Computerized provider order entry implementation: no association with increased mortality rates in an intensive care unit
Computerized physician order entry (CPOE) has been touted as an essential component to improving patient safety since the publication of the Institute of Medicine report To Err is Human: Building a Safer Health System. So when Children’s Hospital of Pittsburgh (CHP) reported an increase in mortality of critically ill children after their CPOE implementation it caught the nation’s attention. This attention came even though a cause and effect relationship had not been established. In particular, the study caught the attention of Children’s Hospital and Regional Medical Center (CHRMC) in Seattle, Washington who had implemented the same CPOE system about a year after CHP. CHRMC set out on its own investigation to determine whether CPOE impacted PICU patient mortality in the immediate postimplementation period and the 13 months following implementation.
The study period included a total of 26 months, 13 months pre and 13 months postimplementation, during which time 2533 patients were admitted to the PICU. Data from the13 months preceding implementation established the baseline patient mortality rate which was compared to the 5-month and 13-month postimplementation mortality rates. Data (demographic, admission source, primary diagnosis, crude mortality and PRISM III risk –adjusted mortality) was abstracted from the proprietary database, PICUEs 3 to determine mortality rates.
The authors described ways in which the CHRMC implementation resembled and differed from that of CHP. The similarity noted was the use of real time decision support in the form of allergy checking, dose checking and custom rules. A significant difference was that CHP did not have order sets available in the critical care setting at go-live as did CHRMC. Prior to implementation, CHRMC order sets were evaluated for appropriateness and usefulness by clinical leaders from the various divisions.
CHRMC found that there was no difference in the PRISM III score for the pre and post implementation groups. Differences between the groups were noted in the areas of age, length of stay, and primary diagnosis.
Preimplementation: Age --- 90.5 months, Length of stay --- 4.74 days, Primary diagnosis --- > asthma > pneumonia
Postimplementation: Age --- 83.2 months, Length of stay --- 4.16 days, Primary diagnosis --- > cancer
The 26-month standardized mortality ratio was (SMR) 0.87. The preimplementation SMR was 0.98 and the postimplementation SMR was 0.77 revealing no significant reduction in the risk of mortality.
When comparing the preimplementation group with the 5-month postimplementation group they found no significant difference in age and length of stay but found that the preimplementation group was sicker. Again there was no significant change in the risk of mortality.
CHRMC study did not reveal any significant change in mortality rates and therefore, differed from what was seen at CHP in the Han et al. report. The report acknowledges that CHRMC benefited from the experiences of CHP. CHRMC worked in collaboration with CHP for many months prior to implementation and applied CHP’s lessons learned. The author’s stated, “The differences in our study suggest that implementation issues (more order sets, sentences, code-set filtering, ability to get medications directly from the medication-dispensing system in emergent cases) rather than inherent issues with the CPOE itself or the underlying high risk of a particular software system are the primary risk factors affecting mortality during implementation of CPOE.”
They offer the following suggestions to institutions that are planning a CPOE implementation:
- have physician approved order sets available at go-live;
- develop tools that will streamline order entry;
- implement a quick registration process to expedite order entry for incoming patients;
- establish a process that allows the dispensing of meds, without an order, in emergent situations;
- emphasize that CPOE augments verbal communication, it does not replace talking;
- conduct detailed current state workflow analysis; and
- use sufficient resources to maximize system functionality
Finally, they emphasize the importance of organizations sharing knowledge and expertise to “hasten the safe deployment of pediatric CPOE.”