Predictors of Transfusion Requirement Among Patients Who Undergo Hysterectomy for Benign Disease

Abstract

Objective: The purpose of this study was to determine preoperative factors that could predict the need for blood transfusion during hysterectomy. Design: The study was designed as a retrospective case–control study. Methods: A retrospective review of hysterectomies from January 2000 to July 2005 was performed. Transfused (137) and randomly selected nontransfused (304) patients were analyzed in an unmatched case-control manner. Preoperative hematocrits (HCTs), types of hysterectomy, surgical indications, physician–estimated blood loss (EBL), uterine weights, Charlson Comorbidity Index scores, and other demographic factors, in transfused and nontransfused patients were analyzed. Results: Mean HCT was lower in the transfused group than in the nontransfused group (34.0 versus 38.8; p<0.001). Patients with a preoperative HCT <30 required transfusion more often than those with a preoperative HCT ≥30 (78% versus 25%; odds ratio [OR] 10.6). Mean EBL was greater in the transfused group than in the nontransfused group (870 mL versus 291 mL). Patients having surgeries with EBLs>1000 mL had an OR for transfusion of 52.1. There was an increased incidence of transfusion when the surgical indication was fibroid growths or menorrhagia rather than prolapse (OR 3.2; confidence interval [CI]: 1.4–7.4), and the risk of needing a transfusion was higher in patients who had had abdominal hysterectomy than in those who had had vaginal or laparoscopic hysterectomy (OR 3.4; C.I. 2.0–5.7). There were no differences in baseline age, race, body mass index (BMI), or tobacco use between transfused and nontransfused patients. Conclusions: Low preoperative HCT, particularly an HCT<30, is associated with increased risk of the need for a transfusion. Anemic patients are more likely to have open procedures and higher EBLs for larger uteri. (J GYNECOL SURG 28:1)

Introduction

With a large number of medical and procedural alternatives, the number of hysterectomies in the United States has declined since 2003. Hysterectomy, however, remains the second most commonly performed surgical procedure in the United States following cesarean delivery.¹ The Nationwide Inpatient Sample of the Healthcare Cost and Utilization Project reported 516,025 hysterectomies for benign disease in 2007, and 30%–40% of women<65 years old have undergone hysterectomy.^1,2

Hysterectomy is a relatively safe procedure. Complications that occur are often directly related to the surgical technique, anesthesia, or medical disorders that occur during the perioperative hospital stay. The eVALuate study compared complications from hysterectomies for benign conditions performed via laparoscopic, abdominal, and vaginal routes.³ In that study, 56 of 1380 patients, or 4%, required transfusion. Another study by Hillis et al. demonstrated that women undergoing hysterectomy with a uterus>500 g were 1.6 times more likely to develop operative or postoperative complications, specifically cuff cellulitis and the need for blood transfusion.⁴ Other studies have examined the potential relationships among race, uterine anatomy, and other comorbidities.⁵

Perioperative blood transfusion carries with it specific risks, such as viral or bacterial infection, acute lung injury, transfusion reactions caused by immune response, or medical errors in cross-matching.⁶ Complications from transfusions can be potentially be avoided. A Cochrane review of preoperative medical management with gonadotropin-releasing hormone (GnRH) analogue therapy demonstrated that hemoglobin (Hb) and hematocrit (HCT) were significantly improved, and that uterine volume, uterine gestational size, and fibroid growth volume were all reduced.⁷ This study sought to determine what specific preoperative factors, primarily what preoperative HCT value, could be used to predict the need for blood transfusion during hysterectomy for benign disease.

Materials and Methods

This study was approved by the institutional review board at MetroHealth Medical Center, Cleveland, OH. Using electronic medical records, a systematic query was made regarding all types of hysterectomies performed at MetroHealth Medical Center by any member of the gynecology staff from January 2000 to July 2005. Procedures were identified and categorized based on current procedural terminology (CPT) codes. Surgical techniques and approaches varied by individual surgeons and were dependent upon patient characteristics.

The subjects of this study were women who underwent abdominal, vaginal, laparoscopic-assisted, or total laparoscopic hysterectomy at MetroHealth Medical Center during the study period, who had complete documentation of their procedures and postoperative hospitalizations in the systemwide electronic medical records. Exclusion criteria were gynecologic cancer diagnosed preoperatively or intraoperatively, and incomplete documentation of procedures and postoperative hospitalizations. An initial review showed that 152 patients were supposed to perioperative blood transfusions, but, after further investigation, only 137 patients actually received these transfusions. A control group was selected randomly from the nontransfused patients, using random number assignment. Charts were abstracted in a 2:1 unmatched case-control manner for the initial 152 patients, therefore yielding a control group that was greater than 2:1. A database of de-identified patient information was created.

