Computerized clinical decision support (CDS) systems have the potential to improve patient outcomes through their impact on clinical decision making. To assess whether CDS systems can improve quality metrics, researchers at Beth Israel Deaconess Medical Center in Boston conducted an analysis of the impact of UpToDate, an online CDS system from Wolters Kluwer Health, on outcomes of care. Beth Israel Deaconess Medical Center’s CIO and CMIO John D. Halamka, MD, sat down with CMIO to explicate the study results, and recount the provider’s daily experiences with clinical decision support systems.
The data conundrum
"In today's clinical practice, physicians are challenged by the overwhelming amount of data," says Halamka. Typically, physicians have approximately 12 minutes to see a patient and document all the data. During that time, the cloud can bring in data from the health information exchange, and the institution needs to ensure that physicians are properly documenting for meaningful use, ICD-10, new forms of reimbursement and Physician Quality Reporting System (PQRS), he points out.
"Based on the data and metrics that physicians are being measured on today, it's been impossible to practice without having that information in front of them," Halamka says. "We have tried to distill all these data in the EHR into knowledge and wisdom, and all that requires clinical decision support rules."
To make these data actionable and not overwhelming for providers, healthcare institutions often customize the technology. For instance, Beth Israel has approximately 2,000 customized decision support rules, and Brigham & Women's Hospital—which is 50 feet away—has about 2,000 completely unique rules. "Ideally, a cloud-based decision support service could offer a few elements of customization, and if providers at Brigham or Beth Israel queried the cloud service, they should probably get the same answer," Halamka says.
Many rules may result in many alerts. To combat this concern, Halamka stresses that alerts need to be "so rare that when one fires, a physician knows that he or she needs to act on it. You have to be very careful about alert fatigue," he says, adding that the 2014 meaningful use standards are more focused on reducing alert fatigue than on customization.
When selecting which metrics a provider chooses to track, Halamka says it's important to first select those that are government-mandated, which are typically areas of focus by PQRS or the Centers for Medicare & Medicaid Services (CMS), as they are most commonly computed and therefore, easiest to measure.
Impact on outcomes
Alerts and reminders only represent subsets of decision support interventions. The capability to retrieve answers to detailed clinical questions with actionable, evidence-based recommendations represents another form of decision support. Several studies have demonstrated that clinicians have many questions at the point of care—and most of these questions go unanswered. Recent research has shown that answering those questions can have a measurable and clinically important impact on the quality and efficiency of care.
One such study, conducted by Thomas Isaac, MD, MBA, MPH, of the division of general medicine and primary care and his colleagues at Beth Israel, examined the relationship between the adoption of UpToDate and three main outcomes: risk-adjusted length of stay; risk-adjusted mortality; and performance on standard quality process metrics in the period from 2004 to 2006. The study was published in the February issue of the Journal of Hospital Medicine.
"Using UpToDate as evidence-based medicine, we can assist physicians with how to best treat diabetics with hypertension, for example," says Halamka. "With this type of trusted resource at the point of care, we should achieve better outcomes, lower cost, higher quality and better safety. UpToDate is such a trusted resource at Beth Israel Deaconess."
The study sought to answer the question: "Will evidence-based medicine at the point of care improve quality, safety and efficiency of care? And, it did," Halamka says.
In the study, Isaac and colleagues examined six common medical and surgical conditions: acute myocardial infarction, congestive heart failure, pneumonia, gastrointestinal hemorrhage, stroke and hip fracture. To assess hospital performance, they employed the Hospital Quality Alliance (HQA) process measures. Overall, they found that hospitals that had adopted UpToDate achieved significantly