ANN ARBOR, Mich. -- When an older person gets hospitalized for pneumonia, where's the best place to care for them? New research findings about deaths and health care costs in such patients fly in the face of conventional wisdom - and could change where doctors decide to treat them.
Seniors with this common lung infection, the researchers show, had a better chance of surviving if they went to an intensive care unit rather than a general hospital bed.
And despite the ICU's reputation as a high-cost place to care for patients, the costs to Medicare and hospitals were the same for both groups.
The research, published in the new issue of JAMA by a University of Michigan Medical School team, focuses on those patients on the "bubble" - those who doctors could send to either an ICU bed or a general bed depending on their judgment.
Since pneumonia sends hundreds of thousands of seniors to the hospital each year, costing taxpayers billions of dollars, even a small difference in mortality risk and cost for some of those patients could make a big difference. The researchers looked at data from 1.1 million hospital stays at 2,988 hospitals between 2010 and 2012.
"With several recent studies suggesting that too many people are going to the ICU when their risk of death is low, we were surprised that there was a benefit to ICU admission for these patients," says Colin Cooke, M.D., M.Sc., M.S., the study's senior author and an intensive care specialist and health care researcher at U-M. "Now, our challenge is to do further work to determine just which patients will get the greatest benefit, and to determine what about ICU care makes a difference."
Surprises in the data
Cooke and his colleagues, including first author and pulmonary and critical care fellow Thomas Valley, M.D., used Medicare data to see how many hospitalized patients with pneumonia survived and what their care cost.
At first, their results suggested the expected: Patients with an ICU stay were more likely to die, and their care cost more than a general bed stay, even taking into account differences between patients' backgrounds and underlying conditions.
But then they used statistical techniques to focus in on pneumonia hospital stays where the choice of bed type appeared to be truly "discretionary" - on the borderline of needing intensive care, and up to a doctor's judgment. In all, about 13 percent of the patients were placed in an ICU bed only because they lived closest to a hospital that happened to place a high percentage of its pneumonia patients in ICU beds.
It was among these patients that the researchers found a nearly 6 percent improvement in survival associated with ICU admission for pneumonia. In all, 14.8 percent of those who went to an ICU died within 30 days, compared with 20.5 percent of those placed in a general bed.
The cost of caring for these patients was about the same no matter which kind of bed they were in. Medicare paid hospitals an average of about $9,900 for the ICU patients and $11,200 for the non-ICU patients. Hospitals routinely accept less from Medicare than what it actually costs to care for patients with that form of insurance - but even these costs were about the same, $14,100 for ICU patients and $11,300 for non-ICU.
For older pneumonia patients, and their loved ones, the results suggest that asking the medical team about the possibility of escalating to ICU-level care is completely acceptable, and may remind clinicians of the potential benefit of ICU care for patients with pneumonia, Cooke notes.
And for policymakers and hospital administrators, the new findings reiterate that instead of focusing on whether America needs more or fewer ICU beds, additional research should be done on how best to use the ones we have, perhaps by better identifying which patients most need an ICU bed.
More research needed
The findings don't apply to patients who clearly need an ICU - those who can't breathe on their own and need mechanical ventilation, for instance - nor to those who have low risk of developing complications from pneumonia in the hospital.
"It's very clear that there are some pneumonia patients who absolutely need to be in an ICU, and some who may not even need a hospital stay at all," says Cooke, an assistant professor of pulmonary and critical care medicine. "We need to understand more about whether ICUs are being overused, or perhaps underused, for patients in the middle of the spectrum of severity, where physicians could reasonably disagree on whether they should be in an ICU or a general bed."
This is especially important given the risks that ICU care can present -- including drug-resistant infections and the potential harms associated with invasive monitoring and procedures that intensive care teams are more likely to perform.
The authors point out that their findings need to be tested with a randomized controlled trial, which their statistical techniques try to replicate but can't match. And, they caution that they don't have data on the bills that individual doctors sent Medicare, nor the costs that occurred after a patient was discharged, just the hospital side of payment.
The team is evaluating if the ICU is beneficial for other conditions, including chronic obstructive pulmonary disease, congestive heart failure and heart attack. They hope to do more to determine what characteristics made pneumonia patients most likely to do well after an ICU stay, and what factors make hospitals more or less likely to put "discretionary" pneumonia patients in an ICU bed.
"It's very rare in medicine that we find something that saves lives and doesn't cost more," says Cooke. "But perhaps this is one of them."
In addition to Cooke and Valley, the paper's authors include fellow Medical School faculty Michael Sjoding, M.D., and Theodore Iwashyna, M.D., Ph.D., and Andrew Ryan, Ph.D. of the U-M School of Public Health. Cooke and several of his co-authors are members of the U-M Institute for Healthcare Policy and Innovation, the Michigan Center for Integrative Research in Critical Care and the Center for Health Outcomes and Policy. The research was funded by the National Institutes of Health (HL007749), the Agency for Healthcare Research and Quality (HS020672) and the VA Health Services Research and Development Service.
Reference: JAMA Sept. 22/29, DOI 10.1001/jama.2015.11068 An editorial is available.