Purpose: We aimed to evaluate differences in the interobserver reliability and accuracy of sequential organ failure assessment (SOFA) scoring between nurses and residents.
Methods: Eight nurses and eight residents independently scored 24 randomly selected patients. Intraclass correlation coefficients (ICCs) for the reliability of total SOFA scoring were calculated. The residents' and nurses' SOFA scores were compared with a gold standard to assess accuracy.
Results: The overall ICC of the total SOFA score was 0.87 (nurses 0.89, residents 0.86) for a single measurement. Residents tended to assign higher total SOFA scores than did nurses, without a statistically significant difference (7.01 ± 4.43 vs. 6.72 ± 4.27, P > 0.05). The mean bias between the nurses' and the gold standard total SOFA scores was -0.16 ± 1.86 and the 95% confidence limit of agreement was -3.8 to +3.49. The mean bias between the residents' and the gold standard total SOFA scores was -0.39 ± 1.81, and the 95% confidence limit of agreement was -3.95 to +3.16. The percentage of accurate data for the total SOFA score was 47.4% for nurses and 51% for residents (P > 0.05). Although not statistically significant, the major error rate (≥2 point deviation from the gold standard score) was higher for nurses than for residents (29.16 and 23.43%, P > 0.05). Accuracy of scoring individual organ systems was similar for the two groups; however, the major error rate in the cardiovascular system score was higher for nurses.
Conclusion: Interobserver reliability was good and mean SOFA scores were not significantly different between nurses and residents. The accuracy of SOFA scoring was moderate for both groups; however, although the difference was not statistically significant, the major error rate was higher for nurses than for residents.