Effects of Pharmacologic Venous Thromboembolism Prophylaxis in Benign Hysterectomy.

This retrospective cohort study included patients in the Michigan Surgical Quality Collaborative (MSQC) who underwent hysterectomy for benign, non-obstetric indications from July 2012 to June 2015, after which time VTE prophylaxis data was no longer collected by the collaborative owing to a low rate of VTE. Patients were excluded if SCDs were not used or clearly documented, if anticoagulation was not documented, or if the patient had a personal history of VTE. Pharmacologic anticoagulation included heparin and low molecular weight heparin, and cases where the regimen was listed as “other” were excluded. Perioperative use was considered from 12 hours prior to the patient being in the operating room through the end of their surgery. MSQC is funded by Blue Cross and Blue Shield of Michigan/Blue Care Network, and includes patients from all insurance payers. Sixty-two participating hospitals in the state were included in this study. At each site, data available in the medical record are abstracted from charts by trained and dedicated nurses. Patient characteristics, intraoperative processes of care, and 30-day postoperative outcomes from patients at member hospitals are routinely collected by these abstractors continuously. Regarding collection of post-operative complications If the electronic medical record including office notes, emergency room visits, and documentation of re-admission or re-operation is insufficient for the 30 days following surgery, three attempts are made to contact the patient followed by a letter. Detailed methods of the registry’s data collection have been described previously. 3 , 12 Race categories of Black, White, and Other used in this publication were pre-determined by the database. We stratified patients into two separate groups based on surgical approach, abdominal (ABD) and minimally invasive (MIS). For each group, an identical type of analysis was performed. The MIS group included laparoscopic, vaginal, robotic-assisted, and laparoscopic-assisted vaginal hysterectomy. Within each group, we further identified patients who received mechanical VTE prophylaxis alone and those who received dual prophylaxis as our groups for comparison. Our primary outcome was estimated blood loss (EBL), defined by the database as milliliters of blood loss observed by clinical team and abstracted from the medical record using sources such as the anesthesia record, operative report, and intra-operative nursing record. Secondary outcomes included perioperative transfusion, operative time, VTE, surgical site infection (SSI), readmission, reoperation, and death. Operative times were reported as a continuous variable in minutes from the start of the surgery (incision) to the closing of the skin incision. Surgical site infections were defined by the Centers for Disease Control and Prevention criteria. In brief, this includes infections that occur within 30 days of surgery and are comprised of superficial surgical site infection that involves only skin and the subcutaneous tissue of the incision, deep incisional surgical site infection that involves the fascial and muscle layers, and organ space surgical site infections that include “any part of the body deeper than fascial/muscle layers that is opened or manipulated during the operative procedure.” 13 Post-operative transfusion, re-admission, re-operation, VTE, and death outcomes all occurred within 30 days of surgery. Clinical characteristics including age, race, body mass index (BMI; kilograms/square meters), smoking status, American Association of Anesthesiology (ASA) classification score (defined as a dichotomous variable as ASA class >3 or not), and presence of co-morbid conditions including hypertension, congestive heart failure (CHF), and Chronic Obstructive Pulmonary Disease (COPD) were abstracted during chart review by clinical data abstractors at each respective hospital. Age was dichotomized at age>60 and BMI >40 because prior work has shown these cutoffs to be associated with VTE in gynecologic surgery 8 and increased post-operative complications 14 , respectively. Caprini scoring, used for stratifying risk of VTE in surgical patients, informed many of the additional clinical characteristics examined in this population. We were not able to calculate individual Caprini scores as this database did not collect all of the necessary characteristics including personal history of thrombophilia or stroke within the last month. Pre-operative characteristics including anemia (defined as hematocrit < 36 %), thrombocytopenia (defined as platelets< 150 K/μL), and transfusion were included due to possible association with avoidance of pharmacologic prophylaxis, perioperative hemorrhage and need for blood transfusion. A hospital-specific rate of pharmacologic prophylaxis use was calculated from abstracted data. Descriptive and comparative statistics of demographics, co-morbidities, and operative details were analyzed. For bivariate analyses, Chi-square tests or Fisher’s exact test were used to compare categorical variables. For continuous outcomes, Student’s t test was used. To perform propensity score matching, we first did a multivariable logistic regression to determine the propensity score for a subject being assigned to the treated group (dual prophylaxis). Covariates for propensity score calculation included age>60, race, BMI >40, hypertension, congestive heart failure, smoking, ASA classification, preoperative transfusion, preoperative anemia or thrombocytopenia, length of stay, use of post-operative prophylaxis, and hospital teaching status. A caliper matching algorithm with caliper size 0.0001, 1:1 ratio without replacement was used in the MIS analysis and 2:1 caliper matching with caliper size 0.0001 without replacement in the ABD analysis. The ABD group underwent 2:1 matching due to its smaller sample size. Relative risks were obtained via conversion of a bivariate logistic regression model and the ratio of the event probabilities. The 95% confidence intervals were based on the ratio of the event probabilities such that all confidence intervals have feasible boundaries including the 0 incidence outcomes in the exposure group. For both the ABD and MIS analyses, clinical characteristics were compared in an identical manner with two exceptions. Clinically and in the data set, length of stay is longer for abdominal hysterectomy and the length of stay categories were separated and compared as such. Additionally, the number of patients with an “exception to pharmacologic prophylaxis” was much higher in the MIS group. A variety of reasons were permitted for listing an “exception” to pharmacologic prophylaxis including home use of anticoagulation, active bleeding, allergy to medication, history of heparin-induced thrombocytopenia, and intra-operative administration of heparin. The causes of the exceptions were not clear and the patients with an exception were in both the dual prophylaxis and single prophylaxis group. Because exceptions were listed frequently in the MIS group, this was used in the propensity score match for the MIS group. The abdominal group had only 50 patients with an exception listed and thus it was not included in the propensity score match. We used SAS Version 9.4 (SAS Institute, Cary, NC) for all analyses. Comparisons were considered significant if p<.05. The institutional review board granted “not regulated” status to this study (HUM00073978) because the study is based on a de-identified database.

The database included 26,525 patients who underwent hysterectomy during the study period. Patients with gynecologic cancer (n=3,051, 11.5%), non-gynecologic cancer (n=96, 0.4%), or obstetric (n=58, 0.2%) indications for hysterectomy were excluded. There were another 1,243 patients excluded because of either incomplete documentation regarding anticoagulation (n=206), lack of SCD use (n=412), or a personal history of VTE (n=625). Therefore, our study included 16,828 patients in the MIS group with 2,017 receiving both pharmacologic and mechanical prophylaxis and 5,249 in the ABD group with 648 receiving both pharmacologic and mechanical prophylaxis. Details of patient selection are highlighted in Figure 1 . For the MIS hysterectomy analysis, patients receiving dual prophylaxis were more likely to be >60 years of age, be of Black race, have hypertension, have preoperative anemia, have surgery at a teaching hospital, and to be non-smokers than those receiving mechanical prophylaxis alone ( Table 1 ). Those receiving dual prophylaxis peri-operatively were more likely to receive post-operative prophylaxis and more likely to have an exception listed for prophylaxis. In unadjusted analyses, those receiving dual prophylaxis had higher EBL and longer operative time ( Table 2 ). With propensity score matching, there were 1,803 pairs (1:1 ratio) matched on all clinical characteristics. Following matching, operative time was significantly longer (mean difference 18.3 mins; 95% CI, 13.8 – 22.8) for those receiving dual prophylaxis compared to mechanical prophylaxis alone. None of the other perioperative outcomes, including EBL, VTE, or blood transfusion, differed with addition of pharmacologic prophylaxis. For the ABD hysterectomy analysis, patients receiving dual prophylaxis were more likely to have age > 60 years, ASA class greater than or equal to 3, surgery at a teaching hospital, length of stay greater than 4 days, exposure to post-operative pharmacologic prophylaxis, and “other” listed for race. They were also less likely to be smokers compared to those receiving only mechanical prophylaxis alone ( Table 3 ). In unadjusted analyses, those receiving dual prophylaxis had higher EBL and longer operative time in addition to being more likely to receive post-operative transfusion or undergo re-operation ( Table 4 ). With propensity score matching, there were 1,168 cases receiving mechanical prophylaxis alone matched to 616 cases receiving dual prophylaxis. These matched cohorts did not differ in demographics or clinical characteristics. Among those receiving dual prophylaxis, the operative time was significantly longer (mean difference 15.3 mins; 95% CI, 9.0 – 21.6) and the average estimated blood loss was significantly greater (mean difference 54.5 cc; 95% CI, 16.9 – 92.1) than those receiving mechanical prophylaxis alone. Postoperative blood transfusion, VTE, and re-operation did not differ with the addition of pharmacologic prophylaxis. There was wide variation in the use of heparin across hospitals. For ABD hysterectomy, 72.5% (45 hospitals) used dual prophylaxis for less than 10% of cases and 3% (2 hospitals) used pharmacologic prophylaxis for >90% of their cases. For MIS, 79% (49 hospitals) used pharmacologic prophylaxis for <10% of cases and only one site used it for greater than 90% of its cases.

