population. The results were unchanged.
To avoid skewing due to the inclusion of very aged patients, we divided the population into cohorts of patients aged 65 to 75 and 76 to 86 years for our primary proportional-hazards analysis (SAS version 6.12, SAS Institute). We used logistic regression to measure possible selection or referral bias by calculating predicted mortality rates for patients who received thrombolytic agents and for those who did not. The logistic model, stratified by age cohort, predicted mortality rates on the basis of individual clinical covariate data while recoding the “thrombolytic” variable to “no thrombolytics,” thus integrating all clinical variables while neutralizing the effect of thrombolytic therapy.
The original data set contained 234 769 admissions for 210 996 patients, of whom 31 267 were admitted in transfer from other hospitals, 81 499 were admitted to hospitals with on-site percutaneous angioplasty capability, and 26 899 had clinical exclusions. We also excluded 6156 patients older than 86 years and 5337 patients whose charts lacked ECG tracings.
In the remaining group of 59 838 patients, 18 659 patients (32% of those aged 65 to 75 and 30% of those aged 76 to 86 years) met our ECG criteria, reflecting the preponderance of non–Q wave MIs among the elderly.13 Older patients had more exclusions for poor functional status and thrombolytic contraindications (Table 1⇓). The final study cohort included 5191 patients aged 65 to 75 years and 2673 patients aged 76 to 86 years, representing 48% and 34% of patients, respectively, with eligible ECGs. Multivariate analysis excluded 1.5% of the study cohort (117 patients) due to missing data. Missing rates for clinical variables were <5%, with the exception of serum albumin and chest radiographs, which were missing for 5% and 23% of patients, respectively. Median follow-up for survivors was 906 days.
Seventy-four percent of study patients aged 65 to 75 years and 60% of patients aged 76 to 86 years received thrombolytic agents within 4 hours of presentation. In the younger thrombolytic group, 77.1% of patients received tissue plasminogen activator (t-PA) and 22.3% received streptokinase compared with 72.6% and 26.3%, respectively, in the older group. Demographic and clinical characteristics of study patients are presented in Table 2⇓. In both age cohorts, patients who received thrombolytic agents were considerably healthier than those managed with heparin and aspirin alone, as previously recognized,14 with a greater proportion of Killip class I and II MIs and lower rates of anterior MI. However, thrombolytic patients had somewhat larger MIs on ECG, with ≈0.5-mm greater maximum ST-segment elevation. Logistically predicted mortality rates markedly favored thrombolytic patients. Among patients aged 65 to 75 years, the 30-day predicted mortality rate for patients treated with thrombolytic agents was 7.8% versus 9.6% for patients not treated with thrombolytic agents; among patients aged 76 to 86 years, the predicted rates were 13.6% versus 15.3%, respectively.
The relationship between thrombolytic therapy and unadjusted mortality rate at 30 days after discharge differed markedly by age. In the younger cohort, patients treated with thrombolytic agents had a crude mortality rate of 6.8% compared with 9.8% for patients not treated with thrombolytic agents. Among older patients, the ratio was reversed: the crude mortality rate for patients treated with thrombolytic agents was 18.0% compared with 15.4% for patients not treated with thrombolytic agents. The 1-year mortality rate was determined in large part on the basis of underlying comorbidity and illness severity, as suggested by the logistically predicted mortality rates: in the younger group, the crude 1-year mortality rate was 10.9% among patients treated with thrombolytic agents versus 17.7% for patients not treated with thrombolytic agents compared with 25.6% and 31.4%, respectively, in older patients.
Crude 30-day Kaplan-Meier survival curves (Figure 1⇓) showed that in the younger cohort, mortality rates were virtually identical during the first 4 days for the groups who received or did not receive thrombolytic agents, with a progressive thrombolytic advantage thereafter. In the older cohort, there was an immediate survival disadvantage within the first 2 days for patients treated with thrombolytic agents.
In proportional-hazards analysis of 30-day survival rates, the interaction between age as a continuous variable and thrombolytic therapy was highly significant (P<0.001), a finding that was confirmed with stratified analysis (Table 3⇓). For patients aged 65 to 75 years, the use of thrombolytic agents was associated with a consistent survival advantage, with hazard ratios ranging from 0.76 (P=0.02) to 0.88 (P=0.29), depending on the statistical model. For older patients, thrombolytic therapy was associated with a survival disadvantage, with a hazard ratio ranging from 1.29 (P=0.01) to 1.38 (P=0.003).
The addition to the comprehensive statistical model of a centered term for the interaction between age and thrombolytic agents (Figure 2⇓) showed that the hazard ratio for thrombolytic therapy was 1.00 at age 74.3 and increased by a factor of 1.056 per year (95% CI 1.030 to 1.083, P<0.001). The hazard ratio was 0.60 (95% CI 0.44 to 0.82, P<0.001) at age 65 and 1.36 (1.13 to 1.64, P=0.001) at age 80. The results were essentially identical when the simple multivariate model was used.
