While our team at 978-bed St. Vincent's Hospital Manhattan managed to successfully install a new pharmacy information system under a brisk schedule, the more remarkable story is that we have a strategic vision for electronic health records and we are getting there one step at a time.
As a Leapfrog Group survey found last year, the proportion of healthcare organizations with computerized physician order entry, or CPOE, remains at 4%.
Costs and ongoing financial chal-lenges have limited the spread of CPOE, but we can leverage limited resources to build both the electronic health record and CPOE.
The old pharmacy information system at St. Vincent's was due to sunset at the end of 2003. As part of our overall information systems plan, we took the opportunity to install a new pharmacy system with clinical decision support. The new system would be the core component of medication management that could significantly reduce medication errors and enable CPOE in the future.
As director of medical informatics at St. Vincent Catholic Medical Centers--an eight-hospital system in the New York metropolitan area--I work full time with information systems. My job includes facilitating clinical software selection, development and maintenance; managing the corporate Internet and intranet sites; and acting as a physician liaison. Translated from geek-speak, I am the human two-way interface between clinicians and the IT department. The new pharmacy system project has been my responsibility.
With the executive leadership's sanction, a steering committee was created to make all high-level decisions. In early 2003, the committee assembled a multidisciplinary software selection team. The selection team included representatives of all the system's major stakeholders: phar-macy, physicians, information systems, nurses, supplies and performance improvement.
Drug-drug and drug-allergy interactions, dose-range check-ing and other clinical decision support tools were musts on our wish list.
Other selection criteria included functions required by government regulation or the Joint Commission on Accreditation of Healthcare Organizations; clinical note and intervention documentation; reporting; clinical decision support; longitudinal patient records; support for order hold/suspend/resume, i.e., stopping orders for patients who go to surgery; ability to require a complete patient profile before drug order approval; outpatient prescription dispensing; interoperability (with the patient registration system, the laboratory, hospital billing, McKesson Corp.'s drug-dispensing robot and ultimately with CPOE); electronic medication administration record; support; and advanced billing.
Of the four products that made the final cut, we chose Cerner Corp.'s PharmNet. In the end, interoperability trumped best of breed. With an integrated electronic health record as the ultimate goal, we saw the pharmacy project as one piece of a bigger puzzle.The pharmacy system we chose would have to work seamlessly with a clinical data repository, Web-enabled patient results reporting and CPOE.
Implementation planning started in June 2003 and the installation started in earnest in August. Four members of the team were temporarily assigned full time to the installment: one member from the pharmacy; one from the vendor; and two from information systems.
My principal assignment was dose-range checking, but I filled in where needed for drug database development (unit parameters and pricing), adverse drug effect rules and medication label alerts.
I configured dose-range checking to recognized guidelines for age and weight. We chose Clinical Pharmacology as our dosing reference.
As a physician in our information systems department, I was perfectly situated to collaborate with our colleagues in pharmacy. I was able to consult with one of the pharmacists in real time whenever there was a question about dose-range checking, unit pricing or alerts. I clocked so many hours in the pharmacy that I was made an honorary member of the department, no small feat for a physician.
We converted the pharmacy conference room into a home for the installation team. Network cables snaked their way between the workstations, laptops and printers. The cozy quarters facilitated close collaboration among the team and instant messaging permitted off-site team members to resolve issues quickly. Weekly conference calls kept the project's management on track, while monthly steering committee meetings addressed larger issues, like prioritizing concerns about "going live."
Though we survived a successful go-live last December, weekly conference calls continue to address new or outstanding questions. Short-term next steps include clinical documentation for pharmacist interventions, continued maintenance and development of clinical decision support, the electronic medication administration record and the future expansion of PharmNet to other hospitals in the system.
In a separate but related project, I am working as part of a team to map some 1,500 tests from our laboratory system to a physician's view of patient results reporting. Long-term, we plan to have bar-coding at point of care and CPOE.
Some of the lessons we learned that will be applied to the expansion to other hospitals in the system include:
- Adequate resources and staff from the vendor and client are critical.
- Installing a major clinical system requires a multidisciplinary effort. Medication management is a process that by its very nature involves physicians, pharmacists and nurses.
- Time devoted to training is time well spent and can provide added benefits. Training can serve as a venue for testing. During staff training, the staff helped refine order entry and decision development. Pharmacy managers and supervisors who served as trainers became the system's super-users, speeding system adoption and enabling "just-in-time" training after going live.
- A culmination of the above but worthy of mention by itself is that training the pharmacy staff is as important as development and testing. People may have to take a hiatus from the installation process to train the pharmacy staff appropriately.
- No matter how much time is allotted to it, drug label design will take longer than you think. One or two drafts will not be enough.
- All functions should be tested before going live. Testing must go beyond conventional unit and integrated testing. Clinical decision support and reporting should be adequately examined.
- Select and build to enable CPOE, even if it's not planned in the immediate future. Rebuilding or modifying interfaces can be painful, costly and sometimes dysfunctional. Measure twice and cut once works for IT as much as it does for home improvement.
- Share knowledge with other clients of your selected software. Other users are usually happy to offer advice that could minimize the angst of the first installation.
- Small working teams are more productive than large, cumbersome ones. The larger steering committee provided vision and direction. The smaller implementation team was empowered to problem-solve in real time. The real take-home lesson for the electronic health record is: "We can get there from here."
Kenneth Ong, M.D., is director of medical informatics at St. Vincent Catholic Medical Centers, New York. He is an assistant professor at New York Medical College; an active board member in the New York State chapter of HIMSS; and a former deputy commissioner at the New York City Department of Health. He is a fellow of the American College of Physicians, Infectious Disease Society of America, and the New York Academy of Medicine.