Risk Factors for Blood Transfusion in Women Undergoing Hysterectomy for Benign Disease

Abstract

Objective: The purpose of this study was to identify risk factors associated with transfusion in women undergoing hysterectomy for benign disease. Methods: A retrospective chart review of women undergoing hysterectomy between June 2007 and July 2009 was conducted. Demographic data, pre- and postoperative hemoglobin (g/dL) values, surgical indication, estimated blood loss (EBL), route of hysterectomy, uterine weight, and perioperative complications were recorded from a departmental surgical database and electronic medical records. A logistic regression analysis was performed to identify independent risk factors for intra- and postoperative blood transfusion. Results: A total of 377 women were included. The overall rate of transfusion was 6.6%. On univariate analysis, the transfusion group had a significantly lower mean age (41.7 versus 44.8, p=0.03), lower pre- (9.28 versus 12.1, p<0.00) and postoperative hemoglobin value (7.1 versus 10.2, p<0.001), higher EBL (870 mL versus 289 mL, p<0.001), and higher mean uterine weight (830 g versus 421 g, p=0.0005). Logistic regression identified only preoperative hemoglobin (odds ratio [OR] 3.0, 95% confidence interval [CI] 1.9, 4.6, p<0.001) and EBL (OR 1.004, 95% CI 1.002, 1.005, p<0.001) as significant risk factors for transfusion. Route of hysterectomy revealed a trend toward a significantly lower rate in the robotic group (0/41) compared with open (19/185), vaginal (2/69), and laparoscopic (4/47) groups, p=0.07. Conclusions: In women undergoing hysterectomy, preoperative anemia is the dominant risk factor for transfusion. (J GYNECOL SURG 28:1)

Introduction

Hysterectomy remains one of the most common surgical procedures performed in women, with an annual rate of ∼600,000 in the United States alone.¹ Despite a decrease in the blood transfusion rate of 15% reported in the landmark Collaborative Review of Sterilization Study (CREST),² probably resulting from the identification of the human immunodeficiency virus (HIV) and fears of blood contamination, rates ranging from 2.2%³ to 11%⁴ still identify the need for transfusion as one of the most frequent complications of elective hysterectomy.

Several studies have evaluated risk factors for transfusion following elective hysterectomy. A retrospective study of 236 women undergoing abdominal hysterectomy in Australia from 1993 to 1995 reported an overall 4.7% transfusion rate and identified preoperative anemia (Hb <12.0 g/dL) and prolonged operative time as the only significant risk factors, with a trend toward an association with larger uterine size.⁵ Similarly, in a study evaluating whether preoperative autologous blood donation is justified for patients undergoing elective hysterectomy, preoperative anemia (mean admission hemoglobin [Hgb] 11.9 g/dL versus 13.2 g/dL), resulting from autologous donation, was identified as the major independent risk factor for the observed transfusion rate of 9.9%. Age, type of hysterectomy, duration of surgery, and uterine weight did not show any significant association.⁶ The one study that did show a twofold increase in the probability of transfusion with uterine weight >500 g did not evaluate preoperative Hgb values, which probably confounded the results significantly.⁷

The surgical route by which a hysterectomy is performed may impact transfusion rates. A higher incidence of bleeding complications has previously been reported with vaginal as opposed to abdominal hysterectomy.^2,8 A recent retrospective cohort study evaluating the effect of obesity on vaginal hysterectomy outcomes reported a significantly higher transfusion rate in both the obese (11.4%) and nonobese (16.6%) groups compared to studies performed in the past 2 decades of abdominal hysterectomy.⁹ With the introduction of laparoscopic, and now robotic, hysterectomy, the impact on transfusion rates with other routes of hysterectomy is not well studied. The potential improvement in energy sources and methods of vessel occlusion also needs further study.

The primary aim of this retrospective cohort study, therefore, was to evaluate current risk factors for transfusion in women undergoing hysterectomy for benign disease in a tertiary care, inner-city, academic medical center where all routes of hysterectomy are performed. One aim was to determine cutoff points for preoperative hemoglobin values and intraoperative estimated blood loss (EBL) that predicted a risk of transfusion. Data gleaned from this study could assist in preoperative counseling and management.

