Factors Associated With Hemostatic Agent Use During Laparoscopic Hysterectomy

Accepted Manuscript Factors Associated with Hemostatic Agent Use During Laparoscopic Hysterectomy Andrea Kakos, M.D., Valerie Allen, M.D., James Whiteside, M.D.

PII:

S1553-4650(16)31034-2

DOI:

10.1016/j.jmig.2016.08.827

Reference:

JMIG 2939

To appear in:

The Journal of Minimally Invasive Gynecology

Received Date: 27 June 2016 Revised Date:

20 August 2016

Accepted Date: 23 August 2016

Please cite this article as: Kakos A, Allen V, Whiteside J, Factors Associated with Hemostatic Agent Use During Laparoscopic Hysterectomy, The Journal of Minimally Invasive Gynecology (2016), doi: 10.1016/ j.jmig.2016.08.827. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Factors Associated with Hemostatic Agent Use During Laparoscopic Hysterectomy Andrea Kakos, M.D., Valerie Allen, M.D., James Whiteside, M.D.

RI PT

Department of Obstetrics and Gynecology, University of Cincinnati, College of Medicine, Cincinnati, Ohio

M AN U

SC

Corresponding Author: James Whiteside, MD Division of Female Pelvic Medicine and Reconstructive Surgery University of Cincinnati College of Medicine 231 Albert Sabin Way Cincinnati, OH 45267-0526 office: 513-475-8588 [email protected] Disclaimer: The authors report no conflict of interest.

Presentation information: Poster Presentation Society of Gynecologic Surgeons 42nd Annual Scientific Meeting Palm Springs California, April 10-13, 2016

TE D

Key words: hemostats, fibrin sealants, hysterectomy

EP

Précis: Surgical metrics related to operative bleeding are not associated with use of hemostatic agents during laparoscopic hysterectomy.

AC C

Short version of title: Hemostatic agent use in laparoscopic hysterectomy

1

ACCEPTED MANUSCRIPT

ABSTRACT Study Objectives: To evaluate factors associated with the use of hemostatic agents during traditional laparoscopic or robotic hysterectomy

RI PT

Design: A retrospective cohort trial (Canadian Task Force III)

Setting: A single medical center, Cincinnati Ohio, 8/1/2013 – 7/31/2014

with and without the use of hemostatic agents

SC

Patients: Women undergoing traditional laparoscopic or robotic hysterectomies

Intervention: The use of hemostatic agent at the time of hysterectomy

M AN U

Measurements: Patient characteristics and pre- and postoperative metrics were recorded for each subject. Associations between categorical variables were analyzed using Chi-square testing while continuous variables were analyzed using ANOVA. Modeling of study variables to predict hemostatic agent use was

TE D

performed using Chi-square-assisted interaction detection (CHAID) methods. Main Results: The study sample included 176 cases performed by 30 surgeons. In our sample, 42% of minimally invasive hysterectomies were performed with

EP

the surgical robot (RALH); the remainder of minimally invasive hysterectomies by approach was as follows: Total laparoscopic (TLH) 27%; laparoscopic-assisted

AC C

vaginal hysterectomy (LAVH) 16%; laparoscopic supracervical hysterectomy 15%. 46% (81/176) of cases recorded use of a fibrin hemostat; 26% (46/176) involved an alternative hemostat and 28% (49/176) of cases did not use any hemostat. By surgical approach, no hemostatic agent use was seen most often among LAVH, alternative hemostats were most often used during TLH. RALH and LSCH were most associated with fibrin-based hemostats. Use of any

2

ACCEPTED MANUSCRIPT

hemostatic agent did not result in clinical significant blood loss relative to cases where no product was used. The study variable identified most predictive of hemostat use by the CHAID regression tree model was surgeon identity.

RI PT

Conclusions: Hemostatic agent use during traditional laparoscopic and robotic hysterectomy does not appear to be associated with operative bleeding but is

AC C

EP

TE D

M AN U

SC

related to surgeon identity.

