Accurate ID of nonsalvageable trauma patients improves trauma center performance metrics

Key clinical point: The ACS TQIP proxy definition of DOA should be used to exclude unsalvageable patients from peer-to-peer benchmarking and performance improvement efforts.

Major finding: Just 0.9% of patients identified as being unsalvageable by the PROXY criteria went on to survive.

Data source: Retrospective analysis of 223,643 trauma patients in the ACS TQIP database.

Disclosures: Dr. Byrne and his coauthors reported having no financial disclosures.

View on the News

High-quality data needed

Byrne et al.’s study evaluating three criteria for identifying unsalvageable trauma patients demonstrates the critical role that high-quality data can play in quality improvement efforts. Using data from the American College of Surgeons Trauma Quality Improvement Program (ACS TQIP), they developed a proxy measure that resulted in less than 1% of patients being incorrectly classified as unsalvageable.

Having an accurate, easy-to-calculate model for predicting survival is essential in applying these findings at the point of care, as the decision of whether or not to perform a resuscitative thoracotomy must be made expediently. An accurate model also allows resources to be concentrated on those who might derive the most benefit and minimizes the known harms, particularly to health care providers, and costs associated with resuscitative thoracotomies. Finally, an accurate model is necessary to ensure that benchmarking accurately reflects quality of care rather than case mix, and that efforts are appropriately directed toward those centers whose adjusted mortality is above expected, whether or not they are outliers.

Future efforts should be directed toward evaluating whether implementation of this proxy measure into clinical practice improves not just survival but survival with good functional status at a longer term end point.

Dr. Lillian S. Kao is an ACS Fellow and associate professor of surgery at the University of Texas Health Science Center at Houston.




LAKE BUENA VISTA, FL – When does no sign of life mean a patient is unsalvageable?

A study has found that up to 33% of patients who local providers determined had no signs of life went on to live, and 10% of patients whose heart stopped before reaching the hospital actually survived.

Historically, there’s been significant variation across trauma centers and registries of how unsalvageable patients are identified. This presents a problem when measuring trauma center performance, particularly when you consider that 25% of deaths occur within 15 minutes of arrival at high-volume trauma centers, Dr. James P. Byrne, with the University of Toronto, explained.

“Variation within inclusion and exclusion criteria can lead to big differences in risk-adjusted trauma center mortality. Therefore, there’s a need for the adoption of a single-best definition for unsalvageable patients,” he said at the annual scientific assembly of the Eastern Association for the Surgery of Trauma (EAST).

To that end, three case definitions of the unsalvageable patient were proposed based on data from the 2012-2013 American College of Surgeons Trauma Quality Improvement Program (ACS TQIP) database. They were no signs of life as determined by local providers (NSOL), prehospital cardiac arrest (PHCA) as entered into local trauma registries, and a PROXY for death established by the ACS TQIP and defined as an emergency department heart rate of 0 and an ED systolic blood pressure of 0 and a Glasgow Coma Scale motor component of 1.

Over the study period, 223,643 patients from 192 trauma centers met the inclusion criteria of at least 16 years of age, blunt or penetrating mechanism of injury, and known hospital discharge status. In-hospital mortality was 7.2%.

NSOL and PHCA, had positive predictive values low enough (66.58% and 89.71%) such that 33% and 10% of patients meeting these criteria went on to survive.

The PROXY had excellent predictive utility for death (PPV 99.09%), with just 0.9% of PROXY patients going on to survive (22/2,424), he said.

To test its validity, the investigators looked more closely at the 2,424 patients who met PROXY criteria (15% of all deaths). PROXY patients mostly fell into two distinct groups: severe multisystem blunt injury caused by motor vehicle collision (MVC) and penetrating trauma to the head or chest caused by firearm, Dr. Byrne said. The median time to death was 8 minutes, with 87% dying in the ED.

Among the 22 unexpected PROXY survivors, 77% had penetrating trauma. Most had isolated injuries to the heart, lung, or large blood vessels and underwent thoracotomy (71%) or open cardiac massage (35%).

“We feel these patients are adequately explained for the most part, even though they come in without vital signs, as patients that have some chance for survival with prompt hemorrhage control surgery or cardiac repair,” Dr. Byrne said.

PROXY patients with penetrating injury rather than blunt injury were more likely to be male (90% vs. 73%), younger (34 years vs. 44 years), and fall victim to firearms or stabbing than an MVC (88% & 12% vs. 59%).

Overall, patients with penetrating injury were 10 times more likely to meet PROXY criteria than those with blunt injury (5.7% vs. 0.6%; P < .001).

“The characteristics of the PROXY patients, as well as the significant association with penetrating trauma, is something that we know to be true for patients who die early from trauma,” Dr. Byrne said.

In contrast, the 13,659 patients who died without meeting PROXY criteria were older (mean age, 60 years), 72% had severe head injury from falls or MVC, and their median time to death was 52 hours.

“We feel these patients represent alternative trajectories to death that are not predicted by presenting characteristics,” he said. “This actually lends construct validity to the PROXY definition since it was able to isolate patients who died early, while excluding those patients who died later and might have a chance for a modifiable outcome.”

Finally, a hierarchical linear model that calculated risk-adjusted mortality was used to look at the influence of including nonsalvageable patients on trauma center performance. Based on the model, 36 trauma centers (19%) were below-average performers and 29 (15%) were above-average performers.

After excluding PROXY patients, 64% of trauma centers changed rank, 17% by three or more positions, but only two centers changed outlier status. The latter suggests that inclusion of unsalvageable PROXY patients would have a minor impact on risk-adjusted mortality used for peer-to-peer benchmarking. However, their inclusion could have a big impact on benchmarking at centers that receive unsalvageable patients more frequently and therefore, PROXY should be used to exclude them from registries, Dr. Byrne said.

Next Article:

   Comments ()