Welcome to NQMC. Skip directly to: Search Box, Navigation, Content.
Expert Resources > Expert Commentary
Expert Commentary |
Have Comments? E-mail them to us. |
| Perspective |
Hospital Process Measures: Only One Piece of the Quality Improvement Puzzle
By: Rachel M. Werner, MD, PhD
Although hospital quality has improved over the past decades, (1,2) it remains uneven and often inadequate. (3) Too often patients do not receive potentially life-saving treatments, such as aspirin when admitted for a heart attack or angiotensin-converting enzyme (ACE) inhibitor after hospitalization for heart failure. The omission of proven therapies is often cited as evidence of inadequate hospital quality that may lead to preventable deaths.
In response to these quality deficits, the Centers for Medicare & Medicaid Services (CMS), along with other health care organizations, participate in the Hospital Quality Alliance, a large scale public-private collaboration that seeks to make performance information on all acute care non-federal hospitals accessible to the public, payers, and providers of care. (4) At the time of its inception, the performance measures used by Hospital Quality Alliance evaluated hospital quality on processes of care for patients with common conditions, including acute myocardial infarction, heart failure, and pneumonia. In April 2005, the CMS began publishing hospital performance and rankings based on these measures on their website, "Hospital Compare" (http://www.hospitalcompare.hhs.gov/).
The performance measures adopted for use in Hospital Compare captured performance across one domain of care: the process of health care delivery. While some research has documented an association between higher adherence to care guidelines and better outcomes of patients who receive that care, (5,6) there has been limited evidence demonstrating that hospitals performing better on process measures also perform better on other important domains of care. Thus, we questioned whether the process measures used in CMS's Hospital Compare are predictive of hospitals' quality beyond the processes that are directly measured?
In an effort to explore this issue, my colleague, Eric Bradlow, and I examined the correlation between each U.S. hospital's performance on process measures from Hospital Compare and risk-adjusted mortality rates, another common metric of hospital quality. (7) We found that although there was a correlation between mortality rates and process measures, only a small part of the difference in mortality rates across hospitals was explained by differences in process performance. For example, compared to bottom-performing hospitals on the heart attack measures, those that were top-performing (with process performance that was 8 percentage points higher) had 30-day mortality rates that were 0.7 percentage points lower. For heart failure and pneumonia, the mortality reduction from better process performance was smaller. Though we found modest mortality reductions, we also found large variations in mortality across hospitals. The interquartile range of 30-day mortality was 12.7% to 19.2% for acute myocardial infarction (AMI); 7.9% to 12.1% for heart failure; and 9.8% to 15.5% for pneumonia. Thus, it seems that a hospital's process performance alone does not convey information about hospital quality that extends beyond what is actually being measured. Furthermore, our findings were consistent with those of other researchers investigating similar questions. (8)
There are numerous reasons for the relatively weak predictive power of these common metrics. First, these discrete metrics measure just that--discrete (and relatively brief) episodes during a hospitalization, and, thus, it may not be surprising that the metrics do not capture information about the entire hospitalization. Second, because process measures alone are not designed to predict mortality (in fact, some, like pneumococcal vaccination, may not be directly related to mortality), and because the variation in process performance across hospitals is not large in some cases, it should not be surprising that we do not find a stronger relationship between process and outcomes. In fact, the magnitude of the mortality difference we find is larger than one would expect if the differences in mortality were only due to the differences in process measure themselves. (9) Third, as always, there may be inaccuracies in the data itself that prevented us from finding a stronger relationship, if it exists.
Some may interpret these findings as evidence that these performance measures are ineffective metrics for improving quality. I disagree. Administering proven therapies to all patients when they need them is a worthy and important goal. The improved compliance with process measures known to improve individual-patient outcomes indicates some degree of success of the Hospital Quality Alliance's public reporting initiative.
However, alone, these quality targets are not enough and are potentially a risky proposition. While laudable, they also have limited utility in quality improvement initiatives as they provide insufficient information to fully inform the public, payers and providers about hospital quality. It is potentially risky if this focus on a single domain of quality draws attention away from, and thus has adverse consequences for, other domains of quality.
Ideally, we would reward hospitals based on overall hospital quality. In the absence of such a measure, however, process measures should be a part of quality improvement incentive systems. Yet, a policy of rewarding hospitals based only on single (or even a limited number of) domains does not make sense.
Author
Rachel M. Werner, MD, PhD
University of Pennsylvania and Philadelphia VA, Philadelphia, Pennsylvania
Disclaimer
The views and opinions expressed are those of the author and do not necessarily state or reflect those of the National Quality Measures Clearinghouse™ (NQMC), the Agency for Healthcare Research and Quality (AHRQ), or its contractor ECRI Institute.
Potential Conflicts of Interest
Dr. Werner states no conflicts of interest.
NQMC Resources
| What's New
| Feedback
| Accessibility
| NQMC Disclaimer
| Privacy Policy Notice
| USA.gov
| Adobe Reader
Date Modified: 11/16/2009
