Efficacy and safety of a restrictive blood transfusion protocol in gynecologic surgical patients

Graphical abstract


Introduction
Major abdominal surgeries for gynecologic indications, both benign and malignant, are often for the purposes of abnormal uterine bleeding or in patients with existing anemia. These surgeries often can incur excessive operative blood loss. Blood transfusion efficacy and safety are relevant and understudied in these populations.
In 2016, the American Association of Blood Banks (AABB) recommended a hemoglobin cutoff of 7 g/dl for transfusion of red blood cells in hemodynamically stable and critically ill patients. A threshold of 8 g/ dl is recommended for patients undergoing cardiac or orthopedic surgery or those with cardiovascular disease (Carson et al., 2016). Blood transfusions are not benign interventions and carry the risks of infection, hemolytic reactions, immunosuppression, transfusion-related acute lung injury, and alloimmunization. Transfusion related immune modulation is a process that has been demonstrated in various patients with solid tumors, and is thought to be associated with duration of storage of blood products, though it is poorly understood (Al-Refaie et al., 2012;Harlaar et al., 2012). Additional research has shown increased tumor recurrence rates and nosocomial infection rates in patients receiving blood transfusions thought to be associated with iatrogenic immunosuppression (Theodoraki et al., 2014).
Evidence to support restrictive blood transfusion protocols varies depending on the study population. With respect to non-surgical patients, the Transfusion Requirements in Critical Care randomized control trial specifically enrolled critically ill patients and demonstrated 30-day mortality was similar in patients receiving a liberal versus restrictive transfusion protocol. Moreover, 30-day mortality was significantly lower in patients receiving the restrictive protocol in the less-critically ill study population (Hebert et al., 1999). Various studies looking at patients with cardiac comorbidities who are undergoing major surgery suggest a restrictive transfusion protocol may be harmful in these cases (Hovaguimian and Myles, 2016;Carson et al., 2015). Data regarding orthopedic surgical patients and restrictive transfusion is mixed and in part seems to be dependent on the comorbidities of the population (Martin et al., 2017).
In non-cardiac surgical patients broadly, restrictive protocols have been shown either not affect or decrease perioperative complications and postoperative morbidity (Martin et al., 2017). In a large study of surgical oncology patients, multivariate analyses revealed intraoperative transfusions negatively impacted postoperative outcomes (Al-Refaie et al., 2012). With respect to gynecologic oncology specifically, existing evidence suggests that use of restrictive blood transfusion protocols may decrease surgical site infection or have no affect on postoperative morbidity (Mark et al., 2019;Prescott et al., 2019;Boone et al., 2014).
Our primary aim was to evaluate the efficacy of a restrictive blood transfusion protocol initiated at a large academic hospital by comparing blood transfusion data pre-and post-initiation of our institutional restrictive blood transfusion protocol. Secondarily, we aimed to compare postoperative outcomes before and after protocol initiation in a gynecologic surgical population. Uniquely, our postoperative complication analysis included a clustered analysis to detect differences more robustly in complication rates in this population by combining surgical site infection, wound disruptions, pulmonary, renal, cardiovascular, and CNS complications, DVT rates and 30-day mortality into one outcome. We hypothesized that implementing a restrictive blood transfusion protocol in a gynecologic patient population was both effective at reducing the number of units of blood given and safe for this patient population.

Materials and methods
This study was designed as a quality improvement effort using a quasi-experimental design. We conducted a retrospective chart review of the Gynecologic Oncology Longitudinal Data Collection and Utilization Project (GOLD-CUP) institutional database and cross referenced this with the ACS National Surgery Quality Improvement Program (NSQIP) to review patients undergoing major abdominal surgery by the gynecology and gynecologic oncology services. Inclusion criteria included all major abdominal surgeries performed by the gynecologic and gynecologic oncology services for gynecologic indications from January 1, 2017, to December 31, 2019. Patients were excluded if they had no longitudinal data for chart review.
On July 1, 2018, our institution implemented a restrictive blood transfusion protocol based on the AABB guidelines, recommending against blood transfusion in hemodynamically stable patients when hemoglobin is above 7 g/dL, unless the indication was cardiovascular (Blood Products Transfusion: Indications-Adult-Inpatient/Ambulatory/ Emergency Department-Clinical Practice Guideline, 2021). In our preprotocol group, we included patients who underwent surgery from January 1, 2017, to June 30, 2018, and compared these to patients in the post-protocol group who underwent surgery between July 1, 2018, to December 31, 2019.
Thirty-day postoperative outcomes were collected. Outcomes for the transfusion data included number of patients who received a blood transfusion and number of units given. Postoperative complication data included reoperation on the day of admission, surgical site infections, wound disruptions, pulmonary, renal, central nervous system (CNS), and cardiovascular complications, as well as deep venous thromboses (DVT), readmissions, and 30-day mortality. Demographic data were also collected. Surgical data including wound class, route of surgery, pathology, and emergent status were also collected. The 7-system clustered analysis combined surgical site infection, wound disruptions, pulmonary, renal, cardiovascular, and CNS complications in addition to DVT rates. The 8-system clustered model combined the same complications as the 7-system model in addition to 30-day mortality.
Transfusion and postoperative complication data were then analyzed using the statistical program Libxlsxwriter (Version 1.9.48 (277)). Categorical variables were analyzed using chi-squared and Fisher's exact tests. Continuous variables were analyzed using student t-tests. A clustered analysis was also completed to further examine the significance of surgical complications (Erekson et al., 2011;. A p-value of < 0.05 was considered statistically significant. This study was deemed exempt by the Institutional Review Board.

