A hospital's low 30-day mortality rates for three key diagnoses might be an indicator of good survival rates for patients who receive hospital treatment for other conditions as well, researchers say.
How can you tell a good hospital from a bad hospital? Aside from anecdotal information about bad coffee and surly nurses, we now have actual data to help make the distinction.
Roughly three months from now starting Oct. 1, the second wave of value-based incentive payments—or penalties as some hospital providers prefer to call them—kick in.
And for the first time, 25% of the score affecting 1.25% of a hospital's Medicare payments will depend on how well those providers kept certain patients alive for at least 30 days after discharge.
The risk-adjusted measure, however, only counts those patients admitted for heart failure, pneumonia, or heart attack, which represent only 13% of all senior hospitalizations and which are the only ones currently reported on Hospital Compare.
Under authority of the Patient Protection and Affordable Care Act, federal regulators picked these diagnoses because researchers believe that better care practices keyed to those conditions, can keep those patients alive longer.
It wasn't known, at least until this week, that a hospital's low mortality rates in these three diagnoses might translate to good survival rates for patients treated at that hospital for other conditions as well.
Now, however, as far as keeping patients alive, it appears from a research paper in JAMA Internal Medicine that there are very good hospitals and those that are, well, bad hospitals, or places from which if you're discharged, you're less likely to survive 30 days, no matter what condition sent you to the hospital to begin with.
The Brigham and Women's Hospital researchers, who include big name researchers such as surgeon and New Yorker contributor Atul Gawande, MD; Karen Joynt, MD; and Ashish Jha, MD and others, found that hospitals with low 30-day mortality rates among patients with those three conditions tend to have low 30-day mortality rates among patients with nine other common medical conditions, and a composite of 10 surgical conditions, too.
I asked Joynt in an interview this week to confirm this "good hospital/bad hospital" assessment of the article's bottom line.
"Yes," she replied. "We found that the publicly reported measures are actually a pretty good surrogate for what's going on overall [in any hospital]. It's not perfect. But if you go to a hospital that does well in acute myocardial infarction, heart failure, and pneumonia, they're much more likely to do well overall."
Indeed, the researchers said so right near the end of the article: "The broader findings from our work support the notion that there may be actually be 'good' and 'bad' hospitals and that performance on a manageable set of key indicators can help identify such institutions."
So, what's going on at these good hospitals, I asked.
Joynt explains that the finding "seems to have something to do with leadership and culture in a hospital, beyond [the performance] of any individual physician or service line."
She adds that unlike the three main conditions now publicly reported, the other conditions they studied don't come with a high chance of post-discharge death.
"If you think about the people we're including in this analysis—for example surgical patients and medical patients—they are coming in for things where death rates are not that high, 1% or 2%, like arrhythmia or chest pain, or some of the gastrointestinal disorders," Joynt says. "The majority of patients that you bring in to do an elective surgical procedure shouldn't be ones that you would expect would end badly."
Hospitals that, do well on 30-day mortality for one of the three conditions, say pneumonia, are likely to do well in the other two. This much was already known.
See Also: Best and Worst Hospitals for 30-Day Mortality
But I wondered if that might not be simply because hospitals have been aware for the last three years that those conditions were ripe to be targeted for measurement.
What we didn't know, until now, is that performance is largely extendable.
"Top-performing hospitals on the publicly reported conditions had more than a five-fold higher odds of being in the best quartile of overall risk-adjusted hospital mortality compared with other hospitals," the report says. "Odds ratios remained statistically significant when we considered medical and surgical mortality separately."
The researchers' data was encompassing. It includes about 6.7 million admissions among Medicare beneficiaries at 2,322 hospitals in 2008 and 2009.
They explored 30-day mortality rates for nine conditions: stroke, arrhythmia, chronic obstructive pulmonary disease, respiratory tract infection, sepsis, urinary tract infection, gastrointestinal bleed, renal failure and esophagitis/gastroenteritis.
For a composite of 30-day mortality after surgical procedures, they chose procedures that were performed at least 10,000 times per year and which had a median mortality of more than 2%. Those included coronary artery bypass grafting, aortic valve repair, above-knee amputation, colon resection, small bowel resection, exploratory laparotomy, and pulmonary lobe resection, abdominal aortic aneurysm repair, esophagectomy, and pancreatectomy.
There was more from their conclusions.
"Top performing hospitals on the publicly reported conditions had 81% lower odds of being in the worst quartile of overall mortality." And these patterns also "persisted for the medical and surgical composites."
The correlations between low mortality hospitals versus high mortality ones were even more stark than those traditionally used, such as hospital size and teaching status.
So what is really going on? Are these hospitals' admitting physicians being secretively selective about which patients to admit for care, stealthily rejecting [moving to hospice care perhaps] those who look like they're at death's door?
Or are these low-mortality hospitals perhaps fortunate to have a mix of healthier patients? That's unlikely, because these researchers screened for that by adjusting for multiple comorbidities.
It's important and fair to point out that the hospital industry has repeatedly voiced strong objections to being scored on whether patients die, saying that 30-day mortality is not a valid measure of hospital quality.
"We continue to believe these measures do not have adequate reliability and should be removed from the program altogether," AHA executive vice president Rick Pollack wrote June 20 in a 58-page letter to the Centers for Medicare & Medicaid Services. The letter is in response to a request for comments to the proposed FY 2014 Inpatient Prospective Payment System rule.
My sense is that at this point, CMS will not take the hospital industry's side, especially with the pressure from many stakeholders to move faster toward measuring outcomes instead of processes.
So now the task is to figure out the secret sauce of these low mortality hospitals..
Perhaps when more consumers realize that these are important metrics, they and the payers who back them will raise the issue, asking questions of those hospitals whose 30-day mortality scores show "worse than national average."
It's too late for hospitals to affect the mortality component of the scores affecting those October 1 payments. It's also too late to affect payments starting Oct. 1, 2015; That performance period ends in three days on June 30.
But it might not be too late to change the score for FY 2016. That 21-month performance period doesn't end for another year, June 30, 2014.