Safety communications warning against use of power

Transcription

1 Original Research Minimally Invasive Hysterectomy and Power Morcellation Trends in a West Coast Integrated Health System Eve Zaritsky, MD, Lue-Yen Tucker, BA, Romain Neugebauer, PhD, Tatiana Chou, MD, Tracy Flanagan, MD, Andrew J. Walter, MD, and Tina Raine-Bennett, MD, MPH OBJECTIVE: To examine trends in minimally invasive hysterectomy and power morcellation use over time and associated clinical characteristics. METHODS: We conducted a trend analysis and retrospective cohort study of all women 18 years of age and older undergoing hysterectomy for benign conditions at Kaiser Permanente Northern California collected from electronic health records. Generalized estimating equations and Cochran Armitage testing were used to assess the primary outcomes, hysterectomy incidence, and proportion of hysterectomies by surgical route and power morcellation. Logistic regression analysis was used to assess secondary outcomes, clinical characteristics, and complications associated with surgical route. RESULTS: There were 31,971 hysterectomies from 2008 to 2015; the incidence decreased slightly from 2.86 (95% confidence interval [CI] ) to 2.60 (95% CI ) per 1,000 women (P,.001). Minimally invasive See related editorial on page 976. From the Department of Obstetrics and Gynecology, the Permanente Medical Group, Oakland, Roseville, and Richmond, and the Division of Research and Regional Women s Health, Kaiser Permanente Northern California, Oakland, California. Presented at the American Association of Gynecologic Laparoscopists Annual Meeting, November 10 13, 2013, Washington, DC. The authors thank Fiona Sinclair, PA-C, MHS, for assistance with data acquisition and validation and Peter Heinlein, MD, for advice and consultation on the database. Each author has indicated that he or she has met the journal s requirements for authorship. Corresponding author: Eve Zaritsky, MD, Department of Obstetrics and Gynecology, Kaiser Permanente Northern California Oakland, 3600 Broadway, Oakland, CA 94611; eve.f.zaritsky@kp.org. Financial Disclosure The authors did not report any potential conflicts of interest by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved. ISSN: /17 hysterectomies increased from 39.8% to 93.1%, almost replacing abdominal hysterectomies entirely (P,.001). Vaginal hysterectomies decreased slightly from 26.6% to 23.4% (P,.001). The proportion of nonrobotic laparoscopic hysterectomies with power morcellation increased steadily from 3.7% in 2008 to a peak of 11.4% in 2013 and decreased to 0.02% in 2015 (P,.001). Robot-assisted laparoscopic hysterectomies remained a small proportion of all hysterectomies comprising 7.8% of hysterectomies in Women with large uteri (greater than 1,000 g) were more likely to receive abdominal hysterectomies than minimally invasive hysterectomy (adjusted relative risk 11.62, 95% CI ) and laparoscopic hysterectomy with power morcellation than without power morcellation (adjusted relative risk 5.74, 95% CI ). Laparoscopic supracervical hysterectomy was strongly associated with power morcellation use (adjusted relative risk 43.89, 95% CI ). CONCLUSION: A high minimally invasive hysterectomy rate is primarily associated with uterine size and can be maintained without power morcellation. (Obstet Gynecol 2017;129: ) DOI: /AOG Safety communications warning against use of power morcellation during hysterectomy issued by the U.S. Food and Drug Administration (FDA) in 2014 may have a negative influence on minimally invasive hysterectomy rates, patient safety, and surgical outcomes. 1 4 Investigators from Michigan and Florida reported decreasing use of laparoscopic routes and both increased complication and 30-day hospitalization rates from 2013 to ,6 Wright et al 7 used data from 2013 to 2015 from the Perspective database, which included more than 500 U.S. hospitals and demonstrated declines in the use of power morcellation ( %) with a decrease in the proportion of laparoscopic hysterectomies ( %, P,.001) and increase in abdominal hysterectomies ( %, P5.004). 996 VOL. 129, NO. 6, JUNE 2017 OBSTETRICS & GYNECOLOGY Copyright ª by The American College of Obstetricians