Results
A total of 739 patients were included. There were 290 people in the pre-protocol group and 449 patients in the post-protocol group. The mean age of patients in each group were not different: 57.6 years old in the pre-protocol group, 59.2 years old in the post-protocol group. Body mass index (BMI) was the only significantly different characteristic between the pre-and post-protocol groups with the post-protocol group having a higher BMI of 33.1 compared to 31.8. All other demographic and clinical characteristics were not different between the groups, including medical comorbidities. With respect to surgical characteristics, significantly more patients underwent surgery for oncologic indications compared to benign indications in both the pre-and postprotocol group (56.2% vs. 72%, p < 0.0001). Additionally, there were significantly more elective than emergency surgeries both before and after the initiation of the restrictive transfusion protocol (p < 0.0001) ( Table 1).
When blood transfusion data were analyzed, a similar number of patients received blood transfusions in both groups (9.3% vs. 10.6% p = 0.57). However, significantly fewer units of blood were given postprotocol initiation. For every patient who received a transfusion preprotocol, 2.66 units were administered compared to 1.2 units after the protocol was initiated (p = 0.003) ( Table 1).

Discussion
This study found that the implementation of a restrictive blood transfusion protocol is effective in reducing the number of units of blood transfused. Moreover, we demonstrated that postoperative complications do not differ in a gynecologic surgical population after the initiation of this protocol. These findings support that adoption of restrictive blood transfusion protocols are not harmful to patients in this population.
Restrictive blood transfusion protocols have been studied in cardiac, vascular, orthopedic, and oncologic surgical patients, or those who are critically ill or have cardiovascular disease. In gynecologic oncologic patients undergoing surgery, one study of 582 patients showed no difference in postoperative infections, thrombotic events, or mortality between patients who received restrictive blood transfusions (Boone et al., 2014). A different study of a gynecologic oncology population showed a reduction in superficial and deep surgical site infections after the implementation of a restrictive transfusion protocol, though the authors comment on the concomitant implementation of a surgical site infection task force introduced during that time (Mark et al., 2019). In a study similarly modeled to this one that compared the administration of blood products before and after the initiation of a restrictive blood transfusion protocol as well as postoperative adverse events, Prescott et al. demonstrated a decrease in surgical site infection and no difference in 30-day mortality, VTE or cardiac rates between groups (Prescott et al., 2019). Our data expands on these findings to include analysis of reoperation on day of admission, reintubation, renal insufficiency, delirium, and readmissions. Additionally, by clustering these postoperative complications and again showing no difference between the pre-and postprotocol groups, we have more robustly demonstrated the safety of a restrictive blood transfusion protocol in a gynecologic surgical population.
In this study, we included both benign and oncologic patients undergoing major abdominal surgery for gynecologic indications. We found no difference in postoperative complications including infection, thrombotic disease, or mortality. These results were further bolstered by clustering complications as one outcome to account for low incidence of these complications. These data provide more evidence to support the use of restrictive blood transfusion protocol in a gynecologic patient population in the existing landscape of growing evidence.
One limitation of this study was the inclusion of both benign and malignant indications for gynecologic surgery, which could confound the rates of postoperative complications, as there was a larger Table 1 Descriptive characteristics of pre-protocol and post-protocol group.

Group
Pre-Protocol Post-Protocol  proportion of benign indications in the pre-protocol group. Additionally, both elective and emergent surgeries were included, though the vast majority were elective. Indications for blood transfusions and resulting complications in an emergent population could affect our overall comparison. Another limitation of this study involves the lack of data detailing timing of blood transfusion, estimated blood loss and reason for transfusion as well as preoperative optimization with blood transfusions or iron. This is in part because gynecologic patient data was abstracted from a larger surgical dataset initially collected by the institutional blood bank, and therefore did not encompass all variables pertinent and of interest to gynecology in particular. To our knowledge, there is no data regarding the use of restrictive transfusion protocols in strictly a benign gynecologic surgical population. Additional research is needed to parse out any differences in postoperative complications or mortality in this population, as chronic or severe acute anemia is often an indication for major gynecologic surgery in this population. More longitudinal data could be beneficial to elucidate long term outcomes and complications in specifically an oncologic population. The effects of immune modulation after transfusion are not well understood with respect to longevity or potential interaction with subsequent immunosuppressive therapies or surgeries.
Our results demonstrated that a restrictive blood transfusion protocol was successful at significantly decreasing units of blood transfused based on evidence-based guidelines from the AABB. Furthermore, our results showed that adopting a restrictive blood transfusion protocol did not increase perioperative complications and is safe in a gynecologic surgical population. Based on our results, we plan to continue to adhere to and promote a restrictive blood transfusion protocol in this patient population.
As a quality improvement effort, this study was able to effectively compare transfusion, surgical, and postoperative data before and after the initiation of a restrictive blood transfusion protocol in a gynecologic patient population. We were able to obtain data for this patient population with the use of an institutional quality improvement database cross referenced with NSQIP for accuracy. This process provided for a robust representation of our patient population leading to clinically applicable results. Moreover, the inclusion of a clustered postoperative complication analysis has, to our knowledge, not been implemented in comparable studies and further reinforces our conclusions and adds strength to the existing body of evidence at large.

Conclusions
We conclude that at an institutional level, the implementation of a restrictive blood transfusion protocol is not only effective in decreasing blood product utilization but also safe as demonstrated by no increase in postoperative complication rates in a gynecologic surgical population.