Abstract & Commentary
By Jonathan Abrams, MD, Professor of Medicine, Division of Cardiology, University of New Mexico, Albuquerque. Dr. Abrams serves on the speaker's bureau for Merck, Pfizer, and Parke-Davis.
Source: Peterson ED, et al. Association Between Hospital Process Performance and Outcomes Among Patients with Acute Coronary Syndromes. JAMA. 2006;295:1912-1920.
Crusade (can rapid risk stratification of unstable Angina Patients Suppress Adverse Outcomes With Early Implementation Of The ACC/AHA Guidelines) is a large quality improvement initiative evaluating the performance of multiple hospitals with respect to current NSTEMI guidelines. This is an evaluation of 350 US hospitals, with a goal of determining variation of hospital performance for a large number of individual care processes; an additional goal is to identify hospital characteristics that are predictive of higher adherence to contemporary guidelines. CRUSADE is a voluntary observational data collection and quality improvement initiative that began in 2001. Enrolled hospitals collect and submit clinical information regarding the care and outcomes of NSTEMI with high risk clinical features. Extensive data extraction and statistical analyses are critical to the success of CRUSADE. Web-based collection tools were used and multiple procedures to monitor and improve data quality were employed. Complete data collection has been documented in audits of the CRUSADE sites.
The study population consists of 427 hospitals enrolling 78,000 patients with NSTEMI between January 2001 and September 2003. Participating hospitals in the analysis had to have at least 40 cases and 1 death during the study period. The final number of study patients was 65,000 at 350 CRUSADE hospitals. Nine ACC/AHA Class 1 guideline criteria for NSTEMI were utilized, including 4 process of care measures (aspirin, beta-blocker, heparin, and glycoprotein 2B-3A inhibitors within 24 hours). In addition, 5 discharge regimens were assessed (aspirin, beta-blocker, clopidogrel, ACEI, and lipid medication). Deaths within 24 hours were excluded from the analysis. Composite adherence scores were calculated at the patient and hospital level. Clinical characteristics, medications, procedures, and mortality were compared among hospital and patients. Correlation of hospital adherence rates, mortality, procedures, medications, and patient clinical characteristics were assessed. Multiple markers were evaluated and many statistical analyses were carried out, including sensitivity analyses. Patients with or without elevated biomarkers and high risk elderly patients > 65 were also separately assessed.
The results indicate that 74% of all treatment care decisions were consistent with current guideline recommendations. Composite guideline adherence scores among the hospitals varied widely, with median composite adherence scores ranging between 82% in the highest performing hospitals to 63% in those in the lowest adherence quartile. Variability in use of recommended therapies were assessed, with the highest composite adherence hospitals having the highest average performance for acute, discharge, and secondary prevention metrics. Aspirin had the lowest degree of variance, while use of clopidogrel at discharge and in-hospital use of glycoprotein 2B-3A inhibitors showed considerable variance (up to 2-3 fold) among the hospitals. Secondary prevention measures were 20-30% higher in the highest performing quartile of hospitals versus the lowest. Predictors of outcomes were higher in institutions with cardiac revascularization facilities, as well as those with a higher proportion of patients treated primarily by a cardiologist. There was a significant inverse correlation between use of individual care processes and in-hospital mortality; the "highest associations between use and mortality were observed for acute intravenous glycoprotein 2B-3A inhibitors, discharge clopidogrel, and discharge lipid lowering agents."
Overall, adherence to the 9 ACC/AHA guidelines demonstrated a negative association in hospital mortality. Mortality rates in patients who had a documented MI were slightly higher than in the broader population of acute coronary syndromes (5.3% vs 4.9%). In-hospital combined death/MI rates decreased as a function of guideline adherence. After adjustment for mortality rates and adherence, there was a differential mortality for ACS of 6.3% in the lowest quartile compared to 4.1% in the highest quartile, P = < 0.001. The odds ratio for in-hospital mortality in the highest vs the lowest hospital adherence quartiles was 0.81. For every 10% increase in overall guideline adherence there was a corresponding 10% decrease in mortality.
