Strategies aimed at bolstering patient safety frequently incorporate Clinical Decision Support Systems (CDSS), and for good reason. Medication errors, particularly those stemming from drug-drug interactions (DDIs), are a significant concern, with studies indicating that a substantial percentage of hospitalized patients encounter potentially harmful drug combinations. Computerized Provider Order Entry (CPOE) systems have evolved to include sophisticated drug safety software equipped with safeguards against incorrect dosing, duplicate therapies, and DDIs. These systems generate alerts that serve as a crucial form of decision support, widely disseminated across healthcare settings.
However, the effectiveness of DDI alerts is hampered by inconsistencies in their presentation, prioritization, and the underlying algorithms used to detect DDIs. Healthcare providers are often inundated with a barrage of alerts, many of which may be irrelevant, and a lack of standardization exists regarding the optimal implementation and prioritization of these alerts. To address this, organizations like the US Office of the National Coordinator for Health Information Technology have compiled lists of ‘high-priority’ DDIs to guide CDSS development. These lists have gained traction and are being adopted in CDSS across various countries.
Beyond medication safety, other patient safety systems, such as Electronic Drug Dispensing Systems (EDDS) and Bar-Code Point-of-Care (BPOC) medication administration systems, play a vital role. Often integrated to create a ‘closed-loop’ system, these technologies computerize and connect each step of the medication process, from prescription to administration. BPOC systems utilize RFID or barcodes to automatically verify medication identity against patient information and prescriptions at the point of administration. This integration with CDSS offers a powerful mechanism for preventing medication administration errors right at the patient’s bedside. While adoption rates are still growing due to technology and cost considerations, research demonstrates the significant efficacy of these systems in reducing medication errors. Many of these systems can be effectively combined with CPOE and CDSS to minimize prescribing errors related to drug allergies, excessive dosing, and unclear orders. It’s important to acknowledge that even with advanced technologies, errors can occur if healthcare providers bypass or circumvent these systems, highlighting the need for comprehensive implementation and user training.
CDSS also extends beyond medication-related safety, providing reminder systems for various critical medical events. For instance, CDSS designed for blood glucose monitoring in intensive care units (ICUs) have proven effective in reducing hypoglycemia incidents. These systems automatically prompt nurses to perform glucose measurements based on patient-specific protocols, considering demographics and glucose level trends.
Overall, CDSS integrated with CPOE and other healthcare technologies have demonstrated considerable success in enhancing patient safety by mitigating prescribing and dosing errors, providing automated warnings for contraindications, and improving drug-event monitoring. Patient safety is increasingly recognized as a fundamental objective of virtually all CDSS implementations, regardless of their primary function.
Streamlining Clinical Management with Decision Support Systems
Clinical guidelines and care pathways, while essential for standardized and evidence-based practice, often face challenges in implementation and clinician adherence. The traditional approach of disseminating guidelines with the expectation of automatic adoption has proven insufficient. However, CDSS offer a powerful solution by embedding these guidelines directly into the clinical workflow. The rules and protocols outlined in clinical guidelines can be encoded into CDSS, transforming them into actionable tools within electronic health records (EHRs).
These CDSS can take various forms, including standardized order sets for specific conditions, alerts triggered by patient-specific protocols, and reminders for necessary tests. Furthermore, CDSS facilitate the management of patients enrolled in research or treatment protocols, aiding in tracking, order placement, referral follow-up, and ensuring preventative care measures are implemented.
CDSS can also proactively identify patients who are non-adherent to management plans, require follow-up, or are eligible for research studies based on predefined criteria. A notable example is a CDSS implemented at the Cleveland Clinic, which provides point-of-care alerts to physicians when a patient’s record aligns with clinical trial criteria. This system streamlines recruitment by prompting users to complete eligibility forms, obtain consent, forward patient charts to study coordinators, and generate patient information sheets about relevant clinical trials.
Cost-Effective Healthcare through CDSS Implementation
The implementation of CDSS offers significant potential for cost containment within healthcare systems. This is achieved through various mechanisms, including optimizing clinical interventions, reducing inpatient lengths of stay, and utilizing CPOE-integrated systems to suggest more cost-effective medication alternatives. CDSS can also minimize redundant testing by flagging duplicate orders. For example, a CPOE rule implemented in a pediatric cardiovascular ICU limited the frequency of blood count, chemistry, and coagulation panels, resulting in substantial cost savings without compromising patient length of stay or mortality.
