April 15, 2008
Computerizing Trigger Tools
This month’s issue of the journal Pediatrics has an article (Takata 2008) on adverse drug events in pediatric patients that has received considerable media attention. It demonstrated that adverse drug events occur much more frequently in this patient population than previously published. It also was one factor that led Joint Commission to publish a new Sentinel Event Alert on pediatric medication errors. Takata and colleagues modified the IHI Trigger Tool for Measuring ADE’s to apply to the pediatric population. They then studied its use in 12 children’s hospitals and found a rate of 11.1 ADE’s per 100 patients, a frequency about 5 times higher than typically reported in studies using other methods of detection in the pediatric population. They found that 22% of the identified ADE’s were potentially preventable, 17.8% could have been identified earlier, and 16.8% could have been mitigated more effectively.
The trigger tool they developed was obviously very effective in identifying ADE’s. Less than 4% of the ADE’s they identified by use of the trigger tool had been identified through the usual occurrence reporting systems at the participating hospitals.
We do endorse the use of trigger tools as part of your quality improvent/patient safety programs but we need to move our thinking to another level. Does it really help us to know that 17.8% of the ADE’s consisted of pruritis after exposure to a drug to which the patient had no previously known allergy? Not likely. (It could help if the drug given was one to which the patient had a previously known allergy or if the drug had been given without an appropriate indication. Also it would be useful be sure that the patient’s allergy list was immediately updated so that the patient does not get the offending drug a second time.) But the chance that quality improvement will occur after follow up on many such cases is small and it still takes at least 20 minutes on most of the chart reviews. So the real focus needs to be on preventable ADE’s.
The real value of the Takata paper on use of a trigger tool for identifying adverse drug events is the potential for developing real-time triggers that help us identify such ADE’s before harm actually comes to the patient or at least early enough so that immediate interventions can at least mitigate any harm. That’s where we should be focusing our energies.
Trigger tools typically rely on flags developed from information coming from pharmacy or laboratory data sources. Many of the triggers in the IHI trigger tools can be used to identify things that potentially need intervention in real time. For example, a rising serum creatinine should both raise the alert that a potential nephrotoxic agent may need to be stopped or that dosage regimens for renally-excreted drugs may need to be adjusted. It’s easy to think of multiple potential possible actions in response to changes in serum sodium, potassium, glucose, hemoglobin, white blood count, platelet count, PT, PTT, INR, etc. Prescription of an anti-diarrheal agent could alert staff to consider the possibility a patient may have developed a C. difficile infection. Prescription of a hi-alert medication (eg. narcotic) or a rescue medication (eg. flumazenil) might also identify patients needing closer attention.
We are all wary as we develop and refine our CPOE (computerized physician order entry) systems that bombarding physicians with too many alerts can be counterproductive. But many of the alerts can be directed to other staff (pharmacy, nursing, etc.). Would it not be more productive to have a QI staff nurse spending 20 minutes identifying a potential preventable complication or mitigating it early rather than doing retrospective chart review? Of course, we can’t answer those questions conclusively until we have developed and validated a set of “real-time” triggers but it sure makes sense to move as much of the process to real time as possible.
We previously mentioned the study done at Northwestern (Szekendi 2006) that utilized such an active electronic trigger tool to identify potential adverse events early (while patients were still hospitalized). They developed such a system using certain abnormal laboratory results and pharmacy data as triggers and demonstrated such a system can identify many adverse events that might have otherwise gone unreported. In addition, it has the more important feature of being able to identify issues in real-time or near-real-time. They gave some specific examples of how the system actually led to interventions that prevented or mitigated harm.
A further advantage of a real-time surveillance system noted in the Szekendi study is that the investigators had the opportunity to speak directly to caregivers who were often able to provide critical information or context about the the case that would not ordinarily be available during retrospective chart review.
