RAND Office of Media Relations
(703) 413-1100, ext. 5117
January 13, 2004
Referring infants with very low birthweights to hospitals that treat a large number of similar cases is an unreliable way to direct patients to the highest quality medical providers, according to a new study headed by RAND Corporation researchers.
The findings challenge the widely followed practice of encouraging patients to seek out the most experienced hospitals for their medical care when no direct information is available about a hospital’s track record in treating patients with a similar condition.
"Our study shows that direct quality indicators, such as patient death rates, are much better at predicting quality than indirect barometers such as volume of patients," said Jeanette Rogowski, lead author of the study and a senior economist at RAND. "These findings have broad implications for the health system as we try to improve medical quality."
Studying patients treated at 332 hospitals nationwide over a five-year period, researchers found that hospitals that treated the largest number of very low birthweight infants were not always the hospitals that had the lowest rate of infant deaths.
"We found there is no factor better in predicting which hospitals would have low mortality rates for these small babies than their actual experience in the preceding years," Rogowski said. "The number of cases treated by an individual hospital was not a very good predictor of which hospitals would have the best results."
The findings are published in the Jan. 14 edition of the Journal of the American Medical Association. In addition to researchers from RAND Health, other authors of the report are from the University of Vermont College of Medicine, Dartmouth College, the National Bureau of Economic Research and the Vermont Oxford Network. Support for the study was provided by the U.S. Agency for Healthcare Research and Quality.
Infants who are under 3 pounds at birth—typically premature infants—require prolonged care in neonatal intensive care units that are found in many hospitals across the nation. Previous studies have shown that death rates differ significantly among the very low birthweight infants treated in the nation’s neonatal intensive care units, even when researchers consider the severity of patients’ medical problems.
Insurance companies and others have been interested in identifying quality indicators that can be used to direct patients with high-risk medical conditions to higher quality hospitals. Given the paucity of reliable, risk-adjusted outcome information for most clinical procedures and medical conditions, groups have looked for measurements that can act as a barometer of quality.
Some groups have begun to substitute measurements of hospitals' patient volume on certain illnesses or procedures as a proxy for outcome information, based on research that has linked higher patient volumes with better end results such as fewer deaths and complications.
For example, the Leapfrog Group, a Business Roundtable-sponsored collaborative of large employers and insurers, recently started an "evidence-based" referral standard for five surgical conditions and two neonatal conditions, including very low birthweight babies.
The group based its standards for very low birthweight infants on the number of patients treated, requiring that infants likely to be born with low birthweights be delivered at hospitals with a regional neonatal intensive care unit with an average patient load of 15 or more per day.
To test the strategy employed by the Leapfrog Group and other health payers, the RAND-led study examined the experiences at hospitals that were a part of the Vermont Oxford Network from 1995 to 2000. The cooperative group collects information from about 40 percent of the nation’s neonatal intensive care units and accounts for about half of all very low birthweight infants.
Researchers analyzed information about the death rate among very low birthweight infants at the hospitals, including information about the severity of each child's condition.
Researchers found that patient volume was associated with better patient outcomes, with busier hospitals generally having lower death rates. But researchers found that past experience was a far better predictor of which hospitals would have lower death rates in the future.
Hospitals that ranked in the lowest 20 percent for very low birthweight infant mortality between 1995 and 1998 also were found to have significantly lower-than-average mortality in 1999-2000. Meanwhile, hospitals in the highest 20 percent for mortality rates from 1995 to 1998 had significantly higher-than-average mortality rates in 1999 to 2000, according to researchers.
Viewed by another measure, the difference in mortality between the best and the worst hospitals was more than five times larger when hospitals were ranked on past mortality rates rather than on past volume. In addition, researchers reported that basing a patient referral system on historical mortality experience would save an estimated 115 lives annually in the Vermont Oxford Network, versus 11 saved lives for a system based upon historical volume standards, a 10-fold difference.
Other authors of the study are Jeffrey Horbar from the Department of Pediatrics at the University of Vermont College of Medicine and the Vermont Oxford Network; Douglas Staiger from Department of Economics at Dartmouth College and the National Bureau of Economic Research; Michael Kenny of the Biometry Division of the University of Vermont; Joseph Carpenter from the Vermont Oxford Network; and Jeffrey Geppert from the National Bureau of Economic Research.
RAND Health is the nation's largest independent health policy research organization, with a broad research portfolio that focuses on health care quality, costs and delivery, among other topics.