Antimicrobial agents (AMAs) play a pivotal role in treating infectious diseases. On the other hand, inappropriate use of AMAs could cause multidrug-resistant bacteria [1, 2]. Pharmacists explore efficient methods to reduce the improper use of AMAs through educational intervention [3], prescription recommendations based on pathogen identification [4], and multi-faceted intervention [5]. Oral AMAs such as Quinolones, Tetracyclines, and Cefdinir are pivotal in treating various infectious diseases. MCs such as magnesium oxide, ferrous citrate, and sucralfate are frequently used as antacids and cathartics. When AMAs are taken simultaneously with MCs, they chelate, reducing their absorption and therapeutic efficacies [6,7,8]. Despite the well-known risks, the prescriptions that order co-intake of AMAs and MCs are still seen in clinical practice. While pharmacists have often been alerted to this risky interaction, it is not sufficient to reduce those prescriptions.

Clinical Decision Support Systems (CDSSs) have emerged as powerful tools in the healthcare landscape, significantly enhancing the ability to prevent adverse drug interactions. These systems utilize comprehensive databases and sophisticated algorithms to provide real-time alerts to healthcare providers about potential drug-drug interactions, thereby preventing patient harm. Numerous studies have demonstrated the efficacy of CDSSs in improving medication safety and optimizing therapeutic outcomes. For instance, research has shown that the implementation of CDSSs can reduce the incidence of serious drug-drug interactions and associated adverse events by alerting clinicians to potential interactions at the point of care [9,10,11]. In this study, we focused on the interaction between AMAs and MCs to facilitate the appropriate use of AMAs.

We investigated the effectiveness of pop-up alerts integrated into our hospital’s ordering and pharmacy dispensation support systems in reducing the prescriptions that ordered co-intake of AMAs and MCs. To this end, we implemented our ordering system HAPPY ACTIS (CANON MEDICAL SYSTEMS CORPORATION, Tochigi, Japan) to pop up when the assigned AMAs (Quinolones, Tetracyclines, and Cefdinir) were ordered, regardless of whether or not the prescriptions contained MCs (Fig. 1A, B). We also implemented our hospital’s pharmacy dispensation support system YUNICOM-EX (Yuyama MFG Co., Ltd., Osaka, Japan) to pop up when the assigned AMAs or MCs were prescribed for the patient who was on MCs or the assigned AMAs, respectively (Fig. 1A, C). The detailed settings for pop-up alerts in the ordering and pharmacy dispensation support systems were shown in the Supplemental Materials. We then surveyed the number of prescriptions ordering co-intake of AMAs and MCs and the number of pharmacist inquiries to prevent those prescriptions between Feb 2016 – Jan 2017 (Pre), Feb 2017 – Jan 2018 (Post 1y), and Feb 2018 – Jan 2019 (Post 2y). This study was approved by the St. Mary’s Hospital Institutional Review Board (approval number: 19–1006). All statistical analyses were performed using EZR software [12]. Proportions were compared using Chi-square test or Fisher's exact test for every pair, and P values due to repeated multiple testing were corrected using the Bonferroni method. P value < 0.05 was regarded as statistically significant.

Automated pop-up alerts implemented in our ordering and pharmacy dispensation support system. A The flow of pop-up alerts in our ordering and pharmacy dispensation support system. B, C)The displays of pop-up alerts on ordering (B) and pharmacy dispensation support system (C)

We first examined the effect of the pop-up alerts on the number of prescriptions ordering co-intake of AMAs and MCs during Pre, Post 1y, and Post 2y. There was no difference in the percentage of prescriptions including AMAs and MCs between before and after the introduction of pop-up alerts, indicating that the pop-up alerts did not abolish physicians to order AMAs (Fig. 2A). On the contrary, the proportion of prescriptions ordering co-intake of AMAs and MCs was significantly decreased after the introduction of pop-up alerts as compared with before the implementation of pop-up alerts (Fig. 2B). Of note, the proportion of prescriptions ordering co-intake of AMAs and MCs was significantly lower in the Post 2y than Post 1y (Fig. 2B). We further analyzed the trend in the proportion of prescriptions ordering co-intake of AMAs and MCs, categorized by type of AMAs. The proportion of prescriptions ordering co-intake of Quinolones, Tetracyclines, or Cefdinir and MCs was significantly reduced respectively (Fig. 2C). Of note, the ratio of prescriptions ordering co-intake of tetracyclines and MCs was dramatically decreased (Fig. 2C). In our hospital, Tetracyclines are frequently used for osteomyelitis, skin and soft tissue infection, and molecular targeted therapy-induced skin disorders. The reduction of prescriptions ordering co-intake of AMAs and MCs is thought to reflect the usefulness of the induction of pop-up alerts in our hospital’s ordering system.

The effectiveness of the implemented automated pop-up alerts in reducing the prescriptions ordering co-intake of AMAs and MCs. A-D The proportion and number of prescriptions containing both AMAs and MCs among prescriptions containing AMAs (A), prescriptions ordering co-intake of AMAs and MCs among prescriptions containing both AMAs and MCs (B), prescriptions ordering co-intake of AMAs and MCs categorized by Quinolones, Tetracyclines and Cefdinir (C), prescriptions in which pharmacists prevented co-intake of AMAs and MCs among prescriptions ordering co-intake of AMAs and MCs (D), before and after implementation of the automated pop-up alerts. The difference in significance of the ratio was analyzed by the Chi-square test (for A, B) and Fisher’s exact test (for C, D), followed by the Bonferroni method. P value < 0.05 was regarded as statistically significant

We implemented the ordering system to alert for all cases when the physicians ordered the assigned AMAs, regardless of whether the physicians ordered the MCs. On the other hand, we did not implement the ordering system to notify whether other physicians had ordered the MCs before. Therefore, we also implemented the pharmacy dispensation support system to alert when the MCs were included in the prescription of the specific patient (Fig. 1A). If the pharmacists find the prescriptions ordering co-intake of AMAs and MCs, they can inquire about preventing the co-intake of AMAs and MCs to the physicians. We examined the proportion of prescriptions in which pharmacist inquiries prevented the co-intake of AMAs and MCs among prescriptions ordering co-intake of AMAs and MCs. The proportion of prescriptions in which pharmacist inquiries prevented the co-intake of AMAs and MCs significantly increased after the implementation of pop-up alerts compared to before the implementation of pop-up alerts in the pharmacy dispensation support system. Of note, the ratio of prescription inquiry for co-intake of AMA and MCs was significantly higher in Post 2y than in Post 1y (Fig. 2D). This result indicates that pop-up alerts in pharmacy dispensation support system promoted the pharmacist’s activity for preventing the co-intake of AMAs and MCs.

In recent years, CDSS has significantly contributed to reducing adverse drug interactions through automated alerts. On the contrary, previous studies have demonstrated that excessive alerts can overwhelm physicians, overriding a substantial proportion of alerts including those that could prevent serious drug interactions [13,14,15]. Indeed, although our pop-up alerts on the ordering system significantly reduced the prescription ordering co-intake of AMAs and MCs, many of those prescriptions are still seen, indicating that pop-up alerts were often overridden. A prolonged study is required to fully understand our dual alert system’s long-term effects. Additionally, external validity is lacking because this study was conducted in only one hospital. A multicenter study is required to evaluate the reproducibility and usefulness of our dual alert system.