Developers pushing the boundaries of clinical decision support (CDS) services want the technology to do more and, curiously, less.

That is, more as in more capabilities, improved ease-of-use and greater sensitivity to a patient's specific clinical situation. And that's less as in fewer repetitive or unwanted nuggets of advice, where too much inbound information creates a condition decision support experts describe as "alert fatigue."

Health IT vendors, government-backed investigators and federal agencies are now addressing some of these trickier issues as new clinical decision support features have begun to surface, while other areas of research won't become mainstream for a few years.

EHR systems may provide decision support technology on their own or in conjunction with a third-party product or service.

The developments come at a time when federal directives, such as the government's incentive plan to spread electronic health records (EHR), encourage greater use of CDS. EHR systems may provide decision support technology on their own or in conjunction with a third-party product or service.

The Department of Health & Human Services will be looking for some degree of CDS capability as part of its meaningful use criteria for EHR adoption. While the requirement established for the current round of meaningful use will strike few as particularly taxing, demands on providers will likely escalate in the 2013- 15 timeframe, when the second stage of the incentive plan kicks in.

The Affordable Care Act, meanwhile, may also influence the uptake of CDS"at least tangentially. The healthcare reform law calls for the creation of accountable care organizations (ACOs) and supports the medical home concept, models of advanced care coordination that will rely on the notifications and database analytics promised by CDS technologies.

In their efforts to qualify for the federal health IT incentives, physicians and hospitals might be tempted to deploy just enough decision support to check off a compliance box. But health IT officials believe providers have an opportunity to do much more than that"namely, improve their level of care while keeping expenses under control.

"If organizations are going to become accountable for cost and quality outcomes, perhaps the most sustainable and scalable way of doing that is by properly designing their decision support," said Dr. Scott Weingarten, chief executive officer of Zynx Health, a decision support vendor.

Better rule coding
CDS has been around for decades. Early on, it was mostly about flagging harmful drug-drug and drug-allergy interactions, but over time the technology began providing clinicians prompts for age- and sexrelated services such as mammograms and colon screenings.

"That was clinical decision support for about the first 10-plus years of electronic health records," said Dr. Sarah Corley, chief medical officer at NextGen Healthcare Information Systems, an EHR vendor.

In the past five years, however, the healthcare field has extended the capabilities of decision support. For starters, EHRs have extended decision support to include not only drug-drug and drugallergy interaction, but also drug-food, drug-lab, and drug-disease conflicts. In addition, suggestions for treatment have become more commonplace.

"We now see vendors including more information on recommended care for chronic diseases," Corely said.

Despite those developments, industry and government are working to advance the state-of-the-art further. In one case, a research consortium funded through the Agency for Healthcare Research and Quality (AHRQ) is focusing on making CDS easier to deploy in EHRs. A key challenge: turning evidence-based guidelines"originating as free text" into machine-readable code an EHR system can readily crunch.

"A lot of physicians don't know how to translate a guideline into a rule or set of rules in EHR," said Dr. Blackford Middleton, corporate director clinical informatics R&D at Partners Healthcare and assistant professor at Harvard Medical School. He estimates that as few as a tenth of 1 percent of all practicing physicians can tackle the translation task.

Middleton is the principal investigator with the consortium, which also includes representatives from Brigham and Women's Hospital, the Regenstreif Institute and the Veterans Health Administration. The effort focuses on a knowledge management portal and a library of rules teased out of clinical guidelines. The concept is to let providers remotely subscribe to a set of rules centrally maintained in the knowledge repository.

Here's how the system works: Once a provider subscribes to a set of rules, the provider's EHR system creates a Continuity of Care Document for a given patient and sends de-identified data to the consortium's CDS engine. The engine then compares new data on clinical interventions on the patient against the rules, generates recommendations, and provides guidance back to the provider's EHR.

Middleton isn't the only health IT official focused on decision support rule-making. Dr. Steven Brown, director of Health & Medical Informatics at the Department of Veterans Affairs, called the need to generate coded data from free-text health guidelines a priority. "Having coded information is pretty high on the list" of CDS improvements, Brown noted.

The Systematized Nomenclature of Medicine"Clinical Terms (SNOMEDCT) is now sufficiently robust that a domain ontology may be applied to free text, Brown said. The VA now conducts research in this field: Brown pointed to an ongoing study by the Consortium for Healthcare Informatics Research that explores methods for extracting meaning out of text. CHIR pulls together investigators from multiple VA sites.

More speed, less noise
Health IT developers also point to greater speed and specificity as important objectives in furthering decision support.

In addition to taking the translation task out of the hands of providers, the Partners consortium's CDS project would create a centrally maintained rule set that, over time, could become a nationwide resource and one that could be quickly updated.

NextGen's Corely, who works with the consortium, said the goal is to provide a system that keeps pace with rapidly evolving medical research. "For those quickly changing guidelines, we will have a mechanism to provide clinical decision support to physicians at the point of care," she said.

The system's use of patient-specific data also takes a step toward individualizing recommendations, which Middleton identified as an important objective. A generic rule on diabetes may not apply to a particular diabetic patient, given variables such as co-morbidities.

"I believe we need to make clinical decision support more sensitive to the context of the patient," Middleton said. "Right now we are using a very coarse instrument, a very dull blade, frankly."

Some decision support developers, however, contend their technologies already offer a level of customization sufficient to navigate the nuances of individual patients. Weingarten said Zynz Health recognizes that most patients in hospitals example, and uses "branching logic" that accounts for patients with co-morbidities.

At the Veterans Affairs Department, decision support is built into its computerized patient record system (CPRS), which provides a cover sheet on each patient detailing medical problems, allergies, medications and other health data. When a caregiver opens the cover sheet, a list of reminders"typically 30 to 70 separate reminders"is evaluated for the patient; the provider sees see only those that are due.

Reminders generated by the clinical reminder system are patient specific. They include general health reminders, suc