Clinical validation of clinical decision support systems for medication review: A scoping review

The PrescIT platform: An interoperable Clinical Decision Support System for ePrescription to Prevent Adverse Drug Reactions and Drug-Drug Interactions

INTRODUCTION

Preventable medication errors have been proven to cause significant public health burden, and ePrescription is a key part of the process where medication errors and adverse effects could be prevented. Information systems and "intelligent" computational approaches could provide a valuable tool to prevent such errors with profound impact in clinical practice.

OBJECTIVES

The PrescIT platform is a Clinical Decision Support System (CDSS) that aims to facilitate the prevention of adverse drug reactions (ADRs) and drug-drug interactions (DDIs) in the phase of ePrescription in Greece. The proposed platform could be relatively easily localized for use in other contexts too.

METHODS

The PrescIT platform is based on the use of Knowledge Engineering (ΚΕ) approaches, i.e., the use of Ontologies and Knowledge Graphs (KGs) developed upon openly available data sources. Open standards (i.e., RDF, OWL, SPARQL) are used for the development of the platform enabling the integration with already existing IT systems or for standalone use. The main KG is based on the use of DrugBank, MedDRA, SemMedDB and OpenPVSignal. In addition, the Business Process Management Notation (BPMN) has been used to model long-term therapeutic protocols used during the ePrescription process. Finally, the produced software has been pilot tested in three hospitals by 18 clinical professionals via in-person think-aloud sessions.

RESULTS

The PrescIT platform has been successfully integrated in a transparent fashion in a proprietary Hospital Information System (HIS), and it has also been used as a standalone application. Furthermore, it has been successfully integrated with the Greek National ePrescription system. During the pilot phase, one psychiatric therapeutic protocol was used as a testbed to collect end-users' feedback. Summarizing the feedback from the end-users, they have generally acknowledged the usefulness of such a system while also identifying some challenges in terms of usability and the overall user experience.

CONCLUSIONS

The PrescIT platform has been successfully deployed and piloted in real-world environments to evaluate its ability to support safer medication prescriptions.

Impact of pharmacist-evaluated clinical decision support system alerts on potentially missing or inappropriately prescribed proton pump inhibitors at hospital discharge: a retrospective cross-sectional study

BACKGROUND

Proton pump inhibitors (PPIs) are among the most prescribed drugs. A clinical decision support system (CDSS) could improve their rational use.

AIM

The impact of an electronic algorithm (e-algorithm) implemented in a CDSS on potentially missing or inappropriately prescribed PPIs at hospital discharge, its specificity and sensitivity, and the outcome of the alerts issued were analysed.

METHOD

An e-algorithm continuously monitored patients of a tertiary care hospital for missing or inappropriate PPIs. Following relevance assessment by a pharmacist, the alerts raised were either displayed in the patients' electronic record or dismissed. After a three-month period, all adult patients' records were retrospectively reviewed for missing or inappropriate PPIs at discharge. The results were compared with a corresponding period before CDSS introduction. Sensitivity, specificity and outcome of alerts were quantified.

RESULTS

In a 3-month period with 5018 patients, the CDSS created 158 alerts for missing PPIs and 464 alerts for inappropriate PPIs. PPI prescribing was proposed 81 times and PPI termination 122 times, with acceptance rates of 73% and 34%, respectively. A specificity of 99.4% and sensitivity of 92.0% for missing PPIs and a specificity of 97.1% and a sensitivity of 69.7% for inappropriate PPIs were calculated. The algorithm reduced incidents of missing PPIs by 63.4% (p < 0.001) and of inappropriate PPIs by 16.2% (p = 0.022).

CONCLUSION

The algorithm identified patients without necessary gastroprotection or inappropriate PPIs with high specificity and acceptable sensitivity. It positively impacted the rational use of PPIs by reducing incidents of missing and inappropriate PPIs.

Identifying the risk of exercises, recommended by an artificial intelligence for patients with musculoskeletal disorders

