Abstract
Introduction:
Routine crossmatch of packed red blood cells (pRBCs) is completed preoperatively at many centers despite conflicting evidence on the incidence of blood transfusions with renal transplantation. In the current economic climate, resource adjudication should be judicious and medically appropriate. The objective of this study was to determine the incidence, timing, and predictors of early postoperative pRBC transfusion in patients undergoing renal transplantation.
Methods:
A retrospective review of all patients undergoing renal transplantation at our institution from January 2013 to May 2016 was performed. Demographic, biochemical, and clinical parameters were recorded. The primary outcome was early postoperative transfusion, defined as an intraoperative transfusion or within 2 days of surgery. Multivariable logistic regression was performed to identify associations with early postoperative transfusion.
Results:
We identified 428 patients during the study period (average age 55 years, 60% male, 30% obese, 67% deceased donor, and 43% preoperative antithrombotic use). Forty (9.3%) patients required early postoperative transfusion (mean: 2.8 pRBCs/transfusion) and most did not require blood urgently. Only 20 (4.7%) patients required a transfusion intraoperatively or on the same day of surgery. Lower preoperative hemoglobin (per g/L unit: odds ratio [OR]: 0.943), female gender (OR: 2.752), and preoperative antithrombotic use (OR 2.369) were associated with a need for early postoperative transfusion.
Conclusion:
Transfusion in the early postoperative period following renal transplantation was less than 10%, suggesting that routine crossmatch may not be necessary for all patients. Preoperative hemoglobin, female gender, and preoperative antithrombotic use were associated with increased risk and may be useful to risk-stratify patients who require crossmatch.
Introduction
As part of the standard perioperative planning for renal transplantation, many centers routinely crossmatch 2 units of packed red blood cells (pRBCs). Contemporary evidence suggests that transfusion rates are quite variable, ranging between 10% and 75%. 1 –5 This variability is due in part to differing postoperative durations: some are reported up to 30 days postoperatively, making it difficult to determine the real-world necessity for up-front preoperative crossmatch for intraoperative and immediate postoperative use. As a result, the need for a routine crossmatch prior to all renal transplantation is likely overstated, and the opportunity exists to tailor this to a patient’s individualized risk profile.
Considering the increasingly resource and financially constrained medical system, it is imperative that physicians responsibly order tests, investigations, and treatments. 6 Although crossmatching a unit of blood may seem relatively innocuous, 1 unit of pRBC costs $522 to $1183 USD. Up to half of this cost is related to the acquisition cost, in addition to the unmeasured labor costs associated with completing the crossmatching process. 7,8 Consequently, routine crossmatch for all patients undergoing renal transplantation may be adding a substantial systemic cost to renal transplantation.
Evidenced-based guidelines for preoperative crossmatching of pRBCs do not currently exist to guide physicians in recognizing patients at high risk of perioperative bleeding. The goal of this study is 2-fold: to identify the incidence and timing of transfusion in the early postoperative period following renal transplant and to identify clinical and biochemical factors that are associated with the need for blood transfusion in this period.
Methods
Design, Setting, and Study Population
We completed an observational retrospective cohort study of all patients who underwent renal transplantation at our tertiary care hospital between January 2013 and May 2016. On average, our center performs 100 to 110 deceased donor transplants and 35 to 50 living donor transplants a year. All patients with end-stage renal disease are considered for transplantation at our center. Patients with a body mass index (BMI) >40 may be required to lose weight prior to consideration, based on the assessment of the consulting transplant surgeon. Patients with severe vascular disease may not be candidates for transplantation or may require revascularization prior to transplant based on assessment by vascular surgery. A consecutive patient sample of all patients was taken over the study duration, and over the duration of the study period and until publication, there were no notable changes to our transplant practice, patients, or approach. Inclusion criteria for our study included adult patients who underwent elective live donor renal transplant or deceased donor renal transplant (donation after cardiac death and neurologic determination of death). Patients with 2 (or more) prior failed renal transplants were excluded from our analysis due to the operative complexity and higher risk for transfusion (patients with a single previous transplant were included, as the second transplant was into the contralateral fossa). We conducted this study according to a prespecified protocol that was approved by the Institutional Review Board at St Michael’s Hospital. Reporting of this study follows the guidelines outlined for observational studies.
Data Collection and Outcome Measures
The primary outcome was the incidence of pRBC blood transfusion in the early postoperative period, defined as transfusions given intraoperatively or within 48 hours following renal transplantation surgery (postoperative day [POD] 0-2). This time interval was selected to capture patients urgently requiring blood intraoperatively and immediately postoperatively for the duration that a unit of crossmatched pRBC is viable for use at most institutions. In a post hoc analysis, this was extended to POD 3 to broaden the generalizability to institutions which allow for use up to 72 hours.