2 These studies showed an increase in abdominal hysterectomies after the FDA warnings, focused on short-term trends in minimally invasive hysterectomy, and lacked detailed clinical information to contextualize changes. We conducted a trend analysis and retrospective cohort study of women undergoing hysterectomy for benign conditions in a large integrated health care delivery system to understand how clinical characteristics are associated with minimally invasive hysterectomy and power morcellation. The primary objective of the study was to describe the trend in use of minimally invasive hysterectomy from 2008 to We wanted to examine changes after an initiative to increase minimally invasive hysterectomy and after the FDA issuance of safety warnings against use of power morcellation in The secondary objectives were to assess clinical and surgical characteristics associated with minimally invasive hysterectomy and power morcellation and assess relative risk of complications. MATERIALS AND METHODS In 2008, Kaiser Permanente Northern California implemented a quality improvement program to address the high rates of abdominal hysterectomies (62% in 2007) coupled with large numbers of lowvolume surgeons using limited to no minimally invasive approaches. The quality improvement program, entitled the Minimally Invasive Hysterectomy Initiative, included the following components: 1) leadership engagement to set goals and monitor outcomes, 2) surgical training, 3) surgeon tracking, and 4) use of data on hysterectomies to track outcomes and drive performance improvement. The obstetrics and gynecology departments in the 15 medical facilities across the region are led by chiefs who work together to establish regional priorities to improve the quality of care for patients. After review of regional data on the proportion of minimally invasive hysterectomies, defined as vaginal and laparoscopic hysterectomies, a desired benchmark of 60% was set in 2008 and was gradually increased to 90% in Robotic-assisted laparoscopic hysterectomy adoption was not part of the initiative. Rather, leadership came to a consensus to limit benign robotic hysterectomies to complex disease and to a small core group of surgeons with the commitment to retrospectively study and understand the appropriate indications. The initiative included Kaiser Permanentesponsored continuing medical education courses on laparoscopic approaches and attendance at external courses was promoted. In addition, proctoring on laparoscopic cases was required for surgeons who did not have residency training in laparoscopic techniques. As a result of the evidence demonstrating better surgical outcomes and higher minimally invasive hysterectomy rates in surgeons with higher hysterectomy volumes, the initiative also focused on decreasing the number of operating surgeons. 8,9 This was achieved through a process of tracking in which service chiefs reviewed hysterectomy caseloads to identify high-volume gynecologic surgeons and encouraged low-volume surgeons to voluntarily limit their clinical practice to office-based or obstetrics only or a combination of office practice coupled with obstetrics. As a result of the salaried payment system for physicians, this did not have an effect on physician income. Over the course of the initiative, the number of surgeons performing hysterectomies decreased from 416 in 2008 to 234 in Chiefs of services reviewed data on