3

ACCEPTED MANUSCRIPT

1 2

INTRODUCTION: Hemostasis is a hallmark of good surgical technique. Hemostatic agents are among a variety of mechanisms available to surgeons to achieve hemostasis

4

during surgery. Fibrin-based products as hemostatic agents became

5

commercially available in the late 1990’s evolving since that time into more roles

6

including tissue sealants and adhesives.1 Examples of commercially available

7

flowable hemostats include Floseal (Baxter Healthcare Corporation, Deerfield,

8

Illinois USA) and Surgiflo (Ethicon Inc., Somerville, NJ, USA). Tisseel (Baxter

9

Healthcare Corporation, Deerfield, Illinois USA) and Evicel (Ethicon Inc.,

M AN U

SC

RI PT

3

10

Somerville, NJ, USA) are examples of fibrin-based sealant products. Both

11

products have been used in gynecologic surgery.2

To date, there is little comparative clinical evidence between classes of

13

products or product manufacturers.3 No identified clinical evidence favors one

14

product and there is likewise nothing to support the routine use of these products

15

in gynecologic surgery.4 Given the cost of commercially available hemostatic

16

agents and the paucity of evidence to recommend their routine use in

17

gynecologic surgery, responsible use of traditional hemostasis techniques such

18

as pressure or cautery could pose an opportunity for cost savings.

EP

AC C

19

TE D

12

Use of hemostatic agents during traditional laparoscopic or robotic

20

hysterectomy appears to have increased although the justification is not clear

21

particularly in light of recognized lower bleeding complications using these

22

surgical approaches.4 Use of these agents to seal denuded surfaces and to

23

reinforce the vaginal cuff has been unsubstantially reported between surgeons.

4

ACCEPTED MANUSCRIPT

Evidence to endorse this role has not been identified although it does correspond

25

with the relatively higher risk of cuff dehiscence using a total laparoscopic

26

approach.5

27

RI PT

24

We sought to understand factors associated with hemostat use during

laparoscopic hysterectomy and compare clinical outcomes between cases with

29

and without these products. It is our hypothesis that use of hemostatic agents

30

during traditional laparoscopic or robotic hysterectomy is associated with

31

measures linked to operative bleeding.

MATERIALS AND METHODS:

34

Sample

35

TE D

33

M AN U

32

SC

28

A retrospective chart review was performed for patients who have undergone laparoscopic hysterectomy with and without the use of hemostatic

37

agents. Billing data and procedure codes were used to identify all hysterectomy

38

cases with and without the use of hemostatic agents in our institution from

39

8/1/2013 – 7/31/2014. Study inclusion criteria was any woman who underwent a

40

robotic or traditional laparoscopic hysterectomy, for any indication, during the

41

study period.

42 43

AC C

EP

36

Our initial data collection was ordered by surgical date and total procedure cost. Given concerns that a bias could be introduced by this sequential order a

5

ACCEPTED MANUSCRIPT

44

second random sample was collected. Randomization was performed with a

45

computer-based random number generator.

Demographic features collected include race, age and BMI, menopausal

47

status, smoking status, and indication for hysterectomy. Significant medical co-

48

morbidities, such as history of bleeding diathesis and prior pelvic surgeries were

49

also recorded. Pre operative use of anticoagulation, intraoperative estimated

50

blood loss (EBL), pre- and post-operative hemoglobin, as available, as well as

51

the need for a blood transfusion were collected. Uterine pathology, weight,

52

hospital length of stay, and post-operative complications (if available) were

53

collected. That information was entered into an EXCEL (Microsoft, Redmond

54

OR, USA) database.

55

Data Analysis

SC

M AN U

TE D

56

RI PT

46

Preliminary frequency analyses were carried out across the variables collected relative to hemostatic agent use. Chi-square tests of independence

58

were used for categorical variables and independent t-tests were used for

59

continuous variables testing for differences across the variables relative to

60

hemostat agent use.

AC C

61

EP

57

Chi-square-assisted interaction detection (CHAID) methods were used to

62

create a growing tree analysis. Exhaustive CHAID analysis was performed to

63

determine the best model for identifying factors associated with hemostat use.

64

Cross validation of the final model was performed using a 5-folded k-fold

65

analysis.

6

ACCEPTED MANUSCRIPT

66

CHAID model specification details are reported in the Appendix 1. CHAID analysis began with the entire patient study cohort and searched across the

68

collected variables for patient subgroups that most powerfully predicted the

69

dependent variable (hemostatic agent use). The cohort was then divided based

70

on the best predictor variable creating “child” nodes. This process continued to

71

search each child node for the next best predictor of the dependent variable until

72

the CHAID model stopping rules were applied creating end nodes.