In women undergoing benign hysterectomy, after propensity score matching, there was a significant increase in operative time for both minimally invasive (18 minutes) and abdominal (15 minutes) hysterectomy and a significant increase in EBL (55mL) for abdominal hysterectomy in patients who received dual prophylaxis. There was no apparent benefit in adding pharmacologic prophylaxis. The rates of VTE were low in both minimally invasive (0.20%) and abdominal hysterectomy (0.55%) and did not differ with the addition of pharmacologic prophylaxis. There was no suggestion that the addition of pharmacologic prophylaxis lowered the risk of potentially morbid complications or hospital utilization. We found statistically significantly longer operative time for those receiving pharmacologic prophylaxis, which has not been previously described in the literature. While a 15–19 minute difference in operative time may not be large enough to have a meaningful clinical impact for an individual patient, we do feel it is notable that this represents an approximately 10% increase in operative time for patients who received dual prophylaxis. Quality improvement programs are increasingly considering longer operative time as a quality indicator due to its association with adverse perioperative outcomes. 15 We hypothesize that operative time could serve as a surrogate for difficulty obtaining adequate hemostasis when heparin is administered pre-operatively, translating to increased difficulty with visualization due to ongoing blood loss during surgical dissection or requiring additional time to observe, suture, coagulate, or apply hemostatic agents. In a prior study of 9,000 women undergoing benign hysterectomy where 92% of the patients received heparin prophylaxis, pre-operative and post-operative administration were compared. Post-operative administration of heparin was associated with a lower risk of bleeding complications including blood loss ≥ 1,000 mL, post-operative bleeding within vaginal vault, wound, or intra-abdominal spaces, and post-operative hematoma. 10 Similar to previous studies, we found a variety of practices for administration of pharmacologic prophylaxis across hospital sites. Most hospitals appeared to take an all-or-none approach, providing pharmacologic prophylaxis for >90% of their hysterectomies or <10%. Furthermore, some characteristics associated with higher or lower likelihood of receiving pharmacologic prophylaxis in our unmatched cohorts were not expected. Smokers were more likely to have SCDs alone and patients with pre-operative anemia were more likely to receive pharmacologic prophylaxis. These findings suggest that use of pharmacologic prophylaxis is more commonly based on hospital-based protocols than individualized patient characteristics. Further research would be useful in determining how decisions are currently being made at both the physician and hospital level with regards to benign hysterectomy in otherwise low-risk patients. Strengths of our study include that it is a large, multi-institutional cohort. This statewide database, regularly audited for accuracy, collects data from dozens of hospitals of varying size, socioeconomic and academic settings, and location, which increases generalizability. The availability of 30-day outcomes allowed us to capture the majority of clinically significant post-operative outcomes. Limitations of our study include its retrospective nature. While we suspect that the decision to give pharmacologic prophylaxis was primarily based on hospital-level guidance, there is the potential for selection bias in which surgeons choose a therapy based on individual patient-level factors that cannot be captured in a database study. For example, personal history of thrombophilia is utilized in Caprini scoring, but it was unavailable in our database to use as a covariate in the propensity score match. However, it is our hope that in our exclusion of patients with a personal history of VTE, many patients with a known thrombophilia would also be excluded given that the diagnosis of a thrombophilia is often made following a VTE. We excluded malignant and obstetric indications given these patients likely have both a higher likelihood of VTE and of receiving pharmacologic prophylaxis. However, we were not able to account for the variability in operative time for complex benign indications such as