AABB2026: Transfusion Safety Officers Urged to Act on Early Warning Signals

Hospitals may already be collecting the data needed to prevent the next transfusion error. The challenge, one transfusion safety leader said, is that too little of it currently triggers sufficient action. 

A root cause analysis can explain a transfusion error in detail, but by the time one is opened, the patient has already been exposed or harmed. In the session “From Event Review to Predictive Safety: How Transfusion Safety Officers Use Data to Prevent the Next Transfusion Error,” Lee Anne Sorto, DNP, FNP-C, urged transfusion safety officers and patient blood management professionals to look further upstream. She called for acting on the frequent, low-level deviations that come before serious events. 

“Surveillance Theater” 

Sorto’s central argument was that monitoring without a planned response accomplishes little. “Data without a defined response is surveillance theater,” she said. 

She framed the issue with the safety pyramid used by quality teams. Serious safety events sit at the top. Near misses that never reach the patient sit at the base: barcode scan failures, rejected specimens, incomplete documentation and out-of-temperature product loss. These happen often, which makes them useful. Trended over time, they are early warnings that could be acted on before harm occurs. 

To decide which near misses are worth tracking, Sorto offered five tests. A useful leading indicator is: 

• Frequent enough to show a trend. 

• Specific to a process, location, shift or staff group. 

• Timely, available before the next event. 

• Actionable, with an accountable owner and response. 

• Balanced, paired with safety and delay measures. 

Timeliness matters most for getting people to change, she said. Feedback on an event from four or five months ago is easy to dismiss as “just an accident.” 

When Good News Isn’t Good 

Sorto’s most striking example was a low number that turned out to be a warning. For several years, her organization’s suspected transfusion reaction rate held at 0.2% to 0.3% of products transfused. Quarter after quarter, committee minutes treated it as a success. National averages, however, run closer to 1% to 2%. 

When medical directors reviewed patient charts, they found reactions that clinicians had recognized and documented in their notes, including possible TACO and TRALI. These cases were never referred to pathology, so none was worked up or classified. Talking with clinicians revealed why. Many remembered an old workflow in which reporting a reaction meant blood could be withheld for days. In fact, the current initial workup takes about 20 minutes, and transfusion often resumes quickly. 

The lesson, Sorto said, is to check whether a falling rate means real improvement or under-recognition. That means comparing denominators, actively looking for missed cases, and reinforcing the steps to recognize, stop, report and notify. “We can’t investigate something that we’re never notified about,” she said. 

Building Safety That Outlasts One Person 

Sorto presented PBM and TSO work as one safety system and said it must not depend on a single champion. She recommended clinical decision support built around precise risks, with override rates monitored. She also advised using single-unit default orders with visible reassessment. 

It is useful to start with a one-page dashboard of six signals: patient identification, specimen integrity, administration reliability, emergency transfusion, waste, and inventory and reaction surveillance. Each signal should have an owner, a threshold and a response. Teams should also check what a dashboard leaves out.  

Finally, every intervention needs balancing measures to catch harm that shifts elsewhere. These cover safety, timeliness (such as order-to-issue time and treatment delay), utilization, clinician and nursing burden and equity across sites and vulnerable populations. 

Start with One Step 

Sorto urged attendees not to take on everything at once. She recalled a medical director’s advice early in her career: kick over one rock and deal with what crawls out, rather than kicking every rock in the garden. She proposed a 30-day playbook: 

1. Days 1-5: Choose one signal, confirm the data source and name an owner. 

2. Days 6-10: Validate the baseline, map the workflow and identify exceptions. 

3. Days 11-20: Set a trigger, pilot a response and collect balancing measures. 

4. Days 21-30: Review results, adjust, then standardize or stop. 

Her takeaway was five steps: pull one trend, define one trigger, name one owner, test one response and report one outcome.