Implementing CPOE--one pill doesn't cure all ills.
And even when implemented, a CPOE system does not guarantee an enhancement of care. That was the case when Guthrie Clinic, Ltd., a physician multispecialty group with over 20 regional offices, and Guthrie Healthcare System, a hospital organization with three acute care hospitals, partnered to create Guthrie Health, an integrated delivery system.
The system cares for patients in the region of Northeastern Pennsylvania and the Southern tier of Upstate New York and the management team wanted to connect the various care settings with one patient chart.
To accomplish this task, they selected an enterprise-wide electronic health record (EHR) that includes an inpatient CPOE system. Robert Packer Hospital (RPH), the health care system's tertiary, non-university teaching hospital, installed CPOE as the lead project of the EHR.
Although RPH does not have a closed medical staff policy, the clinic employs more than 90 percent of the physicians with hospital privileges. The mutual decision by the clinic CEO and hospital CEO to move forward with EHR followed a top-down approach.
After making the decision, they worked with vice presidents and department heads to convince all involved of the necessity of the EHR with CPOE. At the same time, Guthrie Health identified core values of patient-centered-ness, teamwork, and excellence. The EHR project related well to all three values.
Although some physicians initially pushed back against the mandatory CPOE, the majority accepted the requirement as a step toward a paperless system. Some of the strongest critics volunteered to serve on the physician oversight committee to help improve the CPOE. This committee met once or twice a month for the first two years and facilitated discussion between physicians, information services and hospital administration.
Medication errors continue
With regard to safety, the initial implementation of CPOE corrected the medical errors associated with physician order legibility. However, the hospital's pharmacy and therapeutics committee soon identified other sources of medication error.
Although the physician placed orders in a computer system, the nursing medication administration record (MAR) remained in a paper format and required the hospital floor unit clerk to transcribe medications from the computer onto a piece of paper.
Further, although the computer system allowed physicians to indicate a specific time for administering medications this information was not transferred to the nursing paperwork.
Another source of error arose from the interface between the vender of the CPOE and the pharmacy-information system from another vendor. The CPOE-pharmacy interface between different systems did not transfer all the information entered in CPOE fields by the physician because the pharmacy system did not offer the same data fields.
Finally, administration errors still occured at the patient bedside. A consultant examined our process and recommended that we proceed with implementation of the CPOE vendor's pharmacy system and electronic medical administration record (eMAR). (The original pharmacy information system license would soon expire and require either renewal from the old vendor or the purchase of a new system.)
In addition, the pharmacy system will be implemented with a bar coding system to be used at the time of administration. The consultant also recommended improving the decision support system within our CPOE.
In an attempt to promote CPOE acceptance, physicians originally had the opportunity to have all their preprinted orders converted into order sets. Although this made the transition from paper orders to CPOE less painful, fewer than 20 of the original hundreds of order sets were evidence-based. Evidence-based order sets will help standardize care and should improve quality.
The cardiology and pulmonary medicine specialty sections created order sets that will comply with Centers for Medicare and Medicaid Services core indicators for congestive heart failure and pneumonia respectively. The trauma surgery section used evidence-based order sets to promote protocols.
However, it appears that a third-party vendor of content will provide the most efficient mechanism for creating and maintaining evidence-based order sets for the entire organization.
After a vendor of content for the order sets is selected, each of the speciality sections within the hospital will evaluate the evidence-based order sets from the vendor and determine if any modifications will be needed. The vendor will provide periodic updates to reflect publications in medical journals that change standards of care.
Another problem detected within our CPOE system relates to the documentation of patient diagnosis. The evidence-based order sets are useless without proper diagnosis within the computer system's problem list to trigger the order sets and other best practice alerts.
At the current time our vendor's native problem list uses the ICD-9 listing as vocabulary. The hospital's physicians uniformly rejected ICD-9 terminology because of the limited scope of diagnosis terms. Other possible options include SNOMED or Intelligent Medical Objects (IMO). Both systems support a larger number of diagnoses to allow physicians to use descriptive clinical vocabulary for diagnosis. Either vocabulary system links diagnoses to a digital code within the CPOE system that triggers the appropriate order sets.
CPOE should theoretically improve a physician's ability to avoid ordering harmful medications by safety links. The CPOE checks laboratory values through an interface to the laboratory information system when the physician orders certain medications. For example, the CPOE checks the creatinine level when the physician orders a medication dependent on real clearance.
CPOE also offers alerts for medication allergies when placing an order. Unfortunately the CPOE system does not magically determine the patient's allergies to prevent ordering of a medication with a documented allergic reaction.
When the hospital implemented CPOE, the allergy documentation process for nursing occurred on paper. The nursing department quickly modified the process so that all patients admitted to the hospital had medication allergies entered into the CPOE system database.
As the electronic health record expands through implementation in our network of regional primary care offices, the allergy database will become more robust and hopefully prevent administration of allergy-provoking medications in the hospital setting.
CPOE down time
Our medical staff initially hoped that our CPOE system would be able to assist with the Joint Commission on Accreditation of Healthcare Organizations' mandated medication reconciliation process. However, our CPOE needed an update from the vendor to accomplish this task.
We determined that a paper work-around must be established until an update of the CPOE became available. This paper workaround resulted in a great deal of physician resentment. At some point within the next 18 months we expect implementation of the electronic health record ambulatory documentation module. This will allow medications prescribed by our regional office physicians to be documented in the patient's electronic health records.
When the patient enters the hospital, this medication list will become available upon admission and allow reconciliation at the time of admission as well as at time of transfer between levels of care within the hospital.
Finally the patient's discharge medications from the hospital can be compared to the pre-admission outpatient medications. For patients entering the hospital from providers outside our network, the patient's medications will need to be entered by a nurse or physician at the time of admission when the updated software becomes available.
Since implementing the CPOE system, we had two significant down times when the hospital lost all computer systems, including the admissions/discharges/transfers (ADT) system provided by another vendor. The CPOE brings patients into the system from an interface with the ADT system.
Although the CPOE system quickly returned to functional status both times, physicians could not place orders because the CPOE system did not recognize the patient's existence within the hospital because of the delay in processing of ADT entries through the interface.
In addition, most physicians and nurses never expected a significant amount of down time. As a result, finding paper order forms became challenging. Of note, our consultants recommended we establish a data center off campus with a distant back-up facility with duplicate information of the data in case of a system failure, such as flood or fire.
In the end, the purchase and implementation of a CPOE system decreased the physician order legibility errors but other sources of error persisted. We now understand that the CPOE system must be integrated with the hospital pharmacy system and the nursing medication administration record and combine bar coding at administration to further reduce medication errors.
To optimize decision support, the CPOE system needs an accurate problem list to trigger evidence-based order sets to standardize care and reduce unexplained variability. Finally, the patient's allergy database must be complete to prevent administration of harmful medications.
Burdett R. Porter, MD, CPE, FIPP is co-chair of the Guthrie Health Inpatient Clinical Advisory Committee, a multi-disciplinary group dealing with electronic health record implementation at hospitals within the Gutbrie Healthcare System, in Sayre, Pa. He can be reached at 570-882-2413 firstname.lastname@example.org
1. Kaushal, R; Shojania, K; Bates, D: Effects of Computerized Physician Order Entry and Clinical Decision Support Systems on Medication Safety. Arch Intern Med. 2003; 163:1409-1416.
By Burdett R. Porter, MD, CPE
Diagram 1 Error Classification Prescribing 5% Transcribing 35% Dispensing 6% Administering 48% ADRs 1% Other 5% Note: Table made from pie chart.