Comprehensive analysis of a medication dosing error related to CPOE

From Clinfowiki
Revision as of 14:00, 14 May 2007 by Ajcollins (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Horsky J, Kuperman GJ, Patel VL. Comprehensive analysis of a medication dosing error related to CPOE. J Amer Med Informatics Association. 2005;12. 377-382. Introduction: This case report of a serious medication error highlights the need for a comprehensive approach to analysis of failures in interaction between humans and information systems (IS). Multiple analyses were conducted which identified errors that methods more limited in scope would have missed. Specific recommendations were made for changes in interface, workflows and training to reduce the risk of similar errors. Case Description: An elderly ICU patient with renal failure had a low serum potassium. Provider A ordered KCl as an IV bolus, then decided to give it as an additive to IV fluids (IVF) instead. He entered an order for 100 mEq of KCL in 1 L IVF at a rate of 75 ml/hr. The previous IV bolus order was not discontinued, but a similar order from a previous day was cancelled instead. The pharmacy noted the order for 100 mEq was higher than allowed and Provider A decreased it to 80 mEq. The order didn’t contain a stop time or total volume to deliver, resulting in delivery of 216 mEq over 36 hours, in addition to the 40 mEq bolus. Provider A then signed out to Provider B, with instructions to check the patient’s potassium level. Provider B noted the most recent potassium level, which was the same one Provider A had addressed, without noting it wasn’t current. Provider B ordered an additional 60 mEq. The patient recieved 316 mEq of KCl over 42 hours, causing severe hyperkalemia which was successfully treated. Methods: Several data collection and interpretation methods were used. The timeline was established from computer logs, followed by visual and cognitive evaluation of order screens and sign-out notes. A semistructured questionnaire was used to interview the clinicians. Data from each analysis was used to assist with interpretation of results from the other analyses. Results: The medication error resulted from several factors, including physician errors in using the order system, absence of automated safeguards, and uncertainty in managing unusual ordering scenarios. The CPOE system didn’t allow limiting IVF by volume, but only by duration, defaulting to 7 days. The IVF order screen has a “Total Volume” field, but this specifies the size of IVF bag to be used, rather than the volume administered. Order screens for IV bolus and IVF administration are visually similar but with important functional differences that allowed misinterpretation. Provider A apparently assumed he had ordered a specific volume of fluid, which he had not. He did enter a limit of 1 L in the free text section, but this wasn’t visible to the system for dosing purposes. Previous IV orders weren’t visible at the time of entering new orders. There was also confusion about the date of the potassium level, with Provider B not realizing the level was old. Although the system caught the high concentration of KCl in the initial order, it didn’t note the administration time was excessive. Finally, there was inadequate provider training, with multiple attempts to enter orders before successful entry. Recommendations included changing ordering screens to avoid confusion, listing active medications including IVF when ordering, more clearly indicating when labs aren’t current, adding alerts for multiple potassium orders and prolonged potassium administration, and improving order entry training. Conclusions: The medical error resulted from multiple failures in the interaction between humans and the CIS, which could only have been identified and corrected through a multifaceted analysis. Increasing complexity of information systems require more comprehensive analyses to prevent similar errors.

Andrew Collins