Methods

After obtaining institutional review board approval, a surgical database, maintained by the department of obstetrics and gynecology at Virginia Commonwealth University Medical Center, was audited to obtain a list of all hysterectomies performed for benign disease between June 2007 and June 2009. This time period was chosen because it represented the year prior to, and following, the introduction of the da Vinci S robotic system, which allowed the opportunity to evaluate the complication rates of this new surgical technology. This study is an extension of a previous report that focused on the impact on route of hysterectomy after introduction of robotic technology.¹⁰ Electronic medical records for each patient were then accessed to extract the following data: demographic variables that included age, race, and insurance status; operative reports to determine surgical indication, route of hysterectomy (open, vaginal, laparoscopic, or robotic), and intraoperative complications; surgical pathology reports for uterine weight and final pathologic diagnosis; all laboratory data including pre- and postoperative Hgb values; and all inpatient records pertaining to the surgical admission and subsequent readmissions for any complications, specifically auditing physician and nursing notes for evidence of transfusion of any blood products. These data were all transferred onto de-identified spreadsheets for analysis that was performed using SAS 9.1.3 software (SAS Institute, Inc, Cary, NC).

The decision to transfuse patients intraoperatively was at the discretion of each attending surgeon and was based on the knowledge of preoperative Hgb status and the estimated intraoperative blood loss. The decision to transfuse any patient postoperatively was based on a composite of hemoglobin value and symptoms of anemia that included orthostatic hypotension and/or subjective complaints of dizziness, shortness of breath, or severe fatigue.

Univariate analyses were performed to identify the significant factors that influenced transfusion. Chi-square and t-tests were used to compare categorical and continuous variables, respectively. Based on the univariate analyses, a pool of factors were chosen and a multivariate logistic regression analysis was used to identify those variables that remained significant in the presence of other factors. Odds ratios (ORs) along with 95% confidence intervals (CIs) were calculated for categorical factors. For the continuous factors, at the observed prevalence of transfusion, the optimum cutoff points (with the most discriminatory power) were obtained. All analyses for route of hysterectomy were by intention to treat. A p-value <0.05 was considered statistically significant.

Results

A total of 377 hysterectomies were performed during the study period, of which 204 (54%) were open, 71 (18.8%) vaginal, 51 (13.5%) laparoscopic, and 41 (10.7%) robotic. Overall, 25 (6.6%) required a blood transfusion of which 19 (10%) were in the open, 2 (2.9%) in the vaginal, 4 (8.5%) in the laparoscopic, and 0 in the robotic groups, respectively (p=0.07). When open hysterectomy was compared with all other methods of hysterectomy, a significant difference in transfusion rates existed (p=0.04). Intraoperative EBL (mean±standard deviation [SD]) was significantly higher for open (402 mL±363 mL) and laparoscopic hysterectomy (306.5 mL±255.5 mL) than for vaginal (206.5 mL±218 mL) and robotic hysterectomy (66.5 mL±70 mL), p<0.0001.

Table 1 presents a univariate comparison of demographic variables, surgical indications, mean pre- and postoperative hemoglobin values, EBL, and uterine weight in the transfusion versus no transfusion groups. Subjects in the transfusion group were significantly younger, more likely to be black, and had a lower mean pre- (9.28 g/dL versus 12.1 g/dL, p<0.001) and postoperative Hgb value (7.1 g/dL versus 10.17 g/dL, p<0.001), higher mean EBL (870 mL versus 289 mL, p<0.001) and higher mean uterine weight (830 g versus 421 g, p=0.0005) than the no-transfusion group. A detailed description of each transfusion subject, including timing and number of units transfused, is presented in Table 2.

Table 1.

Demographic and Surgical Data in Transfusion versus no Transfusion Groups

	Transfusion	No transfusion
Variables	n=25	n=342	p-Value
Age^*	41.76 (95% CI 38.9, 44.5)	44.79 (95% CI 44.0, 45.5)	0.03
Race			0.05
Black	20	218
White	3	112
Other	2	11
Preop hemoglobin^*	9.28 (95% CI 8.57, 9.99)	12.1 (95% CI 11.9, 12.3)	<0.001
EBL (mL)^*	869.6 (95% CI 747.7, 991.4)	289.4 (95% CI 256.3, 322.5)	<0.001
Postop hemoglobin^*	7.10 (95% CI 6.46, 7.75)	10.17 (9.99, 10.3)	<0.001
Surgical Indication			0.06
Leiomyomata	16	187
DUB	2	71
Cervical dysplasia	0	19
Pelvic pain	2	43
Pelvic mass	3	16
TOA	2	5
Molar pregnancy	0	1
Route of hysterectomy			0.07
Open	19	185
Vaginal	2	69
Laparoscopic	4	47
Robotic	0	41
Uterine weight (g)^*	830 (95% CI 610, 1049)	421 (95% CI 361, 481)	0.0005

Mean (95% confidence interval).

