Sequential Organ Failure Assessment in pandemic planning

Introduction

The H1N1 pandemic has highlighted the importance of reliable and valid triage instruments for scarce resources during periods of high demand. Christian and colleagues have proposed a triage protocol that utilizes a Sequential Organ Failure Assessment (SOFA) score >11 to exclude patients from critical care resources quoting an associated mortality of more than 90% [1]. We sought to assess the hospital mortality associated with this SOFA threshold and the resource implications of such a triage protocol.

Methods

This retrospective cohort study included consecutive ICU patients admitted to any one of our three tertiary-care adult multisystem ICUs from January 2003 to December 2008. Patients were excluded if they were admitted for routine postoperative monitoring (ICU stay <48 hours) or postoperative cardiac surgery. SOFA was collected daily by an electronic bedside clinical information system (QS; GE Medical Systems).

Results

A total of 10,204 patients (69,913 patient-days) were included. Mean age was 59. Mean admission APACHE was 19.1. Mortality was 25%. Median ICU LOS was 4 days. A total 13.4% of the cohort (representing 9% of total patient-days) had an initial SOFA >11. Mortality in patients with an initial SOFA score >11 was 59% (95% CI 56%, 62%). Figure 1 demonstrates increased mortality associated with SOFA >11 during the ICU stay to a maximum of 78% (95% CI 68%, 86%) on day 14. The mortality associated with an initial SOFA >11 across diagnostic categories (ICNARC) varied from 29% for poisoning to 67% for neurological patients. Mortality associated with an initial SOFA >11 was lowest for those patients 18 to 20 years old (37%) and highest for those >80 years old (75%). Mortality exceeded 90% when the initial SOFA was >20. However, only 0.2% of patients had an initial SOFA >20.

Figure 1

Hospital mortality associated with SOFA >11 during the ICU stay.

Conclusions

A SOFA score >11 was not associated with a hospital mortality >90% at any time during the ICU stay. Age and diagnostic category represent potential modifying factors in the association of SOFA >11 and hospital mortality. Only a small proportion of patients have the extreme initial SOFA values associated with a hospital mortality >90%, limiting the usefulness of SOFA as a triage instrument for pandemic planning.