[Accuracy of sequential organ failure assessment score in emergency department physicians in Beijing: a multicenter investigation study].

Objective: To investigate the accuracy of sequential organ failure assessment (SOFA) scoring in emergency physicians in Beijing.

Methods: Emergency physicians from 8 hospitals in Beijing in January 2018 were demanded to complete a SOFA questionnaire which was developed on "wenjuanxing" website and submit via cell phone. All participants were divided into urban center group (UC group) and no-urban center group (NUC group) based on the hospital's location. The accuracy rate of components and total score of SOFA along with the mistakes were evaluated, and the results of the two groups were compared.

Results: (1) The questionnaire was sent to 217 emergency physicians of the 8 hospitals, and 197 qualified questionnaires were received with 109 of NUC group and 88 of UC group, respectively, the total response rate was 90.8%. Compared with those from NUC group, UC physicians had older ages [years: 37 (32, 42) vs. 34 (29, 40), Z = -2.554, P = 0.011] and higher education level [postgraduate degree 76.1% (67/88) vs. 40.4% (44/109), χ = 25.327, P < 0.001], and more of them experienced SOFA scoring [62.5% (55/88) vs. 45.9% (50/109), χ = 5.409, P = 0.020]. Other baseline characteristics such as gender, working years, professional title and training experience were not different between the two groups. (2) The accuracy rate of total SOFA score was 62.4% (123/197) in the whole cohort, and UC group was lower than that of NUC group, but the difference was not significant [56.8% (50/88) vs. 67.0% (73/109), χ = 2.141, P = 0.143]. While comparing the accuracy of individual variable/system of SOFA, the accuracy rate of norepinephrine of UC group was much higher than NUC group [80.7% (71/88) vs. 66.1% (72/109), χ = 5.235, P = 0.022], but the accuracy of Glasgow coma scale (GCS) was much lower in NUC group [38.6% (27/70) vs. 81.6% (71/87), χ = 30.629, P < 0.001]. Other variables of SOFA were not different between the two groups. Based upon the results of all submitted questionnaires, 566 mistakes were identified. It was indicated that the mistakes per capital was 2.9 in the whole cohort and in the two groups. The first type mistakes which caused by carelessness (including calculating error, filling error, choosing error) were 233 times. The calculating error in norepinephrine from NUC physicians was higher than the UC group [33.9% (37/109) vs. 19.3% (17/88), χ = 5.235, P = 0.022], there was no significant difference in any other first type mistakes between the two groups. The total second type mistakes caused by misunderstanding of SOFA (including using wrong variables, not using the worst value within 24 hours, and incorrect GCS score) were 333 times in the whole cohort. GCS error [61.8% (42/88) vs. 16.9% (14/109), χ = 32.292, P < 0.001], and using urine output per hour instead of urine output per 24 hours [15.9% (14/88) vs. 4.6% (5/109), χ = 7.162, P = 0.007] were much higher in UC group than NUC group.

Conclusions: The total accuracy of SOFA scoring in the investigated emergency physicians of 8 hospitals in Beijing was not good. Mistakes causing by carelessness or misunderstanding of score rules were similar. It is necessary to apply strict training in SOFA scoring.