rates of minimally invasive hysterectomy by facilities across the region on a quarterly basis to identify high-performance facilities and disseminate information on best practices region-wide. The leadership structure and engagement also facilitated prompt region-wide reaction to the FDA safety communications on power morcellation in Leadership coordinated removal of power morcellators from operating rooms in all facilities and training of surgeons on alternate surgical techniques such as use of manual morcellation vaginally or through suprapubic or umbilical incisions with containment bags. We report minimally invasive hysterectomy trends over the time period of the initiative to Data were abstracted from Kaiser Permanente s electronic health record and regional claims systems. Oversight and approval to conduct this study were provided by the institutional review board of the Kaiser Foundation Research Institute. Validation of data for the study was carried out in several steps including separate chart reviews to validate procedure codes for hysterectomy route, other procedures, morcellation status, complications, and clinical characteristics such as body mass index (BMI, calculated as weight (kg)/ [height (m)] 2 ), parity, and uterine weight. In addition, study investigators selected a random sample of 2% of charts (n5600) from the final cohort for chart review and found that in 95% of cases (95% confidence interval %), electronically abstracted data were consistent with chart review data. For the primary study outcome, hysterectomy trends, we identified all women 18 years of age and older who underwent hysterectomy for benign gynecologic conditions from January 2008 to December 2015 at a Kaiser Permanente Northern Californiaowned hospital or contracted facility using the VOL. 129, NO. 6, JUNE 2017 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates 997

3 Table 1. Hysterectomy Incidence and Number of Hysterectomies, Year Variable Women aged 18 y or older 1,386,749 1,367,981 1,373,439 1,401,480 Total no. of hysterectomies 4,806 4,840 4,681 4,664 Hysterectomies for cancer Hysterectomies for benign indications 3,969 3,946 3,801 3,748 Benign hysterectomy incidence (95% CI) 2.86 ( ) 2.89 ( ) 2.77 ( ) 2.67 ( ) Benign hysterectomies by surgical route and power morcellation Abdominal 2,389 2,027 1,558 1,056 Nonrobotic laparoscopic with power morcellation Nonrobotic laparoscopic without power ,366 morcellation Robotic with power morcellation Robotic without power morcellation Vaginal 1,045 1,074 1, Supracervical hysterectomies k Abdominal Laparoscopic CI, confidence interval. * Trend in annual incidence; linear models were fitted using generalized estimating equations. Cochran Armitage testing was used to assess for linear trend in the proportion of hysterectomies by surgical route over time. Women with at least 1-month membership in the specified year. Excludes hysterectomies for cancer (endometrial54,837, ovarian or tubal51,517, cervical5640, other gynecologic54,532, and nongynecologic peritoneal cavity51,287), pregnancy-related5479, and route not known515. Rate per 1,000 women (excluded women with hysterectomy in previous years). k Supracervical abdominal hysterectomy proportion was the portion of abdominal hysterectomies. Supracervical laparoscopic hysterectomy proportion was the proportion of laparoscopic hysterectomies and included robotic hysterectomies and excluded laparoscopically assisted vaginal hysterectomies. International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) and the Current Procedural Terminology (CPT) codes. Hysterectomy cases were classified by surgical route as: 1) abdominal including total, supracervical, and minilaparotomy (ICD-9-CM: 68.3, 68.39, 68.4, 68.49, 68.9; CPT: 58150, 58152, 58180); 2) laparoscopic including