Higher mortality due to DNR rates results in classification of some hospitals as performance outliers, researchers found. Some ethical concerns include the following:

• hospital reimbursement and reputation can be adversely affected due to higher DNR rates;

• not all deaths can be prevented with high-quality care;

• important metrics such as functional status and symptom control are difficult to measure.

Higher “do not resuscitate” (DNR) rates resulted in some hospitals being classified as performance “outliers,” found a recent study.1

“[The Centers for Medicare & Medicaid Services] and a lot of other organizations spend a lot of time trying to develop outcome measures to help hospitals deliver higher-quality care, mostly based on administrative claims data,” notes Jeffrey T. Bruckel, MD, MPH, the study’s lead author and a cardiovascular fellow at University of Rochester (NY) Medical Center.

The researchers used the California State Inpatient Database because it captures the “early DNR variable.” “We decided to see if it varied between hospitals, and if it would impact hospital outlier status for heart failure,” says Bruckel.

Researchers found that DNR status did affect hospital risk-adjusted heart failure mortality metrics. “I don’t think it surprised any of us that patients at different hospitals had different DNR rates,” says Bruckel. Patients with DNR orders have increased in-hospital mortality, which can skew the hospital’s quality metrics. “If you don’t account for it in your risk models, then you might classify hospitals as outliers which might not be,” says Bruckel. Some key findings include the following:

• among 55,865 patients from 290 hospitals, 12% had a DNR order;

• hospitals with higher DNR rates had higher mortality;

• including DNR in risk models resulted in reclassification of 9.3% of “outlier” hospitals.

Some predictors, such as the presence of early DNR status, that could have potentially been included in the model were excluded, either because they weren’t available or are difficult to measure. “They have had to make some decisions, just by virtue of the way the outcomes are created,” says Bruckel. “And it turns out that those things can actually be pretty important indicators of patients who carry risk.”

If the excluded predictors are spread unevenly throughout hospitals, this can skew results further. “Some hospitals might take care of patients that have higher rates of one thing or another, that aren’t necessarily captured in the models,” explains Bruckel.

Bruckel says important questions for CMS, in light of the study’s findings, include: “Are models performing adequately? Are they really measuring what we want to measure? Or are we penalizing hospitals arbitrarily?”

“The bottom line is, if you take care of patients who elect to be DNR for whatever reason, you shouldn’t be indiscriminately penalized,” says Bruckel.

Unintended Consequences

Generally speaking, the goal of reporting quality metrics to hold hospitals accountable is laudable, says Kenneth Covinsky, MD, MPH, a clinician-researcher in the division of geriatrics at University of California, San Francisco.

“But as a community, we do have an obligation to make sure that our quality measures are accurate and actually do lead to better care for patients,” says Covinsky.

One issue is that metrics on in-hospital deaths don’t take into account that not all deaths can be prevented with high-quality care. Some patients are admitted to hospitals at the end of life, when the ability of medical intervention to impact mortality is limited. “In these cases, an equally important goal is to provide care that enhances the well-being of the patient, controls symptoms, and respects the patient or family,” says Covinsky.

The finding that the inclusion or exclusion of DNR orders in hospital mortality quality metrics strongly influences hospital quality rankings “warns us about potential unintended consequences of hospital mortality metrics,” says Covinsky.

The issue is the inclusion of patients whose deaths are not driven by quality of care, but rather, are dying because they are at the end of life. Thus, a hospital that recognizes when patients are at the end of life and delivers good palliative care can be classified as a performance outlier because of a higher death rate. By the same token, a hospital that provides ineffective care and discharges patients to a skilled nursing facility, with the deaths occurring at the skilled nursing facility, can be wrongly deemed as delivering high-quality care.

One reason quality metrics focus on mortality rates is because they’re easy to measure. “We may need to focus on measures that better represent quality, such as functional status and symptom control — even though they are more difficult to measure,” concludes Covinsky.


1. Bruckel J, Mehta A, Bradley SM, et al. Variation in do-not-resuscitate orders and implications for heart failure risk-adjusted hospital mortality metrics. JACC Heart Fail 2017; 5(10):743-752.


• Jeffrey T. Bruckel, MD, MPH, Division of Cardiovascular Medicine, University of Rochester Medical Center, Rochester, New York. Email: jeffrey.bruckel@gmail.com.

• Kenneth Covinsky, MD, MPH, Division of Geriatrics, University of California, San Francisco. Phone: (415) 221-4810 ext. 4363. Email: ken.covinsky@ucsf.edu.