Clinical decision support systems: Promoting evidence-based medicine
Clinical Decision Support System (CDSS) assists clinicians at the point of care. This unique support system forms a significant part of the knowledge management technology via its capacity to support clinical processes from diagnosis and investigation through treatment and long-term care.
Clinical decision support systems (CDSS) provide clinicians and other care givers with information that is intelligently filtered and presented to enable improvement of treatment outcomes. The Institute of Medicine has long recognised problems with healthcare quality, and for more than a decade has advocated using electronic CDSS to improve quality. However, one must not forget that these health Information Technology (IT) applications are a means to improve healthcare quality, not an end in themselves. Electronic medical records (EMRs) with Computerised Provider Order Entry (CPOE) digitise clinical information. But there can be no major improvements in the quality of care from the use of health IT without proper implementation and use of CDSS.
Every year, more than a lakh death are reported in US due to medical errors, including drug interactions. For India, these numbers would be in several multiples of the US figures. Any of the several types of CDSS tools can prevent most of these mishaps. Examples include a pop up alert to potential drug interaction; clinical prediction rules to assess the risks of particular medication for a patient subset; clinical guidelines for treatment of diseases; or reminders for timely follow-up. EMRs lay the foundation for patient safety and healthcare quality improvement but CDSS is the tool that delivers these goals.
Early CDSS systems were derived from expert systems research. Engineers strived to create programs with rules that would allow it to ‘think’ like an expert clinician when confronted with a patient subset. Thereafter, these systems evolved so that they could be used to assist clinicians in decision across routine tasks, warning clinicians of potential problems and providing suggestions as per programmed rule sets.
Common features of CDSS systems that are the knowledge base (e.g., compiled clinical information related to diagnoses, various types of drug interactions, allergies and evidence-based guidelines), a smart program linking that knowledge with patient-specific information and a data bus for porting such information into the CDSS engine and convey relevant information outputs (e.g., lists of possible diagnoses, drug interaction alerts or preventive care reminders) back to the clinician.
CDSS can be built into the EMR or hosted as a standalone application available on the network, or on a mobile device, like a smart phone-based dosage calculator. Its knowledge base may be available centrally for local use as needed, or locally with constant upgrades. Practically, a deployed CDSS could utilise any of these computational architectures, methods of access or devices. The way these elements are constructed and deployed will depend on the type of clinical automation systems in place, local vendor solution availability, clinical workflow, security issues and funding constraints.
CDSS has the ability today to support physicians and clinical care givers at all key stages in the care delivery pathway, including preventive care, diagnosis, treatment, monitoring and follow-up. They can include order sets created for particular diseases or patient subsets, based on evidence-based guidelines and/or customised as per individual clinicians’ preferences, provide access to guidelines and other external clinical content that provides information relevant to particular patients/diseases, clinical reminders for preventive care and alerts that flag potentially dangerous clinical situations needing attention. Commonly, CDSS is used for addressing needs, such as assisting accurate diagnoses, screening for preventable diseases or averting adverse drug events.
A key feature of CDSS is the level of control the user has over the decision to use CDSS recommendations. Others include whether the CDSS recommendations are set up to be displayed on demand so that users have full choice to access it as also to accept it. These two control aspects are related and connect closely with how the CDSS advice matches a clinician’s intention. CDSS are designed to:
Remind clinicians of things they need to do, but should not have to remember
Provide information whenever care givers are unsure about the next course of action
Assist to correct errors clinicians may make and
Recommend that the clinicians change their plans.
The users’ reactions to CDSS may differ with these diverse intents. An analogy can be seen when a user employs the calendar functions on the PC, an alarm automatically pops up a reminder of something the user intends to do. This automatic notification is the most helpful feature of the calendar application. The spell checker in MS Word provides advice as well as corrects errors, doing so ‘automatically’ while one writes, or ‘on demand’ after one finishes and then accesses the function. Similar Word features make suggestions about changing what has been done. The grammar checker is often accessed on demand to correct grammatical errors. It can also suggest sentence revision, which may be ignored. Many users access the help function in MS Word. However, for most people the automatic appearance of the help wizard (like an automated decision support alert) creates ‘reminder fatigue,’ which leads to it being disabled. Automated alerts therefore should be few and relevant. Users are more amenable to on demand information.
CDSS challenges depend largely on how closely the system is tied to what the clinician already intends to do. Clinicians may initially want a set of clinical reminders, and after review assessments agree that they need some more. The timing of reminders is the key issue after the user has agreed to be reminded. Questions like ‘should reminders for preventive care pop up in advance of the patient visit or during the patient’s visit’? are debated. Other key user issues are speed and ease of access. Even though most clinical users recognise the need for information, they are willing to access it only if it can be accessed without wasting time or effort. Difficult or time-consuming access drives away potential users from CDSS.
The key issue in pointing out errors or suggesting that physicians change what they had planned to do for the patients is balancing the clinicians’ desire for autonomy with improving patient safety. A related issue revolves around how much control users have over the manner they respond to CDSS suggestions. This relates to whether clinicians are forced to accept the CDSS suggestion, the ease with which they can ignore it and the amount of effort needed to override the advice.
As EMRs join the mainstream, the base for truly achieving the advantages of electronically managed and recorded clinical workflow is put in place. There is really no meaning in creating electronic records without using the vast storehouse of digitally available clinical content to guide and enable truly evidence-based care. An oft quoted example in the West is the importance of prescribing Aspirin in the first 24 hours following Acute Myocardial Infarction (AMI). Hospitals switching from paper-based systems to EMRs with CDSS (in this particular instance with a drug order set for AMI) saw their Aspirin prescribing rates go up from 30 per cent to more than 90 per cent. The system, by suggesting what is clearly established in the evidence base, was just preventing physician oversight, an all too common phenomenon in a busy emergency room.
It is, however, relevant to note that for CDSS to achieve its stated purpose of improving patient outcomes, the system needs to be properly designed, carefully implemented and judiciously used. Systematic evaluations of processes of implementation, user satisfaction, and other factors that may affect the outcome of the intervention should be done by experts in the field. There is a need for qualitative evaluations of the physician-CDSS interaction and its impact on the clinician, the workflow, and other organisational processes and outcomes, in order to continuously refine the system and maximise its benefits. Several institutions in the West have treated CDSS as an out of the box system to deploy and use as-it-is and suffered from poor usage and negligible benefits in the long run. Others who created a system of constant feedbacks and reviews guiding system redesign and iterations, created a strong culture of CDSS usage leading to improved safety and clinical outcomes. Last but not the least, hospitals must beware of sketchy systems, which depend on non-validated content and poorly designed logic and interaction engines. These systems can actually cause clinical harm, and already have entered the realm of accepted causes of iatrogenic injuries.