The AABB Recommendation Guidelines for red blood cell and platelet transfusion have been available for several years.1-3 While general plasma indications have been available for decades from organizations such as the College of American Pathologists, the American Society of Anesthesiologists and the New York State Council on Human and Blood Transfusion Services, the AABB Multidisciplinary Guidelines Panel reviewed plasma product use and could not recommend for or against transfusion of plasma in most clinical situations evaluated.4-7 The AABB Panel did suggest that plasma be transfused into patients requiring massive transfusion and those with warfarin therapy-related intracranial hemorrhage.7 More recently, the American Red Cross has published an updated edition of a compendium of transfusion practice guidelines in which general plasma dosing information is primarily based upon recommendations proposed over a decade ago.6,8-10 Despite these recommendations, studies have shown a wide variation in the use of plasma products without a consensus for indications of their use, dosing or clinical efficacy.11-13 However, it has been shown that when hospitals have institutional guidelines for plasma transfusion, there is evidence of compliance, suggesting a role for transfusion services to monitor and educate, regarding the use of plasma products.14

A commonly utilized method of evaluating plasma transfusion efficacy is to monitor the change in coagulation study laboratory values. However, this method has not proven to be a consistent predictor of plasma transfusion utility. Silbert et al. reported in 1981 that only 47% of patients showed modest changes in abnormal coagulation test results following the transfusion of plasma.15 In 2006, Abdel-Wahab et al. reported only the post-transfusion normalization of the international normalized ratio (INR) in 0.8% of patients (95% confidence interval (CI)) and the halfway normalization being achieved in only 15% of patients (95% CI).16 Other studies have reported a greater reduction in the INR in patients with higher pre-transfusion INR values, but minimal reductions in lower pre-transfusion INR ranges.10,17,18

Although plasma products are still commonly utilized for the reversal of vitamin K-antagonist therapy (warfarin), some studies have suggested the use of vitamin K alone to treat non-life-threatening bleeding and for the reduction of the INR in pre-surgical patients.19-21 While some studies have suggested the use of prothrombin complex concentrates (PCCs) as an alternative to plasma transfusion for the emergent warfarin reversal, at least one study reports a higher risk for thromboembolic events in patients receiving 4-factor PCCs, compared to plasma.22-26 Still other authors suggest there is insufficient data to support the routine use of PCCs rather than plasma for warfarin reversal.27,28

At our institution, the computerized provider order entry (CPOE) system requires the ordering provider to include an indication for transfusion on the plasma order request form. Prior studies have not effectively stratified the predicted post-transfusion INR specifically by the order indication or by the presence or absence of vitamin K co-administration. In this study, we aim to evaluate the observed change in the INR secondary to the plasma transfusion, stratified by the concomitant use of vitamin K (phytonadione) or not, the number of plasma units transfused, the plasma order indication given by the ordering provider and the patient's body mass index (BMI), with the goal of defining determining factors, alone or in combination, predictive of a significant INR reduction.

Approval was obtained from the institutional review board prior to initiating this study. We performed an observational retrospective cohort analysis of the data at the University of Tennessee Medical Center through electronic review of available medical records. The data was collected on hospitalized patients who received plasma transfusions from January through December 2018. The data set excluded all pediatric patients, outpatient transfusions, patients transfused during a surgical procedure and patients transfused under a massive transfusion protocol.

The institution's product inventory includes fresh frozen plasma (FFP), plasma frozen within 24 h after phlebotomy (PF24), plasma held at room temperature for up to 24 h (PF24RT24), and 5-day cold plasma. Regardless of the type of plasma product transfused, each product is considered equally effective for providing plasma proteins in comparable proportions. Our institution is a Level I Trauma Center and plasma units are pre-thawed for immediate availability as a component of emergency preparedness. Pre-thawed units (5-day cold plasma) are issued preferentially to reduce product wastage. The plasma product issued from the blood bank is an ABO-compatible plasma unit with the shortest product outdate. Plasma units vary in total unit volume from 293 to 358 mL (mL). While the institutional recommendation for plasma transfusion is to administer 10 to 15 mL/kg, the individual ordering provider selects the number of units (plasma doses) to be transfused, given the clinical setting.

The data file was initially reviewed for duplicate order information. All duplicate data points were removed. The original data set included information for 516 plasma transfusion orders from 329 inpatient hospital encounters with 316 unique patients. Our CPOE system requires the ordering provider to select an indication for plasma transfusion from a defined choice list, which includes the active bleeding/coagulation factor deficit, invasive procedure with INR > 1.5, multiple transfusions, reversal of anticoagulation, thrombotic thrombocytopenic purpura, thromboelastography, therapeutic apheresis and other. Orders with indications related to thrombotic thrombocytopenic purpura, thromboelastography and therapeutic apheresis were removed from the data set to isolate only patients whose clinical indication for plasma transfusion was based on the pre-transfusion INR value. For the purpose of this study, if during a single hospital encounter multiple plasma transfusion orders were placed, only the initial plasma transfusion order for that hospital encounter was included in the study to prevent a data skew due to a potential cumulative effect from successive transfusions. Thus, all subsequent orders following the initial plasma transfusion order from the same patient encounter were removed from the data set. At our institution, the estimated average time of order-to-transfusion is approximately 1 hour. In an attempt to ensure that the laboratory test values for the post-transfusion INR were reflective of the plasma transfusion, post-transfusion INR samples obtained less than 2 h or more than 24 h from the time the plasma order was placed, as well as INR values listed as incalculable or greater than 20.0, were removed from the data set. Based on all the exclusion criteria, the final numbers resulted in 276 inpatient encounters from 266 unique patients. The data was then divided into patients who received only the plasma product (Group 1; n = 177) and patients who received both the plasma product and vitamin K (Group 2; n = 99).