SC

RI PT

67

73

75

RESULTS:

M AN U

74

The billing query resulting in 217 gynecological cases with hemostatic agent charged and 1727 cases without hemostatic agent charged, from this pool

77

only laparoscopic and robotic hysterectomy cases were considered. One

78

hundred ten sequential cases were selected from this pool, however one case

79

was excluded in data analysis for being a missed conversion to open

80

hysterectomy. The second random sample of 100 cases was developed with 33

81

duplicates identified between this and the first sequential sample. After

82

comparing the two samples and finding no remarkable differences across the

83

included measures (data not shown), the pooled 176 cases composed the study

84

sample unless otherwise indicated. See Figure 1.

AC C

EP

TE D

76

85

Thirty surgeons were associated with the 176 hysterectomies in our

86

sample. The most common hysterectomy approach in our sample was the robot

87

at 42%, followed by total laparoscopic (TLH) at 27%, laparoscopic-assisted

7

ACCEPTED MANUSCRIPT

(LAVH) at 16% and laparoscopic supracervical hysterectomy at 15%. 81 (46%)

89

cases involved the use of a fibrin hemostatic agent and 46 (26%) cases involved

90

an alternative hemostatic agent (Table 1). By surgical approach, no hemostat

91

use was seen most often among LAVH, alternative hemostats were most used

92

during TLH. RALH and LSCH were most associated with fibrin-based

93

hemostats.

SC

94

RI PT

88

Table 2 lists the perioperative variables collected by what hemostatic agent they received during the surgery. Overall, there were no identified

96

differences across clinical variables by whether or not these products were used

97

although product use did vary by hysterectomy approach and EBL.

99

In our sample, LAVH was associated with significantly more blood loss (approximately 100cc) than the other surgical approaches. LAVH was also most

TE D

98

M AN U

95

associated with no hemostatic agent use. Longer OR times were noted in the

101

RALH group (median time 125 minutes, interquartile range 110-151 minutes).

102

The change in hemoglobin was not significant by hemostatic agent product used

103

or hysterectomy approach. Uterine weight was not associated with hysterectomy

104

approach. A patient history of bleeding diathesis as well as pre-operative

105

anticoagulant use by type of hemostatic agent used were not significant. There

106

was a strong association between product use, approach to hysterectomy, and

107

surgeon.

AC C

EP

100

108

Surgical energy used to perform the hysterectomy was analyzed by

109

bleeding and hemostatic agent use. Surgical cases utilizing the Harmonic Ace ®

8

ACCEPTED MANUSCRIPT

(Ethicon Endo Surgery, Inc., Cincinnati OH) were more likely to get alternative

111

hemostatic agents. Those cases that utilized a PKTM Gyrus (ACMI) were more

112

likely to get fibrin hemostats. LigaSureTM (Medtronic, Minneapolis MN), Enseal

113

(Ethicon Endo Surgery, Inc., Cincinnati OH) and/or a combination of

114

electrocautary were not significantly associated with any one type of hemostatic

115

agent.

SC

116

RI PT

110

The study variable identified most predictive of hemostat use by the CHAID regression tree model was surgeon identity (overall R2 0.69; K fold R2

118

0.65; surgeon identity accounted for 78% of the model). This finding was

119

consistent no matter which sample cohort we ran the analysis on (e.g. the 109

120

original cases, the 67 random (non duplicative charts), or the total 176 cases).

121

The surgeon identity never contributed less then 75% to any CHAID model to

122

explain hemostatic product use across the analyzed traditional laparoscopic and

123

robotic hysterectomies.

TE D

M AN U

117

126

CONCLUSION:

The most important finding from this study is that the use of hemostatic

AC C

125

EP

124

127

agents at the time of traditional laparoscopic or robotic hysterectomy is not

128

associated with aggregate differences in intra and postoperative bleeding metrics

129

or length of stay relative to like hysterectomies done without routine use of

130

hemostats. This finding leads us to reject our hypothesis in favor of the null. The

131

surgeon, above all else, determined whether a hemostatic agent was used during

9

ACCEPTED MANUSCRIPT

our reviewed cases. Our findings support the idea that surgeons are using these

133

products as prophylaxis. Indeed, several reviewed operative notes reporting

134

applying the product to “denuded peritoneal surfaces” without any indication of

135

bleeding at these sites.

136

RI PT

132

A Cochrane review on fibrin sealants concluded these products are

efficacious in reducing postoperative blood loss and perioperative transfusions,

138

with the evidence being strongest in orthopedic surgery.6 Data from within other

139

surgical specialties showed only a non-significant trend toward these endpoints.

