Han YY, Carcillo JA, Venkataraman ST, Clark RSB, Watson RS, Nguyen TC, Bayir H, Orr RA. Unexpected increased mortality after implementation of a commercially sold computerized physician order entry system. Pediatrics. 2005 Dec; 116(6): 1506-12
The authors report that in response to the IOM To Err Is Human: Building a Safer Health System report and the Leapfrog Group patient safety standard bechmarks, the Children’s Hospital of Pittsburgh implemented a hospital-wide commercially sold CPOE system in October of 2002. CHP was one of the first children’s hospitals in the United States to attain 100% electronic physician order entry. They also note that there is increasing discussion in the medical literature of unintended consequences and impacts of CPOE on patient outcomes. They proposed to evalute mortality rates of a specific group of high risk patients before and after instituting CPOE, the null hypothesis being that patient outcome would improve after this implementation.
The study population was all children admitted to CHP by interfacility transport during an 18 month period extending from October, 2001 to March 2003. CHP comprehensively implemented the CPOE system simultaneouly with its complementary clinical applications platform in October, 2002. It was felt that this patient population represented a first-encounter cohort to the CHP system that required urgent processing of admitting and treatment orders. Additionally, each patient’s presentation was assigned a Pediatric Risk of Mortality (PRISM) score which was used to calculate predicted mortality rates for the study population. The authors also noted “Changes to health care team dynamics and the manner by which bedside care was delivered ...”.
A total of 1942 children were admitted to CHP via interfacility transport during the study perod, 1394 before and 548 after CPOE implmentation. The most common primary presentations were for respiratory/airway issues (42.6%), infectious disease (34.9%) and CNS/neuromuscular problems (19.4%). 1102 (56.7%) of patients were admitted to the ICU. There were no essential differences in the clinical and demographic characteristics of the two groups. The overall unadjusted mortality rate was 3.86% (75 patients) but was 2.80% before CPOE implementation and 6.57% after. Based on PRISM scores observed mortality was better than predicted before CPOE implementation but worse than predicted after. In a regression model adjusted for PRISM score CPOE was independently associated with an increased risk of mortality (OR = 3.71, 95% CI 2.13 – 6.46), second only to shock (OR = 6.24) as an independent predictor of mortality in the study population. The same analysis not adjusted for PRISM score still resulted in a mortality OR of 3.28 for CPOE.
The authors document a number of observations. The standard admission and order implementaion process was significantly changed after CPOE implementation. In the ICU setting, the electronic order process was less efficient, more time consuming and more labor intensive, resulting in delays in diagnostic testing and initiation of therapies. The organization of bedside care was significantly altered as one physician had to be exclusively doing order entry and a nurses had to computer activate all orders, resulting in less bedside personnel deliviering actual patient care. Centralization of medications in the pharmacy and the need for pharmacy processing of all orders, a primary component of the ADE function of the PCS, resulted in serious delays in medication administration. The number of patients who received antibiotics and vasoactive drugs within national guideline-recommended timelines was decreased by greater than 50% after CPOE implemetation. Finally, direct physician-nurse interaction, deemed crucial to patient care and management in a critical care time-dependent setting, was fundamentally and negatively impacted by CPOE.
The authors conclude that implementaions of CPOE systems “are typified by contingencies and proceed in a far from linear manner. They are part and parcel of organizational dynamics that ... cannot be foreseen, let alone predicted” (1). They make a number of associated observations. This is a retrospective study and definitive statements about causal relationships cannot be made. The study population may be unique and different subpopulations may demonstrate differing results. The study period included the CPOE/clinical applications implementation period and may represent findings associated with institutional change, not necessarily CPOE intrinsically. The results may be due to the imbalance in the two study periods. The findings may be secondary to issues with the clinical application system and not specifically CPOE and finally, the observed association between CPOE implementaion and increased mortality could be due to an unidentified confounding variable.
Comment The extrapolation of ADE and medical error reduction by CPOE to improved patient outcomes is unproved. CPOE is a tool that holds great potential promise for improving health care delivery. More outcomes based research, however, is needed to assess the impact of CPOE on the delivery of clinical care.
1. Aarts J, Doorewaard H, Berg M. Understanding omplemantation: The case of a computerized physician order entry system in a large Dutch university medical center. J AM Med Inform Assoc. 2004; 11: 207 – 216.