EBL, estimated blood loss; DUB, dysfunctional uterine bleeding; TOA, tubo-ovarian abscess; CI, confidence interval.

Table 2.

Description of Each Transfusion Case

							Transfusion
Age	Race	Procedure	EBL (cc)	Path/est Weight (g)	Preop Hgb	Postop Hgb	Preop	Intraop	Postop	Final path primary DX
31	B	TVH/LOA	1500	100	8.1	6	0	0	3	Normal
44	B	Ex-lap, abdominal supracervical hyst	200	225	8.2	6.5	0	0	1	Leiomyomata
37	B	LAVH	750	680	8.8	7.6	0	1	0	Leiomyomata
34	B	TAH	50	240	8	5.9	0	0	1	Leiomyomata
41	B	TAH/RSO	1000	1130	8.2	6.1	0	0	2	Leiomyomata
41	B	TAH/BSO	2500	1700	14.3	6.7	0	1	1	Leiomyomata
37	B	SCH/LOA/bladder bx/cystotomy repair	1500	1440	13.4	9.3	0	2	0	Leiomyomata
41	B	TAH	3300	620	11.9	10.2	0	3	1	Leiomyomata
45	H	TAH/LSO	400	695	10.9	5.5	0	0	1	Leiomyomata
31	B	TVH	600	170	7.4	6.1	0			Leiomyomata
42	B	Ex-lap, TAH, LSO, omental LN biopsy, cystotomy	250	3670	8.1	7.6	0	0	2	Leiomyomata
49	B	Attempted TVH converted to TAH	300	315	7.6	6	0	0	1	Leiomyomata
42	B	Ex-lap, TAH	1200	2280	8.6	7.2	0	2	3	Leiomyomata
45	B	Ex-lap, TAH, BSO, LOA	600	287	7.4	9.2	1	2	0	Leiomyomata
42	B	Ex-lap, TAH	200	340	7.1	5.7	0	0	1	Leiomyomata
40	B	Ex-lap, TAH	200	440	7.7	8.9	0	1	0	Leiomyomata
50	B	TAH/BSO/LOA/appy	800	835	8.2	6.2	0	1	1	Adenomyosis
49	B	TVH converted to TAH/BSO/LOA	1000	275	9.3	6.7	0	0	1	Leiomyomata
51	B	TAH/BSO/LOA	1600	2200	8.8	8.1	0	2	0	Endometrial polyp
43	W	Total laparoscopic hysterectomy, LSO	40	120	8	7.2	0	1	0	Normal
40	W	LAVH	100	60	14	7.4	0	0	2	Adenomyosis
35	W	LAVH	750	90	12	5.9	0	0	1	Leiomyomata
52	O	Laparoscopic SCH converted to laparotomy; RSO	500	440	8.7	7.3	0	1	0	Adenomyosis
49	B	TAH, BSO	1500	2085	9.8	7.8	0	2	1	TOA
33	B	TAH; RSO; left salpingectomy	900	315	7.6	6.6	0	0	1	Leiomyomata

TVH, total vaginal hysterectomy; LOA, lysis of adhesion; EBL, estimated blood loss; EST, estimated; preop, preoperative; postop; postoperative; Hgb, hemoglobin; DX, diagnosis; lap, laparotomy; LAVH, laparoscopic-assisted vaginal hysterectomy; TAH, total abdominal hysterectomy; RSO; right salpingo-oophrctomy BSO, bilateral salpingo-oophrectomy; SCH, supracervical hysterectomy; bx, biopsy; ex-lap, exploratory laparotomy; LSO, left salpingo-oophrectomy; appy, appendectomy; LN, lymph node.

A logistic regression analysis that controlled for age, race, EBL, uterine weight, and preoperative Hgb was performed to determine independent risk factors for transfusion. Route of hysterectomy was not able to be included in this model, as the frequency of transfusion in each group was not high enough to determine if it had a significant impact on other variables. The only significant associations that remained in the regression analysis were preoperative anemia, which conferred a threefold increased risk of transfusion (OR 3.005, 95% CI 1.9, 4.6, p<0.001), and higher EBL (OR 1.004, 95% CI 1.002, 1.005, p<0.001). For hemoglobin values and EBL, at the observed prevalence of transfusion (6.6%), the optimum cutoff points were 11.89 g/dL and 334.2 mL, respectively.

