As we were doing some research on safety issues dealing with cancer chemotherapy, we came across an interesting story “Frantic 43 hours to fix chemo overdose” (Khalik 2009). It’s a news article from Singapore in November 2009 about the successful search for an antidote for a patient inadvertently given a massive overdose of 5-FU (5-fluorouracil). The patient, being treated for a nasopharyngeal carcinoma, was to receive the total 5-FU dose over 5 days but the IV pump had erroneously been programmed to administer the dose over 5 hours. The patient’s oncologist was notified very shortly after the dose had been administered and happened to recall a poster presentation at the American Society of Clinical Oncology (ASCO) the previous June that addressed a new drug (vistonuridine) developed as an antidote for 5-FU. A series of frantic phone calls ensued, complicated by time zone differences and occurrence on a weekend, as they tried to get the US company that developed the new antidote. They successfully procured the drug but then had to rush through bureaucratic red tape and health affairs authorities to use a drug that was not yet approved for use in Singapore. The patient apparently survived that potentially fatal overdose and the hospital apologized to the patient and family, covered the medical expenses, and investigated to implement steps to prevent similar future occurrences (Koh 2010). The Singapore Ministry of Health issued an alert to other hospitals to check their procedures for administering chemotherapy via pumps.
From the patient safety perspective, the key factor was a problem programming of the pumps. In fact, there were actually two cases and it was the first case (administering several days worth of a much less toxic agent to a different patient) that alerted the pharmacists that the above patient might have inadvertently received the overdose of 5-FU. The hospital was using pumps that looked very similar but some were programmed in milliliters per hour and others in milliliters per day. Despite a double check by two pharmacists, the error was not noticed until the first patient happened to call the hospital and tell them that her infusion had gone in over several hours rather than several days.
When we read this story, it sounded eerily familiar so we searched through our archive. We did not have to search far. In last week’s Patient Safety Tip of the Week “Publicly Released RCA’s: Everyone Learns from Them” we made reference to one of the few good publicly available RCA’s in the past, one we discussed in our September 11, 2007 Patient Safety Tip of the Week “ ”. That was an RCA done by ISMP Canada on an almost identical event in 2006! The full RCA of that incident is available online (and is an excellent example of the way to do an RCA, utilizing both the Canadian Root Cause Analysis Framework and the Veteran’s Administration root cause analysis tools).
That case involved a 43 year old woman who was being treated for an advanced nasopharyngeal carcinoma with a chemotherapy protocol of high-dose fluorouracil and cisplatinum. The chemotherapy was to be given intravenously by an infusion pump over a 4-day period. However, the pump was inadvertently programmed to infuse the entire contents over a 4-hour period instead. The patient died as a result of the mishap.
The discussion on the causal chains leading to incorrect programming of the infusion rate is excellent and focuses on many of the individual issues we have mentioned in previous Tips of the Week. The bedside nurses had to perform an incredibly complex calculation and they did not have at hand all the information needed to make the calculation. And the double check process failed as well. The nurse doing the double check did not have a calculator readily available so did the complex calculation on paper and “in her head”. Much as we described in last week’s column, the independent double check should be done without knowing what or how the first individual did the calculation and should be done by someone who has the time and an undistracted environment in which to do the calculation and check. All too often the nurse (or pharmacist or even physician) doing the double check is called away in the midst of another task to do the double check and never ends up doing it correctly. One very important point is that neither the first nurse nor the one doing the double check did a “mental approximation”. That’s where you do a simple calculation in your head that would have said “this infusion is going to be done in 4 hours”, not in the 4 days that were intended.
The medication label printed in the pharmacy not only had several features not in keeping with good human factors design principles but also included unnecessary information that may have increased the opportunity for false confirmation of the infusion rate. The pharmacists also were not familiar with how the pumps are programmed. (Note also that this RCA was probably done before widespread adoption of smart pumps because today’s smart pumps would likely also have a built-in drug library that included dose range limits). And the patient education process did not include review of pump data input, which might have been an additional opportunity to notice the mistake. There was also lack of feedback from the pump that further reduced the opportunity to notice the incorrect infusion rate. The description about how several advanced technologies were poorly integrated and actually led to increased cognitive workload is particularly informative. The RCA notes that use of checklists would help in this complex process and there should be ways to incorporate the formulas and calculations into the ordering process.
They also discussed issues with the chemotherapy protocol and the lack of standardization across multiple cancer centers in the system. The protocol lacked any section on what to do in the event of misadministration of the drug.
All the above occurred against a background where the potential toxicity of the regimen and the lack of an effective treatment for overdose did not appear to be fully appreciated by numerous members of the healthcare team. Much had to do with the dearth of published literature on fluorouracil overdosage. That is probably still largely the case and in the Singapore case it was only by chance that the oncologist had recalled the poster presentation about the possible antidote.
Many of the most useful comments pertain to the response to the incident. The recommendations expand upon some of the important points we included in our July 24, 2007 Patient Safety Tip of the Week “Serious Incident Response Checklist”. They also showed how multiple other individuals would have had similar difficulties programming the pump and nicely illustrated how a facility needs to assemble a representative group of end-users during the pre-purchase phase of any investment in new technology in order to identify potential usability and safety issues.
The recommendations made in the RCA are well thought out and do take into consideration possible unintended consequences. The recommendations extend well beyond the facility and include recommendations for manufacturers, regulators, safety organizations, etc. A followup to that RCA was also published in the August 20, 2007 ISMP Canada Safety Bulletin. And a summary also appears in the September 20, 2007 Medication Safety Alert from ISMP (US).
This is an excellent RCA and a model for doing a good RCA. In fact, our sole criticism has to do with their prioritization of causal statements. The treatment protocol design was ranked #15 (out of 16) on their list. We would have ranked it at or near the top of the list. A high-reliability organization would assume that something could go wrong with the infusion process and ensure that, if it did go wrong, irreparable harm would not take place. We see this as akin to the concentrated potassium chloride issue. We understand that convenience of the patient must have been important in development of the chemotherapy protocol but inclusion of 4 days worth of fluorouracil (plus a single high dose of cisplatinum) in a single IV bag enabled the fatal outcome when subsequent errors occurred. A “safety culture” would likely have said “what is the highest dose that a patient could tolerate in one day (or less) if there was inadvertent administration of the infusion?”. A safety culture would have designed the protocol with sublethal dosages that would have protected the patient in the event of “what can go wrong will go wrong”. It also would not have put the healthcare workers at the “sharp end” in a situation none of use would want to be.
There are incredibly valuable lessons in both the Singapore case and the Canadian case. Your organization can benefit from reviewing the issues and recommendations here regardless of whether you perform cancer chemotherapy or not.
Update: See also our April 2010 What’s New in the Patient Safety World column for new papers on medication incidents involving cancer chemotherapy.
Frantic 43 hours to fix chemo overdose.
By Salma Khalik
Tue, Nov 17, 2009
The Straits Times
Elroy Koh (medical student blog) Sunday, March 28, 2010
Chemotherapy mix-up - there is power in an apology
ISMP Canada. Fluorouracil Incident Root Cause Analysis: Follow-Up. ISMP Canada Safety Bulletin 2007; 7(4) 1-4. August 20, 2007.
ISMP Canada. Root Cause Analysis. Fluorouracil Incident Root Cause Analysis. May 22, 2007 (full RCA)
ISMP (US). Fluorouracil error ends tragically, but application of lessons learned will save lives. ISMP Medication Safety Alert. Acute Care Edition. September 20, 2007