CDSS can also provide real-time information on drug formularies and insurance coverage, guiding clinicians towards cost-effective treatment options. In Germany, a hospital developed a drug-switch algorithm integrated into their CPOE system to address errors in drug substitutions when patients were switched to hospital formularies. This CDSS automated a significant portion of drug switch consultations with high accuracy and no errors, improving patient safety, reducing clinician workload, and lowering costs for healthcare providers.
Optimizing Administrative Functions with CDSS
CDSS extend their benefits to administrative tasks within healthcare, providing support for clinical and diagnostic coding, procedure and test ordering, and patient triage. Sophisticated algorithms can refine lists of diagnostic codes, assisting physicians in selecting the most appropriate codes for billing and data analysis. A CDSS designed to improve the accuracy of ICD coding in emergency departments utilized an anatomographical interface linked to ICD codes, enabling physicians to more efficiently and accurately identify diagnostic admission codes.
Furthermore, CDSS contribute to enhanced clinical documentation quality. An obstetric CDSS, featuring an enhanced prompting system, demonstrably improved the documentation of labor induction indications and estimated fetal weight. Accurate documentation is crucial for effective clinical protocols. For instance, a CDSS was implemented to ensure proper vaccination for patients post-splenectomy. However, it was discovered that a significant number of patients with splenectomy in their EHR lacked documentation on their problem list, which triggered the CDSS alert. A supplemental CDSS was subsequently developed to improve problem list documentation of splenectomy, thereby enhancing the effectiveness of the original vaccination CDSS.
Advancing Diagnostic Capabilities with Decision Support
Diagnostic decision support systems (DDSS) represent a specialized category of CDSS focused on clinical diagnosis. Historically, DDSS have functioned as computerized consultation tools, processing user inputs and outputting lists of potential diagnoses. However, the adoption of DDSS has been slower compared to other CDSS types, partly due to physician skepticism, accuracy limitations stemming from data gaps, and integration challenges requiring manual data entry. These limitations are being addressed through improved EHR integration and standardized terminologies like Snomed Clinical Terms.
A successful example of DDSS is a system developed for diagnosing peripheral neuropathy using fuzzy logic. This system, utilizing 24 input fields encompassing symptoms and test results, achieved a high degree of accuracy in identifying different types of neuropathies and normal cases. While valuable in settings with limited access to specialists, there is also a growing need for DDSS that can augment specialist diagnostics. DXplain, a reference-based DDSS, provides probable diagnoses based on clinical manifestations and has shown to improve diagnostic accuracy among family medicine residents in clinical trials.
Given the recognized prevalence of diagnostic errors, particularly in primary care settings, CDSS and related IT solutions hold immense promise for enhancing diagnostic accuracy and efficiency. The emergence of non-knowledge-based techniques like machine learning in diagnostic systems is paving the way for even more precise and reliable diagnoses. The Babylon AI powered Triage and Diagnostic System in the UK exemplifies the potential of AI in this domain, while also highlighting the ongoing development needed to realize the full potential of these systems.
Image-Enhanced Diagnostics: The Role of CDSS in Imaging
In the realm of medical imaging, knowledge-based imaging CDSS are predominantly used for image ordering. These systems assist radiologists in selecting the most appropriate imaging tests, provide reminders of best practice guidelines, and alert to contraindications, such as contrast agent allergies. An interventional CDS for image ordering at Virginia Mason Medical Center effectively reduced the utilization of lumbar MRI for low back pain, head MRI for headache, and sinus CT for sinusitis. This CDS required providers to answer a series of appropriateness questions prior to ordering, and importantly, suggested alternative imaging options when an order was denied. RadWise®, a commercial CDS, guides clinicians to the most relevant imaging order by analyzing patient symptoms against a comprehensive database of diagnoses and providing point-of-care recommendations.
