Purpose: Ranson's criteria have proven a useful predictor of mortality in
acute pancreatitis. The Ranson variables, however, are not specific for
pancreatitis but rather general markers of organ system dysfunction. We
therefore studied Ranson's original criteria as a predictor of mortality in
non-pancreatic critical illness.
Methods: Over a 14 month period, all patients admitted to the surgical
intensive care unit (SICU) for >=48 hours, excluding those with the diagnosis
of pancreatitis, were included in the study. Ranson's criteria at admission
(Age, WBC, glucose, LDH, AST) and at 48 hours (BUN increase, hematocrit
decrease, calcium, PO2, base deficit, fluid sequestration) were
recorded on all patients. The hematocrit was adjusted by 3.3 percentage points
for each unit of blood transfused. PO2 values were analyzed
regardless of FiO2. In addition, we corrected for the effect of
supplemental O2 using a ratio of PO2/FiO2.
As per Ranson's original paper, patients were stratified to one of four groups
based on the total number of signs (0-2, 3-4, 5-6, >6). Hospital mortality
was compared with the number of Ranson's signs at admission, at 48 hours and
with the total criteria.
Results: 107 patients met the criteria for the study and had all 11 Ranson's
criteria measured. The mean age was 70.6 (16-94 15.2), with 45 females and 62
males. Ten surgical services contibuted to the cohort (30 vascular, 29 general
surgery, 22 orthopedic and 26 other). The mean SICU length of stay (LOS) was 4.6
days (2-34) and the mean total LOS was 13.6 days (3-44). Thirty-nine patients
(36%) had 0-2 Ranson's signs, 51 (48%) had 3-4 signs, 15 (14%) had 5-6 signs and
2 patients (2%) had >6 signs. Mortality for patients with 0-2 Ranson's signs
was 2.5%; 3-4, 9.8%; 5-6, 13.3% and >6, 50% (p=0.07). Admission Ranson's
signs alone were poor predictors of mortality (p=0.9) while those 6 signs
collected at 48 hours were predictive at p=0.06. As independant variables, base
deficit and fluid sequestration >6L were the best predictors of mortality (p<0.01
and p<0.03 respectively).
When analyzed using the correction for FiO2, mortality rates
changed for each group as follows: 0-2, 0%, 3-4, 7.5%, 5-6, 18% and >6,33%
and became statistically significant (p < 0.05).
Conclusions: Ranson's criteria can predict mortality in non-pancreatic
critical illness with results parallelling that of Ranson's original work in
acute pancreatitis. The accuracy of the system is improved by modifying PO2
values for supplemental oxygen. The simplicity, accessibility and familiarity of
the Ranson variables facilitates the utility of this scoring system.