Musculoskeletal disorders (MSDs) impact people globally, cause occupational illness and reduce productivity. Exercise therapy is the gold standard treatment for MSDs and can be provided by physiotherapists and/or also via mobile apps. Apart from the obvious differences between physiotherapists and mobile apps regarding communication, empathy and physical touch, mobile apps potentially offer less personalized exercises. The use of artificial intelligence (AI) may overcome this issue by processing different pain parameters, comorbidities and patient-specific lifestyle factors and thereby enabling individually adapted exercise therapy. The aim of this study is to investigate the risks of AI-recommended strength, mobility and release exercises for people with MSDs, using physiotherapist risk assessment and retrospective consideration of patient feedback on risk and non-risk exercises. 80 patients with various MSDs received exercise recommendations from the AI-system. Physiotherapists rated exercises as risk or non-risk, based on patient information, e.g. pain intensity (NRS), pain quality, pain location, work type. The analysis of physiotherapists' agreement was based on the frequencies of mentioned risk, the percentage distribution and the Fleiss- or Cohens-Kappa. After completion of the exercises, the patients provided feedback for each exercise on an 11-point Likert scale., e.g. the feedback question for release exercises was "How did the stretch feel to you?" with the answer options ranging from "painful (0 points)" to "not noticeable (10 points)". The statistical analysis was carried out separately for the three types of exercises. For this, an independent t-test was performed. 20 physiotherapists assessed 80 patient examples, receiving a total of 944 exercises. In a three-way agreement of the physiotherapists, 0.08% of the exercises were judged as having a potential risk of increasing patients' pain. The evaluation showed 90.5% agreement, that exercises had no risk. Exercises that were considered by physiotherapists to be potentially risky for patients also received lower feedback ratings from patients. For the 'release' exercise type, risk exercises received lower feedback, indicating that the patient felt more pain (risk: 4.65 (1.88), non-risk: 5.56 (1.88)). The study shows that AI can recommend almost risk-free exercises for patients with MSDs, which is an effective way to create individualized exercise plans without putting patients at risk for higher pain intensity or discomfort. In addition, the study shows significant agreement between physiotherapists in the risk assessment of AI-recommended exercises and highlights the importance of considering individual patient perspectives for treatment planning. The extent to which other aspects of face-to-face physiotherapy, such as communication and education, provide additional benefits beyond the individualization of exercises compared to AI and app-based exercises should be further investigated.Trial registration: 30.12.2021 via OSF Registries, https://doi.org/10.17605/OSF.IO/YCNJQ .

[INTERPOLAR-prospective, interventional studies as part of the Medical Informatics Initiative to improve medication therapy safety in healthcare]

Medication analyses by ward pharmacists are an important measure of drug therapy safety (DTS). Medication-related problems (MRPs) are identified and resolved with the attending clinicians. However, staff resources for extended medication analyses and complete documentation are often limited. Until now, data required for the identification of risk patients and for an extended medication analysis often had to be collected from various parts of the institution's internal electronic medical record (EMR). This error-prone and time-consuming process is to be improved in the INTERPOLAR (INTERventional POLypharmacy-Drug interActions-Risks) project using an IT tool provided by the data integration centers (DIC).INTERPOLAR is a use case of the Medical Informatics Initiative (MII) that focuses on the topic of DTS. The planning phase took place in 2023, with routine implementation planned from 2024. DTS-relevant data from the EMR is to be presented and the documentation of MRPs in routine care is to be facilitated. The prospective multicenter, cluster-randomized INTERPOLAR‑1 study serves to evaluate the benefits of IT support in routine care. The aim is to show that more MRPs can be detected and resolved with the help of IT support. For this purpose, six normal wards will be selected at each of eight university hospitals, so that 48 clusters (with a total of at least 70,000 cases) are available for randomization.

Evaluation of Triple Whammy Prescriptions After the Implementation of a Drug Safety Algorithm

BACKGROUND AND OBJECTIVE

The term triple whammy (TW) refers to the concomitant use of non-steroidal anti-inflammatory drugs, diuretics, and angiotensin system inhibitors; this combination significantly increases the risk of acute kidney injury (AKI). To prevent this serious complication, we developed an electronic algorithm that detects TW prescriptions in patients with additional risk factors such as old age and impaired kidney function. The algorithm alerts a clinical pharmacist who then evaluates and forwards the alert to the prescribing physician.

METHODS

We evaluated the performance of this algorithm in a retrospective observational study of clinical data from all adult patients admitted to the Cantonal Hospital of Aarau in Switzerland in 2021. We identified all patients who received a TW prescription, had a TW alert, or developed AKI during TW therapy. Algorithm performance was evaluated by calculating the sensitivity and specificity as a primary endpoint and determining the acceptance rate among clinical pharmacists and physicians as a secondary endpoint.

RESULTS

Among 21,332 hospitalized patients, 290 patients had a TW prescription, of which 12 patients experienced AKI. Overall, 216 patients were detected by the alert algorithm, including 11 of 12 patients with AKI; the algorithm sensitivity is 88.3% with a specificity of 99.7%. Physician acceptance was high (77.7%), but clinical pharmacists were reluctant to forward the alerts to prescribers in some cases.

CONCLUSION

The TW algorithm is highly sensitive and specific in identifying patients with TW therapy at risk for AKI. The algorithm may help to prevent AKI in TW patients in the future.

21st century (clinical) decision support in nursing and allied healthcare. Developing a learning health system: a reasoned design of a theoretical framework