Triggers are generally positive results, events or occurrences. But equally important are the negative ones, i.e. the absence of something that is expected. For instance, the lack of a digoxin level within a certain time frame after a change in digoxin dosing, particularly in a patient with renal impairment, should serve as a red flag. And obviously, the lack of an evidence-based intervention (eg. no beta blocker in a patient with a recent MI) should be flagged for intervention or at least investigation.
While the lack of full integration of electronic data systems currently limits trigger tools and active surveillance systems to a limited set of potential adverse events, future advances will likely expand our ability to identify early a larger number of potential adverse events. Hospitals should prioritize events based on their quality improvement/patient safety initiatives and goals, regulatory considerations, and even financial considerations (since almost every adverse event in a hospitalized patient has financial as well as human consequences). It could be a great tool for improving your hi-alert medication program (eg. anticoagulants, narcotics, sedating agents, insulin, neuromuscular blocking agents, etc.). One group (Hartis 2005) demonstrated use of specific triggers to detect and improve warfarin-related adverse events. Itcould also be used to identify potential look-alike/sound-alike (LASA) medication errors. It could be used to flag circumstances where a patient’s fall risk or DVT risk or even decubitus risk has increased because of events occurring subsequent to admission. It should be noted that such systems are also clearly of potential utility in meeting goals in various pay-for-performance programs.
More and more hospital IT systems are now incorporating electronic medical records and many are including test-based reports (radiology reports, EKG reports, etc.). Some are even mandating that all history & physicals, admission and discharge summaries, consults, and progress notes be entered electronically. The capability of performing text searches and use of techniques such as fuzzy logic have the potential to dramatically expand the data sources that could be used in trigger tools and real-time electronic surveillance systems.
Trigger tools have a place today in your patient safety programs but clearly the wave of the future will be in electronic trigger tools or real-time electronic surveillance programs that integrate rules-based alerts and reminders into your CPOE systems.
By the way, that recent Joint Commission Sentinel Event Alert on pediatric medication errors discusses many of the reasons children and infants may be more prone to medication errors and has many excellent recommendations on things your organization can and should be doing to prevent adverse events related to medication errors. And for those interested in another pediatric trigger tool, see some of the fine work done on the Canadian Paediatric Trigger Tool (
Takata, GS, et al: Development, Testing, and Findings of a Pediatric-Focused Trigger Tool to Identify Medication-Related Harm in US Children’s Hospitals. Pediatrics 2008; 121:e927-3935. http://pediatrics.aappublications.org/cgi/content/full/121/4/e927
Joint Commission. Sentinel Event Alert. Preventing pediatric medication errors. Issue 39, April 11, 2008 http://www.jointcommission.org/SentinelEvents/SentinelEventAlert/sea_39.htm
Institute for Healthcare Improvement. Trigger Tool for Measuring Adverse Drug Events.
Szekendi, M K; Sullivan, C; Bobb, A; Feinglass, J; Rooney, D; Barnard, C; Noskin, G A Active surveillance using electronic triggers to detect adverse events in hospitalized patients. Quality & Safety in Health Care 2006; 15(3):184-190 http://qshc.bmj.com/cgi/content/abstract/15/3/184?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&author1=Szekendi&andorexactfulltext=and&searchid=1&FIRSTINDEX=0&sortspec=relevance&resourcetype=HWCIT
Hartis C, Gum MO.; Lederer JW. Use of specific indicators to detect warfarin-related adverse events. American Journal of Health-System Pharmacy 2005; 62(16):1683-1688
The Development of the Canadian Paediatric Trigger Tool for Identifying Potential Adverse Events. http://www.longwoods.com/product.php?printable=Y&productid=17671
Matlow A, Cronin G for the CAPHC Paediatric Trigger Tool Research Group.
Reducing Harm in Paediatric Care: Learning about Adverse Events using a Validated Canadian Paediatric Trigger Tool. Presentation at the CAPHC annual conference. Montreal, October 14, 2007
Update: See also our October 28, 2008 Patient Safety Tip of the Week “More on Computerized Trigger Tools”.