The future of imaging diagnostics is increasingly intertwined with non-knowledge-based CDS and AI-driven precision radiology, also known as ‘radiomics’. With medical images constituting a growing portion of medical data, technologies that aid in extraction, visualization, and interpretation are crucial. AI technologies, particularly deep learning (DL), are proving capable of uncovering insights from image data that surpass human capabilities. Companies like IBM Watson Health and Google are at the forefront of developing AI-powered tools for tumor detection, medical image interpretation, diabetic retinopathy diagnosis, and Alzheimer’s diagnosis, among others. IBM Watson’s ‘Eyes of Watson’ exemplifies this trend, combining image recognition with text recognition for comprehensive decision support.
Several studies have demonstrated that AI performance in image analysis is approaching or even matching the level of human experts. For instance, Google’s deep learning algorithm for diabetic retinopathy detection achieved performance on par with board-certified ophthalmologists. Similarly, a Stanford study showcased a CNN for arrhythmia detection on electrocardiograms that surpassed the accuracy of the average cardiologist. While some experts speculate that computers may handle the majority of diagnostic imaging interpretation in the coming decades, for now, these AI-powered systems serve as valuable augmentations to clinicians’ toolsets, enhancing their diagnostic capabilities.
Laboratory and Pathology: CDSS for Enhanced Diagnostic Interpretation
CDSS applications extend to laboratory testing and interpretation, providing alerts for abnormal lab results, a common feature in EHR systems. Beyond simple alerts, CDSS can enhance the utility of lab tests, potentially reducing the need for riskier or more invasive diagnostic procedures. In Hepatitis B and C testing, AI models are being developed to combine multiple lab tests, imaging, and gene tests to achieve greater diagnostic accuracy compared to non-invasive lab tests alone. CDSS also serve as interpretation tools, particularly when test reference ranges are highly personalized based on factors like age, sex, or disease subtypes.
Pathology reports are pivotal in guiding decisions across medical specialties. CDSS are being utilized for automated tumor grading in pathology, achieving high accuracy in areas like urinary bladder tumor grading and brain tumor classification. Numerous other applications exist, including computerized ECG analysis, automated arterial blood gas interpretation, protein electrophoresis reports, and CDSS for blood cell counting, demonstrating the broad utility of decision support in laboratory and pathology diagnostics.
Empowering Patients: Patient-Facing Decision Support Systems
The rise of Personal Health Records (PHRs) marks a significant shift towards patient-centric care, with CDSS functionalities increasingly integrated into these platforms. CDS-supported PHRs are instrumental in facilitating shared decision-making between patients and providers, overcoming information barriers that may hinder patient engagement in their own care. PHRs, often extensions of EHR software or standalone web/mobile applications, can establish bidirectional information flow between patients and providers.
Early PHRs, like the “Patient Gateway,” primarily offered patients a view-only dashboard of medications and lab results, along with communication tools. Modern PHRs are evolving to allow patients to actively contribute to their health records, with patient-entered information impacting EHR data. Vanderbilt University’s MyHealthAtVanderbilt PHR exemplifies this integration, offering disease-specific educational materials and tools like a “Flu Tool” to guide patients with flu-like symptoms in seeking appropriate care. Symptom tracking and collection of diverse health data, from allergies to medication information, are common features of PHRs. Furthermore, PHRs and patient monitoring applications are increasingly designed to integrate data from wearable health devices, providing actionable insights for healthcare providers.
Diabetes care provides a compelling example of this integration. Systems utilizing wearable glucose monitors that transmit data to devices like Apple HealthKit and integrated with EHRs and PHRs like Epic MyChart are enabling providers to remotely monitor patient glucose trends, facilitating timely interventions and communication. Pilot studies have demonstrated improvements in provider workflow, patient communication, and overall quality of care through these systems. Similar integrated PHR/EHR, wearable technology, and CDSS systems are being deployed across various medical fields, including cardiology, hypertension management, sleep apnea treatment, and palliative care, signifying a growing trend towards technology-enabled, patient-centered healthcare.
The evolution of PHRs towards interactive tools with CDSS capabilities reflects a growing emphasis on shared decision-making and patient empowerment. PHRs that function solely as data repositories are increasingly seen as inadequate, particularly by patients who desire more active involvement in managing their health. The integration of decision-making support tools into patient-facing technologies represents a crucial step towards a more collaborative and effective healthcare paradigm.
