Skip to Main Content

Seattle Children’s Hospital once sued after being left out of insurance plans. Massachusetts officials have touted the value of care provided at the state’s community hospitals. In Houston, some health plans exclude the city’s renowned research hospitals.

At a time when insurers are steering patients away from expensive academic medical centers, a new study counters the idea that the quality of care is consistent across hospitals, concluding that major teaching hospitals have lower mortality rates for older patients than community hospitals.


Using millions of Medicare records, researchers found that the 30-day mortality rate — the percentage of patients who died within 30 days of hospitalization and one common way to gauge quality — was 8.3 percent at major teaching hospitals, compared with 9.2 percent at minor teaching hospitals and 9.5 percent at non-teaching hospitals. The figures accounted for differences in patient populations and hospital characteristics.

That 1.2 percentage point spread translates into one fewer patient dying for every 83 patients the hospitals see, said Dr. Laura Burke, one of the study authors.

The researchers discovered similar differences when they narrowed the data down to examine specific conditions and surgical procedures.


“Obviously teaching hospitals are largely considered to be more expensive,” said Burke, an instructor at the Harvard T.H. Chan School of Public Health and an emergency medicine physician at Beth Israel Deaconess Medical Center in Boston. “But when thinking about cost, taking into account outcomes needs to be part of the discussion. If you are completely excluding teaching hospitals, that could have some implications for quality.”

The study was published Tuesday in the Journal of the American Medical Association.

It relied on data from 21.4 million hospitalizations among Medicare patients, and the researchers acknowledged it’s not clear if the pattern they found would hold for younger patients. James Robinson, a health economist at the University of California, Berkeley, who was not involved with the study, noted that mortality might not be as good a quality measure for younger patients, who are more likely to go to the hospital to deliver a baby or have their gall bladder removed instead of after a stroke. They are less likely to die overall, Robinson said, so death rates are not as relevant to them.

Still, he and other outside researchers said the analysis was well-done and would add to the discussion about how the nation’s health care system should balance the cost and quality of care.

If large academic medical centers do offer better quality, Robinson said, then how much more should they be able to charge? Does a 1.2 percentage point difference in mortality rates mean they can charge double? He said that more expensive hospitals should have to justify their prices.

Researchers have found that, in some cases, the most expensive hospital in an area will charge up to 60 percent more than the cheapest hospital for an inpatient stay.

“If the Four Seasons charges more than the Hilton, they have to show that they’re better,” Robinson said. “I’m a little bit concerned that this study could be misinterpreted as, ‘Well, quit telling us to be more efficient.’”

Burke said she and her colleagues pursued the research because of the narrative, sometimes promoted by insurers and policymakers, that teaching hospitals were charging more than non-teaching hospitals but not necessarily producing better results. Studies from two decades ago found that academic medical centers were providing better care, and more recent studies focused only on individual procedures or conditions. No one had looked at the system holistically in recent years, and large teaching hospitals were being penalized more by Medicare for their higher readmission rates — a potential sign of poorer quality.

In addition to finding lower mortality rates at academic medical centers at 30 days, the researchers discovered similar results after seven and 90 days, which Burke said was surprising.

In an email, a spokeswoman for America’s Health Insurance Plans said that “health plans focus on building innovative, high quality, and cost-effective networks that best serve the varying needs of their customers, including academic and non-academic hospitals.” She said AHIP wouldn’t comment on the study specifically, but pointed to another report that found no difference in quality between larger and narrower networks.

For the new study, researchers also compared mortality rates for 15 common conditions and six surgical procedures. Major teaching hospitals had lower mortality rates than non-teaching hospitals for 13 of the conditions, including congestive heart failure, hip fracture, and respiratory disease; for sepsis and stroke, the results were comparable. Among the surgeries, the large academic hospitals had lower mortality rates for open abdominal aortic aneurysm and colectomy.

The lesson, Burke said, was not that everyone should be heading to academic medical centers for all their care; in fact, non-teaching hospitals made up almost three-fourths of the hospitals included in the research. Instead, she wondered whether there were lessons from the teaching hospitals that could be replicated at all hospitals.

This paper did not explore possible reasons for the differences in mortality, but Burke said she and colleagues are investigating that now.

The new research could complicate the debate about the narrow networks of doctors and hospitals insurers increasingly offer to patients in hopes of controlling costs. So far, researchers have found they have not hurt the quality of care.

Simon Haeder, a political scientist at West Virginia University who was not involved with the new paper, concluded as much when he and colleagues studied the impact of narrow network plans offered through the Affordable Care Act in California. But he said he thought the findings from Burke and her team likely extended beyond the Medicare population to all patients.

So how does he square the different findings of the papers?

Haeder noted that the researchers were using different measures of quality, and that his research was only focused on California plans. Regardless, he said the new study should be considered as insurers create networks for their patients.

“The real hard tradeoffs come in when we start thinking about, what do we do when a provider is high-quality and high-cost?” he said. “For an individual, they don’t care what it cost is — they want the best care and the best results. But at a societal level, we need to start thinking about whether things are worth paying for, and that’s a really challenging discussion to have.”