LAKE BUENA VISTA, FLA. – A multimodal intervention founded on prompt peer-to-peer review increased adherence to restrictive red blood cell transfusion guidelines without increasing mortality.
During the intervention, if patients were transfused outside of established hospital guidelines, all clinicians taking care of the patient received an e-mail notification within 72 hours of transfusion. This included the ICU staff, primary team, intern, resident, fellow, nurse practitioner, and attending, said Dr. Daniel Yeh, a trauma and critical care surgeon at Massachusetts General Hospital in Boston whose signature appears on the e-mail along with the endorsement of the surgeon-in-chief, anesthetist-in-chief, director of the MGH Critical Care Center, codirector of blood transfusion services, and the urology chief.
The e-mail blast was coupled with a 1-hour educational lecture during surgery grand rounds on the potential harms of and indications for blood transfusion, a surgical ICU didactic lecture, and monthly division-wide reports.
Prior to the intervention, providers felt they were probably doing pretty well in terms of using a more restrictive transfusion strategy, Dr. Yeh said. After all, there are strong guidelines in all the professional societies and 15 years of high-quality, level 1 evidence from the pivotal 1999 TRICC (Transfusion Requirements in Critical Care) trial to last year’s data in traumatic brain injury (TBI) patients (JAMA 2014;312:36-47) showing that a lower transfusion threshold is noninferior, if not superior, to a higher transfusion threshold.
“Yet, anecdotally in my own practice and objectively in observational data, we see what can only be euphemistically described as a knowledge-practice gap,” he said at the annual scientific assembly of the Eastern Association for the Surgery of Trauma (EAST). “Basically, we weren’t practicing what we were preaching.”
A record review of 144 patients from January to June 2013 found that fully 25% of all transfusions in stable, low-risk ICU patients had a pretransfusion hemoglobin (Hb) trigger > 8.0 g/dL.
Further, 5% of patients received the old standby 2-unit transfusion without an intervening hemoglobin measurement, which has been recommended against by all major medical societies, Dr. Yeh said.
The overtransfusion rate, defined as a post-transfusion level > 10 g/dL, was 11%. Most of this was accounted for by these 2-unit patients, he added.
Stable, low-risk patients comprised 33% of all transfusions during the review period and were identified using liberal exclusion criteria including anyone with visible bleeding or suspected internal bleeding, hemodynamic instability, ischemia (myocardial, intestinal, or peripheral vascular), or who were high risk (defined by a history of coronary artery disease, coronary artery bypass grafting, or congestive heart failure).
After the October 2013 to March 2014 intervention, the percentage of stable, low-risk patients with a Hb trigger of > 8.0 g/dL declined from 25% to 2% (P < .001), with a 1-month audit of 15 patients showing a rebound up to 17% 6 months after the intervention ended, Dr. Yeh said.
“There was no difference in the patients who got a single transfusion; however, the ones who got multiple transfusions during their hospital stay decreased, so that is where we believe most of the decline came from,” he explained.
Among the 137 postintervention patients, the average pretransfusion trigger decreased significantly (7.6 g/dL vs. 7.1 g/dL; P < .001), before rising to 7.3 g/dL 6 months post intervention.
The number of monthly transfusions declined 35% from 47 units to 31 units (P = NS) and the overtransfusion rate fell from 11% to 3% (P = .004).
There were no significant differences in maximum lactate, maximum troponin, median ICU or hospital length of stay, although ICU discharge Hb (8.6 g/dL vs. 8.2 g/dL; P = .087) and hospital discharge Hb (9.0 g/dL vs. 8.6 g/dL; P = .006) were lower in the intervention period, Dr. Yeh said.
No significant differences were seen post intervention in 30-day readmission (23% vs. 16%; NS) or overall mortality (4% vs. 9%; NS).
“Transfusions were not really temporally related to the death nor did we attribute any of these deaths to symptomatic anemia,” he said.
Limitations of the intervention include the possibility of a Hawthorne effect, use of crude clinical outcomes, and the lack of concomitant control in the ICU, although the partial regression seen 6 months after the intervention suggests something about the intervention was working, Dr. Yeh observed.
Discussant Dr. Laura J. Moore of the University of Texas Health Science Center at Houston, congratulated the authors on an interesting and pertinent study, but questioned why they utilized a Hb threshold of 8 g/dL when the transfusion trigger was set at 7 g/dL in TRICC and in the Villanueva et al. study in acute upper gastrointestinal bleeders (N. Engl. J. Med. 2013;368:11-21).