Author + information
- Received August 30, 2015
- Revision received February 29, 2016
- Accepted March 8, 2016
- Published online May 24, 2016.
- Emily M. Bucholz, MD, PhD, MPHa,b,
- Neel M. Butala, MD, MBAc,
- Sharon-Lise T. Normand, PhDd,e,
- Yun Wang, PhDe and
- Harlan M. Krumholz, MD, SMf,g,h,∗ ()
- aDepartment of Medicine, Boston Children’s Hospital, Boston, Massachusetts
- bYale School of Medicine and Yale School of Public Health, New Haven, Connecticut
- cDepartment of Internal Medicine, Massachusetts General Hospital, Boston, Massachusetts
- dDepartment of Health Care Policy, Harvard Medical School, Boston, Massachusetts
- eDepartment of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
- fSection of Cardiovascular Medicine, Department of Internal Medicine, Yale School of Medicine, New Haven, Connecticut
- gRobert Wood Johnson Foundation Clinical Scholars Program, Yale School of Medicine, New Haven, Connecticut
- hSection of Health Policy and Administration, Yale School of Public Health, New Haven, Connecticut
- ↵∗Reprint requests and correspondence:
Dr. Harlan M. Krumholz, Department of Internal Medicine, Yale School of Medicine, 1 Church Street, Suite 200, New Haven, Connecticut 06510.
Background Guideline-based admission therapies for acute myocardial infarction (AMI) significantly improve 30-day survival, but little is known about their association with long-term outcomes.
Objectives This study evaluated the association of 5 AMI admission therapies (aspirin, beta-blockers, acute reperfusion therapy, door-to-balloon [D2B] time ≤90 min, and time to fibrinolysis ≤30 min) with life expectancy and years of life saved after AMI.
Methods We analyzed data from the Cooperative Cardiovascular Project, a study of Medicare beneficiaries hospitalized for AMI, with 17 years of follow-up. Life expectancy and years of life saved after AMI were calculated using Cox proportional hazards regression with extrapolation using exponential models.
Results Survival for recipients and non-recipients of the 5 guideline-based therapies diverged early after admission and continued to diverge during 17-year follow-up. Receipt of aspirin, beta-blockers, and acute reperfusion therapy on admission was associated with longer life expectancy of 0.78 (standard error [SE]: 0.05), 0.55 (SE: 0.06), and 1.03 (SE: 0.12) years, respectively. Patients receiving primary percutaneous coronary intervention (PCI) within 90 min lived 1.08 (SE: 0.49) years longer than patients with D2B times >90 min, and door-to-needle (D2N) times ≤30 min were associated with 0.55 (SE: 0.12) more years of life. A dose–response relationship was observed between longer D2B and D2N times and shorter life expectancy after AMI.
Conclusions Guideline-based therapy for AMI admission is associated with both early and late survival benefits, and results in meaningful gains in life expectancy and large numbers of years of life saved in elderly patients.
Performance measurement and public reporting are integral to improving care for patients with acute myocardial infarction (AMI) (1–3), yet quantifying the importance of quality measures in real-world populations is challenging. Many AMI guidelines were derived from clinical trials that enrolled younger, healthier patients than the general population with AMI (4–9), and the evidence for early treatment was largely based on short-term outcomes. Only 4 trials have examined long-term survival (5,10–12). Similarly, observational studies in older populations have focused on short-term outcomes (13–20). Little is known about the magnitude and persistence of these survival benefits over the long term.
The relationship between these therapies and long-term outcomes may be best characterized by evaluating life expectancy in recipients and non-recipients of these therapies. Life expectancy places the benefit in terms of absolute, rather than relative, terms and is an easily understood metric of the magnitude of the benefit. No prior studies have evaluated the association between the use of AMI guidelines and life expectancy, because to do so requires a large study sample, long-term mortality data, and detailed clinical information. We used data from the Cooperative Cardiovascular Project (CCP), which included detailed medical record abstraction and over 17 years of follow-up, to determine the relationship between 5 AMI admission guidelines (aspirin on admission, beta-blockers on admission, acute reperfusion therapy, door-to-balloon [D2B] within 90 min, and door-to-needle [D2N] within 30 min of arrival) with long-term survival and life expectancy after AMI in elderly patients.
