CDS systems seek "real life" data | Behavioral Healthcare Executive Skip to content Skip to navigation

CDS systems seek "real life" data

November 1, 2010
by Dennis Grantham, Editor-in-Chief
| Reprints
Clinical decision support systems developers say vast data warehouses are needed for individualized treatment

Everyone is familiar with evidence based protocols (EBPs), the research-based guidelines or procedures used by clinicians to select and guide the course of individual treatment. EBPs originated to offer reliable guidance to clinicians, without the need to track ever-growing amounts of research data and patient findings. Clinical decision support (CDS) systems provide a means of applying guidance like this more globally and systemically, ideally resulting in more consistent, high-quality care. CDS functionality built into many EHR systems typically enables organizations to select and install knowledge bases (i.e., drug allergies/interactions) as well as “rule sets” (vetted treatment protocols, local treatment standards, etc.).

In operation, CDS systems process physician or clinician inputs and patient data according to rule sets and knowledge bases installed in the EHR system. Then, they communicate information that, according to the rules, is designated as important to clinical decision making. CDS functionality is very versatile and flexible. Organizations have found that they can develop rule sets for many purposes. For example, CDS can serve practical and administrative needs with rules that help ensure all data, assessments, and documentation required by a payer for a specific patient diagnosis are completed at the time a claim is filed.

Clinical leaders can modify decision rules to ensure that new assessment criteria or treatment steps are used for all new patients with a particular diagnosis. For prescribers, it means e-prescriptions will automatically be checked for possible contraindications, allergies, or interactions. To an authorizations team, it might mean creating reminders to clinicians to complete evaluations needed to support treatment authorization or reauthorization. Of course, the “holy grail” of CDS is about improving the quality of patient care by providing physicians and clinical personnel with more personalized, symptom-specific diagnostic and treatment information.

Ideally, CDS systems are seen as critical components in creating EHR-based “learning systems” that continually incorporate clinical advances, apply them to ever-growing and more detailed patient databases, and increase clinicians’ abilities to predict-based on patient data and demographics-which of a range of interventions will offer the best outcome.


While many medical provider groups have made progress in developing CDS rule sets that improve care quality for patients, such definitive rules for behavioral healthcare have proven far more difficult to develop. The reason, simply put, is a lack of data to stand behind them.

Dennis morrison, phd, ceo of centerstone research institute

Dennis Morrison, PhD, CEO of Centerstone Research Institute

According to Dennis Morrison, CEO of Centerstone Research Institute, the behavioral healthcare field lacks the high-quality quantitative testing and measurement data available to general medicine. This makes it very difficult to develop stronger rule sets, what he calls “the hard part” of a better CDS system for behavioral healthcare.
At present, Morrison says that most organizations are using the next-best tools available, typically vetted treatment protocols and guidelines developed by research collaboratives. These provide good general guidance for the treatment process, but fall short of the CDS ideal of helping clinicians provide effective, individualized treatment.

Deo Garlock, Director of Behavioral Health Informatics at Duke University Health System (Durham, N.C.), agrees. He says that current best practices and treatment protocols are based on data that are sometimes decades old, generated in studies that serve too many widely varying objectives-everything from clinical trials to public health.

Deo garlock, director of behavioral health informatics at duke university health system

Deo Garlock, Director of Behavioral Health Informatics at Duke University Health System

“The data for treatment guidelines are generally too old and generic to effectively apply to an individual patient," he explains. "Determining probability of treatment outcomes is difficult and requires a lot of data and sophisticated statistical analyses.”