Clinical Decision Support: What It Is and how to use It

Clinical Decision Support (CDS) is an important element in improving health care delivery. Given the dramatic variation in health care costs from one locale to another (the Dartmouth Atlas experience), prompting rank-and-file physicians with standard-of-care guidelines (one way of implementing CDS), at the point of care, will go a long way to normalizing how health care is delivered across the country.

CDS is also a Meaningful Use measure, and integration of CDS into Electronic Health Records (EHRs) is a Certification requirement. There are many organizations who offer CDS systems, and when these modules are linked in to other “core” EHR systems (particularly in hospitals, where collections of separate technologies are often combined together in order to achieve Meaningful Use EHR Incentive payments), then use of CDS can be demonstrated.

As we build our own CDS capabilities, and engage in discussion with multiple other vendors to understand the various approaches to this issue, it has become apparent that CDS means different things to different people. It would be helpful to take a step back and look at how CDS can be integrated into physician workflow.

What is CDS?
Clinical Decision Support can take many forms. The simplest form it can take is alerting a physician about drug-drug and drug-allergy interactions at the time of creating prescriptions. Though simple, this kind of alerting can be very important and can have a significant impact on quality of care.

CDS can also be a prompting dashboard, tightly linked with Clinical Quality Measures (CQM) – when Mrs. Hernandez comes in for a visit for an acute bronchitis, the clinician is prompted that she is a diabetic and that her most recent glycohemoglobin (HbA1c) was performed 10 months ago and was 8.3%. With display of that fact (a CQM-derived finding), and a visual prompt that she has poorly controlled diabetes and is past-due for follow up (showing what the targets and standards of care are for this situation) – this would be one way of implementing CDS.

Additionally, the availability of resources for the physician – reference items that the physician can go search (e.g. UpToDate), or interactive libraries with quick dashboards (e.g. the Cochrane Library), or even more interactive real-time probability displays of likely diagnoses, given presenting symptoms (e.g. Evidence-Based Solutions) – are all different ways of offering more review-based decision support that can be quite valuable.

Integrating CDS into workflows
The different kinds of CDS have different places where they are useful. When designing a CDS system, it is important to appreciate the physician workflows. From this understanding, one can look at CDS as in-line CDS and review-oriented CDS.

Imagine this typical scenario: the physician is in an exam room with a patient, using an EHR deployed on a wireless notebook computer. It’s a busy day – there are 2 other patients waiting to be seen in adjacent rooms, becoming impatient. The patient in the next room has just opened the door, and asks the medical assistant, “has the doctor forgotten about me? When am I to be seen?” The physician doesn’t want to do anything that slows her down. How can CDS work in this kind of environment?

In-line CDS
Some kinds of prompting and alerts are appropriate to deploy during the seeing-the-patient-right-now workflow. These should not be items that interrupt the workflow (or, at least, can be dealt with quickly). These kinds of CDS elements would be drug-drug and drug-allergy interaction alerting if new medications are added or prescriptions created. Display of CQM-based prompts in a quickly-viewable dashboard (“Mrs. Hernandez is a diabetic not at target with control,” or “Mrs. Smith is past due for her mammogram”) would also fit into this workflow. Such prompts should be on a patient-specific dashboard, so that their information is visible from within that patient’s chart.

Review-oriented CDS
Additional resources may be helpful to the physician after the patient interaction has taken place – after clinic, even from home, later. This would be more along the lines of a physician stating “I saw Mr. Gonzalez today with a condition I want to know more about – what is the standard of care here? What is the evidence-based best-practice way of proceeding? Are there any new tests or medications that would be good to learn about?”

Ideally, during the in-line process (seeing the patient in the exam room, with time pressure present), it would be nice to allow the physician to bookmark the chart in a way that can be reviewed later. Then, when the pace-of-clinic is relieved, the bookmarks can be reviewed in order that review-oriented CDS can be accessed. This might include literature searches, second-opinion gathering (e.g. via a physician forum), review of PQRI or other items that would need to be addressed (and can be triggered by after-the-visit messaging to other members of the clinical team, such as medical assistants or nurses). This might even be a place where Continuing Medical Education (CME) activities can be brought in.

Conclusions
Clinical Decision Support is a broad conceptual category that includes many different ways in which best-practice recommendations can be implemented at the point of care. It is helpful to distinguish between in-line CDS and review-oriented CDS, as the kinds of support appropriate in each setting is different. One wants decision support to be something useful, valued and sought-after, and not an impediment to the time-pressured workflow faced in clinical practice. With good design, robust CDS can be built into EHRs in ways that support both kinds of needs.

Robert Rowley, MD
Chief Medical Officer
Practice Fusion EMR