The CCP is a national program conducted by the Health Care Financing Administration (now Centers for Medicare & Medicaid Services) that reviewed medical records from all U.S. Medicare fee-for-service beneficiaries hospitalized at nongovernmental acute care hospitals with a principal discharge diagnosis of AMI (International Classification of Diseases, Ninth Revision, Clinical Modification, code 410) between January 1994 and February 1996 (N = 234,769). AMI readmission code 410.x2 was excluded (21,22). Professional teams at centralized data centers abstracted medical charts for patient demographics, medical history, clinical presentation, and receipt and timing of procedures and medical therapies.
The Yale University institutional review board approved this study.
We included all patients ≥65 years with a confirmed AMI defined by elevations in cardiac enzymes (creatine kinase-myocardial band or lactate dehydrogenase), chest pain before admission, and diagnostic changes on electrocardiography (ST-segment elevation or new pathological Q waves). We excluded patients hospitalized outside of the 50 states, patients who developed AMI during surgery or other acute hospitalizations, and patients who arrived by interhospital transfer. We also excluded patients whose records could not be linked to data on hospital characteristics from the American Hospital Association, information about socioeconomic status from the 1990 U.S. Census, and information on vital status from the Medicare Denominator files. Our final study cohort included 147,429 patients (Figure 1).
In addition to criteria listed in the preceding text, we used separate eligibility criteria for each guideline. Patients were considered to be “ideal candidates” if they were eligible for a particular therapy and had no therapy-specific contraindications per American Heart Association/American College of Cardiology guidelines (Table 1). We excluded ideal candidates for aspirin and beta-blockers who died within 2 days of hospitalization, and ideal candidates for acute reperfusion therapy who died within the first day of hospitalization, to ensure that all eligible patients were alive long enough to receive these therapies. For the D2B and D2N guidelines, we required that eligible patients presented within 12 h of symptom onset with electrocardiographic evidence of ST-segment elevation AMI or left bundle branch block. Patients had to have undergone percutaneous coronary intervention (PCI) within 3 h or fibrinolytic therapy within 2 h of presentation.
Definitions of variable
We evaluated the association between life expectancy after AMI and 3 guideline-based admission therapies (aspirin within 48 h of admission, beta-blockers within 48 h of admission, and acute reperfusion therapy [fibrinolytic agents or PCI] within 12 h of admission) and 2 reperfusion guidelines (D2B within 90 min and D2N within 30 min of hospital arrival). We obtained information on receipt and timing of therapies from the medical record data in CCP. To ascertain vital status over 17 years of follow-up, we linked the CCP to the Medicare Denominator files from 1994 to 2012 using Medicare beneficiary health insurance claim numbers to obtain dates of death. Because nearly all patients >65 years of age who qualify for Medicare remain enrolled for life, death information was considered complete for all patients enrolled in CCP.
Baseline characteristics of eligible patients receiving and not receiving each admission guideline therapy were compared separately using chi-squared tests or Student t tests. Conditional hazard ratios for the intervals 0 to 30 days, 30 days to 1 year, 1 to 5 years, and 5 to 17 years were calculated using marginal Cox proportional hazards models. This approach accounts for clustering of patients within hospitals and adjusts for hospital site in order to account for differences in the quality of care delivered across hospitals. Only patients alive at the start of the interval were included in the hazard ratio calculations. Models were then repeated adjusting for patient and hospital characteristics.
Covariates were selected using a combination of prior literature and clinical judgment, and included patient demographics, medical history, admission findings, and hospital characteristics. Because patients receiving 1 guideline-based therapy were more likely to receive other therapies, we also adjusted for other admission therapies. Patients with missing data for systolic blood pressure were assigned the cohort’s median systolic blood pressure, as well as a binary dummy variable denoting missing data. When covariates were contraindications to receiving a therapy (i.e., congestive heart failure for beta-blockers on admission), they were not included in models.
