July 15, 2008
We’ve already written about the dangers of heparin flushes three times since December 2007 and – guess what – they are back in the news again. This time there is a slightly different nuance but the end result is the same – multiple neonates were inadvertently exposed to dangerously high doses of heparin via heparin flushes. Somewhere between 14 and 17 neonates apparently were given heparin flushes with solutions containing about 100 times the expected concentration of heparin.
In our December 2007 and April 2008 What’s New in the Patient Safety World columns we pointed out many of the lessons learned from prior incidents in Indiana and California. ISMP (the Institute for Safe Medication Practices) has done a great service in its September 21, 2006 Newsletter and November 29, 2007 Newsletter disseminating useful tips on prevention of incidents related to heparin flushes and most of their recommendations are in the above columns. Then in our May 2008 What’s New in the Patient Safety World column we discussed the UK NPSA’s alert “Risks with Intravenous Heparin Flush Solutions”. Their recommendations were based not only on a thorough root cause analysis of several cases similar to those reported in the US, but also a thorough search into the evidence base on the utility (or lack thereof) of heparin flushes in many situations. The latter led to the conclusion that heparin flushes have no advantage over normal saline for maintaining peripheral intravenous catheters (evidence is less clear for arterial catheters or central venous catheters).
It is our opinion that one of the biggest problems is that heparin flushes are often not considered “drugs” or “medications”. They (along with oxygen and intravenous fluids) are not treated with the same “respect” that is used when dealing with a chemotherapy agent or insulin or other drugs. We clearly need to think of the heparin flush not only as a drug/medication but in fact as a high risk medication. Joint Commission’s 2009 National Patient Safety Goals would appear to exempt the heparin flush from its more rigourous standard (which does not apply to use of anticoagulants in a fashion not expected to significantly alter coagulation parameters). Well, these cases are good examples of what can happen even when you don’t expect the coagulation parameters to be altered. And inclusion of heparin in the flush solution also exposes patients to the risk of allergy and heparin-induced thrombocytopenia.
We don’t have enough details about the current Texas case to be certain, but we wonder whether the prior episodes may have actually made the current episode more likely. In all the prior cases, the person administering the heparin flush had drawn it up from a multidose vial. In the current case, multidose vials of heparin were not in floor stock. Rather, pharmacy was mixing up the heparin flush solutions and the error apparently happened during pharmacy’s mixing procedure and was not identified by a double check. We suspect they may have been doing this procedure with the good intention of removing this highly error-prone procedure from nursing staff. Unfortunately, it seems this just moved the “smoking gun” from one type of provider to another. If so, that is another lesson relearned: when you implement what you think is a good solution to a patient safety problem, you need to have some “measurement” or other methods of both demonstrating the solution leads to the desired outcome and some way of looking for undesirable unintended consequences.
The incident highlights another point we have made over and over – double checks are very weak safety interventions. From all industries we know that the error rate when a supervisor checks someone else’s work may be 10% or higher. Note that we don’t know what influence the double check has on the error rates of the original person. It is quite conceivable that the original person may make more errors if they feel that their errors will be intercepted by a second reviewer. We have certainly seen in some technology solutions that staff become so confident in the computer’s ability to capture errors that their own vigilance may wane. The UK NPSA study did acknowledge the controversy regarding double checks but notes that the literature supports a medication error reduction of about 30% when using a double check system.
In response to the current Texas incident, the Leapfrog Group released a statement again encouraging all hospitals to adopt CPOE and encouraging consumers to ask whether hospitals have CPOE before seeking care there. While we are, of course, huge advocates for CPOE, the Texas incident frankly would not likely have been prevented by CPOE. Note that we say “not likely”. The only way CPOE could have prevented this would be if heparin flushes required a specific physician order. Then, at the time of order entry, an alert could appear with links to the evidence base on use of heparin flushes in various settings.
This is clearly a situation where those who are ‘looking for blood” and someone to blame are way off base. Just look at some of the crude comments appearing about pharmacists and nurses and physicians and hospitals in many of the blogs in the LA Times and Wall Street Journal related to the heparin flush incidents. The two Texas pharmacists in the current episode are the two people in this entire country most likely to never have this happen again! Just as was the case with nurses getting fired after giving concentrated potassium solutions intravenously, looking at the role of individuals here takes away from getting to the root causes and coming up with viable solutions that need to be applied in all healthcare settings to prevent recurrence of these unfortunate events. Punishing those individuals would simply pave the way for a severely flawed system to be perpetuated.
So what are we to do? One solution is to use commercially available prefilled heparin flush syringes. However, those syringes must be clearly labeled in such a fashion that they are unlikely to be confused with or mistaken for similar-sized syringes filled with other medications. We have strong doubts that such a distinctive syringe can really be created. Using barcoding medication safety systems would improve that capability as long as the barcode on the prefilled syringe correctly matched the order in the system (which, in turn depends on the reliability of getting the correct concentration of heparin and correct barcode label at the factory).
Previous ISMP recommendations noted the hazards of stocking automated medication dispensing machines, especially on neonatal and pediatric units because they are especially high-risk patients. They noted that a double check of all medications as they leave the pharmacy is important but not fool-proof, as the current case demonstrates. They also made a renewed call for vigilance for look-alike/sound-alike (LASA) items and strongly recommended use of barcoding as a patient safety tool. They pointed out that even when bedside barcoding is not available, barcode systems are available for many of the automated dispensing machines to help ensure proper stocking of those machines. Previous NPSA recommendations on injectable medications included things such as only stocking low-dose heparin products in areas other than pharmacy, correctly labeling all syringes, and using double checking systems when preparing medication for injection and when administering such injections.
But perhaps the most important solution is to stop treating heparin flushes as something other than what they really are – drugs – and require a formal physician order for their use. Each institution clearly needs to look at their policies for use of heparin flushes and determine for which type of catheters and which type of patients is the evidence base solid enough to merit use of a heparin flush over a saline flush. Most facilities will find that they are currently using heparin flushes for many situations in which the evidence base is lacking. The UK NPSA alert recommends that heparin flushes should never be considered “routine” and should be used only when specifically prescribed.
And regarding the evidence base or lack thereof for some of the situations, this is where funding of multicenter randomized controlled trials by NIH or AHRQ should come in. No drug company is likely to sponsor such a study (since its conclusion could well be that the drug has no use in these situations) and many researchers will consider this too “mundane” an academic endeavor. However, this is exactly the kind of patient safety initiative that the Institute of Medicine (IOM) probably had in mind when it recommended legislation and creation of various methods of promoting clinical patient safety research.
We hope that the hospital in Texas will widely share the lessons learned from their root cause analysis on this incident. How many times have we already said “that could happen at almost any hospital today”? Let’s solve this vexing issue like the concentrated potassium solution issue was solved.