A review of other medical and surgical complications was performed in the study cohort to determine if the need for transfusion was associated with any other morbidity. No significant differences in rates of infectious complications (urinary tract infection, abscess, cuff cellulitis, pneumonia, or wound infection) or visceral injury existed between patients who did, and those who did not, receive a transfusion (Table 3).

Table 3.

Comparison of Surgical and Medical Morbidities Between Transfusion and No Transfusion Groups

	Transfusion	No transfusion
Condition	n=25 (%)	n=342 (%)
Urinary tract infection	0	2 (0.58)
Abscess	0	1 (0.29)
Cuff cellulitis	1 (4.0)	4 (1.2)
Cystotomy	1 (4.0)	7 (2.0)
Urinary retention	0	4 (1.2)
Respiratory complication	0	3 (0.87)
Wound breakdown/infection	1 (4.0)	8 (2.3)
Enterotomy	0	1 (0.29)
Ileus	1 (4.0)	9 (2.6)

Discussion

In this retrospective cohort study, the overall rate of 6.6% (10% for abdominal and 2.9% for vaginal hysterectomy) identified blood transfusion as the most common perioperative morbidity in women undergoing hysterectomy at the department of obstetrics and gynecology at Virginia Commonwealth University Medical Center. Although lower than the 15.4% rate of transfusion for abdominal hysterectomy and the 8.3% rate for vaginal hysterectomy that was reported in the largest series evaluating outcomes after hysterectomy for benign disease,² the potential for blood transfusion remains a significant variable that doctors and patients must consider before proceeding with surgery.

The dominant risk factor that was identified as a major contributor to intra- and postoperative transfusion was preoperative anemia, a variable that conferred a threefold higher risk. In contrast, EBL was only found to have a marginal association. It makes intuitive sense that, even in the face of surgical hemorrhage, a robust starting hemoglobin value will protect most women from the need for blood products. It was demonstrated that a starting hemoglobin value <11.89 mg/dL presented a significantly higher risk of needing a transfusion. Whereas other studies have similarly identified “anemia” as an associated risk factor, ^5,11,12 this study was able to determine an absolute value below which the risk of transfusion increased significantly. Knowledge of this cutoff value is helpful for surgical planning and provides a benchmark to achieve prior to elective surgery. Aggressive use of preoperative medical therapies such as progesterone agonists, oral contraceptive pills, gonadotropin-releasing agonists, and iron supplementation, all of which have demonstrated efficacy in reducing blood loss and increasing hemoglobin values, should be encouraged, to reach this threshold.^13–17

Determination of a threshold value for a “safe” starting hemoglobin value also may assist in cost-conscious efforts to reduce standard orders for blood typing and screening for benign hysterectomies, a routine practice at the department of obstetrics and gynecology at Virginia Commonwealth University Medical Center. Whereas rare cases of unexpected hemorrhage may occur, this study demonstrates that the risk of transfusion is extremely unlikely in women without preoperative anemia. An alternative algorithm is proposed whereby only women with a starting Hgb value <11.89 g/dL and those with intraoperative blood loss >330 mL have blood typing and screening. In addition, the recommendation to omit a routine postoperative hemoglobin check, proposed by Kohli et al., is strongly corroborated by this study's data in women who have no transfusion risk factors.¹¹

Whereas demographic variables and mean uterine weight were not found to have any independent association with transfusion risk, a trend toward an association was found when comparing each individual route of hysterectomy, and was significantly lower when the nonlaparotomy routes were grouped together and compared with the open route. The observed risk of transfusion in the vaginal and robotic cohorts was much lower than in the open and laparoscopic groups. This could be a reflection of surgical case selection whereby the easiest cases with the lowest risk of bleeding were assigned to these two modes of access. It is also possible, however, that the 0% transfusion rate in the robotic group was attributable to improved visualization and coagulation of vessels with instruments that have increased dexterity. In a report of perioperative outcomes of 152 robotically assisted hysterectomies for benign cases with complex pathology, Boggess et al. reported a 0% transfusion rate.¹⁸ Similarly, a 0% transfusion rate was reported from 256 patients recruited from five community practices who underwent robotic hysterectomy for benign conditions involving high uterine weight and complex pathology. In this series, the median uterine weight was 453 g with a range of 250 g to 3020 g, suggesting that the advantages provided by robotic instrumentation, not the case selection, was responsible for the observed improvement in transfusion risk,¹⁹

The major limitations of this study were the retrospective design and biases that are introduced through medical record review. An attempt was made to minimize data bias by reviewing the actual operative reports and medical records as opposed to relying on CPT and International Classification of Diseases–9 codes to determine surgical procedures, surgical indication, and medical complications. Because of the extensive nature of required documentation for the administration of any blood products by the blood bank, physicians, and nurses at Virginia Commonwealth University Medical Center, there should be an accurate calculation of transfusion rates in this series. The other limitation was sample size. Although similar in numbers to other published series regarding tranfusion risk for abdominal hysterectomy, larger numbers for each route of hysterectomy are required to determine if a significant difference exists between each method.