We estimated life expectancy from the time of admission using a 4-step process. First, we limited the sample to eligible patients and fit a marginal Cox proportional hazards model with a dichotomous variable for guideline receipt. Proportional hazards assumptions were checked using Schoenfeld residuals, examined graphically and tested formally. Second, we plotted the 17-year expected survival curves from the Cox models for recipients and nonrecipients. Third, we extrapolated the survival curves to age 100 using exponential models. The constant hazard for the exponential model was specified as the average hazard over the last 2 years of follow-up, and the median age of each cohort (i.e., aspirin eligible, beta-blocker eligible) was used to determine the number of years for extrapolation to age 100 years. We selected exponential models because we lacked information on the shape of the survival curves and thus opted for a model with a constant hazard that does not make assumptions about changes to the hazard function over time. Finally, mean life expectancy estimates were calculated by adding the areas under the proportional hazards and exponential survival functions (Online Figure 1). Ninety-five percent confidence intervals (CIs) for the mean were calculated in the same manner using the upper and lower confidence bounds of the expected survival curves. Mean years of life saved by each admission guideline were calculated as the difference in life expectancy between patients receiving and not receiving the guideline. We calculated standard errors from the upper and lower confidence bounds of the survival curves to show uncertainty and significance of estimates. The percentage of life-years saved was defined as the ratio of years of life saved to life expectancy in nonrecipients.
To evaluate whether receipt of admission guideline therapy is associated with longer life expectancy after AMI independent of other characteristics, we repeated the life expectancy calculations described above adjusting for patient and hospital characteristics. To plot the adjusted survival curves, we used overall covariate frequencies or means among all eligible patients. By using the same covariate values to plot the survival curves of recipients and nonrecipients, and thus forcing both groups to have the same risk factor profile, we were able to estimate the adjusted gains in life expectancy associated with each guideline for the “average” eligible patient. Survival curves were again extrapolated to age 100 using exponential models, and life expectancy was calculated by summing the areas under these curves.
We repeated the unadjusted and adjusted life expectancy calculations for aspirin, beta-blockers, and acute reperfusion therapy with age interactions included in the models to determine whether the survival benefits associated with each therapy were consistent across all 5-year age groups. Additionally, we repeated the analyses for aspirin and beta-blockers as 3-level variables (never received, received on admission only, received on admission and discharge) as a sensitivity analysis to determine whether patients receiving these medications on admission only had similar life expectancies to patients receiving them on both admission and discharge. For these analyses, only patients surviving to discharge were included, and all survival estimates were calculated from the time of discharge. Similarly, we repeated the D2B and D2N calculations broken down by 30-min intervals to determine whether patients with greater delays in reperfusion lost more years of life after AMI than those with shorter delays. All analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina).
As a sensitivity analysis, we repeated the Cox proportional hazards models and life expectancy estimates for intravenous (IV) nitroglycerin, a therapy that is unlikely to have long-term benefits, and compared these estimates to those of aspirin and beta-blockers. Unlike aspirin and beta-blocker therapies, there are no established criteria for identifying ideal candidates for nitroglycerin therapy. Initially, we excluded patients with the following contraindications to nitroglycerin only: hypotension (systolic blood pressure <100 mm Hg), bradycardia (heart rate <50 beats/min), and tachycardia (heart rate >100 beats/min) (23). In a second analysis, we also excluded patients with a history of stroke or anemia because these populations may be at higher risk of adverse outcomes after nitroglycerin therapy due to side effects like methemoglobinemia and increased intracranial pressure (24,25), and because the Food and Drug Administration advises caution when using nitroglycerin formulations in these populations (26,27). Additionally, we excluded patients with heart failure because they likely represent a separate subgroup in which nitroglycerin is commonly prescribed but mortality is also significantly higher. All analyses were specified a priori, and both sets of analyses are included for transparency.