total, supracervical, and laparoscopically assisted vaginal (ICD-9-CM: 68.31, 68.41, 68.51; CPT: , 58550, , ), and robot-assisted (17.4x and any other code for hysterectomy); and 3) vaginal hysterectomy (ICD-9-CM: 68.5, 68.59; CPT: 58260, , 58267, 58270, 58275, 58280, ). Laparoscopic cases were subdivided into cases with and without power morcellation. Women with pregnancy-related diagnoses, radical hysterectomy (ICD-9-CM: 68.6x, 68.7x; CPT: 58200, 58210, 58285, 58548, 58956), or cancer of the reproductive tract or peritoneal cavity identified at hospital discharge or outpatient diagnoses 30 days before and 30 days after hysterectomy or in the Kaiser Permanente Northern California Cancer Registry before through 3 months of the hysterectomy were excluded. Secondary study outcomes assessed included clinical and surgical characteristics and complications. Surgical characteristics and complications are not consistently available for hysterectomies done outside of Kaiser-owned facilities and are not included in the retrospective cohort analysis. Clinical characteristics included age (18 39, 40 49, 50 59, 60 years or older), race ethnicity (white, Hispanic, African American, Asian, other), and income (defined as low if address of residence was in a census block with 20% or greater households with income below the federal poverty level), parity, and BMI (24.9 or less, , , 40 or greater). Primary indication for surgery was classified as leiomyoma, abnormal bleeding, prolapse or incontinence, endometriosis, ovarian neoplasm, or other based on ICD-9-CM and International Classification of Diseases, 10th Revision, Clinical Modification codes recorded as the principal discharge diagnosis. Hysterectomies for dysplasia or endometrial hyperplasia were classified as precancerous conditions. Abdominal and laparoscopic cases (excluding laparoscopic-assisted vaginal hysterectomy) were classified as supracervical hysterectomy (yes or no). Additional procedures performed at the time of 998 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates OBSTETRICS & GYNECOLOGY

4 Year Total P* 1,423,193 1,446,804 1,526,049 1,617,221 11,542,916 4,992 5,046 5,210 5,179 39, , ,447 4,008 4,131 4,163 4,205 31, ( ) 2.86 ( ) 2.73 ( ) 2.60 ( ) 2.77 ( ), ,904, ,048,.001 1,829 2,038 2,351 2,602 12,116, , , ,924, , ,281,.001 hysterectomy (yes or no) including oophorectomy (either unilateral or bilateral), salpingectomy, and procedures for prolapse and incontinence were identified. We also created a variable any procedure coded as yes if any of these additional procedures were performed. Medical comorbidities were assessed using Fig. 1. Proportion of hysterectomies by surgical route and morcellation, Cochran Armitage testing was used to assess for linear trend in the proportion of hysterectomies by surgical route over time. FDA, U.S. Food and Drug Administration. Zaritsky. Power Morcellation and Minimally Invasive Hysterectomy Rates. Obstet Gynecol VOL. 129, NO. 6, JUNE 2017 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates 999 Copyright ª by The American College of Obstetricians

5 the Charlson Comorbidity Index; a score of greater than 1 was considered increased risk. 10 Characteristics of the hysterectomies including operative time (time from start of incision to incision closure) and postoperative length of hospital stay (time left the operating room to time of discharge from the hospital) were obtained. Uterine weights as a proxy for uterine size (500 or less, 501 1,000, and greater than 1,000 g with or without adnexa included in the specimen) and use of power morcellation (yes or no) were obtained from the dictated pathology reports and the operative notes (respectively) using natural language processing with I2E. Complications associated with the hysterectomy were identified using ICD-9-CM, International Classification of Diseases, 10th Revision, Clinical Modification, and CPT codes documented within 7 days of surgery and classified as: operative (bladder, ureteral, intestinal, vascular); medical (venous thromboembolism, cardiovascular, pulmonary, renal); and perioperative (hemorrhage, gastrointestinal obstruction, fistula, blood transfusion, wound). The primary surgeon s route-specific hysterectomy volume per surgical year was determined and grouped into three categories (low 1 10; medium 11 20; high greater than 20). 