The Power of Better Clinical Decision-Making: Driving Data & Best Practices to the Point of Care

Swift, optimum clinical decision-making depends on healthcare systems delivering the right information at the right time to the right caregiver. Model systems across the country are exploiting a host of technologies such as computerized physician order entry (CPOE), clinical decision support (CDS) and electronic medical records (EMRs), to ensure this better than anyone else while increasing quality, disseminating best practices and decreasing costs. To create comprehensive, intuitive systems, several of the country's most proactive health systems have formed extensive partnerships with health IT vendors to benefit from the vendor's connectivity and software system development knowledge base.

The benefits of clinical IT are phenomenal. Consider:

• A typical 300-bed hospital can save more than $10 million by deploying a today's standard EMR, says Marc Probst, CIO of Intermountain Healthcare in Salt Lake City. The estimate swells to an impressive $14.7 million with a Stage 4 EMR. The Stage 4 EMR is a concept that includes knowledge sharing and advanced decision-support capabilities.*
• University of Pittsburgh Medical Center Presbyterian Hospital (UPMC of Pittsburgh) decreased its rate of medication errors by 55 percent since deploying a positive patient identification (PPID) system to match patients to their medications.
• Duke University Health System in Durham, N.C., reports that its physicians order the appropriate initial tests for 98 percent of patients as part of the initial presentation since it deployed CPOE several years ago. Such solutions are a tall order, even for health IT giants. IT changes, and improves, healthcare across the board; and genuine change in the practice of medicine is seldom easy. "The goal isn't simply to get physicians to use computers; it's to provide them with easy-to-use technology that supports the way they think, automates routine processes and helps them deliver better, more efficient care. That's how you build a system that physicians embrace," shares Michael Russell, MD, associate CIO at Duke University Health System.
• Results at Partners HealthCare in Boston of a decade-long EMR rollout to 3,200 physicians and CPOE deployment at all 11 hospitals include a 55 percent reduction in serious medication errors on the inpatient side and identification of 120 critical and 600 subcritical results requiring follow-up in the average outpatient month.

  *Note: Intermountain Healthcare uses its own EMR adoption stage model, not a HIMSS Analytics model.

Intermountain Healthcare: Moving to a next-generation enterprise clinical information system

Intermountain Healthcare is an EMR pioneer. Comprised of 22 hospitals, the healthcare giant developed HELP, a first-generation EMR, decades ago. The electronic system evolved as healthcare evolved, but after the turn of the 21st century, Intermountain Healthcare had reached a crossroads. HELP and the second-generation HELP2 no longer met the clinical needs of the organization. The health system wanted to expand on its intellectual property and build an enterprise clinical information system (ECIS) that delivers clinical data and advanced decision support.

In 2005, the pioneer partnered with GE Healthcare to develop the system. The 10-year, $10 million partnership is designed to develop an ECIS built on GE Centricity Enterprise Solution. It weds Intermountain Healthcare's rich clinical informatics history with the vendor's broad clinical software suite.

Intermountain Healthcare: Value of EMRs for a Typical 300-bed Hospital

Advanced decision support takes center stage in the team's product development process. "We live off of advanced decision support and data analytics. When we started to prioritize which factors are most important in an ECIS, we ranked advanced decision support and accurate reporting very high. These components, rather than CPOE and data input, are what drove our initial success at Intermountain Healthcare," explains Probst. Basic decision support consists of alerts that consider a few variables to warn caregivers about potential adverse drug events. Advanced decision support, on the other hand, takes into account hundreds of variables to guide optimal patient care. Each advanced protocol takes multiple man-years to develop, says Probst. The benefits, however, are hefty.

Consider for example Intermountain Healthcare's glucose protocol. According to the Journal of American Medical Association, hospitals that successfully manage the glucose of the sickest ICU patients to a very narrow band realize better patient outcomes. Management hinges on hundreds of time and event-driven variables. After years of work, Intermountain Healthcare successfully harnessed the variables to develop its glucose protocol. The number of patients with glucose values in the optimal range after coronary artery bypass surgery (CABG) increased steadily from 28.6 percent in the third quarter of 2002 to 70.8 percent just three years later in the first quarter of quarter 2005-which may help reduce the health system's already low rates of morbidity and mortality associated with CABG. In addition, the availability of IHC's enterprise data warehouse lab information helps researchers conduct in-depth quality assurance and improvement activities.

The health system employs more than 100 advanced protocols that improve patient care. Its acute respiratory distress syndrome (ARDS) protocol has slashed ARDS fatality from 90 percent 10 years ago to 60 percent. "It not only decreased mortality, but also increased quality and reduced costs," shares Probst.