Chart-abstracted sample characteristics for the 147,429 patients eligible for at least 1 guideline-based therapy are presented in Table 2. Of these, 106,928 (72.5%) were eligible for aspirin and 65,957 (44.7%) for beta-blockers (Online Table 1). Patients who received aspirin or beta-blockers were on average younger and more likely to have ST-segment elevation AMIs and lower Killip class than those who did not. These patients were also more likely to present to hospitals with greater annual AMI volume and more cardiac care capabilities.
Of the 19,949 patients eligible for acute reperfusion therapy in our sample, 11,225 (56.3%) received either PCI or fibrinolytic therapy (Online Table 2). Patients undergoing reperfusion therapy had fewer comorbidities and lower Killip class than those who did not. A total of 1,261 patients in our sample received primary PCI, and 12,019 patients received emergent fibrinolytic therapy. Approximately one-third of patients receiving primary PCI had D2B times ≤90 min, and 25.7% of patients receiving emergent fibrinolytic therapy had D2N times ≤30 min.
For all 5 guidelines, survival curves of patients receiving and not receiving each therapy separated almost immediately after admission and remained distinct over the duration of follow-up (Figure 2). Patients who received recommended therapies had significantly lower mortality across all follow-up time points (Online Table 3).
To further characterize the shape of the survival curves, we repeated the Cox models for 0 to 30 days, 30 days to 1 year, 1 to 5 years, and 5 to 17 years (Table 3). Conditional hazard ratios for aspirin showed divergence of the survival curves up to 5 years of follow-up but no added benefit beyond 5 years (hazard ratio: 0.99, 95% CI: 0.96 to 1.02). By contrast, the survival benefits associated with beta-blockers and acute reperfusion therapy persisted throughout the entire 17 years of follow-up, although the magnitude of this benefit decreased over time (Online Table 4). The survival benefits associated with early reperfusion (D2B ≤90 min and D2N ≤30 min) continued to increase up to 1 year after D2B and 5 years after D2N, and then persisted up to 17 years after AMI.
These survival patterns were reflected in the calculated life expectancy estimates (Table 4). For any given therapy, patients receiving the recommended therapies had significantly longer crude life expectancies than those not receiving the measure. Differences in life expectancy persisted after adjustment for patient and hospital characteristics, although the magnitude of these differences decreased (Table 4). After adjustment, aspirin on admission was associated with 0.65 (standard error [SE]: 0.05) years of life saved on average, beta-blockers with 0.45 (SE: 0.06) years of life saved, and acute reperfusion therapy with 0.90 (SE: 0.11) years of life saved among eligible patients. The absolute number of life-years saved was greater in younger patients for all 3 guidelines; however, the percentage of life-years saved was comparable across age groups (Figure 3).
Recipients of aspirin, beta-blockers, and acute reperfusion therapy were significantly more likely to receive additional therapies at discharge. For example, 64% of patients who received aspirin on admission were also prescribed aspirin at discharge, compared with only 33% of nonrecipients. Similarly, 63% of patients receiving beta-blockers on admission versus 19% of nonrecipients also received beta-blockers at discharge. When aspirin and beta-blocker receipt were examined as 3-level variables in sensitivity analyses, the survival benefit was greatest in patients receiving aspirin or beta-blockers on both admission and discharge (Online Table 5). Compared with patients who did not receive aspirin during hospitalization, receipt of aspirin on admission only was associated with 0.35 (SE: 0.05) years of life saved, and receipt of aspirin on admission and discharge was associated with 1.78 (SE: 0.05) years of life saved (Online Table 6). There was no difference in life expectancy between patients receiving beta-blockers on admission only and patients not receiving beta-blockers during hospitalization. However, receipt of beta-blockers on both admission and discharge was associated with a significant number of years of life saved (1.01 [SE: 0.07] years).
Earlier reperfusion was also associated with significant gains in life expectancy. After adjustment, D2B times ≤90 min were associated with 0.98 (SE: 0.47) years of life saved and D2N times ≤30 min were associated 0.52 (SE: 0.17) years of life saved (Table 4). When D2B and D2N times were further divided into 30-min intervals, longer times to reperfusion were associated with greater numbers of years of life lost (Figure 4). Compared with patients with D2B times ≤90 min, patients with times between 121 and 150 min and 151 and 180 min lost 0.95 (SE: 0.56) and 1.72 (SE: 0.65) years of life, respectively. Similarly, compared with patients with D2N times ≤30 min, patients with D2N times between 61 and 90 min and 91 and 120 min lost 0.81 (SE: 0.21) and 0.71 (SE: 0.27) years of life, respectively.
