The current study from the Departments of Critical Care Medicine and Pediatrics at Children’s Hospital of Pittsburgh calls our attention to significant issues with implementation of CPOE that have potential to inhibit the goal of improving patient safety. Citing a study by Upperman, et al. from their own institution, which noted significant improvement in adverse drug events after implementation of CPOE, the authors of the current study state, “Children’s Hospital of Pittsburgh implemented … Cerner’s commercial CPOE system in October 2002 in an effort to become one of the first children’s hospitals in the U.S. to attain 100% CPOE status.”
The study was designed to measure the effect of CPOE implementation on mortality rate for patients transferred into this pediatric tertiary care facility who required “immediate processing … and stabilization orders.” This retrospective study examined an 18-month period of mortality data: Thirteen months before the implementation of CPOE and five months after.
They discovered an unexpected increase in unadjusted mortality rate after the implementation of CPOE from 2.8% to 6.57%, (P<0.001). Mortality odds ratios were calculated, and observed mortality before CPOE implementation was consistently better than after CPOE implementation. Regression analysis with and without Pediatric Risk of Mortality comparisons were performed and CPOE, in addition to other factors, was associated with increased mortality in both analyses. How should this be interpreted?
The authors note several factors having impact on the validity of the study, specifically mentioning that “study design precludes any statements regarding cause and effect,” “[researchers] examined a unique patient population admitted through interfacility transport … (thus) findings may not be generalizable to the hospital experience as a whole,” and “[the] observation period after CPOE implementation was brief.”
Seasonal variability was also a potential confounding factor because CPOE implementation was in October and the observation period post-CPOE extended only five months—to March of the next year. However, a statistical comparison between matched five-month periods supported the association between implementation of CPOE and increased mortality in this population.
In addition, the authors discuss how their institution’s chosen implementation process for CPOE affected the work processes and pattern of care provided to these critically ill patients. Issues such as being unable to enter orders on a patient in preparation of their arrival because they were not yet enrolled in the electronic system hampered the availability of important medications at the time of arrival. Forcing all of the orders to go through their designed CPOE process in an effort to use the error prevention capabilities of the system caused potentially significant delays in the administration of life-saving medications.
The lack of functional order sets further delayed the physician’s ability to efficiently enter orders electronically. These variables may have impacted mortality rates. Further delays were evident due to the fact that a nurse had to activate orders placed by the physician, bypassing some of the efficiency of an electronic system. Similarly, the pharmacy reviewed and processed the order before the medication was available to the nurse for delivery to the patient.
These checks and balances have the potential to prevent errors in medication ordering, but if inefficient, they may not be appropriate for an intensive care setting in every circumstance. Researchers stated the new workflow took the physicians and nurses away from the patient’s bedside, potentially decreasing observation benefits for the patient. The capacity of the system to compute at times seemed “frozen” causing further delays—not necessarily an uncommon occurrence with any electronic system.