Computerized physician order entry (CPOE) and electronic prescribing (eRx) have great potential to reduce medication errors and improve patient safety. The government and other payors have put megabucks into promoting use of both. But are we there yet? We’ve done multiple articles on both the optimistic and pessimistic sides of the question regarding CPOE in general. Most articles on eRx have highlighted positive outcomes, though the reductions in numbers of medication errors have not fully translated into reduction in harm.
In general we have felt that the lack of robust clinical decision support tools has been a major reason for the limited success of CPOE and eRx. And the problem of alert fatigue keeps rearing its ugly head in those systems that do provide robust clinical decision support tools.
We’ve always touted that just by doing away with the handwriting issue we would would likely see improved patient safety with eRx, though we were quick to point out that we’d create some new errors (eg. the “cursor” error where one inadvertently clicks on the wrong item in a list).
But now a new study (Nanji 2011) shows that errors occur in almost 12% of all electronic prescriptions, a rate very similar to the rate of errors in handwritten prescriptions. Nanji and colleagues looked retrospectively at almost 4000 computer-generated prescriptions and found 11.7% of them had at least one error. And about a third of those could have caused potential adverse drug events.
There was, however, substantial variability by vendor (from 5.1% to 37.5%). Omitted information (eg. duration, dose, frequency) accounted for 60% of the errors.
That error rate is actually higher than the 7.6% error rate that Gandhi and colleagues (Gandhi 2005) had found for electronic and handwritten prescriptions combined. The Gandhi study did note that handwritten errors (11%) were more frequent than electronic ones (4.3%) but that difference was not statistically significant.
The Nanji article does, however, point out that many of the errors should be relatively easy to avoid. For example, using forcing functions to prevent omission of key elements or using dose range limits could substantially reduce the number of errors and potential adverse events. Even using a simple calculator could have eliminated 5.6% of the errors. So the promise of e-prescribing is still there – we just have a way to go to get there!
This paper comes on the heels, however, of another disastrous incident in which an infant died because of a computer-related medication error (Graham 2011). In that incident, a data entry error resulted in an automated machine mixing a solution that resulted in a massive overdose of sodium to a premature baby. That IV compounding machine apparently had alerts capabilities but these had not been activated. Of course, as in almost all events with disastrous outcomes, there were other errors and conditions that contributed, such as a disparity between the label on the IV bag and the actual content and fact that a lab technician thought the infant’s blood sodium levels were so high they must be in error.
About a year ago there were a series of incidents related to confusion about weights in computerized medication prescribing systems. ISMP Canada first reported a case in which a chemotherapy agent was given in excessive dosage after the height and weight on a computerized order entry system were transposed (ISMP Canada 2010). They discussed several factors that contributed to the problem and had several excellent suggestions on ways to avoid this type of error. ISMP (US) reprinted this in August 2010 (ISMP 2010). ISMP and ECRI had co-authored an article on the importance of accurate patient weights in a 2009 PPSA Safety Advisory. That article mentioned several cases in which weights were incorrectly entered into computerized systems, confusing pounds with kilograms, resulting in overdosing or underdosing. The PPSA article notes that the weight issue is so important because most of the drugs that have weight-based dosing are hi-alert drugs.
When we have more sophisticated clinical decision support tools and, more importantly, forcing functions and constraints that can help us avoid erroneous order entry, we’ll make the medication ordering phase much safer. However, we still also need better interoperability with other systems to integrate the EMR, lab and pharmacy systems from multiple healthcare organizations to better coordinate care across the entire continuum. So for the time being the promise of e-prescribing is exactly that – a promise. We have a long way to go!
Nanji KC, Rothschild JM, Salzberg C, et al. Errors associated with outpatient computerized prescribing systems. JAMIA 2011; Published Online First: 29 June 2011
Gandhi TK, Weingart SN, Seger AC, et al. Outpatient prescribing errors and the impact of computerized prescribing. J Gen Intern Med 2005; 20: 837-841
Graham J, Dizikes C. Baby's death spotlights safety risks linked to computerized systems. As hospitals rely more on electronic data, worries over potential errors grow. Chicago Tribune June 27, 2011
ISMP Canada. Vulnerabilities of Electronic Prescribing Systems: Height and Weight Mix-up Leads to an Incident with Panitumumab. ISMP Canada Safety Bulletin 2010; 10(5): 1-3 July 31, 2010
ISMP. Electronic prescribing vulnerabilities: Height and weight mix-up leads to dosing error. ISMP Medication Safety Alert! Acute Care Edition 2010; August 26, 2010
Pennsylvania Patient Safety Authority. ECRI, ISMP. Medication errors: significance of accurate weights. Pennsylvania Patient Safety Advisory 2009; 6(1): 10-15 www.patientsafetyauthority.org/ADVISORIES/AdvisoryLibrary/2009/Mar6(1)/Pages/10.aspx