2.2 Construction and performance evaluation of active surveillance programs
The CHPS Drug Evaluation System (Figure 2) is one of the subsystems in the CHPS. It can obtain seven dimensions of clinical information from HIS, including patient retrieval (basic information of patients), test retrieval (test items, test values), medical order retrieval (drug ID), medical record retrieval (admission records, progress notes), diagnosis retrieval, physical sign retrieval, and examination retrieval, these seven dimensions can be easily connected with each other by Boolean logic operators. In this study, we used Boolean logic programming for triggers in medical orders, diagnoses, and progress notes to form retrieval rules. To improve the positive rate, the triggers in the progress notes and medical orders were connected by ”AND”. The rules were then run on the discharged patients to identify and analyze cases with positive triggers but not anaphylaxis, and exclusion rules were set accordingly to improve triggers performance. Finally, active surveillance programs were constructed by combining search rules and exclusion rules (Figure 3).
The positive predictive value (PPV) of the active surveillance programs was calculated as the number of anaphylaxis cases detected by active surveillance programs divided by the number of cases with positive triggers. The incidence of anaphylaxis was calculated as the number of anaphylaxis cases detected by active surveillance programs divided by the number of discharged patients. The performance of the active surveillance programs compared with the spontaneous reporting system was expressed as cases of anaphylaxis detected by active surveillance programs divided by reports of anaphylaxis in the spontaneous reporting system in the same period.