Quality in healthcare can be a slippery concept. But in general, our medical system treats mortality as the ultimate adverse event. Higher mortality is thought to indicate poorer quality care.
But what if death were the appropriate and preferred outcome for an individual?
Consider the hypothetical case of an 87-year-old man named Philip. Philip has a living will: if he experiences a serious medical event, he would prefer to receive comfort measures only. He wants to die at home, if at all possible — the way most Americans say they want to die.
He has expressed a preference to avoid the costs, pain, suffering, and indignity that often accompany the end of life in the modern North American hospital.
Unfortunately, Philip has a severe stroke and is incapacitated. He is rushed to the nearest emergency room.
If the hospital where he receives care knows about his preferences, he will not undergo a painful percutaneous endoscopic gastrostomy procedure and other potentially life-saving interventions. The hospital will give him comfort measures only, stabilize him, and then discharge him.
Suppose he dies a week or two later, at home — as he hoped? Would you count it as a black mark against that hospital’s quality record? Or would you say he had received the highest quality care he could have received?
People may disagree about whether they would want their loved one to refuse a potentially life-saving treatment. Many of us, faced with a parent’s decision to forego more intensive care, would be heartbroken. None of us wants to lose a loved one.
But in a high-quality health care system, a patient’s preferences for less intensive end-of-life care must be respected. In those cases, the predictable end result of less intensive end-of-life care – mortality – must also be accepted as the preferred outcome and not count against a healthcare provider’s quality record.
Mortality after ischemic stroke is a publicly reported measure of hospital-based quality of care. The Centers for Medicare and Medicaid Services (CMS) has developed two risk-adjusted 30-day mortality quality measures for stroke patients. The most recent versions of the measures just went before the National Quality Forum (NQF) for endorsement. (Full disclosure: I am on the committee that evaluated the measures, but all this information is publicly available.)
The measures exclude from the denominator any patients enrolled in the Medicare hospice program any time in the 12 months before admission, including the first day of admission. The rationale: “These patients are likely continuing to seek comfort measures only; thus, mortality is not necessarily an adverse outcome or signal of poor quality care for these patients.”
An ideal risk-adjusted mortality measure leaves only two factors to which differences between a hospital’s observed mortality rate and its expected rate can be attributed: random variation and quality of care. Unfortunately, these stroke mortality measures don’t exclude people who actually elect comfort measures only. Why? Perhaps because that information is not available in claims data. Some have proposed adding codes to show patient preferences for comfort measures only, but that could take years. And this problem is not limited to stroke-related quality measures.
Any rational preference-sensitive decision to limit care that varies at the hospital level is a problem for mortality measures. Unfortunately, we don’t know how often any of these things happen or whether, when they happen, they’re preceded by high or low quality care at the hospital level.
We do know that hospitals with higher rates of do-not-resuscitate orders have higher post-stroke mortality rates. Unfortunately, until we have better methods of taking patient preferences into account, hospitals will continue to be penalized for respecting the wishes of people at the end of life.