8,9 We calculated the annual incidence of hysterectomies from 2008 to 2015 using total number of hysterectomies as the numerator and number of women aged 18 years or older with one or more health plan membership months of the study year as the denominator. Unadjusted trend tests in annual incidence of benign hysterectomy were implemented by fitting a linear model using generalized estimating equations. 11 The test for trend in the incidence of benign hysterectomies over time was implemented by a standard z-test for the null hypothesis that the slope of the linear model was equal to zero using the robust estimate of the standard error using Stata 13. All other analyses were performed using SAS 9.3. The proportion of hysterectomies done by surgical route and power morcellation was calculated for each study year. Cochran Armitage testing was used to assess for linear trend in the proportion of hysterectomies by surgical route over time. We conducted bivariate analyses of clinical characteristics across all surgical routes using x 2 or for categorical variables and Kruskal-Wallis for continuous variables. P values for all overall comparisons were significant at the P,.001 level because of the large sample size and are not shown. Post hoc analyses of surgical characteristics (continuous variables) with comparisons between abdominal and minimally invasive hysterectomy were tested using Wilcoxon- Mann-Whitney. Multiple comparisons of surgical characteristics between laparoscopic hysterectomy with power morcellation and without power morcellation were tested using one-way analysis of variance with the rank-transformed method. Multivariate analyses were conducted using logistic regression to assess clinical characteristics associated with surgical route. The dependent variable was surgical route and was assessed in two separate models comparing: 1) abdominal compared with minimally invasive routes (laparoscopic and vaginal); and 2) laparoscopic with power morcellation compared with laparoscopic without power morcellation. Independent variables controlled for in the models were demographics, parity, BMI, Charlson Index, surgical indication, uterine weight, supracervical hysterectomy, additional procedures performed, surgeon hysterectomy volume, and year of surgery. Multivariate analyses were also conducted using logistic regression to assess the association between surgical route and complications. The dependent variable was occurrence of complications and was assessed in three separate models 1) operative complications; 2) medical complications; and 3) perioperative complications. Surgical route was considered an independent variable. Other independent variables controlled for in the models included age, race, income, BMI, Charlson Index, indication, uterine weight, any procedures (yes or no), and surgeon volume Given the large sample size and multiple comparisons, we considered an adjusted relative risk (RR) of greater than 3 or less than 0.33 to be clinically meaningful. 