Such success does not come easily. It requires managing data in a very discrete fashion and ultimately drove the decision to build, rather than select, a solution. Most vendors offer solid capabilities in terms of basic EMR elements such as automation, connectivity and physician order entry, says Probst. The higher order data analytics that drive advanced decision support; however, remain a works-in-progress. GE shared the same vision as Intermountain Healthcare, making it a perfect fit for the pioneer.

Intermountain Healthcare rolled out the first software based on the partnership in a single hospital ER late in 2008. The state-of-the-art system incorporates a multi-screen composite user interface, standardized terminology and robust, built-in decision support. "We're proving the technology. Physicians do like composite user interfaces," says Probst. One universal dislike is workflow disruptions. The health system strives to minimize workflow disruptions, honing in on essential information.

Patient Data: By the numbers

UPMC: On the eRecord

UPMC is a health IT giant. Comprised of 20 hospitals, UPMC employs 2,500 physicians, with another 2,500 physicians affiliated with the enterprise. In the last five years, UPMC has invested more than $1 billion in health IT, including a $402 million deal with IBM to overhaul its IT infrastructure. "UPMC has more beds in HIMSS Analytics than anyone else," says CMIO G. Daniel Martich, MD. In fact, two UPMC hospitals have attained Stage 6 HIMSS ranking of EHR progress.

UPMC has staged a multi-front EHR project with CPOE and decision support incorporated as separate modules in the eRecord. "Every component-ER, ICU, OR-is necessary, so we're attacking all areas simultaneously, targeting the most critical needs first. For example, UPMC first deployed CPOE in Children's Hospital of Pittsburgh and expanded from there. CPOE adoption hovers near 90 percent in the four hospitals where its been deployed, says Martich. The CMIO attributes the project's successful adoption rate to physician involvement and leadership in the decision on which vendor or home-grown solution to select, as well as in development, training, implementation and post go-live support. The focus and message of why CPOE is being deployed-quality, safety and patient care-can be translated much more easily when physicians are involved from start to finish in the process, says Martich.

eRecord focuses on providing on-time information at the point of care or what UMPC dubs "automating perfection." The before and after differences in patient care are fairly dramatic. Consider for example patients with catheterized bladders, whose risk of infection increases the longer the catheter remains in place. Prior to the decision-support implementation, a patient might remain catheterized until a physician or nurse remembered to remove it. Now, an automated alert pops up 48 hours after insertion to remind caregivers about the risk of catheter-associated infection. eRecord sites report a 60 percent reduction in catheter-associated urinary tract infections. Other alerts focus on cumulative acetaminophen doses as it becomes toxic to the liver when a patient takes more than 4 grams daily. Because acetaminophen is a component of many medications, patients often receive more than 4 grams daily. With CPOE, an alert prompts the clinician to discontinue certain drugs or order others to avoid exceeding the 4 gram mark. The hospitals have realized a tremendous drop in the number of patients exceeding the 4 gram dose.

The other half of the UPMC environment consists of 500 ambulatory care sites. The national ambulatory EMR adoption rate hovers at 4 to 17 percent, but UPMC has achieved a 60 percent adoption rate and expects to roll out Epic Systems Corporation's EpicCare Ambulatory EMR software to all sites within 18 to 24 months. Universal adoption, however, is just one step in the journey.

UPMC aims to unite the inpatient and outpatient records to provide a standard patient view. Currently, physicians access the ambulatory record regardless of locale, but inpatient use is restricted by geographic zones.

The enterprise record is a tall order for a health system that uses 200 clinical applications. UPMC supports 650 interfaces, but the reality is that not every system can be interfaced to every other system. "It would be unscalable among 200 applications," explains Martich. So, UPMC's scalable solution rests on its interoperability or aggregator platform. UPMC's chosen aggregator is dbMotion. According to Martich, dbMotion has shown it can interoperate with multiple different vendor products. The initial plan taps into dbMotion to pull essential clinical data from both the major inpatient and outpatient applications of the enterprise to present physicians with a core set of patient data that includes medications, allergies, labs and immunizations. The capability drives improvement in the transfer of care between inpatient and outpatient settings, says Martich. For patients, it should produce such benefits as reducing unnecessary tests-like being re-stuck for bloodwork performed in separate venues.