In sensitivity analyses evaluating survival after IV nitroglycerin (n = 100,907 in patients without contraindications and n = 67,626 in patients without additional exclusions), recipients had significantly lower crude mortality at all follow-up time points than nonrecipients (Online Tables 7 and 8). After adjustment, the risk of death was higher in the first 30 days for patients receiving IV nitroglycerin. Among the first group, the long-term risk of death was lower for those receiving IV nitroglycerin at all follow-up time points (Online Table 9). However, when the additional exclusion criteria were applied, the long-term risk of death was only lower from 1 year to 5 years after AMI. When quantified in terms of life-years, receipt of nitroglycerin in this group was not associated with improved long-term mortality. Patients receiving nitroglycerin lived 0.10 (SE: 0.06) years longer than those not receiving this therapy (Online Table 10).
Our study provides the first estimation of the lifetime benefit of early treatment with guideline-based care for patients hospitalized with AMI. This study indicates that early benefits persist and enlarge over time, resulting in substantial gains in life-years. For every 1,000 patients treated with aspirin or beta-blockers on admission, 648 and 450 years of life were saved, respectively. For every 1,000 patients treated with fibrinolytic therapy or PCI, 902 years of life were saved; however, these gains differed by timing of receipt. The study provides evidence that intensifies the support for these rapid treatments and estimates what is likely lost by their omission, thereby strengthening the imperative to treat appropriate patients.
Our results extend the findings of previous trials and observational studies by examining AMI admission therapies in the elderly and following patients throughout their lifespan. Several early trials evaluated the prolonged use of aspirin and beta-blockers after AMI and found significant reductions in short-term mortality among patients randomized to these therapies (4–9). We found similar short-term risk reductions in elderly patients with comparable risk ratios to those reported in the trials (Online Table 11). There was a 7% reduction (95% CI: 2% to 12%) in the risk of 30-day mortality and a 10% reduction (95% CI: 7% to 14%) in the risk of 1-year mortality among patients receiving beta-blockers after AMI, which were similar to the 13% for 30-day mortality reported by the MIAMI (Metoprolol in Acute Myocardial Infarction) trial (4) and the 11% for 1-year mortality reported by the ISIS-1 (First International Study of Infarct Survival) trial (5). The similarities in short-term risk estimates between our data and the trials support our risk adjustment methods and suggest that our models adequately adjusted for differences between therapy recipients and nonrecipients.
Our results are consistent with those of prior observational studies in elderly populations. Previous analyses from the CCP reported a 22% decrease in 30-day mortality associated with aspirin administration within 48 h of admission (14), and a 19% reduction in the odds of in-hospital mortality for patients receiving early beta-blocker therapy (19). Similarly, the GRACE (Global Registry of Acute Coronary Events) study found that, during the index hospitalization, aspirin significantly reduced 6-month mortality for non–ST-segment elevation MI patients and beta-blockers significantly reduced 6-month mortality for both non–ST-segment elevation MI and ST-segment elevation MI patients (18). In addition to admission therapies, lower D2B time has also been associated with lower mortality in-hospital (16,20,28) and at 30 days (17) and 1 year (16). Our study extends these findings to the long term and demonstrates a clear dose–response relationship between longer D2B and D2N times with lower life expectancy.
