- Do simple algebra incorrectly (e.g., dose calculations) 2%
- Enter 10 digits into a calculator incorrectly (e.g., enter NDC into computer) 5%
- Failure to notice adverse indicator 10%
- Failure to recognize an incorrect status during inspection 10%
- Error rates under high stress 25%
- Failure to act correctly after 1 minute in an emergency situation 90%
An observation study of cart filling at Brigham and Women's Hospital (Jt Comm J Qual Patient Saf 2006 Feb; 32(2):73-80) indicates that preparation errors occurred 3.6% of the time, of which only 79% were caught by pharmacist inspection.
Another study from the same location (Ann Intern Med 2006 Sep 19; 145(6):426-434) demonstrated a 58% reduction in near misses recorded by a BCMA system that resulted from implementation of bar code scanning of everything that left the pharmacy.
Data output by one vendor of IV workflow systems has rather consistently indicated product selection errors revealed by bar code scanning of products between 5.5% and 6% and enumerates more than ten different types of errors committed in the product use process. (Baxter DoseEdge sales presentation, 2013).
Literature on human bias pretty clearly indicates that we are not dispassionate observers but rather tend to selectively interpret what we see in favor of what we expect to see (http://www.12manage.com/description_confirmation_bias.html).
It is arguable that at least some of this inspection is perfunctory in that it cannot actually determine that the correct dose was made. For example, most IV admixtures represent the injection of one clear liquid into another; we only really inspect the process inferentially in the absence of some kind of automation. Indeed, the ISMP Safety Summit proceedings indicate that ISMP recommends against this kind of inferential checking for high-risk IV's.
Why, then, would we treat pharmacist physical inspection as some kind of gold standard of excellence? Why would we not want to create a standard around some more reliable system? Given our drive for a more clinical practice model, and its inferred reliance on alternative providers for drug distribution activities, why would we not want to create systems that were so reliable that anyone could operate them?
I am not advocating for any particular product here. I am advocating for something better than the system we currently have; one that generates the data that permits us to monitor the process without having to perform it, and brings the kinds of in-process checks necessary to make our current manual check redundant. Because I am concerned that, as long as we are constrained to physically handling every dose, we will be unlikely to reach our potential as professionals, and, more urgently, we will be unable to move the safety needle on behalf of our patients without bringing much better tools and facilities to the table.
What do you think?
Dennis A. Tribble, Pharm. D., FASHP
The opinions expressed in this blog are entirely my own, and not necessarily those of my employer or of the American Society of Health-System Pharmacists.