The primary outcome sought was average preoperative HCT in transfused versus nontransfused patients. For secondary analyses of transfusion risk, subjects were stratified by preoperative HCT>30 and ≤30. HCT<30 was chosen as a cutoff as it is commonly considered to be consistent with anemia. Subjects were also stratified by estimated blood loss (EBL) ≤250, 251–500, 501–1000, and >1000 mL. In addition, route of hysterectomy, indication for surgery, physician-EBL, uterine weight, Charlson Comorbidity Index (CCI) score, tobacco use, age, race, and body mass index (BMI) were analyzed as independent variables for predicting the need for transfusion. Surgical indications were categorized as follows: fibroids; menorrhagia with anemia; prolapse; benign cyst or mass; chronic pelvic pain; or endometriosis.

Univariate comparisons of preoperative factors were made between transfused and nontransfused subjects. Student's t-tests for continuous variables and χ² tests for proportion analysis of nominal data were performed. A logistical regression model was developed to evaluate the impact on transfusion risk of BMI, age, race, tobacco use, anemia, fibroid history, prior abdominal surgery, route of hysterectomy, EBL, and CCI score. The data were analyzed using StatView statistical analysis package from SAS Institute Inc. (Cary, NC, copyright 1992, version 5.0.1).

Results

There were 1911 patients meeting inclusion criteria, 137 (or 7%) of whom were transfused, and there were 304 controls, yielding a total of 441 analyzed charts. Mean preoperative HCT was 34.0 in the transfused group versus 38.8 in the nontransfused group (p<0.001). Patients with a preoperative HCT of ≤30.0 required a transfusion more often than those with an HCT>30.1, (OR 10.6; 95% CI 5.4, 21.0). There were no differences in baseline characteristics of age or BMI between nontransfused and transfused patients. CCI score, EBL, and uterine weights were higher in the transfused group, whereas preoperative HCT values were significantly lower in this group (Table 1).

Table 1.

Patient Characteristics by Transfusion Status

Patient characteristics	Transfused (n=137)	Nontransfused (n=304)	p-Value^*
Age (mean±SD) (y)	46.1±10.9	44.9±8.8	0.230
BMI (mean±SD) (kg/m²)	31.8±9.1	30.9±7.8	0.300
Preoperative HCT (mean±SD) (%)	34.0±5.0	38.8±3.9	<0.001
CCI score (mean±SD)	0.67±1.3	0.33±0.9	0.006
Uterine weight (mean±SD) (g)	583.4±773.8	349.7±451.4	<0.001
EBL (mean±SD) (mL)	869.8±791.7	291.0±224.6	<0.001

p<0.05 represents statistical significance.

t-test.

SD, standard deviation; BMI, body mass index; HCT, hematocrit; CCI, ; EBL, estimated blood loss.

Uterine weights were positively correlated with EBL, and both were also notably greater in the lower HCT group. EBL was greater in the transfused group than in the nontransfused group (mean±standard deviation [SD] 870 mL±792 versus 291 mL±225). OR for transfusion when EBL was 251–500 mL, 501–1000 mL, or>1000 mL were 3.5 (CI: 1.7–4.3), 9.5 (CI 1.7–4.3), and 52.1 (CI: 5.1–11.1) respectively. Secondary analyses also showed increased incidence of transfusion when the surgical indication was fibroids or menorrhagia rather than prolapse (OR 3.2; CI 1.4–7.4) Risk of needing a transfusion was also higher in patients who had had abdominal hysterectomy than in those who had had vaginal or laparoscopic hysterectomy (OR 3.4; CI: 2.0–5.7).

After logistical regression analysis, the preoperative HCT was found to be a significant predictor of transfusion risk. This regression analysis is shown, compared with χ²-derived values, in Table 2. With HCT<30.0, the adjusted OR for transfusion was 9.1 (95% CI: 2.9, 28.6). Other factors that were shown to be significant in the regression analysis were higher CCI scores and higher EBLs. The other factors in the regression analysis did not meet significance.

Table 2.

Transfusion Risk Factors

Patient characteristics	Unadjusted OR (95% CI)^a	Adjusted OR (95% CI)^b	p-Value^b
Abdominal hysterectomy	3.40 (2.0, 5.7)	2.2 (1.0, 5.1)	0.06
EBL 250–500 mL	3.54 (2.0, 6.3)	3.0 (1.4, 6.6)	<0.01
EBL 501–1000 mL	9.50 (5.0, 18.3)	10.2 (4.1, 25.4)	<0.001
EBL>1000 mL	52.10 (20.0, 135.6)	35.2 (9.5, 131.2)	<0.001
Fibroid diagnosis	3.22 (1.4, 7.4)	0.5 (0.3, 1.1)	0.08
HCT<30	10.60 (5.4, 29.0)	9.1 (2.9, 28.6)	<0.001
Prior abdominal surgery	1.10 (0.7, 1.7)	0.9 (0.4, 1.8)	0.73
Tobacco use	1.30 (0.9, 2.1)	0.7 (0.4. 1.3)	0.26

χ² test derived OR.

