High-Cost Hospitals: Because Patients are Sicker? Think Again.

Posted by

Wendy Lynch, Ian Beren, Justin Shaneman and Nathan Kleinman

Posted 12/30/11 on the HCMS Blog

It’s not surprising news that inpatient healthcare costs vary from hospital to hospital; large differences in price for the same procedure are common. But the reasons for variation are less clear. Some hospitals have consistently more expensive fees for identical treatments. However, these differences do not necessarily reflect better care: a recent study found that some high-cost hospitals rank low in quality scores and some high-quality hospitals are relatively low-cost (1). Plus, evidence shows that spending more does not produce better outcomes, higher satisfaction, or more appropriate care (2, 3).

Some plausible reasons for price differences include higher negotiated rates with health plans, delivery of additional or unnecessary services, poor efficiency or management of hospital stays, or several other possible causes. Yet, the most common assumption most of us make when we see price differences among hospitals is that some hospitals have patients that are simply sicker.

A recent HCMS analysis of hospital use by employees of a large, regional employer refutes that assumption.+   The graphic below shows cost per admission at ten different hospitals in the same geographic region according to the severity of illness burden of the patients the hospital treated. Most hospitals fall along this expected cost trend, with a median cost of $2,300 per increasing unit of illness. However, a few had costs above that expected rate of $15,000 to $20,000 per admission.

Represented in a different way, the next graphic demonstrates the projected cost across these ten hospitals for a person with the same illness burden. As we see, the same patient could expect to have an admission cost of about $8,000 at Hospital A versus $36,000 at Hospital J – over four times higher—after receiving the exact same services. Remember, these are real data from ten hospitals most commonly used (50 to 600 total admissions each during one year) by employees of a large, regional employer. The employer now knows that hospitals A, B, and C have the highest level of efficiency, while hospitals H, I, and J have the lowest. Comparing hospital A and J (the lowest cost and highest cost), each were rated highly on four of seven quality measures reported in national quality ratings (4).

The implications for both health plan sponsors and consumers are significant. By placing these cases on an equivalent scale, it is possible to identify which hospitals are delivering cost-efficient care, independent of severity of illness. By removing the argument that patients admitted at expensive hospitals are “sicker,” employers can focus on other causes of variation.

This is important not just because there is a chance to avoid unnecessary cost, but because patients have a better chance of avoiding risk, injury, pain, and suffering. For health plan sponsors, the potential for saving costs on inpatient care appears dramatic, which creates opportunity to contract preferentially with more efficient facilities and encourage consumers to choose equally high-quality, but lower-cost hospitals. An increasing number of employers are providing information about price, cost, and safety to their employees.  Beyond basic consumer-directed strategies, some companies now share financial savings with employees who choose high-quality, lower-cost care; awarding 50% of the difference in cost is one example (5).

When preparing for medical procedures or planning how best to respond to serious events, consumers should be aware of both medical quality and cost. In this population, none of the hospitals scored highly on all quality indicators; few do. However, most would be surprised to learn that some of the lower-cost, less-recognized hospitals in the area have equal if not higher quality ratings than the more recognized, highly-regarded establishments. The combination of timely, useable information and financial incentives may be the just the combination employers and employees need to rein-in costs and improve care.

For those interested in more technical aspects of this analysis, see below.

Measuring level of illness

Analyses such as these always bring up questions about how we know if some patient populations are sicker than others. It’s not a straightforward process when you are looking at data rather than examining a patient. Over the past several years, the HCMS Data Analytics Team developed a comprehensive Health and Utilization Index (HUI) based on the experiences of 3 million people in their Research Reference Database (RRDb).

The HUI score is an indicator of burden of illnesses, absence and injury relative to an average population, normalized to an average score of one. A person or a population , with an average score of 2.0 would have double the average burden of illness, medications, absence and injury incidence. The index measures illness level (both number and severity) on actual claims and diagnoses data, weighted according to the expected contribution to total medical and benefits costs.

HUI is based on the expected contribution of every illness and type of medication to the total cost burden of a population. Each illness and medication class contributes an expected, specific amount to the total score, according to the population average. HUI also assesses the expected level of benefit utilization across the population, such as disability and workers’ compensation. A person’s HUI score is attributable to the sum of an individual’s illness, medication, and benefits utilization. A person with more illnesses, more medications, and more disability payments will accumulate a higher score—unrelated to his or her actual amount paid for treatments. Additionally, a person with a more severe illness will have a higher score than someone with the same number of less serious illnesses. For example, uncomplicated high blood pressure contributes 0.07 to HUI, while lung cancer contributes over 4.5. Based on the overall database population, the added HUI equivalent for uncomplicated high blood pressure would add an expected amount of $208, while the value of lung cancer would be an additional $14,500.

Because each sub-score is based on expected population averages, the totals provide a clean comparison of similar levels of illness burden. Also, by linking illness level to overall cost in the past (over 3 million people), we would expect similar costs for similar levels of HUI. Hospitals seeing patients with higher HUI scores should have costs proportionately higher than hospitals seeing patients with lower HUI scores. As we saw in the first graphic, most hospitals show a consistent level of cost increase per each unit of HUI. However, as we also saw, there are significant variations from that expected trend as well.

References

1.Jha AK, Orav EJ, Epstein AM: Low-quality, high-cost hospitals, mainly in South, care for sharply higher shares of elderly black, Hispanic, and medicaid patients.  Health Aff (Millwood) 2011;30:1904-11.21976334

2. Fisher, E. S.; Wennberg, D. E.; Stukel, T. A.; Gottlieb, D. J.; Lucas, F. L., and Pinder, E. L. (Center for the Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, New Hampshire 03755, USA. elliott.s.fisher@dartmouth.edu). The implications of regional variations in medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med. 2003 Feb 18; 138(4):273-87.

3.Fisher, E. S.; Wennberg, D. E.; Stukel, T. A.; Gottlieb, D. J.; Lucas, F. L., and Pinder, E. L. (Center for Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, New Hampshire 03755, USA. elliott.s.fisher@dartmouth.edu). The implications of regional variations in medicare spending. Part 2: health outcomes and satisfaction with care. Ann Intern Med. 2003 Feb 18; 138(4):288-98.

4.The Leapfrog Group. 2011;(accessed Dec 20, 2011).

5.Torinus JJr: The Company That Solved Health Care: How Serigraph Dramatically Reduced Skyrocketing Costs While Providing Better Care, and How Every Company Can Do the Same, Dallas, TX: BenBella Books; 2010.


+ The technical details of our analysis are below in the “Measuring level of illness” section.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s