Duke: Paperless patient care

Duke University Health System is a paperless pioneer. The health system, which is compromised of an academic medical center, two hospitals and a variety of ambulatory practices, launched its paperless initiative 12 years ago. It developed a clinical data repository to assimilate documents and results in a single location and weaned hospital and clinic-based physicians off the paper chart six years ago.

The paperless record, however, is a complex, multiphase process. After building the repository, Duke University Health System continued to add paper to stored records. In addition, in the interim stage, hand-written notes were scanned into the EMR. The health system accelerated the implementation process by assessing documents groups such as lab data, discharge notes and radiology results and determining that the EMR met each application's electronic signature and security requirements. After that determination, each application was declared paperless, recalls Russell.

In 2004, the health system embarked on its CPOE project, selecting McKesson's Horizon Expert Orders for the project. The software serves as the baseline technology for hospital-wide decision support at the point of care. CPOE use is universal at Duke, with all physicians and physician extenders using the system.

The project incorporates complex decision support. For example, the pediatric functionality evaluates patient's weight, age, gestational age and renal and liver function to determine appropriate medication dosages.

The system standardizes more than 1,000 order sets, which not only improves service and patient care but also better leverages resources. Consider for example a suspected myocardial infarction. The patient requires different numbers of cardiac enzyme tests-three creatine phosphorkinase (CPK) lab tests and two troponin tests. But physicians often order three CPKs and three troponin tests. "Now the order set is standardized, and the right number of tests is always ordered," shares Russell. In fact, the health system reports that 98 percent of patients receive the appropriate order set on initial presentation to the hospital.

The immense undertaking provided an opportunity to re-consider processes. The health system formed multi-disciplinary clinical groups consisting of clinical experts, nurses, pharmacists and other stakeholders as it created its order sets. As physicians proposed order sets, other stakeholders shared downstream implications. If a specific order created downstream challenges, the new, standard order set could be appropriately amended.

More recently, the project launched an ambulatory analytics program; Duke's academic informatics group created Chronic Disease Managers within the clinical data repository. The rules, which tackle specific diseases like type 1 diabetes, are based on national guidelines and internal expertise. The system automates data collection and provides an objective assessment across various groups. That is, each practice or provider can use the system to determine if it met specific clinical goals such as management of hemoglobin A1C.

Partners: From implementation to effective use

Partners HealthCare in Boston is one of the nation's largest integrated healthcare delivery systems. In addition to two academic medical centers, the Partners system includes community and specialty hospitals, community health centers, a physician network, home health and long-term care services and other health-related entities. Its physician network includes 6,000 physicians. By the end of this year, Partners will have completed the rollout of its EMR to 3,200 physicians and deployed CPOE at all 11 hospitals. "It's the end of an era," sums CIO John Glaser.

The conclusion of the implementation phase changes the focus to effectiveness of use. "We know," says Glaser, "that e-prescribing rates vary from 95 percent to 75 percent." Similar variability exists among complete problem lists. The next agenda for the healthcare giant is determining obstacles to full utilization of EMR/CPOE. The barriers range from technical to financial. In some cases, screen redesign or technical changes are required to boost physician acceptance. Other situations demand workflow adjustments. The final incentive comes on the financial front. Ten years ago, Partners adopted a pay-for performance program with a 10 percent incentive for technology adoption. Now, the system may shift the incentive to effective use, including e-prescribing and complete problem lists.

Patient data: The next stage

The U.S. will continue to evolve in the next decade with early and mid-tier adopters reaping benefits from pioneers' projects. Take for example Duke University Health System. As the academic powerhouse collaborates with McKesson, the vendor makes the content available to its other customers. UPMC and IBM created a joint-venture fund aimed at commercializing solutions that may benefit health providers beyond UPMC. "Several concepts are in the development pipeline," confirms Martich.

At Intermountain Health, Probst continues to think big, envisioning an eventual global transformation of healthcare enabled by Stage 4 EMRs. "The difference between current Stage 3 EMRs and Stage 4 systems centers on advanced decision support and data sharing," states Probst. Stage 4 records will provide the ability to share knowledge and protocols among institutions. For example, a Chinese hospital could digitally disseminate SARS best practices to educate and inform providers around the world. 

Next-generation clinical decision-making

Harnessing IT to disseminate patient data and best practices and improve clinical decision-making is a healthcare essential. "The process is relatively well understood. Success requires organizational leadership, workflow changes and good tech choices. But that doesn't mean it's easy. It's a long journey," advises Partners' Glaser. Benefits accrue and accelerate along the way, and they come in all forms-improved patient care, financial savings and greater efficiency. It is the medicine healthcare needs.