Slide Presentations: Monday, October 24, 2011 |

Limited Utility of SOFA and APACHE II Prediction Models for ICU Triage in Pandemic Influenza FREE TO VIEW

Kyle Enfield, MD; Russ Miller, MD; Todd Rice, MD; B. Taylor Thompson, MD; Jonathon Truwit, MD
Chest. 2011;140(4_MeetingAbstracts):913A. doi:10.1378/chest.1118087
Text Size: A A A
Published online


PURPOSE: The recent experience with pandemic influenza raised concerns with limited availability of necessary resources. Previous modeling studies of non-H1N1 influenza using SOFA scoring suggest that many ICU beds are used unnecessarily, but no such studies have been conducted in H1N1 influenza. We hypothesized that SOFA score would not discriminate survivors and non-survivors.

METHODS: Using data collected by the ARDS Network during the 2009 H1N1 pandemic, we calculated SOFA scores at ICU admission and on day 3 and APACHE II at admission. We created receiver operating characteristic (ROC) curves to evaluate the predictive value of these scores for hospital mortality to day 28.

RESULTS: We enrolled 894 patients. 40% of the patients at the time of admission had a SOFA score >11 and 44% had a SOFA score between 8 and 11. The mortality for these two groups was 33.2% and 16.4% respectively. Of those patients who survived to day 3, 29% had SOFA scores greater than 11 and 71% had SOFA scores between 8 and 11. The observed mortality for these groups was 36.1% and 15.0% respectively. The (area under the curve) AUC was 0.647, 0.656, and 0.736 for day 1 SOFA score, day 3 SOFA score, and APACHE II, respectively.

CONCLUSIONS: Mortality was lower than predicted by Day 1 SOFA, Day 3 SOFA, and APACHE II. Neither the baseline nor day 3 SOFA nor APACHE II scores perform sufficiently well to discriminate survivors form non-survivors in pandemic influenza. We conclude SOFA and APACHE II should not be used for triage decisions for patients with H1N1. Explaining why these scores, which have generally good AUCs, did not perform as well in this group of patients will require further study.

CLINICAL IMPLICATIONS: Current predictive models do not perform sufficiently well to be used as triage tools in the setting of pandemic influenza. This research was supported by N01-HR-56179 and N01-HR-56165 through HR-56176.

DISCLOSURE: The following authors have nothing to disclose: Kyle Enfield, Russ Miller, Todd Rice, B. Taylor Thompson, Jonathon Truwit

No Product/Research Disclosure Information

11:00 AM - 12:30 PM




Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543