Impact of computerized physician order entry (CPOE) on PICU prescribing errors

Abstract

Dear Editor, Prescribing errors frequently occur in pediatric intensive care units (PICUs), for which both lowand high-technology solutions have been evaluated [1, 2]. Computerized physician order entry (CPOE) systems, including clinical decision support (CDS), offer the potential to reduce prescribing error rates in PICUs, but only if well designed and implemented [2–4]. We examined frequency, types, and risk factors of PICU prescribing errors in relation to use of CPOE. This prospective study was performed in a 14-bed PICU of a tertiary children’s hospital using a homegrown CPOE system since 2001. Contemporaneously, orders could be handwritten, allowing comparison. Medication orders of all patients admitted between February 2008 and December 2010 were reviewed for prescribing errors. The study was performed in accordance with the institutional review board guidelines. A total of 718 patients were included with 22,280 medication orders, of which 15,136 (68 %) were handwritten and 7,144 (32 %) CPOE. Omission rates were 66 % in handwritten versus 24 % in CPOE orders. Writing by hand was a strong risk factor for omissions, but CPOE did not reduce the error rate to zero. This was mainly due to the possibility to enter free text into the CPOE system. Because an order has to be complete for the purpose of dose checking, we state that free text entry should be minimized. In our study 12,879 orders were complete and could be reviewed for dosing errors (i.e., dose more than 10 % below or above guideline therapeutic range). Dosing error rates were 21 % in handwritten and 21 % in CPOE orders. Strikingly, 81 % of these dosing errors were more than 10 % below rather than above the therapeutic range, as shown in Fig. 1, and mainly concerned underdosing of antibiotics and analgesics for example. Kadmon et al. [4] already advocated that CPOE in a PICU must be accompanied by CDS that checks medication overdosing. We add that prevention of underdosing should also be incorporated. Associations between dosing errors and covariates related to the patient, order, or drug were studied using logistic regression analysis with forward selection. Intermittent dosing was the strongest risk factor for dosing errors [ORadjusted 5.6 (95 % CI 3.2, 9.8)]. Additionally, all routes of administration related to intermittent dosing, e.g., oral, rectal, and pulmonary, were significantly associated with omissions and dosing errors compared to the parenteral route. Currently though, CPOE/CDS development for PICUs mainly aims to support fast, correct, and easy ordering of continuous infusions [5]. We add that support for intermittent dosing regimens is also required. In conclusion, we identified three important additional requirements for CPOE/CDS to further prevent PICU prescribing errors: (1) minimization of any free text entry that could limit CDS for dose calculation and/or dose