A Call for Standardized Rehospitalization Measures and Information Systems
Are rates of rehospitalizations improving or worsening in the United States, in your state, in your community, or in your hospital? These questions are likely to lead not to answers but to a number of other questions. For which group of patients: Medicare, Medicaid, or privately insured? And within what time frame: 15 days, 30 days, or six months from the initial hospitalization? Does the rate take into account preventable rehospitalizations, disease-specific rehospitalizations, or those for any cause? Does it include readmissions to the same hospital or rehospitalizations to any hospital? Even if you can clarify such issues, meaningful information on rehospitalization trends may continue to elude you, given the lack of national standards for measuring them.
While there are plenty of published studies on rehospitalization rates, these typically look at rehospitalizations at a single point in time and for a specific group of patients. For example, in 2009, Stephen F. Jencks, M.D., M.P.H., and his colleagues published data on 30-day rehospitalizations for any cause to any hospital. They found that 20 percent of Medicare beneficiaries who were discharged from the hospital during a 15-month period—October 1, 2003, through December 31, 2004—were readmitted to a hospital, not necessarily the original hospital, within 30 days, at an estimated cost of $17 billion in Medicare payments. Yet, those data represent the performance of the health care system at a particular time, and only for Medicare patients. It is difficult if not impossible to find a comparable study that shows whether the rate Jencks identified represents an improvement or a decline in rehospitalizations over the past five or 10 years.
Two recent studies on hospital rehospitalization rates underscore the problem by providing only snapshots of performance. A study from California's Office of Statewide Health Planning and Development showed that about one-third of California patients were readmitted to the hospital in 2005 for any reason. And the Agency for Health Care Research and Quality (AHRQ) found that about 25 percent of patients returned to the hospital for reasons related to the original condition within two years of a 2006–2007 initial stay.
Several national reports published by groups such as the Medicare Payment Advisory Commission, the National Quality Forum, the Centers for Medicare and Medicaid Services (CMS), and AHRQ, as well as The Commonwealth Fund's national and state scorecards, feature rehospitalization rates, and some have tracked rates over time. Yet, two problems persist: each group has adopted a different definition of rehospitalization, which makes comparisons and benchmarking across states, regions, and hospitals impossible, and most reports are based on data that are more than two years old.
One reason for the measurement conundrum we face today is that basic data are difficult to obtain. For instance, in many states only Medicare data are available and using those databases require special analytic expertise. Many states face significant challenges in getting data to track rehospitalizations for the under-65 population. And for smaller hospitals, where rehospitalizations are not that frequent, the need to collect data over a longer time period to ensure stability in the calculated rates leads to results that are not really actionable when it comes to improving performance.
In light of the scope of this national problem, the lack of timely information on all patients and a standard way to measure rehospitalizations is no longer tenable. Rehospitalizations are prevalent, harmful to patients, and costly. More important, they are a symptom of our fragmented health care system, which is still working within silos. To move toward greater levels of care coordination, integration, and accountability, we must have standard measures to track progress. After all, what gets measured gets done.
The new measures should apply to all patients, regardless of who insures them and where they receive their care. And the country must invest in information systems that can provide the timely data—not five-year-old or even six-month-old data—that are necessary to monitor and take appropriate steps when problems are identified.
It is also clear that the measurement system used significantly affects rehospitalization rates at the hospital level, as well as hospital rankings. If Medicare, states, and other payers adopt different measurement methods, it will create significant problems for providers and make it difficult to use the measures to compare and reward or penalize good and poor performance, respectively. A measurement standardization process among payers (Medicare, states, and other commercial payers) should thus be a high priority.
What Would a Standard Measure Specify?
Standard definitions should specify whether the measure is an "all-cause" rehospitalization or a "condition-specific" measure. The definitions should also specify the time frame (e.g., rehospitalization within one week, 30 days, 60 days, or one year) and reasons for the rehospitalization—whether they are related to the initial admission, unrelated, or considered to be avoidable complications. It should also define the population to be included in calculations of rehospitalization rates, specifying whether those with special circumstances, such as planned readmissions, or certain conditions such as cancer, should be excluded. Risk-adjustment methods should be specified.
The measure should capture all individuals, wherever they enter or re-enter the health care system. Indeed, Jencks and colleagues have shown that 20 percent of patients rehospitalized within 30 days go to a hospital other than the one from which they were originally discharged. In order to identify everyone who is rehospitalized within a certain time frame, an all-payer all-provider database will be essential.
Timely and Actionable Data
Although a standard measure of rehospitalizations is critical, it is not sufficient. To foster appropriate care management, and to provide the right incentives to reward or penalize poor performance, the data need to be current and actionable.
For example, CMS publishes 30-day readmission rates for three conditions: congestive heart failure, acute myocardial infarction (heart attack), and pneumonia care. (Hospital-level comparative data for these measures, as well as state, national, and custom benchmarks, are available on the Commonwealth Fund's benchmarking Web site, WhyNotTheBest.org.) But those data represent a three-year average rolled into one, and the latest rate available covers 2005 to 2008. Although the use a three year timeframe to aggregate data may be necessary to ensure stability and validity in rates, especially for smaller hospitals, given the pace of change in health care delivery, the problem of real time information required for quality improvement needs to be seriously considered and solved.
Health information technology and exchange can help to speed public reporting. Through the comprehensive set of programs being deployed by the Office of the National Coordinator for Health Information Technology, progress is being made toward a solid health information technology infrastructure.
For providers, having rehospitalization data in real time will mean they will be able to follow the health outcomes of their population, regardless of payer. Health information exchange will enable providers to know when patients are rehospitalized elsewhere or return to an emergency room after being discharged. It's critical that this information be available to providers when a patient is admitted or seen, not one year later and not even two weeks later.
In addition to benchmarking and monitoring progress and enabling providers to improve care, 30-day rehospitalizations rates will soon be used to make payment decisions. The Hospital Readmissions Reduction Program, to be established under the Affordable Care Act, calls for reductions in Medicare payments to hospitals for preventable readmissions in the fiscal year beginning on or after October 1, 2012. Medicare will also introduce new payment models: bundled payment and episode-based payments, which will include the three-day window prior to admission, the length of stay, and 30 days following discharge.
The Affordable Care Act also includes provisions for a program that will implement a methodology for tracking avoidable hospital readmissions and calculating savings that result from improved chronic care coordination.
Improving care transitions and reducing rehospitalizations are among the priorities of The Commonwealth Fund's Health System Quality and Efficiency Program. The Commonwealth Fund-supported State Action to Avoid Rehospitalizations (STAAR) initiative led by the Institute for Healthcare Improvement is working with providers, payers, and state agencies in Massachusetts, Michigan, and Washington to redesign care transitions and develop appropriate levers—measurement standards and reporting and payment reforms—to support health system transformation.
National standards for rehospitalizations will enable patients to engage in their care; providers to monitor what works and where they may need to intervene once a problem is identified; regulators to develop effective regulations; and payers to develop appropriate payment incentives to foster high levels of performance. It is imperative that we seize the chance to implement standards now.