Preoperative risk score tool for rAAA 30-day mortality is accurate, new study finds


An entirely preoperative risk-scoring tool for the prediction of 30-day mortality after a ruptured abdominal aortic aneurysm (rAAA) was found to be accurate after a retrospective review of a prospectively collected database at Harborview Medical Center in Seattle.

Jake Hemingway

The calculator has proven to be a valuable aid in preoperative counseling and in decision-making over transfers for treatment, which represented the majority of the patients studied by a research team at the University of Washington (UW). The results were delivered during the 2020 Western Vascular Society (WVS) virtual annual meeting (Sept. 27–29). In addition, the study’s findings have been confirmed at a number of other institutions, UW vascular surgery resident and first-named author Jake Hemingway MD, told attendees.

Ruptured AAAs carry a considerable burden: The high morbidity and mortality associated with them mean a significant mobilization of both pre-hospital and in-hospital resources is often necessary, he said.

Various other risk-scoring tools for rAAAs have varying levels of effectiveness and reproducibility, Hemingway continued, and some are overly complex and not practical for use. “Others rely on intra-operative and postoperative variables that are entirely worthless to the patient prior to the operation. And some of them are simply not accurate outside of the initial study that created them.”

The Harborview risk score, first published in the Journal of Vascular Surgery (JVS) in 2018, is entirely preoperative and based on four factors: age greater than 76, creatinine over 2mg/dL, a pH of less than 7.2, and episodes of hypotension with a systolic blood pressure less than 70mmHg. The original publication of the institution’s risk score (range: 1–4) found a 30-day mortality of 22% if patients had 1 point, 69% for 2 points, 78% for 3, and 100% mortality if all four factors were present.

In the latest study, the investigators sought to examine the outcomes of modern open and endovascular repair of rAAAs as well as the accuracy of the Harborview risk score in predicting 30-day mortality in the modern rAAA patient. They hypothesized that despite advancements in endovascular techniques and improved mortality following rAAA repair, the score would remain an accurate prediction tool.

The study included all patients who presented to the institution with a rAAA from January 2002–December 2018. Participants were divided into three cohorts based around the shifting sands of Harborview’s practice changes: those patients who belonged to the pre-EVAR era (January 2002–July 2007), the pre-risk score use era (August 2007–October 2013), and the modern era (November 2013–December 2018). The risk score itself, of course, is based on the data gathered from the first two cohorts published in JVS in 2018, with the purpose of the latest study to validate the risk score in the third cohort.

During the period analyzed, 417 patients were treated for rAAAs, with 391 undergoing an operation: 133, 152 and 106 from cohorts one, two and three, respectively. The majority (80%) were male and had an average age of 74. Some 61% underwent open surgical repair, with 38% receiving endovascular aneurysm repair (EVAR).

“On univariate analysis, we see that open surgical repair is associated with an increased risk of 30-day mortality (odds ratio [OR] 2.8, 95% confidence interval [CI], 1.80–4.40; p<0.0001), and being repaired in a later cohort—either two or three, as compared to one—is associated with a decreased risk of mortality,” Hemingway said during the Sept. 27 presentation.

“Regarding the risk score, we see that not only is each individual component associated with increased 30-day mortality, but so, too, is the Harborview risk score as a whole [p<0.0001]. Other variables associated with an increased 30-day mortality include heart rate greater than 110 and an INR [international normalized ratio] over 1.8.” Interestingly, Hemingway added, AAA diameter and patient transport by helicopter were not associated with worse mortality.

Furthermore, Hemingway continued, “on multivariate analysis, we found that repair type as well as the Harborview risk score, INR greater than 1.8, and heart rate over 110 all remain statistically significantly associated with increased 30-day mortality, whereas cohort no longer does.”

The investigators also found that over time—moving from cohort one to three—a higher proportion of repairs were carried out using an endovascular approach; there was a slight decrease in risk score for both open and endovascular patients; there was no difference in risk score between the two groups across all three cohorts; and 30-day mortality decreased from 58.6% in cohort one down to 35% in both two and three. “This is driven both by an increased adoption of endovascular techniques, but also by a decrease in the mortality of open repair, which decreased from 58.7% in cohort one to 40% in cohort three,” Hemingway told the WVS virtual gathering.

“When we look specifically at the risk score within the modern cohort, we see that when we look at all patients combined, as well as patients by repair type, that an increase in a Harborview risk score is associated with an increased 30-day mortality [combined: p<0.0001; open repair: p<0.0003; and EVAR: p<0.0001]. Additionally, when we look at the risk score among all of the patients in the modern cohort, we see that a score of 1 corresponds to a mortality of 35%, 2 to 68%, and 3 or 4 is associated with 100% mortality.”

Hemingway then dealt with common critiques of such risk scores—that they are variable or predictive and based off retrospective reviews of already collected data then prospectively applied to patients.

But the Harborview study had a least one unique aspect, he said: “We were actually able to take our previously created retrospective risk score and apply it in a prospective fashion. And what we found is that the areas under the two ROC [receiver operating characteristic] curves—of both the development cohort [the initial study] and the modern cohort—are similar. In fact, the AUC [area under the RUC curve] is 0.75, essentially, in both cases.”

Thus, the research team concluded that the Harborview risk score is accurate among the modern rAAA patients studied. Hemingway explained: “This represents the largest single-institution experience with rAAA in which we saw over triple the odds of 30-day mortality following open surgical repair compared with EVAR. We also saw that following the institution of an EVAR-first approach, mortality dropped from 58% to 35%, which is likely a reflection of both increased adoption of endovascular techniques but also due to improved mortality following open AAA repair.

“We also saw that the Harborview risk score is accurate among the modern cohort, which is consistent with studies performed at other institutions. We found that the majority of patients are transfer, thus only highlighting the importance of an accurate risk score in guiding these transfer decisions; and that the most likely variable missing in the risk score is pH, so it is important to get an ABG [arterial blood gases] early. Lastly, we identified two variables—INR and heart rate—which might be added to the preoperative risk score and could further increase its accuracy.”


Please enter your comment!
Please enter your name here