Conclusions

Strict attention to preoperative Hgb values in women undergoing hysterectomy for benign disease could have a profound impact on the need for transfusion of blood products. This remains a far more significant variable than uterine size, route of hysterectomy, or EBL. Additional studies are needed to determine if advanced technology and instrumentation such as is provided by robotics could have a favorable impact on transfusion risk in those women who may have difficulty optimizing their hemodynamic parameters before surgery.

Footnotes

Disclosure Statement

No competing financial conflicts exist.

References

, Wechter

, Geller

, Nguyen

, Visco

. Hysterectomy rates in the United States, 2003. Obstet Gynecol, 2007; 110:1091.

Dicker

, Greenspan

, Strauss

et al. Complications of abdominal and vaginal hysterectomy among women of reproductive age in the United States: The Collaborative Review of Sterilization. Am J Obstet Gynecol, 1982; 144:841.

Reiter

, Wagner

, Gambone

. Routine hysterectomy for large asymptomatic uterine leiomyomata: A reappraisal. Obstet Gynecol, 1992; 79:481.

Chan

, Ho

, Chen

. Abdominal hysterectomy: Indications and complications. Singapore Med J, 1993; 34:337.

. Blood transfusion requirements for abdominal hysterectomy: 3-year experience in a district hospital (1993–1995) Aust N Z J Obstet Gynaecol, 1997; 37:452.

Kanter

, van Maanen

, Anders

, Castro

, Mya

, Clark

. Preoperative autologous blood donations before elective hysterectomy. JAMA, 1996; 276:798.

Hillis

, Marchbanks

, Peterson

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

Amirikia

, Evans

. Ten-year review of hysterectomies: Trends, indications, and risks. Am J Obstet Gynecol, 1979; 134:431.

Harmanli

, Dandolu

, Isik

, Panganamamula

, Lidicker

. Does obesity affect the vaginal hysterectomy outcomes? Arch Gynecol Obstet, 2011; 283:795.

10.

Matthews

, Reid

, Ramakrishnan

, Hull

, Cohen

. Evaluation of the introduction of robotic technology on route of hysterectomy and complications in the first year of use. Am J Obstet Gynecol, 2010; 203:499e491.

11.

Kohli

, Mallipeddi

, Neff

, Sze

, Roat

. Routine hematocrit after elective gynecologic surgery. Obstet Gynecol, 2000; 95:847.

12.

Gambone

, Reiter

, Lench

. Quality assurance indicators and short-term outcome of hysterectomy. Obstet Gynecol, 1990; 76:841.

13.

Stovall

. Gonadotropin-releasing hormone agonists: utilization before hysterectomy. Clin Obstet Gynecol, 1993; 36:642.

14.

Stovall

, Summit

Jr. , Washburn

, Ling

. Gonadotropin-releasing hormone agonist use before hysterectomy. Am J Obstet Gynecol, 1994; 170:1744.

15.

Lumsden

, West

, Thomas

et al. Treatment with the gonadotrophin releasing hormone-agonist goserelin before hysterectomy for uterine fibroids. Br J Obstet Gynaecol, 1994; 101:438.

16.

Benagiano

, Morini

, Primiero

. Fibroids: Overview of current and future treatment options. Br J Obstet Gynaecol, 1992; 99,suppl7:18.

17.

Audebert

, Madenelat

, Querleu

et al. Deferred versus immediate surgery for uterine fibroids: Clinical trial results. Br J Obstet Gynaecol, 1994; 101,suppl10:29.

18.

Boggess

, Gehrig

, Cantrell

et al. Perioperative outcomes of robotically assisted hysterectomy for benign cases with complex pathology. Obstet Gynecol, 2009; 114:585.

19.

Payne

, Dauterive

, Pitter

et al. Robotically assisted hysterectomy in patients with large uteri: Outcomes in five community practices. Obstet Gynecol, 2010; 115:535.