Health Policy$ense

The Value of Teaching Hospitals

Two papers analyze costs and outcomes

One benefit of teaching hospitals is obvious: they train the next generation of clinicians. But that training comes at a cost: teaching hospitals tend to have higher-priced clinical care than other hospitals. In the era of value-based care, the question is whether the additional spending produces better patient outcomes, and if so, whether the clinical benefit is worth the cost.

We have some answers in two recent papers (here, and here) by a team of LDI Senior Fellows led by Jeffrey Silber and Lee Fleisher. In general, they find that the value of teaching hospitals is greatest for high-risk surgical and medical patients, where better outcomes justify slightly higher costs. Let’s take a look at what they found.

First large national study
To get an apple to apples comparison, the authors tracked costs and outcomes for clinically and demographically similar Medicare patients, utilizing a novel matching approach, who were admitted to 339 major teaching hospitals and 2,439 non-teaching hospitals. The first paper looked at patients admitted for general, orthopedic, or vascular surgery. The second paper focused on patients admitted for one of three common medical conditions: acute myocardial infarction (heart attack), congestive heart failure, and pneumonia.

The “costs” were total hospital utilization costs (rather than negotiated rates or charges), plus emergency department, outpatient, and physician office visits within 30 days of admission. Outcomes included 30-day mortality, ICU utilization, length of stay, and failure-to-rescue rates. Importantly, procedures performed only at teaching hospitals were not included in the analysis because there would be no non-teaching cases to compare.   

For general surgery patients, teaching hospitals had a 0.95% lower mortality rate compared to non-teaching hospitals (4.6% vs 5.6%), at an added cost of $915 ($29,071 vs. $28,113). This yielded an overall value estimate of $965 for a 1% reduction in mortality. Similarly, for vascular surgery patients, teaching hospitals had a 0.39% lower mortality rate at an added cost of $1,392, yielding a value estimate of $3,567 for a 1% drop in mortality. For both surgeries, teaching hospitals tended to have longer lengths of stay, less ICU use, and lower failure-to-rescue rates.  

Interestingly, for orthopedic procedures performed at both teaching and non-teaching hospitals, there were no differences in mortality outcomes, although the overall mortality rates were low. Teaching hospitals also had an added cost of just over $1,200, longer lengths of stay, and higher ICU use. Failure-to-rescue rates remained lower.

Pooling results from all three medical conditions (heart attacks, heart failure, and pneumonia), teaching hospitals had an 1.3% improvement in mortality rates (10.7% versus 12%), at an added cost of $273, yielding a value estimate of $211 for each 1% reduction in mortality. Teaching hospitals tended to have longer lengths of stay, but less ICU use.  

Higher risk patients benefit
But there’s more to the story. The value of teaching hospitals is concentrated among patients at different levels of mortality risk upon admission, with most benefits accruing for high risk patients. For medical conditions, teaching hospitals offered no observable survival benefit for the least risky patients, and resource costs were comparable. As a result, there were no value benefits. In contrast, the highest risk patients had a 3% reduction in mortality at an added resource cost of $1,289—yielding an added value of $427 per 1% decrease in mortality.

Value estimates for vascular surgery were similar to medical conditions: patients in the four lowest quintiles of risk had no survival benefit at teaching hospitals, despite marginally higher cost in most cases. However, the riskiest patients had mortality benefits of 0.9%, at an added cost of $3,652—yielding a value estimate of $4,052 for an additional 1% mortality reduction.

Where to go from here
The higher cost of teaching hospitals is a well-known feature of the health care system. Teaching hospitals invest more in expensive technologies and place an emphasis on workforce development alongside clinical care, which incurs costs which likely go beyond the roughly $16 billion federal and state governments spend supporting graduate medical education. While previous work has examined the additional clinical cost of teaching hospitals and compared quality at teaching and non-teaching hospitals, this is among the first studies designed to quantify the added clinical value of the teaching hospitals—and the first to do so by matching and tracking patients based on clinical characteristics.

From a system perspective, these results suggest more research should be done to guide Medicare patient referral patterns. Complex heart attack, heart failure, pneumonia, and general and vascular surgery patients likely receive high value, cost-effective care at teaching hospitals, where they have better outcomes at a marginally increased cost. Assuming that teaching and non-teaching hospitals can adapt to volume shifts, it may be beneficial to steer some high-risk patients to higher-cost academic medical centers, while guiding lower risk patients to non-teaching hospitals, which achieve comparable outcomes at the same or lower costs.