The ultimate KM challenge: Healthcare informaticists work toward meaningful alerts at the point of care.
"The effect on inappropriate care is not there," he says. "It sounds good, but we haven't figured out how to make it plug and play yet. But if I were in charge of CMS, I would definitely be pushing this too."
Some research studies have shown that clinicians override the majority of medication alerts in computerized physician order entry (CPOE) applications. Hospitals are rapidly implementing CPOE, in part because it is seen as a great patient safety advancement and in part because it is a requirement for federal incentive funding for "meaningful use" of EHRs. But hospitals are finding that "alert fatigue" is a serious problem if clinicians don't find the alerts useful.
If 80 or 90 percent of alerts are being overridden, it gets to the point that clinicians don't even see them anymore, Van Kooy says. "And it makes them believe the system is stupid and hurts the credibility of the effort," he adds.
Patient safety officials are having tough discussions with care teams about where to place the alerts and how they should be developed. "I have seen cases where the physicians turn off most of the alerts, but the pharmacy does not turn them off," Van Kooy says. "The pattern is to put only the most meaningful alerts in front of the physician." The pharmacists aren't necessarily thrilled with that solution, he adds.
The question is where?
Where alerts fit into clinicians' workflow is another challenge. The medication alerts are pretty easy to get in the right place and time: when a doctor has selected a medication and is about to sign the order. But where and when to put other messages is much harder.
Van Kooy asks, "When do you want to have hard-stop messages to keep someone from doing something? Where do you put the message in the diagnosis process? Where do you remind a physician that the patient has rising creatine levels or that platelet levels are dropping?"
CMS' goal, he adds, is to get the adoption rates up and demonstrate that CDS modules do work and then fine-tune them. "For a nurse, it only takes one alert that actually stops them from administering the wrong medication to make them true believers," Van Kooy says. "I have seen someone crying because they were so grateful."
Clinical decision support and public health
Clinical decision support shows great potential in helping doctors and nurses care for individual patients, but it may also play a role in public health reporting efforts. A division of the Centers for Disease Control and Prevention (CDC) has launched a pilot project to begin work on an interoperable, flexible and open architecture for decision support modules in electronic health records (EHR) systems that can serve public health purposes.
The collaboration between CDC, GE Healthcare, the Chicago Department of Public Health and the Alliance of Chicago community health centers "is to see if we can provide public health information at the point of care," explains Ninad Mishra, M.D., CDC clinical decision support lead in the Division of Informatics Policy, Practice and Coordination.
Rolled out May 1 with 20 community health center clinicians in Chicago, the initial study is limited in scope to food-borne illnesses. A patient who shows up at a clinic with abdominal pain or diarrhea would not lead to an alert most of the time. But if there is a known outbreak of a food-borne illness in the area, that would allow the public health agency to activate a warning. Then, when the clinician goes to the next screen to do an order, he or she would see a flag warning about the outbreak.
"Traditionally we send public health notices and alerts via e-mail or snail mail," Mishra says, "but this would allow us to use clinical decision support based on the patient's presentation. That would be much more valuable."
The knowledge repository for the pilot project is Web-based and located at CDC. "If we scale it up, there would have to be several multijurisdictional levels to the knowledgebase, including federal, state and local," Mishra adds.
If the pilot project is deemed successful, many areas of public health have potential for expansion. For instance, if a patient is at high risk of influenza complications and has not been vaccinated, a warning could come up.
Sharing CDS best practices (and content)
One of the challenges of wider implementation of clinical decision support is the ability to share data across health systems to learn about best practices. That is the role of the Clinical Decision Support (CDS) Consortium, which is working to improve the translation of knowledge in clinical practice guidelines into actionable clinical decision support.
Funded since 2008 by grants from the Agency for Healthcare Research and Quality (AHRQ), the CDS Consortium includes researchers from several institutions, including Brigham and Women's Hospital, Harvard Medical School and Partners HealthCare Information Systems, along with the Regenstrief Institute, the Veterans Health Administration, The University of Texas Health Science Center at Houston, Oregon Health Sciences University, Kaiser Permanente, Mayo Clinic, NextGen, Siemens Medical Solutions, GE Healthcare, Duodecim Medical Publications, Philips, Wolters Kluwer Health, Accenture, AT&T, Geisinger, MITRE and the University of Utah Health Sciences Center.
Those organizations are working to create and provide CDS tools and services in electronic health records used in both academic settings and community-based physician office practices.
Speaking at last year's American Medical Informatics Association meeting in Washington, D.C., Tonya Hongsermeier, M.D., principal informatician in the Clinical Informatics Research and Development group at Partners HealthCare (partners.org/cird), said the consortium grappled with the need to create a safe approach to sharing CDS content, while ensuring that intellectual property rights of contributing parties are respected. The consortium drafted legal language for both publishers and users that takes into consideration attribution and rights granted by the publisher to users, and warrants that the publisher fully owns the CDS artifacts submitted. Basically, she said, content contributors retain authorship, users may make derivative works with appropriate attribution and artifacts may not be resold.