The age-related relative difference was minimally changed in extensive sensitivity analyses, which excluded patients who had undergone cardiopulmonary resuscitation before admission (n=155), who had delirium on admission (n=47), or who underwent angioplasty on the day of admission (n=30); recoded to thrombolytic therapy patients who received thrombolytic agents >4 hours after admission; and included patients regardless of mobility status and use of aspirin or heparin, in models that included each of these variables. When we included all 13 955 patients who met Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries trial (GUSTO-I)15 ECG criteria for thrombolysis, who did not have absolute contraindications, and who had symptom onset within 12 hours of presentation, regardless of renal function, initial blood pressure, mobility status, and aspirin or heparin use, the thrombolytic hazard ratio in the cohort aged 65 to 75 years was 1.05 (95% CI 0.89 to 1.24) compared with 1.46 (95% CI 1.26 to 1.70) among patients aged 76 to 86, for a difference of a factor of 1.39.
Both within our study cohort and in a broader multivariate analysis of 21 426 CCP patients <87 years old who received thrombolytic agents, t-PA was slightly more effective than streptokinase (hazard ratio 0.92 in the broader CCP cohort, 95% CI 0.84 to 1.02, P=0.10), without a discernible interaction between age and thrombolytic agent. The survival disadvantage of thrombolytic agents in older patients was unchanged after we limited the analysis to patients who received t-PA.
At hospitals that offered on-site angioplasty (Table 4⇓), within each age cohort, overall rates of reperfusion therapy (with either thrombolytic agents or same-day angioplasty) were nearly identical to those at lower-technology hospitals. Among patients >75 years old, there was a trend toward a 30-day survival disadvantage for the use of thrombolytic agents, with a hazard ratio of 1.39 (95% CI 0.93 to 2.08, P=0.11) compared with same-day angioplasty. This finding should be interpreted cautiously because of potential selection bias that involves primary angioplasty.
Subgroup Analysis Based on Clinical Characteristics
The interaction between age and thrombolytic therapy was stronger for women than for men (Table 5⇓), although the difference was not statistically significant and the crossover age at which there was no thrombolytic effect (hazard ratio 1.00) was essentially the same for both sexes. For women, the hazard for thrombolytic therapy increased by a factor of 1.082 for each year of increasing age (95% CI 1.045 to 1.122, P<0.001), compared with 1.027 (95% CI 0.991 to 1.065, P=0.33) for men. In the younger cohort, several groups tended to have particular benefit from thrombolytic agents, including patients with Killip class II and III MIs, with a greater magnitude of ST-segment elevation, and with no contraindications to thrombolysis. Surprisingly, younger patients with an anterior MI did not have the disproportionate benefit shown in randomized trials, with a thrombolytic hazard ratio of 1.04 (95% CI 0.76 to 1.42). There was a survival advantage among patients with anterior MI aged 65 to 66 years (hazard ratio 0.51, 95% CI 0.19 to 1.39, P=0.19), with a rapidly diminishing benefit with increased age (P<0.001 for age interaction within the subgroup). Among older patients, no clinical subgroup had a survival benefit from thrombolytic agents.
Although the CCP database does not include the cause of death, the data suggest that bleeding and intracerebral hemorrhage account for about one fifth of the survival disadvantage among older patients treated with thrombolytic agents. The CCP database noted red blood cell transfusions (Table 2⇑), which were administered to 7.0% of patients treated with thrombolytic agents aged 76 to 86 years compared with 4.3% of comparable patients not treated with thrombolytic agents. The database also noted new cerebrovascular events, without causes. The stroke rate was 1.4% among younger patients treated with thrombolytic agents and 2.7% among patients aged 76 to 86 compared with 0.8% and 1.5%, respectively, for patients not treated with thrombolytic agents. The survival disadvantage for older CCP patients treated with thrombolytic agents remained significant (hazard ratio 1.31, 95% CI 1.04 to 1.64, P=0.02) after the exclusion of patients with transfusion or stroke.
The observational CCP database allows empiric evaluation of the general consensus that elderly patients with MI benefit from thrombolytic therapy.16 17 The adverse effects of thrombolytic agents (eg, stroke, bleeding, and cardiac rupture) are accepted complications of MI and of the use of thrombolytic agents, so regulatory postmarketing surveillance is unlikely to detect excess deaths.
Our findings suggest that patients with MI who are >75 years old are unlikely to have a clinically significant benefit from the use of intravenous thrombolytic agents and may have a survival disadvantage. The 30-day hazard ratio for this cohort was 1.38 (95% CI 1.12 to 1.71, P=0.003). By contrast, with the use of identical methods for patients aged 65 to 75, we found a hazard ratio of 0.88 (95% CI 0.69 to 1.12), which is concordant with the 0.83 odds ratio for thrombolytic agents versus placebo in a large meta-analysis.3 The survival hazard for thrombolytic agents increased progressively with age (Figure 2⇑).