In this paper, we present a framework for developing a Learning Health System (LHS) to provide means to a computerized clinical decision support system for allied healthcare and/or nursing professionals. LHSs are well suited to transform healthcare systems in a mission-oriented approach, and is being adopted by an increasing number of countries. Our theoretical framework provides a blueprint for organizing such a transformation with help of evidence based state of the art methodologies and techniques to eventually optimize personalized health and healthcare. Learning via health information technologies using LHS enables users to learn both individually and collectively, and independent of their location. These developments demand healthcare innovations beyond a disease focused orientation since clinical decision making in allied healthcare and nursing is mainly based on aspects of individuals' functioning, wellbeing and (dis)abilities. Developing LHSs depends heavily on intertwined social and technological innovation, and research and development. Crucial factors may be the transformation of the Internet of Things into the Internet of FAIR data & services. However, Electronic Health Record (EHR) data is in up to 80% unstructured including free text narratives and stored in various inaccessible data warehouses. Enabling the use of data as a driver for learning is challenged by interoperability and reusability.To address technical needs, key enabling technologies are suitable to convert relevant health data into machine actionable data and to develop algorithms for computerized decision support. To enable data conversions, existing classification and terminology systems serve as definition providers for natural language processing through (un)supervised learning.To facilitate clinical reasoning and personalized healthcare using LHSs, the development of personomics and functionomics are useful in allied healthcare and nursing. Developing these omics will be determined via text and data mining. This will focus on the relationships between social, psychological, cultural, behavioral and economic determinants, and human functioning.Furthermore, multiparty collaboration is crucial to develop LHSs, and man-machine interaction studies are required to develop a functional design and prototype. During development, validation and maintenance of the LHS continuous attention for challenges like data-drift, ethical, technical and practical implementation difficulties is required.

Polypharmacy and precision medicine

Precision medicine is an approach to maximise the effectiveness of disease treatment and prevention and minimise harm from medications by considering relevant demographic, clinical, genomic and environmental factors in making treatment decisions. Precision medicine is complex, even for decisions about single drugs for single diseases, as it requires expert consideration of multiple measurable factors that affect pharmacokinetics and pharmacodynamics, and many patient-specific variables. Given the increasing number of patients with multiple conditions and medications, there is a need to apply lessons learned from precision medicine in monotherapy and single disease management to optimise polypharmacy. However, precision medicine for optimisation of polypharmacy is particularly challenging because of the vast number of interacting factors that influence drug use and response. In this narrative review, we aim to provide and apply the latest research findings to achieve precision medicine in the context of polypharmacy. Specifically, this review aims to (1) summarise challenges in achieving precision medicine specific to polypharmacy; (2) synthesise the current approaches to precision medicine in polypharmacy; (3) provide a summary of the literature in the field of prediction of unknown drug-drug interactions (DDI) and (4) propose a novel approach to provide precision medicine for patients with polypharmacy. For our proposed model to be implemented in routine clinical practice, a comprehensive intervention bundle needs to be integrated into the electronic medical record using bioinformatic approaches on a wide range of data to predict the effects of polypharmacy regimens on an individual. In addition, clinicians need to be trained to interpret the results of data from sources including pharmacogenomic testing, DDI prediction and physiological-pharmacokinetic-pharmacodynamic modelling to inform their medication reviews. Future studies are needed to evaluate the efficacy of this model and to test generalisability so that it can be implemented at scale, aiming to improve outcomes in people with polypharmacy.

Potential Drug-Related Problems in Pediatric Patients-Describing the Use of a Clinical Decision Support System at Pharmacies in Sweden

The clinical support system Electronic Expert Support (EES) is available at all pharmacies in Sweden to examine electronic prescriptions when dispensing to prevent drug-related problems (DRPs). DRPs are common, and result in patient suffering and substantial costs for society. The aim of this research was to study the use of EES for the pediatric population (ages 0-12 years), by describing what types of alerts are generated for potential DRPs, how they are handled, and how the use of EES has changed over time. Data on the number and categories of EES analyses, alerts, and resolved alerts were provided by the Swedish eHealth Agency. The study shows that the use of EES has increased. The most common type of alert for a potential DRP among pediatric patients was regarding high doses in children (30.3% of all alerts generated). The most common type of alert for a potential DRP that was resolved among pediatrics was therapy duplication (4.6% of the alerts were resolved). The most common reason for closing an alert was dialogue with patient for verification of the treatment (66.3% of all closed alerts). Knowledge of which type of alerts are the most common may contribute to increased prescriber awareness of important potential DRPs.

Implementing evidence-based anticoagulant prescribing: User-centered design findings and recommendations

Background

Direct oral anticoagulants (DOACs) are widely used medications with an unacceptably high rate of prescription errors and are a leading cause of adverse drug events. Clinical decision support, including medication alerts, can be an effective implementation strategy to reduce prescription errors, but quality is often inconsistent. User-centered design (UCD) approaches can improve the effectiveness of alerts.

Objectives

To design effective DOAC prescription alerts through UCD and develop a set of generalizable design recommendations.

Methods

This study used an iterative UCD process with practicing clinicians. In three rapid iterative design and assessment stages, prototype alert designs were created and refined using a test electronic health record (EHR) environment and simulated patients. We identified key emergent themes across all user observations and interviews. The themes and final designs were used to derive a set of design guidelines.

Results

Our UCD sample comprised 13 prescribers, including advanced practice providers, physicians in training, primary care physicians, and cardiologists. The resulting alert designs embody our design recommendations, which include establishing intended indication, clarifying dosing by renal function, tailoring alert language in drug interactions, facilitating trust in alerts, and minimizing interaction overhead.

Conclusions

Through a robust UCD process, we have identified key recommendations for implementing medication alerts aimed at improving evidence-based DOAC prescribing. These recommendations may be applicable to the implementation of DOAC alerts in any EHR systems.