MCG Informed Care Strategies May Help Reduce Unexplained Clinical Variation in U.S. Health Care
By William Rifkin, MD, Managing Editor and Physician Relations Specialist at MCG Health
One of the most puzzling conundrums in the U.S. health care system is geographic variation in care. An important aspect of this phenomenon is inpatient utilization rates. Variation in inpatient utilization refers to differences in the percentage of patients presenting to an Emergency Department that are subsequently admitted to inpatient care. Numerous studies have found that this clinical decision fluctuates significantly, even after statistical adjustment for variables such as age, insurance status, and severity of illness.
Geographic region is one of the most studied means by which to categorize clinical variation. In general, patients are admitted more often on the East coast, in big cities and most especially in certain metropolitan regions, such as the New York to D.C. corridor.
Importantly, studies have found that higher intensity of care is not associated with improved clinical outcomes, in fact, in some cases, it is associated with worse outcomes which is not surprising as the risks of inpatient care are well documented.
Looking at specific examples based upon analysis of national data performed by my colleagues at MCG, we can see how this plays out in actual clinical practice by examining variation in highest quintile to lowest quintile admission rates for patients seen in emergency departments with common diagnoses for which inpatient hospital admissions are not automatic, that is clinical judgement is required to determine which patients need inpatient care.
In commercially insured patients, upper quintile to lowest quintile admission rates vary as follows; for renal colic 2.8 fold, for syncope 2.7 fold, and for dizziness 3.6 fold. For Medicare fee-for service insured patients the variation is 2.4 fold for renal colic, 2.8 fold for syncope, and 2.6 fold for dizziness. This sort of upper to lowest quintile comparison identifies many other diagnoses that also have significant variation. Importantly, regions that are on the high-end for one diagnosis tend to also be on the high-end for other diagnoses.
A recent study (Caines, 2016)1 that looked at all-cause admission rates at the county level for Medicare fee-for-service insured patients across the United States confirmed the now familiar patterns of higher and lower utilization. This pattern, generally higher admission rates on the East coast compared to the rest of the country, and in big cities compared to less populated areas, has been a consistent finding for decades. To put into perspective the sheer volume of patients involved, and recalling that in this study all patients had similar insurance coverage, if the highest admission rate quintile counties implemented the consistent use of MCG care guidelines, amongst other changes, such that their admission rate fell to that of the second highest quintile, there would be over 1 million fewer Medicare FFS admissions per year or over 2,800 fewer per day. Considering this magnitude of inpatient admissions, one can recognize the cost and safety implications surrounding this sort of variation.
The identified reasons for this geographic variation are numerous, and include things like perceived norms of practice, physician training, patient expectations, and interestingly, the number of available hospital beds. One means by which to combat this variation would be physician education, for example analysis of this sort of benchmark statistical data in concert with implementation of evidence-based admission criteria (such as those created by MCG). In this way, determination of which patients truly require inpatient care can be standardized, at least to some degree. Reducing this sort of costly and potentially harmful variation is one of the more important challenges facing the U.S. health care system.
1. Caines K, Shoff C, Bott DM, Pines JM. County-level variation in emergency department admission rates among US Medicare beneficiaries. Annals of Emergency Medicine. 2016;68(4):456-60
Photo courtesy Shutterstock/Andrei_R