Our analysis has the strengths and weaknesses of all observational studies. Some eligible patients doubtless did not receive thrombolytic agents for unmeasured reasons, such as spontaneous reperfusion, smaller infarctions, or frailty. Our predicted-mortality analysis suggests that the combined effect of measured factors actually favors patients treated with thrombolytic agents, but selection bias nevertheless limits direct comparisons between thrombolytic and nonthrombolytic populations.
However, there is no reason to believe that selection biases are fundamentally different in adjacent age cohorts. Rates of thrombolytic use are roughly comparable, yet the outcomes in the 2 age cohorts are strikingly different: in the younger cohort, the crude mortality rate for patients treated with thrombolytic agents was 6.8% versus 9.8% for patients not treated with thrombolytic agents, and the respective rates were 18.0% versus 15.4% in the older cohort. This reversal was not explained by measured clinical differences in comorbidity, clinical history, or MI severity.
Our finding of no benefit for thrombolytic therapy in patients >75 years old is consistent with subgroup analyses of randomized trials that compared the use of intravenous thrombolytic agents with primary angioplasty. In both the Global Use of Strategies to Open Occluded Coronary Arteries in Acute Coronary Syndromes trial (GUSTO-IIb)18 and a meta-analysis of smaller randomized trials,19 there was disproportionate benefit for primary angioplasty among patients >70 years old, a finding that may reflect a survival disadvantage for thrombolytic agents in this age group.
However, our results differ from nonsignificant trends in randomized trials of intravenous thrombolysis. Both the meta-analysis of early randomized trials by the Fibrinolytic Therapy Trialists’3 (FTT) and a subgroup analysis2 of the GUSTO-I trial found that patients >75 years old had a nonsignificant trend toward an absolute benefit of 1 to 2 lives saved per 100 patients treated, albeit with diminished relative benefit among older patients: in the FTT meta-analysis, the odds ratio for patients >75 years old was 1.14 times that for patients aged 65 to 74, whereas in GUSTO-I, the odds ratio for patients aged 75 to 85 who received t-PA was 1.18 times the ratio for patients aged 65 to 74. By comparison, in the present study, the adjusted hazard ratio for patients aged 76 to 86 years compared with those aged 65 to 75 was increased by a factor of 1.70 with the simple model and by a factor of 1.57 with the comprehensive model.
The differences between our findings and the results of randomized trials probably are multifactorial. Most important, the CCP is a study of actual effectiveness in nationwide clinical practice. Randomized trials typically enroll healthier patients: even in GUSTO-I, the most inclusive thrombolytic trial, patients aged 75 to 85 years were healthier than the comparable CCP cohort, with lower rates of diabetes, prior bypass surgery, anterior infarction, and Killip class III infarction. Our findings were minimally changed with the use of GUSTO-I inclusion criteria.
In addition, randomized trials enforce consistency in medical treatment. Older patients in nationwide practice hypothetically could receive poorer postthrombolytic care than both patients in randomized trials and younger patients, although CCP quality indicators and discharge medications show no evidence of such discrimination. Randomized trial data do suggest that anticoagulation is a special concern in older patients treated with thrombolytic agents: in GUSTO-I, the mean activated partial thromboplastin time 12 hours after thrombolytic agent administration for patients <65 years old was 78.2±44.5 seconds compared with 103.4±52 seconds for patients 75 to 85 years old, a difference that persisted at 24 hours. Variable anticoagulation or thrombolytic dosing could contribute to excess mortality rates.
Thrombolytic trials also have some inherent methodological limitations. The GUSTO-I subgroup analysis suggested that for patients 75 to 85 years old, front-loaded intravenous t-PA plus heparin is superior to streptokinase plus heparin but could not evaluate the possibility that in this cohort both agents are inferior to heparin and aspirin alone. In the FTT meta-analysis, 55% of patients were enrolled in trials that did not routinely administer heparin and 26% were in trials that did not routinely administer aspirin, so the control group did not have the full benefit of antiplatelet therapy20 or heparin. Randomized trials may be skewed toward younger patients even within the subgroup of patients >75 years old, who represent the tail of a skewed distribution, whereas the CCP reflects the actual age distribution of elderly patients with MI. In addition, elderly patients in research studies may have a shorter symptom-to-presentation interval than do patients in general clinical practice.