15 RESULTS A total of 39,911 hysterectomies were performed from 2008 to 2015; 31,971 (80%) were for benign gynecologic conditions. There were 7,447 hysterectomies associated with cancers (4,837 endometrial, 1,517 ovarian or tubal, 640 cervical, 4,532 other gynecologic, and 1,287 nongynecologic peritoneal cavity) and 493 pregnancy related or route unknown. Although the benign hysterectomy rate fluctuated over the 8-year period, there was an overall decrease from 2.86 (95% confidence interval [CI] ) in 2008 to 2.60 (95% CI ) per 1,000 women (P,.001) (Table 1). There were large changes in the distribution of hysterectomies for benign conditions by surgical route and morcellation over this time period (Fig. 1). From 2008 to 2015, minimally invasive hysterectomies increased from 39.8% to 93.1%, almost replacing abdominal hysterectomies entirely (P,.001) (Fig. 1). The proportion of vaginal hysterectomies decreased slightly from 26.3% in 2008 to 23.4% in 2015 (P,.001). The proportion of 1000 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates OBSTETRICS & GYNECOLOGY

6 Table 2. Patient and Clinical Characteristics by Hysterectomy Surgical Route, Minimally Invasive Hysterectomy Characteristic Abdominal (n58,520) Laparoscopic With Power Morcellation* (n52,109) Laparoscopic Without Power Morcellation* (n512,977) Vaginal (n57,779) Total (N531,385) Age group (y) ,144 (23.6) 263 (5.4) 2,337 (48.2) 1,103 (22.8) 4, ,503 (29.0) 1,277 (8.2) 6,778 (43.7) 2,956 (19.1) 15, ,063 (29.9) 443 (6.4) 2,736 (39.6) 1,660 (24.1) 6, or older 810 (19.7) 126 (3.1) 1,126 (27.3) 2,060 (50.0) 4,122 Race ethnicity White (25.5) 835 (5.4) 6,325 (41.1) 4,298 (27.9) 15,384 Hispanic 1,681 (25.0) 398 (5.9) 2,752 (41.0) 1,890 (28.1) 6,721 African American 1,402 (34.2) 434 (10.6) 1,769 (43.2) 490 (12.0) 4,095 Asian 1,089 (30.3) 314 (8.7) 1,502 (41.7) 694 (19.3) 3,599 Other 422 (26.6) 128 (8.1) 629 (39.7) 407 (25.7) 1,586 Low-income residence 953 (27.0) 209 (5.9) 1,560 (44.2) 809 (22.9) 3,531 Parity Nulliparous 1,559 (37.8) 421 (10.2) 1,944 (47.2) 202 (4.9) 4,126 Multiparous 6,539 (25.1) 1,575 (6.0) 10,483 (40.2) 7,471 (28.7) 26,068 Unknown 422 (35.4) 113 (9.5) 550 (46.2) 106 (8.9) 1,191 BMI (kg/m 2 ) Less than 25 1,924 (25.3) 581 (7.7) 3,036 (40.0) 2,058 (27.1) 7, ,585 (25.4) 662 (6.5) 4,060 (39.9) 2,876 (28.2) 10, ,056 (28.4) 683 (6.3) 4,635 (43.0) 2,398 (22.3) 10, or greater 946 (33.6) 180 (6.4) 1,246 (44.3) 444 (15.8) 2,816 Charlson Index greater than 1 2,161 (28.2) 404 (5.3) 3,109 (40.6) 1,988 (26.0) 7,662 Indication Abnormal bleeding 968 (16.5) 334 (5.7) 3,039 (51.8) 1,529 (26.1) 5,870 Leiomyoma 4,828 (37.6) 1,413 (11.0) 5,058 (39.4) 1,533 (12.0) 12,832 Prolapse or incontinence 305 (6.7) 141 (3.1) 595 (13.1) 3,486 (77.0) 4,527 Endometriosis 467 (31.9) 66 (4.5) 748 (51.2) 181 (12.4) 1,462 Ovarian neoplasm 800 (60.6) 28 (2.1) 461 (34.9) 32 (2.4) 1,321 Other 1,152 (21.4) 127 (2.4) 3,075 (57.2) 1,018 (19.0) 5,372 Precancer diagnosis 458 (21.0) 16 (0.7) 1,196 (54.8) 514 (23.5) 2,184 Uterine weight (g) 500 or less 5,466 (22.1) 1,342 (5.4) 10,962 (44.2) 7,020 (28.3) 24, ,000 1,378 (46.3) 438 (14.7) 1,079 (36.2) 82 (2.8) 2,977 Greater than 1, (69.0) 128 (9.6) 275 (20.7) 9 (0.7) 1,329 Unknown 759 (33.2) 201 (8.8) 661 (28.9) 668 (29.2) 2,289 Supracervical hysterectomy 2,223 (49.5) 1,471 (32.7) 800 (17.8) 0 4,494 Oophorectomy 4,436 (36.6) 559 (4.6) 5,187 (42.8) 1,944 (16.0) 12,126 Prolapse 341 (6.7) 156 (3.1) 876 (17.2) 3,725 (73.1) 5,098 Incontinence 355 (10.4) 113 (3.3) 577 (16.9) 2,362 (69.3) 3,407 Salpingectomy 867 (9.3) 704 (7.5) 6,263 (66.9) 1,534 (16.4) 9,368 Any procedure 5,331 (23.5) 1,273 (5.6) 10,569 (46.7) 5,473 (24.2) 22,646 Operative time (h) 2.0 ( ) 2.8 ( ) 2.4 ( ) 2.1 ( ) 2.3 ( ) 2.0 ( ) 2.4 ( ) 2.3 ( ) Postoperative stay (h) 48.5 ( ) 4.4 ( ) k 4.6 ( ) k 24.1 ( ) 22.2 ( ) 48.5 ( ) 9.2 ( ) 22.2 ( ) Estimated blood loss (ml) 200 ( ) 100 (50 150) k 95 (50 150) k 150 ( ) 100 (50 250) 200 ( ) 100 (50 200) 100 (50 250) BMI, body mass index. Data are n (%) or median (interquartile range). Bold indicates minimally invasive hysterectomy (laparoscopic and vaginal hysterectomies). Overall