The following data points were collected for analysis for each patient encounter that met all inclusion criteria: the INR value within 24 h before the plasma transfusion order (pre-transfusion INR), first INR value within 24 h after the plasma transfusion order (post-transfusion INR), the time interval between the reported pre- and post-transfusion INR values, order indication, number of plasma units ordered, body mass index, the plasma order and the administration of vitamin K, if applicable.

The collected data were analyzed using the SPSS software to derive descriptive and inferential statistics. Statistical analyses were performed utilizing the t-test for equality of means, ANOVA for analysis of variance and regression analysis, as appropriate. A p-value of less than 0.05, utilizing a 95% CI, was considered statistically significant.

The patients presented with an average pre-transfusion INR value of 3.02 (median, 2.22; range, 0.96–15.82) and subsequent average post-transfusion INR value of 1.75 (median, 1.52; range, 0.94–7.88), which was performed at an average of 8.58 h (median, 6.76; range, 2.02–23.40 h) after the transfusion order was placed. The average observed change in the INR value from all plasma transfusions in the defined population was 1.27 (median, 0.54; range, −4.91–23.01), with all positive values indicating a decrease in the pre- to post-transfusion INR.

As seen in prior studies, our study shows that there appears to be no direct relationship between the pre-transfusion INR and post-transfusion INR values related to plasma transfusion.15-18 Prior studies have suggested that vitamin K alone may be effective for warfarin reversal.19-21 Similarly, our study shows, with univariate analysis, that the concomitant use of plasma transfusion and vitamin K resulted in a more significant change in the INR, compared to those who received plasma alone, although not when controlling for the pre-transfusion INR values.

While Holland and Brooks proposed a formula to predict the post-transfusion INR, based on the number of units transfused,10 we found no predictable change in the INR, related to the number of plasma units transfused, as a sole predictor in either study group. The change in INR does increase with the number of plasma units transfused up to three units, however, increasing the dose to four plasma units, with or without vitamin K, did not show significant additional benefit. Nevertheless, the pre-transfusion INR seems to be a predictor of the change observed in the post-transfusion INR when more units of plasma are given with concomitant vitamin K administration.

When plasma transfusion indications were compared, the greatest observed change in the INR was observed in patients requiring plasma for the reversal of anticoagulation and in patients with active bleeding. While other indications for plasma transfusion showed a decrease from the pre- to post-INR values, these changes were minimal, when compared to those patients with active bleeding and those requiring the reversal of anticoagulation. While the indication for multiple blood transfusions showed no direct predictable change in the INR with plasma transfusion, multiple blood transfusions as the given indication is a significant predictor of the post-transfusion INR when concomitant vitamin K is administered to a patient with a greater pre-transfusion INR value.

Surprisingly, the BMI did not reveal a predictive value of the post-transfusion INR alone, or when controlling only for the number of plasma units received. However, the inclusion of the BMI as a predictor variable of the post-transfusion INR, in combination with the pre-transfusion INR, the vitamin K administration, the number of plasma units transfused and the order indication, reveals that giving vitamin K in the setting of either multiple blood transfusions or active bleeding and, interestingly, in patients with only mildly elevated pre-transfusion INR values, is a significant predictor of the post-transfusion INR. It is worth noting that the BMI categories were not equally distributed, with both the underweight and morbidly obese categories accounting for only a small percentage of the total patient population.

This study is a retrospective review. Some limitations of our study include no comparison group for patients receiving vitamin K alone. Furthermore, as previously stated, we utilize a variety of plasma products at our institution: FFP, PF24, PF24RT24 and 5-day cold plasma. While each of these plasma products are generally comparable in the percentage of plasma proteins, there is some variation in the volume of each plasma unit (293–358 mL). The volume variation is a normal, standard feature in blood product processing. While the institutional guidelines for plasma recommend 10–15 mL/kg, the number of units transfused (plasma doses) was at the discretion of the ordering provider. Additionally, the order indication for each transfusion is based on that entered by the ordering provider into the CPOE system. The clinical scenario for each patient was not independently verified.

The effect of plasma transfusion on laboratory values for the INR has not shown to have consistent predictive value. Even when stratified by reported indication for the transfusion and BMI, there appears to be no direct correlation with the predictive change in the INR. However, taking into consideration the combination of multiple factors, such as the use of vitamin K, the degree of the elevation in the pre-transfusion INR and the clinical setting, may aid in a more efficient use of plasma products. As programs for patient blood management are continuing to evolve, established recommendations for the monitoring of plasma utilization would be potentially beneficial. Given the unpredictable effects of plasma transfusion and the great variation in clinical use, the consensus for plasma transfusion is likely to be difficult. However, continued research and review of plasma utilization are essential.

The authors declare no conflicts of interest.