Logistic regression for adjusted OR and p-values.

OR, odds ratio; CI, confidence interval; EBL, estimated blood loss; HCT, hematocrit.

Discussion

To the authors' knowledge, this is the first study to demonstrate the objective clinical marker of preoperative HCT<30.0 as an indicator of increased risk of need for a blood transfusion during hysterectomy for benign disease. The data support previous studies that have found associations with large uterine size, higher EBL, and increased operative times to be associated with increased complications including transfusion.⁸ However, in prior studies, clinical markers of uterine size of 14–18 centimeters (cm) correlating to size >500 g are very subjective and are examiner-dependent.^4,8 Laboratory evaluation of HCT level offers objective results that may be compared more easily across many examiners and institutions. Unlike pathology-measured uterine weight and physician-EBL, preoperative HCT is a concrete value that can assist in clinical decision making.

The authors believe that this study's use of unmatched randomized controls to compare transfused to nontransfused patients was one of its strengths. Randomization of the control sample decreased any possible sample bias, and unmatched controls allowed any differences in baseline characteristics to be noted prior to any analyses. Initially, a 2:1 ratio of controls–cases was used to ensure adequate observational numbers, but ultimately a>2:1 ratio of controls–cases was collected. The authors believe that this did not weaken the study in any way. All factors were analyzed individually as predictors of the need for transfusion, and confounding factors in the logistic regression were also controlled for. The primary outcome findings were not changed when controlling for multiple other factors that often vary clinically among hysterectomy patients.

This study was retrospective and, therefore, it had limitations. There were some missing datapoints among the study patients because of a lack of information in the electronic medical records. Because of the lack of accurate operative times in the electronic records, procedure length was not recorded or compared. This study also does not address issues, such as preoperative GnRH use to decrease uterine size and vascularity, prophylactic uterine-artery embolization before surgery, or intraoperative injection of vasopressin into the uterus during hysterectomy, each of which has been reported to decrease blood loss during hysterectomy.^7,9

Conclusions

From this analysis, it can be concluded that a low preoperative HCT is associated with an increased risk of the need for transfusion, particularly an HCT<30. Anemic patients are also more likely to have higher EBLs for larger uteri. Preoperative treatment of anemia and its causes, such as fibroids and menorrhagia, with a goal of increasing the HCT to a level>30 may minimize potentially avoidable transfusions in these at-risk patients. Surgeons can also adopt newer technologies in intraoperative blood salvage, such as “cell-saver” technology, and use minimally invasive techniques for procedures to decrease operative blood loss in this anemic population, as transfusion also was seen to increase in patients with EBLs>250 mL. Prospective studies to evaluate preoperative medical treatment of fibroids and anemia, and surgical outcomes related to transfusion rates would be helpful for determining clinical significance.

Footnotes

Disclosure Statement

No competing financial conflicts exist.

References

Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP)

www.hcupnet.ahrq.gov. 2009 December 18.

Candiani

, Izzo

, Bulfoni

, Riparini

, Ronzoni

, Marconi

. Laparoscopic vs vaginal hysterectomy for benign pathology. Am J Obstet Gynecol, 2009; 200:368.e1.

Garry

, Fountain

, Mason

, Hawe

et al. The eVALuate study: Two parallel randomized trials, one comparing laparoscopic with abdominal hysterectomy, the other comparing laparoscopic with vaginal hysterectomy. BMJ, 2004; 328:129.

Hillis

, Marchbanks

, Peterson

. Uterine size and risk of complications among women undergoing abdominal hysterectomy for leiomyomas. Obstet Gynecol, 1996; 87:539.

Roth

, Gustilo-Ashby

, Barber

, Myers

. Effects of race and clinical factors on short-term outcomes of abdominal myomectomy. Am J Obstet Gynecol, 2003; 101:881.

Goodnough

. Risks of blood transfusion. Anesthesiol Clin North America, 2005; 23:241.

Lethaby

, Vollenhoven

, Sowter

. Pre-operative GnRH analogue therapy before hysterectomy or myomectomy for uterine fibroids. Cochrane Database Syst Rev, 2001; 2:CD000547.

Unger

, Paul

, Caldito

. Hysterectomy for the massive leiomyomatous uterus. Obstet Gynecol, 2002; 100:1271.

Okin

, Guido

, Meyn

, Ramanathan

. Vasopressin during abdominal hysterectomy: A randomized controlled trial. Obstet Gynecol, 2001; 97:867.