Red Blood Cell Transfusion Practices: A Paragon of EBM?

We at Sensible Medicine try to be all about evidence-based medicine. Some practices that seem like they are paragons of EBM may, upon closer inspection, be anything but. This dive into current blood transfusion practices might seem a bit “Inside Baseball” but it is important for two reasons. First, the decision to transfuse red blood cells is a common and impactful one. Second, the topic is a great example of how the results of trials can be misread and inappropriately extrapolated and then incorporated into guidelines and electronic health systems to the detriment of patients and doctors (an issue I once published on).

The other thing I found interesting in this article was the use of practice variability as both a red flag and its eradication as a goal. The former seems appropriate. The latter maybe not.

(This article is carefully and extensively referenced here.)

Adam Cifu, MD

---

Many practicing clinicians point to the quick and widespread adoption of restrictive red blood cell (RBC) transfusion practices – transfusion only when a patient’s hemoglobin (Hb) drops below 7g/dL -- as a paragon of evidence-based medicine. Prior to restrictive transfusion practices, clinicians transfused hospitalized patients at a Hb <10g/dL (“10/30” rule), but transfusion practices varied across providers and there was little empiric data to guide clinicians.^1,2 This began to change after the landmark Transfusion Trial in Critical Care (TRICC) study, which showed no mortality difference between critical care patients who received transfusion at either a “liberal” Hb threshold of 10-12g/dL, or a “restrictive” Hb threshold of 7-9g/dL.³ Since the publication of TRICC over two decades ago there have been dozens of transfusion RCT’s that have replicated the findings in different patient populations, and the consistency of the findings from these trials has been summarized in multiple well done meta-analyses.^4-6 In turn, guideline committees from multiple medical societies, including most prominently the AABB^7,8, have supported as “strong recommendations” restrictive transfusion practices in hospitalized patients with anemia.

In response to the accumulated trial data and guideline recommendations, transfusion practice has changed. This has been promoted by hospitals that have programmed the electronic health record (EHR) to monitor clinician adherence to restrictive transfusion practice as an evidence-based quality and safety metric.^9-41 Providers are now discouraged from transfusing under circumstances other than a patient having a single Hb value below restrictive transfusion levels, restrictive transfusion is standard of care for stable hospitalized patients,^42-44 and transfusion practice is standardized.⁴⁵

This narrative of RBC transfusion, from a practice that lacked empiric data to universal acceptance and adherence by clinicians of standardized practice based on data from large well done RCT’s, is why transfusion practice might be considered a paragon of evidence-based medicine. However, there are limitations to the trials and the data that inform restrictive transfusion practices, and these often go undiscussed in the literature and on morning rounds. The result is that there are common misunderstandings of the data and the use of restrictive transfusion.

One of the most common oversights is the recognition that trials supporting restrictive transfusion have been designed to test and compare transfusion strategies (liberal vs. restrictive), and not transfusion of units of RBC’s, or the optimal treatment of anemia.⁴⁶ This distinction is important because clinicians mostly care about and make decisions regarding the risks of harms and likelihood of benefits of a single unit transfusion compared to no transfusion in an individual patient at some Hb level, but data from the transfusion trials doesn’t answer that clinical question.

Another common misunderstanding is that “restrictive transfusion is better than liberal transfusion,” and/or that “transfusion has no benefit.” In fact, the trials have shown there to be no difference between a liberal vs. restrictive transfusion strategy. There are numerous scenarios (e.g. GI bleeding) where clinicians might feel compelled to transfuse a patient above 7g/dL, but if they withhold transfusion under the assumption that restrictive transfusion is better, they may actually end up causing harm to the patient. Moreover, the transfusion trials have overwhelmingly used mortality as the primary outcome. There is far less trial data describing the effect of transfusion strategies on other outcome patients care about (i.e. quality of life, fatigue, etc.), and both biologically plausible arguments and some data (trial and observational) to suggest transfusion (either liberal strategy or single unit) is likely “better” for these outcomes.

There are several other limitations to the transfusion trials that are worth considering. The Hb ranges used to define “restrictive” and “liberal” transfusion thresholds has varied across trials, raising the question of what Hb level even defines these practices in different patient populations. Moreover, in some of the trials separation in the average Hb levels between the trial arms was never achieved, meaning the intervention was never actually tested. Additionally, most of the trials collected adverse event data on transfusion but not on untreated anemia, biasing the results against liberal transfusion. Last, the early transfusion trials never included a usual care group, resulting in practice misalignment^47,48 that may have impacted the findings from those early studies and the scientific direction of future trials. Over time the methodologic precedent of comparing liberal and restrictive thresholds has been repeated such that now restrictive transfusion is usual care. Yet, it’s worth wondering how the body of trial data and the practice of transfusion may have differed if those early trials had included a usual care group.

One reason busy clinicians may be unaware of the limitations of the transfusion trials and misunderstand their data is because hospitals have promoted unified restrictive transfusion across health systems using the EHR. Hospitals are motivated to do this by multiple factors, including the presumed costs-savings associated with reduced transfusion.^49-52 Any cost-savings are likely overestimated because the studies describing the cost benefits of restrictive transfusion have never included a formal cost-effectiveness analysis. Hospitals have also been motivated by the long-standing belief that standardization of care improves safety and quality,^53-58 resulting in health systems applying restrictive transfusion practices to patient populations never studied in any of the large RCT’s. It has long been known that individual clinical, physiological, and biologic factors influence the need for and response to transfusion.⁵⁹ Guideline committees recognize this and recommend that transfusion decisions consider individual patient factors and the overall clinical context, and not just a single Hb value.^8,60 However, this “good clinical practice” recommendation can’t be easily embedded into and monitored by the EHR. Last, the presumed cost-savings and pursuit of standardization in the context of restrictive transfusion practice has resulted in the promotion of single unit RBC transfusion in all clinical circumstances. While there are good arguments for transfusing one unit of RBC’s at a time rather than multiple units, this practice isn’t evidence based, is not related to the data from transfusion trials, and may not be appropriate or optimal in numerous clinical circumstances. By adopting and promoting restrictive transfusion thresholds across health systems using the EHR, hospitals have opted for and placed a value on the simplicity of standardization over promoting clinical decisions driven by the data we have, without considering the adverse consequences of doing so.

Universal and uniform restrictive transfusion practices are not the paragon of EBM. Rather, they are the dogmatic adherence to a practice that isn’t supported by the body of trial data. Transfusion trials have strong data supporting restrictive transfusion strategy being as safe as liberal transfusion strategy with respect to mortality. Yet, the trials and data have limitations and uncertainty that raise many clinical and scientific questions. Is there an optimal Hb transfusion threshold for all patients? Is an optimal universal transfusion strategy even possible? Without recognition of the limitations and uncertainty in our evidence base we can’t answer these questions, and over time that is a formula for medical reversal.⁶¹

---

Micah Prochaska is a clinician scientist and hospitalist physician who studies how hospital-based interventions affect patient reported outcomes after hospital discharge.