Of all studies examining the association between AMI admission guidelines and outcomes, only 4 trials have examined mortality beyond the first year. In the first study by von Domburg et al. (11), the authors investigated the effects of reperfusion therapy (streptokinase with or without coronary angioplasty) on life expectancy in 533 patients with AMI. They found that life expectancy was increased by 2.8 years in patients receiving reperfusion therapy compared with those receiving conventional therapy. Similarly, in the ISIS-1 trial, patients with suspected AMI randomized to atenolol had a 15% decrease in mortality in the first week, which narrowed to 11% at 1 year and became nonsignificant at 20 months, suggesting beta-blockers were associated with early, but not late, survival benefits (5). The ISIS-2 trial and the GISSI-1 (Gruppo Italiano per lo Studio della Streptochinasi Nell'Infarto Miocardico-1) trial demonstrated that the survival curves for patients receiving streptokinase and aspirin separated shortly after admission and then remained parallel over the next 10 years, again suggesting that the survival benefits associated with these therapies were largely limited to the short term (10,12). By contrast, we found that the survival curves for aspirin, beta-blockers, and acute reperfusion continued to widen over 5 years of follow-up, suggesting that patients receiving these therapies were still accruing a survival advantage well beyond the first year after AMI. Interestingly, there did appear to be a trend towards a long-term effect for patients receiving streptokinase (discharge to 10-year risk ratio 0.84 [95% CI: 0.65 to 1.08]) in the GISSI-I subgroup who were randomized within the first hour, but these analyses may have been underpowered to detect an effect with only 1,126 patients.
There are several potential explanations as to why recipients continued to accrue a survival advantage over time. First, early receipt of PCI or fibrinolytic therapy decreases myocardial ischemia and thus can reduce both the size of the infarct and subsequent remodeling (29,30). Preservation of myocardium can improve both regional and global ventricular function, and reduce the rates of heart failure and reinfarction after AMI (31). These persistent benefits may also be seen with early administration of aspirin and beta-blockers. Within minutes of administration, aspirin prevents platelet activation and interferes with platelet adhesion, which may reduce infarct size, progression, or reinfarction (32–35). Similarly, beta-blockade decreases overall myocardial oxygen supply and can minimize myocardial injury (32,36,37). Thus, some of the long-term benefit observed may be due to the immediate effects of aspirin and beta-blockers in reducing the severity and extent of myocardial death. Second, continued use of therapies like aspirin and beta-blockers explains much of the long-term survival advantage observed with these therapies. Recipients of aspirin and beta-blockers were significantly more likely to receive these therapies at discharge and may have been more likely to receive other therapies such as statins, ACE inhibitors, and calcium channel blockers. The 3-level analyses for aspirin and beta-blockers demonstrate that the long-term survival benefits of these therapies were most pronounced in patients receiving these therapies on admission and at discharge. Patients receiving aspirin on admission, but not discharge, had a smaller but still significant benefit from the early therapy, whereas patients receiving beta-blockers on admission, but not discharge, showed no differences in long-term survival relative to patients not receiving beta-blockers at all. Interestingly, the only AMI measure to have been retired by the Centers for Medicare & Medicaid Services is measure AMI-6, which recommended beta-blockers within 24 h of admission (38). This measure was retired in 2009 due to overwhelming evidence from COMMIT (Clopidogrel and Metoprolol in Myocardial Infarction Trial), which showed that early beta-blockers after AMI may not be appropriate for certain patient populations (39–41).
Finally, we cannot exclude the possibility of residual confounding by unmeasured confounders, which may explain some of the continued survival benefit if recipients of admission therapies were significantly healthier or at lower risk of adverse outcomes than nonrecipients. Although we adjusted for a number of demographic, clinical, and hospital characteristics, there are several additional confounders that we were unable to measure, such as receipt of other medications including anti-platelet agents, receipt of aspirin and beta-blockers immediately prior to hospitalization, additional socioeconomic characteristics, and long-term compliance with aspirin and beta-blockers. Nevertheless, the findings of the nitroglycerin sensitivity analyses support the risk adjustment models used in this study. When patients without our set of exclusions were examined, IV nitroglycerin was not associated with an improvement in life expectancy between recipients and nonrecipients.