endometriosis, adhesive disease, and very large fibroids as there was not a reliable or meaningful way to account for every factor that could impact the length of surgery. The indication for surgery was available in the database but included only those listed pre-operatively. This would exclude any endometriosis that was not known pre-operatively. Furthermore, endometriosis as a diagnosis encompasses a broad spectrum of disease with varying impacts on operative time depending on if it is stage 1 or stage 4, and during chart abstraction, adenomyosis was grouped with endometriosis as a single entity further complicating the impact on operative time. We attempted to maintain generalizability to all benign hysterectomy. Based on patterns of heparin administration, we believe it is likely hospital practices and existing scoring systems played a larger role in the decision to administer pharmacologic prophylaxis, but the role of anticipated complexity of surgery by the surgeon in this decision is an area for future study. Having the benefit of a large population provided by a quality improvement database does come with limitation in regards to granularity of data. EBL is readily available in the medical record, but it is also subject to variations in method of collection and calculation. Variation in the clinical team member responsible for this measurement across the 62 participating hospitals is not known. Hematocrit was the indicator of anemia used in this retrospective study as hemoglobin was not consistently available in the database. There are also data that were too often missing such as uterine weight or not collected such as intraoperative blood transfusion in the database chart abstraction protocol. In conclusion, the addition of pharmacologic VTE prophylaxis to mechanical prophylaxis could be associated with longer operative time regardless of surgical approach and increased EBL in abdominal hysterectomy. Addition of pharmacologic VTE prophylaxis was not associated with a lower incidence of perioperative VTE compared to mechanical prophylaxis alone, but incidence of VTE was very low in both groups. While ongoing prospective research is warranted to further evaluate the benefits and harms of pharmacologic prophylaxis in benign hysterectomy, it seems reasonable to recommend that surgeons make the decision for dual prophylaxis on an individual rather than routine basis, keeping in mind there may be harm in overtreatment.

Venous thromboembolism (VTE) is a potentially preventable perioperative complication with significant morbidity and mortality. Prior to routine VTE prophylaxis and minimally invasive surgical approaches, patients undergoing benign hysterectomy had rates of post-operative DVT and PE of 9.6% and 1.8%, respectively. 1 Routine prophylaxis in addition to increased utilization of minimally invasive surgical approaches has substantially decreased risk for perioperative VTE in benign hysterectomy. More recent estimates of VTE are 0.1–0.3% for laparoscopic hysterectomy 2 , 3 , 4 and 0.24%−0.6% for abdominal hysterectomy. 2 , 5 The American College of Obstetricians and Gynecologists (ACOG) and American College of Chest Physicians (ACCP) recommend mechanical prophylaxis with sequential compression devices (SCDs) or pharmacologic prophylaxis for hysterectomy 6 , 7 . SCD use is evidence-based and associated with extremely low risks 7 , 8 , 9 while use of preoperative pharmacologic prophylaxis has been associated with increased risk for bleeding complications in abdominal hysterectomy. 4 , 10 Current guidelines do not differentiate between abdominal and minimally invasive surgical routes, despite significant differences in both VTE and bleeding risks. 2 , 3 , 4 VTE risk assessment tools, such as the Caprini scoring system, are used to predict risk associated with general surgery procedures, but there is relatively little data for gynecologic procedures. 11 There is significant variability with regard to VTE prophylaxis practice patterns across institutions and providers. 2 , 3 , 4 Gynecologic surgeons and patients would benefit from more tailored guidance regarding VTE prophylaxis recommendations. The aim of our study was to examine whether the addition of pharmacologic prophylaxis to mechanical prophylaxis for benign hysterectomy was associated with changes in perioperative outcomes, including adverse outcomes and reduction in VTE risk.