WASHINGTON, DC — Do-not-resuscitate (DNR) orders are common in patients hospitalized with heart failure—how common varies widely among hospitals—and they have a big influence on centers’ risk-adjusted rate of in-hospital HF mortality, according to an analysis encompassing thousands of patients across California[1].
The findings are based on DNR orders initiated up to 24 hours after admission. Should they also apply to 30-day mortality in hospitalized HF patients, they would likely affect hospitals’ rankings based on that metric, in particular those used by the US Centers for Medicare and Medicaid Services (CMS), Dr Jeffrey Bruckel (University of Rochester Medical Center, NY) told theheart.org | Medscape Cardiology.
The CMS-specified risk-adjusted 30-day mortality metric for heart failure, on which Medicare hospital reimbursement is partly based, does not account for DNR or hospice status. The model assumes, wrongly it seems, that DNR status “would be completely accounted for by patient comorbidities,” he said.
Bruckel is lead author on the analysis of in-hospital mortality by DNR status, based on 55,865 admissions in patients with a primary diagnosis of heart failure at 290 hospitals, which was published September 25, 2017 in JACC: Heart Failure.
Their data came from the 2011 California State Inpatient Database that, perhaps unusually, included whether the patients hospitalized with heart failure had an “early DNR” order.
“There’s still significant unmeasured confounding in hospital mortality measures, and if we’re going to continue using these kinds of measures to judge hospital quality and hospital reimbursement, then we need to be more mindful about how those measures are designed,” Bruckel said.
An editorial accompanying the current analysis explains the 30-day metric’s limitations: “A quality measure is only as good as its signal-to-noise ratio[2]. In the case of 30-day mortality for heart failure, the signal (variation in mortality due to quality differences) is small compared with the noise (variation in mortality due to everything else),” writes Dr Paul Heidenreich (VA Palo Alto Health Care System, CA).
“Bruckel et al have demonstrated that one way to reduce this noise in hospital mortality rates is to consider do-not-resuscitate status.”
Beyond in-hospital mortality, a single-center study recently identified a number of significant relevant predictors not included in the adjustments of the 30-day metrics themselves[3].
One of those predictors of the center’s “higher-than-anticipated” 30-day mortality for heart-failure hospitalizations was DNR status. Of the patients who died within 30 days of discharge, 85% had had DNR orders, reported Dr Robert T Faillace and colleagues September 20, 2017 in the American Journal of Medicine.
“Use of mortality as a complementary measure would be reasonable if paired with evidence-based process measures that have been shown to improve mortality through randomized controlled trials,” Heidenreich writes in his editorial.
“We clearly need to adopt better measures of the quality of heart-failure care. Until then, adjusting mortality for DNR status would be a step in the right direction.”
What Drives DNR Variance Among Hospitals?
In the current study, about 12% of HF hospitalizations were associated with DNR orders put in place up to 24 hours after admission, with wide variance among centers that was not explained by differences in patient mix, the report notes.
It’s tough to discern exactly what is driving that wide variance, according to Bruckel. Because in most cases the DNR status is active on admission, “in theory it should reflect the chronic illness, not necessarily the acute reason for the patient’s hospitalization,” he said. So some of the variance reflects chronic HF severity and comorbidities.
But, he adds, “some of it is probably due to patient demographics, religion, and social and cultural things, and hospital documentation of DNR status is going to vary.”
Not as much attention is paid to limitations of the CMS 30-day mortality metric as to its 30-day-readmission measure of quality of care, according to Dr Gregg C Fonarow (Ronald Reagan-University of California, Los Angeles Medical Center).
The concern, he said in an interview, is that the metrics are “not related to the quality of care that’s being provided.”
After they were introduced as mediators of hospital reimbursement, centers “encouraged physicians to adequately capture code for those specific variables being used in the risk-adjustment model.”
So the metrics “created a financial incentive toward trying to do better in a way that is on the metric rather than in a patient-centered way,” while giving short shrift to process measures, Fonarow said.
“If you have hospitals that are being inappropriately penalized for higher mortality rates because they happen to have a higher proportion of do-not-resuscitate patients, that provides hospitals an adverse incentive to do things that are against patients’ wishes or are not patient-centered,” Bruckel agreed.
Indeed, no love is lost among cardiologists for the CMS application of the 30-day mortality and readmission metrics, but they may have some other value, according to Bruckel.
“I think they’re more useful to track improvement in a particular hospital’s performance than they are for comparing between hospitals, even though that’s how they’re inevitably used.”
In the current analysis, patients with early DNR orders had a 9.9% in-hospital mortality, compared with 2.1% for non-DNR patients, for an adjusted relative risk with a DNR of 3.63 (95% CI 3.17–4.16).
In comorbidity-based covariate analyses that excluded DNR status, hospitals’ risk-adjusted DNR rates correlated with in-hospital mortality (P<0.001). Adding DNR status to the risk adjustment model led to a 17% increase in the model’s predictive power and reclassified the “outlier” status of 9.3% of the hospitals in the analysis. That is, they no longer had a significantly higher- or lower-than-predicted in-hospital mortality.
“The hospital to which a patient was admitted was more strongly associated with receipt of a DNR order than the strongest clinical predictors in the model,” which were advanced age, dementia, and metastatic cancer or leukemia (P<0.001 for each).
“Disastrous Consequences”
Should the 30-day metrics be abolished or, rather, improved on as guides for CMS? “There may be some steps that could be taken to try to improve upon the measures, but the moment that step is taken, you know it’s going to be immediately gamed,” Fonarow said.
For example, he said, “Once you take the higher-risk patients that you’re kind of worried about out of the denominator, you might be highly encouraged to make them, at least temporarily in the first 24 hours, do not resuscitate.”
Indeed, Fonarow said, those metrics intended to rein in costs and improve care processes “are not actually serving their intended purpose and thus are having what ultimately now appears to be disastrous consequences.”
He pointed to a recent report[4] based on Medicare fee-for-service patients suggesting that 30-day mortality after hospitalization for heart failure (and also for acute MI, unstable angina, and stroke) declined gradually from 1999 to 2001, before the metrics were used to “shame and penalize” hospitals.
Since then, there has emerged “a shocking increase in mortality,” he said, specifically a 1.3% absolute increase and 16% relative increase in 30-day postdischarge mortality[5].
Bruckel reports serving as an advisor for AvantGarde Health; disclosures for the coauthors are listed in the paper. Heidenreich reports that he has no relevant financial relationships. Fonarow has recently disclosed “significant consulting for Novartis and modest consulting for Amgen, Janssen Pharmaceuticals, Medtronic, and St Jude Medical (now Abbott).” Neither Faillace nor his coauthors had relevant financial relationships.
Follow Steve Stiles on Twitter: @SteveStiles2. For more from theheart.org | Medscape Cardiology, follow us on Twitter and Facebook.
Tidak ada komentar:
Posting Komentar