This study strengthens the case for why rapid and early delivery of AMI admission guidelines is important and highlights the magnitude of progress made in improving the quality of care for patients with AMI in the United States. Today, rates of adherence to aspirin and beta-blocker guidelines approach 99%, and 95% of D2B times are <90 min (42). However, when CCP was conducted in the mid-1990s, approximately 25% of patients eligible for aspirin and one-half of patients eligible for beta-blockers at admission did not receive these therapies. Similarly, two-thirds of patients received PCI after the recommended 90 min. Our data imply that if rates of adherence in CCP equaled those today, an additional 15,764 years of life could have been saved by treatment with aspirin, 15,065 years by beta-blockers, and 764 years by shortening D2B times to the recommended 90 min. These findings demonstrate the cost of a missed opportunity when eligible patients are not treated.
Although public reporting and pay for performance initiatives have greatly improved the quality of care for AMI patients in the United States, many countries lag behind in implementing these quality measures. Recent reports from Europe, Australia, China, and the Middle East indicate that the percentage of patients with AMI receiving these admission treatments varies widely by country. Rates of adherence to aspirin and beta-blocker guidelines range from <50% in some countries to >90% in others (13,43–47), but rates of adherence to D2B recommendations have been consistently low across countries (25% to 60%) (15,48). Thus, opportunities for improving AMI care in other countries are numerous and may result in large gains in both lives saved and years gained.
First, because this is an observational study and not a randomized trial, patients who received the guideline-based therapies likely differ from those who did not. To address this, we limited the analysis to only ideal candidates and then further adjusted for differences in baseline demographic, clinical, and hospital characteristics. The IV nitroglycerin sensitivity analyses support the contention that risk-adjusted models are sufficient, although patients receiving and not receiving the guideline-recommended therapies may have differed on other unmeasured confounders. Second, the use of guideline-based therapies is strongly associated with the use of other pharmacological interventions. Although we adjusted for other admission guidelines in our analyses, we chose not to adjust for discharge medications because we hypothesized that discharge medications may be intermediate variables rather than confounders in the relationship between admission therapies and long-term survival and thus should not be included in the models. We did evaluate aspirin and beta-blocker receipt as a 3-level variable in sensitivity analyses, which showed that much of the observed benefit was due to both early and late use of aspirin and beta-blockers over the long term rather than one-time use. Third, comparative effectiveness studies are frequently complicated by survivorship bias, whereby patients surviving longer are more likely to receive a particular therapy. We limited analyses to patients who survived the first 24 or 48 h of hospitalization to ensure that patients lived long enough to receive the admission therapies; however, this requirement may limit the generalizability of our findings to all patients. Finally, the quality of AMI care has improved since data for CCP were collected in the mid-1990s. Compared with older therapies, stents and fibrinolytic agents used today have been shown to increase longevity after AMI. As a result, we may be underestimating the life expectancy gap between guideline recipients and nonrecipients.
Although these life expectancy analyses have their limitations, they can only be performed using a dataset such as CCP and will likely never be repeated for several reasons. First, calculation of life expectancy requires complete follow-up over many years, which contemporary datasets currently lack. Second, evaluation of admission therapies requires a sufficient number of patients who did not receive these therapies for comparison. At present, compliance with AMI admission guidelines is >90% in the vast majority of U.S. hospitals, leaving only small numbers of patients for comparison. Finally, adjustment for differences in patient and hospital characteristics between patients requires an immense dataset with detailed clinical information, such as CCP.
We demonstrate that early survival benefits observed in patients receiving AMI admission guidelines are sustained over the long term and result in sizeable improvements in life expectancy (Central Illustration). Similarly, more rapid time to treatment with direct PCI or fibrinolytic therapy predicted substantial gains in life expectancy for patients undergoing acute reperfusion therapy. These findings illustrate that early and rapid delivery of guideline-based admission therapies for AMI is associated with a large number, not only of lives saved early, but also of years gained over the long term.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: In elderly patients with AMI, early administration of aspirin, beta-blockers, and reperfusion therapy improves both short-term survival and years of life preserved.
TRANSLATIONAL OUTLOOK: Although substantial progress has been made in providing guideline-directed care for patients with AMI in the United States, more work is needed to extend similar gains throughout the world.
The authors acknowledge the assistance of Qualidigm and the Centers for Medicare & Medicaid Services in providing data, which made this research possible. The content of this publication does not reflect the views of Qualidigm or the Centers for Medicare & Medicaid Services, nor does mention of organizations imply endorsement by the U.S. government.