140

For example, in a retrospective study on the efficacy of the fibrin sealant Evicel

141

(Ethicon Inc., Somerville, NF, USA) use during robot-assisted partial

142

nephrectomy, no significant decrease in rate of complications, blood loss, or

143

hospital stay was found compared to the control group7. As documented by our

144

study, there is no compelling argument to believe this conclusion is not likely true

145

in gynecologic surgery.

M AN U

TE D

Value is the quotient of care outcome and cost. Seeking high-value care

EP

146

SC

137

is a focus of current health reforms in the United States. One important

148

consideration in the routine use of hemostatic agents is cost. The hospital cost

149

for fibrin sealants such as Evicel (Ethicon Inc., Somerville, NF, USA) and Tisseel

150

(Baxter Healthcare Corporation, Deerfield, Illinois USA) range from one hundred

151

eighty to over five hundred dollars. The hospital cost for hemostatic agents such

152

as Floseal Hemostatic Matrix (Baxter Healthcare Corporation, Deerfield, Illinois

153

USA) can range from nine hundred to fifteen hundred dollars. Other agents can

AC C

147

10

ACCEPTED MANUSCRIPT

be just as costly; Surgical SNoW Absorbable Hemostat, Ethicon Inc., Somerville,

155

NF, USA can range from six hundred to thirteen hundred dollars. Given no

156

identified care outcome improvement with use of these products across our

157

sample these costs pose a value challenge. If half of all minimally invasive

158

hysterectomies performed at our institution (approximately 300 cases) used the

159

Floseal product the minimum hospital cost would total nearly a quarter of a

160

million dollars annually.

SC

Our study is limited by selection bias. We used one institution and two

M AN U

161

RI PT

154

different sampling methods, sequential and random. We recognize that our

163

results may not be generalizable to every hospital. Mitigating our hospital

164

sample limitations, it was reassuring that our study parameters and modeling

165

results were not remarkably difference across the two sample populations

166

(sequential and random). Another limitation is that reported complications were

167

only those recorded in our hospital electronic medical record system.

168

Complications that occurred in private physician offices not requiring readmission

169

to our hospital would be missed. We cannot account for this omission.

EP

There are numerous examples in medicine wherein care practices are

AC C

170

TE D

162

171

associated with no improved clinical outcomes but large differences in cost.

172

Regional variations in clinical practices without apparent differences in clinical

173

outcomes among Medicare patients has been well described. There are settings

174

in which hemostatic agents can provide helpful, perhaps life-saving, benefit to

175

women undergoing gynecologic surgery. The routine use, however, cannot be

11

ACCEPTED MANUSCRIPT

supported on the basis of clinical evidence. The use of hemostatic agents for

177

roles outside of hemostasis is also unsupported by the clinical evidence. It is

178

hoped that our study may render some insight into the use of these products that

179

either justifies their wider utilization or retrains it.

AC C

EP

TE D

M AN U

SC

180

RI PT

176

12

ACCEPTED MANUSCRIPT

181 182

REFERENCES:

183

1

184

Journal of Surgery, 34(4), 632-634. doi:10.1007/s00268-009-0252-7 [doi] 2

RI PT

185

Spotnitz, W. D. (2010). Fibrin sealant: Past, present, and future: A brief review. World

Duenas-Garcia, O. F., & Goldberg, J. M. (2008). Topical hemostatic agents in

gynecologic surgery. Obstetrical & Gynecological Survey, 63(6), 389-94; quiz 405.

187

doi:10.1097/OGX.0b013e31816ff805 [doi] 3

Spotnitz, W. D., & Burks, S. (2012). Hemostats, sealants, and adhesives III: A new

M AN U

188

SC

186

189

update as well as cost and regulatory considerations for components of the surgical

190

toolbox. Transfusion, 52(10), 2243-2255. doi:10.1111/j.1537-2995.2012.03707.x

191

[doi] 4

Wright, J. D., Ananth, C. V., Lewin, S. N., Burke, W. M., Siddiq, Z., Neugut, A. I., . . .

TE D

192

Hershman, D. L. (2014). Patterns of use of hemostatic agents in patients

194

undergoing major surgery. The Journal of Surgical Research, 186(1), 458-466.

195

doi:10.1016/j.jss.2013.07.042 [doi]

5

197

after different modes of hysterectomy. Obstet Gynecol. 2011;118(4):794-801. doi:

198

199

Hur HC, Donnellan N, Mansuria S, Barber RE, Guido R, Lee T. Vaginal cuff dehiscence

AC C

196

EP

193

10.1097/AOG.0b013e31822f1c92 [doi].

