This past October the Healthcare Information and Management Systems Society announced that the Center for Behavioral Health (CBH) in Bloomington, Indiana, won a Nicholas E. Davies Award for excellence in implementation of electronic health records—only the second behavioral health provider organization to do so. Our journey to the EHR has been documented in excruciating detail on HIMSS’ Web site. While EHR implementation is difficult under any circumstances, the “holy grail” of such implementations is the development of a clinical decision support (CDS) system.
A CDS system uses data from the EHR to generate patient/client-specific advice for the clinician. CDS systems are well-established in medical-surgical EHRs partly because medical practitioners have an abundance of quantifiable data available (e.g., measurements and lab values). Behavioral healthcare does not use quantifiable indicators as much as general medical care, instead relying on textual descriptions of services that do not lend themselves to CDS systems.
Despite this obstacle, we were determined to develop a CDS system. We identified two main problems we wanted the system to address. First, our experience has been that the clinical record often does not show a clear link between (1) an assessed need (2) related to a diagnosable condition and (3) possible outcomes through (4) appropriate interventions. The second problem is “clinical drift—i.e., when a clinician addresses needs not identified in the client's approved treatment plan. These services may be clinically helpful or necessary but usually are not defensible in audits. We believed we could improve the quality of care and mitigate financial audit problems if we could help clinicians focus their documentation on the services approved in the treatment plan.
We addressed these problems by developing a new kind of treatment plan and progress note that integrates into our existing EHR. We call our CDS system “PsychRemix.” Clinicians just aren’t asked to complete computer-prompted tasks; the CDS system assists their decision making.
For example, the CDS system has a library of outcomes based on internal practice patterns and industry best practices, all including behavioral descriptors to measure outcomes. Based on the data the clinician enters, the CDS system selects a list of the client's most likely outcomes based on his/her functional level. The clinician can choose one of the listed outcomes, expand the list to choose from a global list of outcomes, or create a new one. Groups of commonly associated outcomes are combined into outcome groups.
For each outcome the clinician chooses, he/she must select one or more interventions from a library of CBH's treatment interventions to achieve the desired result. A sophisticated algorithm uses the client's age, diagnostic-related group, and functional level to generate the list. The clinician can choose one or more of the system-selected interventions, but he/she may not define one because all of the relevant interventions CBH offers are listed, including more than 20 evidence-based treatments. The intervention and desired outcome are selected with the client's participation.
The progress note, which is integrated with the treatment plan, uses similar CDS functionality. To complete a progress note and bill for a service, the clinician must select which problems the treatment session addressed from a list that displays impairments related to the service. For example, if the clinician observed that the client was taking his medications, the CDS system would not allow the clinician to choose “grooming” as the problem addressed by this billable service, but the clinician could select “medication noncompliance” as the problem the session addressed. Only problems and progress on outcomes identified in the current treatment plan can be entered into the progress note. However, newly identified problems can be easily added to the treatment plan.
In the progress note, the clinician must rate the client's progress on each problem addressed in a session using a seven-point scale ranging from “complete” to “much worse.” Those rated “much worse” trigger an alert asking if the change in status warrants a suicide or homicide risk assessment. If so, the CDS system asks for a disposition or plan. If the client is not at risk, the clinician selects “no” and completes the progress note. Our goal is not to dictate practice but to remind clinicians to assess for risk when clients’ progress toward goals takes a turn for the worse in any area of their lives.
To assist the clinician with writing the note, the CDS system presents context-appropriate hot-linked keywords specific to the particular service. These words are verbs that auditors and consultants have told us best describe the action that takes place in the type of service. For example, billing life skills training would present verbs such as “aided,” “coached,” and “trained,” while billing for medication monitoring might present such words as “observed” or “monitored.” When a clinician clicks on a word, it automatically is inserted into the text field that describes the billable service, and the clinician can type the remainder of the sentence around that word. Progress notes now require very little typing compared to the old free-form text blocks, yet the notes are not cookie-cutter in their appearance or content.
Our CDS system offers much more functionality, but this limited explanation at least demonstrates how the system substantially enhances the documentation trail. In fact, in a 1999 Medicaid audit, before we implemented the EHR, missing documentation cost us $550,000 in penalties; in a 2004 audit, after the EHR and CDS system were implemented, we had no financial penalties for missing documentation.