Secondary outcomes included the incidence of: (1) transfusion intraoperatively or on the day of surgery (intraoperative/same day transfusion) and (2) transfusion at any point during the inpatient hospital stay. In addition, predictors of transfusion (early postoperative, intraoperative/same day, and total inpatient stay) were determined. Demographic (age, gender, BMI), clinical (preoperative use of anticoagulants, previous transplantation status, side of transplant, type of transplant, duration of operating time), and biochemical factors (preoperative hemoglobin [Hb] count, platelet count, international normalized ratio, partial thromboplastin time) were collected to determine whether a relationship exists between these factors and the need for transfusion. Obesity was defined as BMI ≥ 30 kg/m2, and baseline use of a preoperative antithrombotic was grouped into a cumulative binary variable, which included at least one of aspirin, clopidogrel, low-molecular-weight heparin, and warfarin. Of note, it is routine practice to perform renal transplantation into the right iliac fossa unless contraindicated (ie, prior renal transplantation, vessel calcification).
Statistical Analysis
Baseline demographics were recorded, and continuous data were reported as mean (standard deviation) and categorical data as a percentage. Evaluation of the clinical, biochemical, and demographic factors and their association with transfusion was assessed using multivariable logistic regression using the R statistical package (The R Foundation for Statistical Computing). Multicollinearity was assessed for model variables at a variance inflation factor >4. To avoid overparameterization, bivariate screening was completed for all variables, and only those with a significance <0.10 were used in the final model. A P value <.05 was considered statistically significant for 2-tailed comparison in the final model.
Results
We identified 428 patients during the study period. The average age was 55 years, 258 (60.3%) were male, 125 (30.5%) were obese, and 182 (42.5%) used antithrombotics preoperatively. The majority of transplants were deceased donor transplants (67.1%) and performed on the right side (92.3%). Twenty-eight (6.5%) patients had a previous renal transplant. The baseline demographics are summarized in Table 1.
Baseline Demographics of the Overall Cohort.
Abbreviations BMI, body mass index; INR, international normalized ratio; PTT, partial thromboplastin time; SD, standard deviation.
Forty (9.3%) patients required early (POD 0-2) postoperative transfusion with a mean of 2.8 pRBCs transfused per patient. A clearly delineated reason for transfusion was found in the medical chart of 21 of these 40 patients. In these cases, the most common reason (52%) was a gradual Hb decline. Including patients requiring transfusion in the first 72 hours (POD 0-3), an additional 14 (3.3%) patients required a transfusion on POD 3. These similarly represented nonacute bleeding with a gradual Hb decline.
Of our secondary outcomes, 20 (4.7%) patients required an intraoperative/same day transfusion, with a mean of 3.1 pRBCs transfused per patient. In addition, 83 (19.4%) patients required a transfusion at any point during their hospital stay, with a mean of 2.4 pRBCs transfused per patient.
On multivariable regression analysis, lower preoperative Hb count (per g/L unit: odds ratio [OR] 0.943, 95% CI: 0.916-0.969, P < .001), female gender (OR: 2.752, 95% CI: 1.318-5.964, P = .008), and preoperative antithrombotic use (OR: 2.369, 95% CI: 1.120-5.150, P = .026) were associated with a need for early postoperative transfusion (Table 2). Longer operative time did not demonstrate a clear association with increased odds of early postoperative transfusion (per minute: OR: 1.005, 95% CI: 0.999-1.011, P = .084).
Odds Ratio by Predictor Variable for Early Postoperative Transfusion in Univariate and Multivariable Analyses.a
Abbreviations: INR, international normalized ratio; OR, odds ratio; Preop, preoperative; PTT, partial thromboplastin time.
a N = 40.
b Statistical significance as a P < .05 for 2-tailed comparison.
For transfusion on the day of surgery and during the inpatient hospital stay, multivariable regression analysis demonstrated that lower preoperative Hb count (per g/L unit: OR 0.934 and 0.953, respectively), female gender (OR: 2.785 and 2.864, respectively), and longer operative time (per minute: OR 1.009 and 1.007, respectively) were associated with transfusions for both end points (Table 3). Antithrombotic use was no longer significantly associated. Living donor transplant type (OR: 0.449) was additionally associated with decreased odds of inpatient transfusion.
Odds Ratio by Predictor Variable for Intraoperative/Same Day Transfusion and Inpatient Transfusion in Univariate and Multivariable Analyses.
Abbreviations: Hb, hemoglobin; INR, international normalized ratio; OR, odds ratio; PTT, partial thromboplastin time.
a Statistical significance as a P < .05 for 2-tailed comparison.
Discussion
Preoperative pRBC crossmatch enables the prompt administration of blood that has a decreased potential for sensitization and adverse events. Routine crossmatching of 2 units of pRBCs is currently the standard practice at our institution and others across Canada. However, in order for this to be an appropriate clinical practice, the incidence of acute bleeding must justify the resource utilization.