6

Carless PA, Henry DA, Anthony DM. Fibrin sealant use for minimizing peri-operative

200

allogeneic blood transfusion. Cochrane Database Syst Rev. 2003;(2)(2):CD004171.

201

doi: 10.1002/14651858.CD004171 [doi].

13

ACCEPTED MANUSCRIPT

202

7

Cohen, J., Jayram, G., Mullins, J. K., Ball, M. W., & Allaf, M. E. (2013). Do fibrin sealants impact negative outcomes after robot-assisted partial nephrectomy? Journal

204

of Endourology / Endourological Society, 27(10), 1236-1239.

205

doi:10.1089/end.2013.0136 [doi]

AC C

EP

TE D

M AN U

SC

RI PT

203

14

ACCEPTED MANUSCRIPT

Appendix 1

k-fold, 5 Folded Exhaustive CHAID 0.05 ML ratio 5 levels 20

AC C

EP

TE D

M AN U

SC

Cross validation Growing tree method Alpha for splitting Chi-square statistic Maximum tree depth Minimum number of cases for parent node

RI PT

Summary of CHAID Model Specifications

ACCEPTED MANUSCRIPT

Table 1: Study Defined Fibrin Hemostats and Alternative (Other) Hemostats

RI PT

Alternative Hemostats Floseal Baxter Healthcare Corporation, Deerfield, Illinois USA Surgicel Fibrillar Absorbable Hemostat, Ethicon Inc., Somerville, NF, USA Surgicel Nu-knit, Ethicon Inc., Somerville, NF, USA Surgicel SNoW Absorbable Hemostat, Ethicon Inc., Somerville, NF, USA, Surgiflo Ethicon Inc., Somerville, NJ, USA

AC C

EP

TE D

M AN U

SC

Fibrin Hemostats Tisseel Baxter Healthcare Corporation, Deerfield, Illinois USA Evicel Ethicon Inc., Somerville, NJ, USA

ACCEPTED MANUSCRIPT

Table 2: Comparison of Surgical Variables to the Classification of Product Used

Age (median, range) ASA (mean, SD) BMI Prior Major Pelvic Surgery*$ Pre-op Dx AUB Pre-op Dx Fibroids Hysterectomy Approach

Product Used Fibrin Hemostatic Other Hemostatic Agent (81) Agent (46) 45 (22-77) 46.5 (29-70) 2.1 (0.6) 1.9 (0.6) 30.3 (7.6) 30.1(7.5) 30.1% 28.3% 59.3% 58.7% 30.9% 50%

SC

No Hemostatic Agent (49) 47(28-72) 1.9 (0.6) 28.0 (6.9) 24.5% 55.1% 44.9%

RI PT

Variable

p Value .73 .11 .2 .74 .9 .11

31.9% 0% 68.1% <.0001 62.1% 6.9% 31% <.0001 16.2% 81.1% 2.7% <.0001 15.4% 73.1% 11.5% <.01 EBL (cc) 123.4 (95.5) 86.2 (101.2) 140.3 (102) .009** Bleeding Noted in OP note 0% 6.2% 10.9% .07 Change in Hb (mean or median?, SD) 1.9 (.86) 2.1(1.2) 1.9 (.89) .6 Transfusion 1 0 0 ..27 OR time (minutes, median, range) 114 (43-258) 115 (64-251) 109 (59-208) .31 LOS (median, range) 1.1 (1-2) 1.0 (0-3) 1.0 (0-2) Uterine Weight (g) 166.9 (120.1) 152.1 (118.3) 191.7 (143.4) .25 Post Op Dx Inactive Endometrium*# 46.9% 34.7% 32.6% .27 # Post Op Dx Active Endometrium* 53.1% 61.3% 60.9% .62 Post Op Dx Fibroids* 71.4% 51.9% 56.5% .07 Significance determined using ANOVA or Pearson Chi-square (*) as appropriate with p< 0.05 regarded as significant. $ Major pelvic surgery in this cohort is defined as history of myomectomy or cesarean section. ** ANOVA significance exist between other hemostatic agent and fibrin, no significance found comparing other hemostatic agent to no hemostatic agent or fibrin to no hemostatic agent. Notably, while a consistent trend, fibrin use was not associated with statistically significant EBL relative to other or no product use in the first non-random sample. # Classification of inactive and active endometrium included 170 of the 176 in the sample.

AC C

EP

TE D

M AN U

TLH LAVH RALH LSCH

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Précis: Surgical metrics related to operative bleeding are not associated with use of hemostatic agents during laparoscopic hysterectomy.