Multiple populations and situations were tested with sensitivity analyses and "the conclusions…were robust." For instance, NSTEMI mortality in the lowest quartile was 7.7% vs 4.3% in those in the highest hospital quartile. In patients > 65, the rate was 8.9% in lowest performing hospitals and 6.1% in the highest quartile; odds ratio of 0.83. Peterson and colleagues found that the outcomes were comparable, including or excluding the analysis to all 427 CRUSADE hospitals. Mortality rates were lower in matched pairs of patients treated at the leading adherence centers; 4.17% vs those in the "lagging quartile centers;" 5.47%, which reflected a 33% difference. Peterson et al conclude that in the overall population of the CRUSADE registry, "up to 25% of opportunities to provide guidelines recommended care were missed in current practice," confirming considerable variability among the centers. They state, "To our knowledge, our study is among the first to link variability in hospital process performance with patient outcomes." Guideline adherence was directly associated with mortality, supporting the use of guideline based process measures as a means assessing the quality of care at an individual institution.
Peterson et al point out that similar results have been suggested in other reports of under-utilization of evidence-based care measures in patients with ACS. They emphasize that multiple process metrics are needed to "fully characterize hospital care practices," as an individual measure does not necessarily predict outcomes. Only CABG availability and direct care by a cardiologist were significantly related to higher adherence rates. Of great importance, they state that there was a "strong dose-dependent association between hospital adherence to care guidelines and acute patient outcomes." Overall, the CRUSADE hospitals data demonstrated that the elderly, minorities, and patients with the most co-morbidities tend to be treated at hospitals with lower measures of adherence, in part because healthier patients are often transferred to high performing hospitals. Peterson et al suggest that adherence performance may be a surrogate marker for hospital culture and overall quality of care. Of interest, Peterson et al point out that use of statins, clopidogrel, and 2B-3A inhibitors may be closer and more accurate markers of hospital outcome than older and established treatments; the registry demonstrated significantly more variation with the newer therapies than in the older ones among the hospitals. They conclude that up to 25% of guideline-based care opportunities are missed in practice, emphasizing that quality assessment approaches are important. The strong association between hospital composite care performance and patient outcomes clearly "supports the central hypothesis of hospital quality improvement, namely, better adherence with evidence based care practices will result in better outcomes for patients who are treated."
This report, which is an extensive and detailed analysis of performance parameters for acute coronary syndromes measures in 350 hospitals. This enormous accomplishment demonstrated what could intuitively be predicted: the better a hospital's overall compliance and adherence to ACC/AHA guidelines, the better the outcomes. The fact that there is a substantial mortality difference between the highest and lowest quartiles of performance is important and disturbing. One would have assumed that mortality in ACS would be equal in almost all hospitals, but there was a significant difference between the highest and lowest level performers. The study emphasizes the use of multiple markers rather than a single variable to assess adherence and hospital performance, and this makes sense as well. Access to revascularization without transfer, and the participation of a cardiologist in the care of ACS patients, turned out to be independent predictors of high quality hospital adherence and performance, supporting the concept that patients do better at institutions that have multiple systems in place for appropriate care (and presumably have a high degree of cardiovascular business).
For the practicing physician, the CRUSADE registry appears to be an indicator of what the future holds. There are multiple, similar initiatives around the country looking at various aspects of cardiovascular care, with efforts to raise the bar for all hospitals, particularly low performers. The CRUSADE experience may or may not be surprising to physicians; it clearly indicates that guideline-driven therapy for acute coronary syndromes is valid and critical to producing the best outcomes. When there is a 20-30% variation in mortality among the 4 quartiles, the message is loud and clear: adherence to Class 1 guidelines is not simply rhetoric, but represents solid adherence to proven performance measures with the assurance that outcomes will be optimal. The era of close surveillance of practice measures is at hand. Peterson et al are to be congratulated for this pioneering report.