In this retrospective study, although the overall rate of transfusion at any point during the hospital stay was 19.4%, only 9.3% required a transfusion in the early postoperative period (<48 hours) and just 4.7% of patients required a transfusion on the day of surgery. Furthermore, of those transfused in the early postoperative period, the majority did not experience an acute bleed. The timing and indication are crucial: having crossmatched blood available is primarily of value in the setting of acute bleeding when a patient cannot wait for a crossmatch to be performed. Another consideration is the relative timing of the expiry of the crossmatched pRBCs (48-72 hours). If a patient is transfused after this period, the pRBCs would have expired without being used and require a repeat crossmatch regardless. In this instance, the initial preoperative crossmatch would again be unnecessary and could be a source of cost-savings. Instead, patients may be more appropriate for a simple group and screen (to detect any antibodies), and then proceed to crossmatch only if antibodies or risk factors for transfusion are present.
The overall incidence of transfusion following renal transplant ranges widely from 10% to 75% in the literature, and our study was consistent with this percentage across the total inpatient hospital stay (19.4%). 1,4,5,9 –13 However, the limitation of applying the existing literature to guide decisions regarding the preoperative crossmatch of pRBC is that the timing of transfusions has been ill-defined. Prior studies routinely report transfusions up to 30 days postoperatively and rarely outline risk factors for early transfusion. Our study is the first to clearly define the timing of blood transfusions relevant to the clinical time line, particularly in the early postoperative phase, while identifying associated risk factors for transfusion.
When considering the implications of these results, concerns regarding the overall transfusion rate and the risk for loss of life have been raised (ie, even if the rate of transfusion is low, it may be worthwhile if it can prevent loss of life). To address this issue, we have contextualized our results versus other major abdominal surgeries (the majority of which also do not undergo preoperative crossmatch). In a recent National Surgical Quality Improvement Program analysis of 19 680 patients undergoing major abdominal surgery, gastric (29.8%), hepatic (32.8%), pancreatic (30.0%) resection, all experienced much greater rates of transfusion. 14 In comparison, this again suggests that the acute need for blood products may be overestimated in renal transplantation, and that it may be safe for some/most patients not to have blood crossmatched preoperatively.
The ability to identify patients at higher risk of transfusion may allow surgeons, nephrologists, and anesthesiologists to make a more informed decision on crossmatching blood as well as to implement strategies to improve modifiable risk factors. Preoperative risk factors of transfusion included Hb count, gender, and preoperative antithrombotic use. These factors are consistent with those reported in the literature: increased risk of transfusion in renal transplant has previously been linked to warfarin/antiplatelet agents, older age, lower preoperative Hb, autosomal polycystic kidney disease, and duration of hospital stay. 1,3,10,15,16 Interestingly, we found that female gender was significantly associated with blood transfusion within our cohort. This has been inconclusively demonstrated in the literature with retrospective studies demonstrating effects in both directions. 1,3,10,15 This finding may be attributable to the effects of anemia, greater hemodilution on average, and a gender-uniform transfusion threshold 17 ; however, these studies, including ours, remain at risk of selection bias due to the retrospective nature and few/single institution designs. Ultimately, the identification of these preoperative factors may be incorporated into a clinical decision-making rule to risk stratify patients prior to their operation for those at high risk of requiring blood transfusion.
Limitations
This study was limited by its retrospective nature at a single site. These may impair the broader generalizability of these results, however the results of this study still represent the most robust and relevant data on the inpatient and early postoperative phase for transfusions currently available in the literature. Although institution-specific data may be needed to feel comfortable with this practice change, these results should prompt other institutions to consider reviewing their own transfusion rates and crossmatching protocols for quality improvement. Furthermore, while our data extended to 2016, we believe that the results continue to be applicable and representative today. There have been no notable changes to our transplant practice, patients, or approach over time, with a consistency over the study period and to publication.
Overall, we observed a low event rate, constraining the number of predictors that could be identified through our study, including subdividing antithrombotic medications. Furthermore, due to routine practice at our institution to perform renal transplantation into the right iliac fossa, transplant side and prior transplantation status were strongly collinear. Therefore, we were unable to examine these factors in more detail (in each model, prior/redo transplantation was used as the more clinically meaningful predictor). Additionally, direct oral anticoagulants were not captured in our database.
Finally, no consistent long-term data were measured or available for sensitization outcomes and thus was beyond the scope of our study. Although these may play a role in the decision-making process, we re-emphasize that the majority of patients who required transfusion did so for a gradual Hb decline and would not preclude crossmatch as necessary after OR. In addition, of those who did require it more urgently, there was often a need greater than 2 units of pRBCs, which exceeded the amount being crossmatched regardless. The primary benefit, thus, would have been only to a small number of patients requiring 1 to 2 units of preoperatively crossmatched blood. Overall, despite the limitations, we believe the predictors from our study represent the most robust and pragmatic risk factors associated with early postoperative transfusion to date to guide crossmatch in this setting.
Conclusion
The data from our retrospective study suggest that the need for transfusion in the acute setting is low. In our cohort, 90% of crossmatched blood went unused as the vast majority of pRBC transfusions occurred after POD 2 or not at all. Lower preoperative Hb, female gender, and preoperative antithrombotic use were associated with an increased risk of transfusion in the early postoperative period. Clinically, these indicators may be useful to risk stratify patients who would benefit from preoperative crossmatch.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