Dr. Bucholz was supported by an F30 training grant F30HL120498-01A1 from the National Heart, Lung, and Blood Institute and by NIGMS Medical Scientist Training Program grant T32GM07205 during the time the work was conducted. Dr. Krumholz was supported by grant U01 HL105270 (Center for Cardiovascular Outcomes Research at Yale University) from the National Heart, Lung, and Blood Institute during the time the work was conducted; has received research grants from Medtronic and Johnson and Johnson through Yale University for the purpose of disseminating clinical trials; and chairs the Cardiac Scientific Advisory Board for United Health. Dr. Normand is a member of the board of directors of Frontier Science and Technology. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute myocardial infarction
- Cooperative Cardiovascular Project
- confidence interval
- percutaneous coronary intervention
- Received August 30, 2015.
- Revision received February 29, 2016.
- Accepted March 8, 2016.
- American College of Cardiology Foundation
- ↵The Joint Commission. Core Measure Sets. 2014. Available at: http://www.jointcommission.org/core_measure_sets.aspx. Accessed March 12, 2014.
- Krumholz H.M.,
- Anderson J.L.,
- Brooks N.H.,
- et al.
- Centers for Medicare & Medicaid Services. Process of Care Measures. 2013. Available at: http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/HospitalProcessOfCareMeasures.html. Accessed March 12, 2014.
- The MIAMI Trial Research Group
- Baigent C.,
- Collins R.,
- Appleby P.,
- Parish S.,
- Sleight P.,
- Peto R.,
- The ISIS-2 (Second International Study of Infarct Survival) Collaborative Group
- von Domburg R.T.,
- Sonnenschein K.,
- Rieuwlaat R.,
- et al.
- Franzosi M.G.,
- Santoro E.,
- De Vita C.,
- et al.
- Krumholz H.M.,
- Radford M.J.,
- Ellerbeck E.F.,
- et al.
- Berger P.B.,
- Ellis S.G.,
- Holmes D.R. Jr..,
- et al.
- McNamara R.L.,
- Wang Y.,
- Herrin J.,
- et al.
- Ramunno L.D.,
- Dodds T.A.,
- Traven N.D.
- O’Connor R.E.,
- Brady W.,
- Brooks S.C.,
- et al.
- Aiyagari V.,
- Gorelick P.B.
- ↵U.S. Food and Drug Administration. Nitrolingual pumpspray (nitroglycerin) spray. Available at: http://www.fda.gov/Safety/MedWatch/SafetyInformation/ucm433470.htm. Accessed February 28, 2016.
- U.S. Food and Drug Administration. Nitrostat. Available at: http://www.accessdata.fda.gov/drugsatfda_docs/label/2014/021134s007lbl.pdf. Accessed February 28, 2016.
- Touchstone D.A.,
- Beller G.A.,
- Nygaard T.W.,
- Tedesco C.,
- Kaul S.
- Marino P.,
- Zanolla L.,
- Zardini P.
- O’Connor R.E.,
- Brady W.,
- Brooks S.C.,
- et al.
- ↵Office of Clinical Standards and Quality. CMS Updates the National Hospital Quality Measure Acute Myocardial Infarction Set for Discharges as of April 1, 2009 [Memorandum]. Baltimore, MD: Department of Health and Human Services. Available at: https://www.cms.gov/medicare/quality-initiatives-patient-assessment-instruments/hospitalqualityinits/downloads/hospitalami-6factsheet.pdf. Accessed February 8, 2016.
- Perez M.I.,
- Musini V.M.,
- Wright J.M.
- ↵Centers for Medicare & Medicaid Services. Hospital Compare. 2014. Available at: http://www.medicare.gov/hospitalcompare. Accessed July 5, 2014.
- Luthi J.C.,
- McClellan W.M.,
- Flanders W.D.,
- Pitts S.R.,
- Burnand B.
- Schiele F.,
- Meneveau N.,
- Seronde M.F.,
- et al.