DETROIT – Benchmarking trauma center performance using traditional measures fails to adequately reflect the aging face of trauma deaths today, a new study suggests.
Increased public awareness of advance directives and an aging population has shifted the demographics of trauma mortality, Dr. Glen A. Franklin said at the annual meeting of the Central Surgical Association.
Dr. Glen A. Franklin
"Traditional measures of trauma center quality based on injury severity or comorbidity indexing may not capture all the issues associated with mortality," he said.
Dr. Franklin and his colleagues at the University of Louisville in Kentucky studied the effects of advance directives and futile care on trauma center mortality by examining all trauma deaths in 2008-2009 to determine the proportion of patients whose outcome could have been changed by medical care.
Of the 5,433 patients treated, 347 died, resulting in a crude mortality of 6.4%. Advance directives were in place in 18% of patients at the time of admission, said Dr. Franklin, director of the university’s critical care/trauma fellowship program.
Care was withdrawn in 147 deaths (42%) at a median of 1.5 days. The median injury severity score was 25 for all patients who died (both those who did and those who did not have care withdrawn). Most deaths (66%) occurred early in the hospital course, within 48 hours or less, and the decision for withdrawal of care was made within that time frame in 24% of cases.
On univariate analysis, only age (69.5 years vs. 48 years) and the presence of an advance directive (69% vs. 31%) were significantly different between patients in whom care was or was not withdrawn. Injury severity score did not correlate with withdrawal of care (odds ratio 1.014), he said.
Patients with advance directives were significantly older than those without advance directives (72 years vs. 53 years) and were three times as likely to have care withdrawn (OR 3.11). The decision to withdraw care was made by family members in 140 cases (95%), by the patient in 6, and by power of attorney in 1.
Of the patients who died, 69 patients (20%) arrived at the trauma center in or rapidly progressed to cardiac arrest, and were placed in the futile-care category. A vigorous quality improvement review process indicated that medical care could not have been reasonably expected to impact survival in 62% (215 deaths), Dr. Franklin said. Only 12 deaths were considered potentially preventable.
When these factors are taken into account, the hospital’s overall crude death rate of 6.4% fell to 3.7% when adjusted for the 147 patients with care withdrawn (200 deaths remaining/5,433 cases) and further to 2.4% when adjusted for the 69 futile care patients (131 deaths remaining/5,433 cases).
According to information from the National Trauma Data Bank, trauma patient volume and injury severity are high at the University of Louisville, but so is the center’s mortality.
"It is this type of [nonadjusted] data that is making it out into the public sector, with really no ability for the layperson to interpret its meaning," Dr. Franklin said.
Risk-adjusted data will soon be available via the American College of Surgeons’ recently launched Trauma Quality Improvement Program (TQIP), he observed. The program mimics the National Surgical Quality Improvement Program (NSQIP) and looks at observed-to-expected mortality ratio in a risk-adjusted fashion. In a recent study using this approach, only 9 of 132 centers in the United States were identified as being above-average performers with regard to elderly trauma after adjustment for case mix (Ann. Surg. 2011;253:144-50), he added.
Dr. Franklin observed that societal, cultural, regional, religious, and racial factors can affect family perceptions of appropriate escalation of care and thus mortality outcomes. For example, emergency medical technicians in Kentucky cannot declare trauma patients dead, and paramedics can do so only under extreme circumstances such as decapitation or rigor mortis. As a result, deceased patients who arrive in the trauma center may be subjected to inappropriate resuscitation attempts, and ultimately figure in the trauma center’s mortality profile, he said. For another example, the decision to withdraw care may be made more conservatively when caring for patients in the Bible Belt, where religious beliefs are more likely to play a role in attitudes toward the end of life.
"We believe all of these factors need to be placed into the data set in some manner to affect trauma center performance, particularly as transparency becomes a key in the quality movement," he concluded.
Invited discussant Dr. Mark Hemmila of the University of Michigan, Ann Arbor, asked whether do-not-resuscitate (DNR) status prior to injury or surgery, data that are collected by both NSQIP and TQIP, might be a reasonable compromise for use as a covariate in risk adjustment of mortality data. He went on to ask whether trauma centers would collect this data accurately or whether there was potential to rig the system.