Comparison of the surveillance effectiveness with different absenteeism collection methods: a pilot study, China, September 2021 to June 2022

Abstract Background The recently launched facial recognition based syndromic surveillance system (FRSSS) by a group needs to address whether the innovative method used in FASSS can result in higher surveillance accuracy, and how to construct appropriate indicators for FASSS. Aim To compare the surveillance effectiveness of absenteeism collected by FRSSS and school physicians and provide a theoretical basis for surveillance indexes constructed for FRSSS. Methods Two schools in Hangzhou, Zhejiang Province were selected (3110 and 3118 students in the first and second semesters, respectively). Grades 1–2 (DARL), 3–6 (DARH), and school-wide (DARX) all-cause absenteeism were reported by FRSSS, and all-cause absenteeism (DARY) and sickness absenteeism (DARZ) were reported by school physicians from September 1, 2021, to June 24, 2022. The sensitivity, specificity and Youden index of these indicators were analyzed through correlation, time series, control chart and event investigation. Results In school A, the sensitivities of DARL, DARH, DARX, DARY and DARZ were 95.0%, 92.9%, 100%, 100% and 100%, respectively; specificities were 88.0%, 92.1%, 80.6%, 78.2% and 76.8%, respectively; Youden indexes were 83.0%, 85.0%, 80.6%, 78.2% and 76.8%, respectively. In school B, the sensitivities of the same five indicators were 100%; specificities were 89.3%, 91.0%, 83.9%, 80.4% and 81.0%, respectively; Youden indexes were 89.3%, 91.0%, 83.9%, 80.4% and 81.0%, respectively. Conclusions The absenteeism collected by FASSS was realistic and had better accuracy of infectious disease detection than absenteeism collected by school physicians. Moreover, classifying total all-cause absenteeism as grade 1–2 and 3–6 significantly improved FASSS infectious disease situational awareness.