There are several possible mechanisms for the apparent lack of thrombolytic benefit among CCP patients >75 years old. CCP rates of stroke21 and bleeding22 among older patients treated with thrombolytic agents are consistent with those in randomized trials. Older patients with MI have slightly lower rates of TIMI 3 flow in angiographic studies and higher rates of 3-vessel disease,3 and consequently, they may have diminished benefit from the use of thrombolytic agents. In addition, the Gruppo Italiano per lo Studio Della Sopravvivenza nell’Infarto Miocardico II (GISSI-2) investigators23 reported an age-related increase in sudden death from mechanical rupture in patients treated with thrombolytic agents. Of 84 patients >70 years old who had no history of prior MI and who died during hospitalization and underwent autopsy, 86% had cardiac rupture compared with 19% of comparable patients <60 years old. Such catastrophes might contribute to the early difference in outcome between older patients treated with thrombolytic agents and those not treated with thrombolytic agents (Figure 1B⇑).
We found that among patients 65 to 75 years old, the nationwide use of thrombolytic therapy in clinical practice had the survival advantage suggested in randomized trials. In the application of the same method to patients 76 to 86 years old, we found that thrombolytic agent use was associated with a crude survival disadvantage and an adjusted hazard ratio of 1.38 (95% CI 1.12 to 1.71, P=0.003). The relative age differential was robust throughout inclusive, wide-ranging sensitivity analyses.
Several years ago, a randomized trial of thrombolytic agent use in the elderly was aborted in large part because clinicians were unwilling to randomize patients to nonthrombolytic therapy.24 CCP data suggest that extrapolation from data for younger patients and reliance on surrogate end points such as angiographic patency may be perilous and that randomized and observational studies of reperfusion therapy that focus specifically on elderly patients with MI are urgently needed.
This work was supported by the Delmarva Foundation for Medical Care (Easton, Md), the HCFA (Baltimore, Md), and the Harry and Jeanette Weinberg Foundation (Baltimore, Md). The analyses on which this report is based were performed under contracts 500-96-P623 and 500-96-P624, entitled “Utilization and Quality Control Peer Review Organization for the State of Maryland and the District of Columbia,” sponsored by the HCFA, US Department of Health and Human Services. We are indebted to the many persons and organizations nationwide who designed and conducted the CCP. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government.
- Received November 12, 1999.
- Revision received December 8, 1999.
- Accepted January 4, 2000.
- Copyright © 2000 by American Heart Association
White HD, Barbash GI, Califf RM, et al, for the GUSTO-I investigators. Age and outcome with contemporary thrombolytic therapy: results from the GUSTO-I trial. Circulation. 1996;94:1826–1833.
Fibrinolytic Therapy Trialists’ (FTT) Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative overview of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Lancet. 1994;343:311–322.
White HD, Van De Werf FJ. Thrombolysis for acute myocardial infarction. Circulation. 1998;97:1632–1646.
Franzosi MG, Santoro E, De Vita C, et al, for the GISSI investigators. Ten-year follow-up of the first megatrial testing thrombolytic therapy in patients with acute myocardial infarction: results of the Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto-1 study. Circulation. 1998;98:2659–2665.
Lee KL, Woodlief LH, Topol EJ, et al. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction: results from an international trial of 41,021 patients. Circulation. 1995;91:1659–1668.
Ridker PM, Hennekens CH. Age and thrombolytic therapy. Circulation. 1996;94:1807–1808.
Global Use of Strategies to Open Occluded Coronary Arteries in Acute Coronary Syndromes (GUSTO IIb) Angioplasty Substudy Investigators. A clinical trial comparing primary coronary angioplasty with tissue plasminogen activator for acute myocardial infarction. N Engl J Med. 1997;336:1621–1628.
O’Neill WW, de Boer MJ, Gibbons RJ, et al. Lessons from the pooled outcome of the PAMI, ZWOLLE and Mayo Clinic randomized trials of primary angioplasty versus thrombolytic therapy of acute myocardial infarction. J Invas Cardiol. 1998;10:4A–10A.
Maggioni AP, Granzosi MG, Santoro E, et al, the Gruppo Italiano per lo Studio Della Sopravvivenza nell’Infarto Miocardico II (GISSI-2) and the International Study Group. The risk of stroke in patients with acute myocardial infarction after thrombolytic and antithrombotic treatment. N Engl J Med. 1992;327:1–6.
Berkowitz SD, Granger CB, Pieper KS, et al, for the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) Investigators. Incidence and predictors of bleeding after contemporary thrombolytic therapy for myocardial infarction. Circulation. 1997;95:2508–2516.
Maggioni AP, Maseri A, Fresco C, et al, for the Investigators of the Gruppo Italian per lo Studio della Sopravvivenz nell’Infarto Miocardico (GISSI-2). Age-related increase in mortality among patients with first myocardial infarctions treated with thrombolysis. N Engl J Med. 1993;329:1442–1448.
Ross AM. The TTOPP study: lessons from an aborted trial. J Myocardial Ischemia. 1990;2:65–69.