The Downside of Accepting Critically Ill Patients in Transfer
Abstract & Commentary
Synopsis: Patients who were transferred directly to the authors’ medical ICU from other hospitals were sicker and had worse outcomes than those who were directly admitted. Benchmarking data generated without taking referral source into account erroneously indicated an excessive death rate and other adverse outcomes.
Source: Rosenberg AL, et al. Accepting critically ill transfer patients: Adverse effect on a referral center’s outcome and benchmark measures. Ann Intern Med. 2003;138:882-890.
In this study, Rosenberg and associates at the University of Michigan assessed the incremental improvement that increasingly more detailed clinical and physiologic case-mix adjustment provided in evaluating the severity of illness of patients transferred to the medical ICU after initially receiving care at another hospital. They sought to determine the magnitude of the "transfer effect"—the fact that such patients are known to be sicker and have worse outcomes than patients initially admitted to the receiving hospital—on benchmarking data as would be collected by the Centers for Medicare & Medicaid Services (CMMS, formerly the Health Care Financing Administration).
Rosenberg et al used records from 4579 consecutive MICU admissions from 1994 through 1998. All data were collected prospectively. Case-mix and severity-of-illness scores were generated using the APACHE system and a daily Acute Physiology Score (APS). The outcome measures were MICU and hospital lengths of stay, MICU and hospital mortality rates, and MICU readmissions. Data from patients admitted directly to the MICU from the emergency department or clinics were compared to similar data from patients transferred from inpatient wards ("floor patients") and patients admitted directly from other hospitals ("transfer patients"). Benchmarking data adjusted for case mix and severity of illness, as used by the CMMS, were determined for this MICU patient cohort, adjusting for case mix and severity of illness.
Transfer and floor patients were sicker both at the time of MICU admission and at MICU discharge than direct-admit patients and had 20-30% greater declines in APS during their stay on the unit. Transfer patients were more likely to be admitted with complex medical conditions such as severe sepsis, acute respiratory distress syndrome, and hepatic failure, and both transfer and floor patients had more comorbid conditions than direct-admit patients. Compared with direct-admit patients, transfer patients had 1.5 times longer MICU stays and remained in the hospital almost twice as long; even compared with floor patients, transfer patients had 20% longer MICU stays. Admission source was a strong, independent predictor of hospital mortality.
Using standard benchmarking procedures that adjust only for case mix and severity of illness, and not for the source of admission, Rosenberg et al calculated that, all other factors being equal (such as efficiency and quality of care), a referral hospital with a 25% MICU transfer rate compared to another hospital with a 0% MICU transfer rate would be penalized by 14 excess deaths per 1000 admissions.
Comment by David J. Pierson, MD
An increasing number of health care providers, health plans, third-party payers, and regulatory agencies collect and disseminate information used to judge and compare the quality of care. Each hospital is compared to its neighbors as well as to broader benchmarks on such things as mortality and ICU length of stay, which are used to make inferences about the quality of care. Potential patients, the media, and others naturally assume that a hospital with higher death rates or ICU lengths of stay than other institutions in the area provides service that is not as good. Therefore, the factors used to determine such data must take into account all aspects of patient demographics, specialization, and other factors in order for them to be as reflective as possible of the true quality of care.
The "transfer factor"—the effect of patients transferred directly to the MICU from ICUs at other institutions—results in skewed data that make the referral hospital’s outcomes look worse than they really are (including, in this instance, 1.4% excess deaths) when standard benchmarking procedures are used. Even when the best available risk-prediction measures are used, there is a substantial underestimation of transfer patients’ resource use and outcomes. As pointed out by Rosenberg et al, this underestimation is almost certainly worse when only administrative or DRG information is used for benchmarking.
One conclusion that could be drawn from this study is that accepting transfer patients from other institutions should be avoided, since such patients make the receiving hospital’s outcomes and quality indicators look worse. Such a conclusion would obviously be a mistake. However, the findings of this study emphasize the importance of using the best possible methods for generating data by which health care institutions are compared, both locally and nationally. Institutions refer complex, critically ill patients to hospitals like the one from which this study’s data were generated precisely because such patients can be better managed there. The health care system must find a way to avoid penalizing referral centers for things that arise as a direct result of the role they play in that system.
Dr. Pierson is Professor of Medicine University of Washington Medical Director Respiratory Care Harborview Medical Center, Seattle.