Today’s Managing Health Care Costs Number is 3:1
The New York Times reported this weekend that “Go to the wrong hospital and you’re three times more likely to die.” Scary! The Plos One research this was drawn from had a more sedate headline: “Quantifying Geographic Variation in Health Care Outcomes in The United States before and after Risk Adjustment.”
We know that health care quality is variable, and we know that the correlation between quality and cost is, uh, not especially robust. This research drew on data from 22 million hospital admissions – both Medicare and younger privately insured patients. –and used 16 different data sources to assess mortality from multiple different disease processes.
Researchers assessed mortalityfor 6 conditions (heart attack, heart failure, stroke, gastrointestinal bleeding, hip fracture and pneumonia) – and counted avoidable complications (pressure ulcers, punctured lungs, central line infections, postop hip fractures and blood clots and infections and wound failure rates, and lacerations) and preventable hospital admissions (diabetes, lung disease, heart failure, dehydration, pneumonia, urinary infection, asthma and diabetic amputation).
The researchers show the hospital and county level of performance (10th % and 90th%ile) – and the differences are startling. Here is the data for heart attack:
Unadjusted hospital level mortality is four times higher at the hospital level at the worst hospital compared to the best – although this difference is diminished when the data is adjusted for demographics (pop) and comorbidities and hospital case volume (system).
But we all would agree that a 2.3 times higher mortality for a heart attack is pretty terrible – so the Times headline is justified –and we should be worried and take action.
But taking action on this data will not be easy.
- The researchers were not allowed to reveal the names of different hospitals – so no one can use this database to figure out what hospital to avoid when you’re getting a hip replacement, or having a heart attack.
- The data is from 2011- and performance at many hospitals has likely changed in the last 6 years.
- There is VERY little correlation in results across these different diagnoses (see the grid below). Hospitals with better outcomes for heart attacks were no more likely than chance to be better at mortality for congestive heart failure –although hospital performance in the mortality metrics other than heart attacks were weakly correlated. There was pretty much no correlation between mortality and safety or prevention indicators – which makes it hard to know what the best performing hospitals are doing compared to poorly performing hospitals.
- This is a single year of data –and we’d need to see that these results were stable over time – and not just the result of randomness.
- We simply can’t send 100% of patients to the top ten percentile of hospitals – there wouldn’t be enough room, and their performance would likely fall.
- Wide geographic variation means that for many Americans, there are no “best” hospitals to seek out without hopping on a plane – hard to do when you are having an acute heart attack!
This research is an important wake up call. We need more such longitudinal research. We should provide incentives for patients to get their care at high value facilities, and we should provide social cues (and incentives) for poorly performing hospitals to do better.
We also need to move to making data from these efforts publicly available. It can be built into transparency and navigation tools – and some patients will move to higher value facilities. Most importantly, the facilities that have higher mortality or complication rates or are less effective at prevention would be shamed into improving their performance.
All graphics or data from Plos One article . This graphic shows that outside of the IQIs (excepting heart attack) there is little correlation of hospital performance among different measures. So finding a hospital with a good score on pneumonia care (IQI 20) does not mean that hospital is likely to be good at preventing pneumonia admissions (PQI11), or have a low rate of accidental lung collapse (PSI06).
EmoticonEmoticon