comparisons across surgical route tested using x 2 for categorical variables and Kruskal-Wallis for continuous variables; P,.001 for all comparisons (P values not shown in table). Comparisons between abdominal and minimally invasive hysterectomy were tested using Wilcoxon-Mann-Whitney for continuous variables; P,.001 for all comparisons (P values not shown in table). Multiple comparison between laparoscopic hysterectomy with and without morcellation for continuous variables were tested using one-way analysis of variance with the rank-transformed method. * Includes robotic-assisted cases. Low-income residence residence in neighborhood where 20% or greater of households with income below federal poverty level. Missing data: low-income residence571, BMI515 kg/m 2, Charlson index516, indication51, operative time574, postoperative stay574, estimated blood loss5937. P,.05. k P$.05. VOL. 129, NO. 6, JUNE 2017 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates 1001

7 Table 3. Multivariable Logistic Regression of Factors Associated With Undergoing Abdominal Hysterectomy and Power Morcellation* Factor Abdominal Hysterectomy (Ref Minimally Invasive Hysterectomy) P Laparoscopic Hysterectomy With Power Morcellation (Ref Without Power Morcellation) P Age (continuous) 1.01 ( ), ( ).15 Race White 1 (ref) 1 (ref) Hispanic 1.12 ( ) ( ).44 African American 1.11 ( ) ( ).27 Asian 1.20 ( ) ( ),.01 Other 0.96 ( ) ( ).75 Low-income residence 0.96 ( ) ( ).01 Parity Multiparous 1 (ref) 1 (ref) Nulliparous 1.29 ( ), ( ),.001 Unknown 1.09 ( ) ( ).01 BMI (kg/m 2 ) Less than 25 1 (ref)] 1 (ref) ( ) ( ) ( ), ( ) or greater 1.73 ( ), ( ).97 Charlson Index greater than ( ), ( ).61 Indication Abnormal bleeding 1 (ref) 1 (ref) Leiomyoma 2.02 ( ), ( ),.001 Prolapse or incontinence 0.62 ( ), ( ).30 Endometriosis 2.26 ( ), ( ).26 Ovarian neoplasm 7.82 ( ), ( ),.01 Other 1.40 ( ), ( ),.01 Precancer diagnosis 0.61 ( ), ( ),.001 Uterine weight (g) 500 or less 1 (ref) 1 (ref) 501 1, ( ), ( ),.001 Greater than 1, ( ), ( ),.001 Unknown 1.12 ( ) ( ),.001 Supracervical hysterectomy ( ),.001 Oophorectomy 1.94 ( ), ( ).59 Prolapse 0.38 ( ), ( ).85 Incontinence 0.92 ( ) ( ).02 Salpingectomy 0.65 ( ), ( ).01 Surgeon volume Medium 1 (ref) 1 (ref) Low 1.67 ( ), ( ),.001 High 0.24 ( ), ( ).79 Ref, reference; BMI, body mass index. Data are adjusted relative risk (95% confidence interval) unless otherwise specified. * Adjusted relative risk estimated using logistic regression with all variables in the table and year of surgery (not shown in table). Supracervical hysterectomy was not included in the model abdominal compared with minimally invasive hysterectomy. Low-income residence includes residence in neighborhood where 20% or greater of households with income below federal poverty level. Primary surgeon s route-specific hysterectomy volume per surgical year (low 1 10; medium 11 20; high greater than 20). nonrobotic laparoscopic hysterectomies with power morcellation increased steadily from 3.7% in 2008 to a peak of 11.4% in 2013 and decreased abruptly beginning in 2014 to 0.02% in 2015 (P,.001). Paralleling this trend, laparoscopic supracervical hysterectomies decreased from 66.9% in 2008 to 8.1% in 2015 (P,.001). Robotic-assisted laparoscopic hysterectomies (with or without power morcellation) remained a small proportion of all hysterectomies performed comprising a larger proportion than abdominal hysterectomies, but only 7.8% (all without power morcellation) of hysterectomies in Of the 31,971 hysterectomies performed from 2008 to 2015, 31,385 were performed in Kaiser 1002 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates OBSTETRICS & GYNECOLOGY

8 Table 4. Unadjusted and Adjusted Risk of Complications Complication n (%) Abdominal Hysterectomy (Ref Minimally Invasive Hysterectomy)* n (%) (Ref) Unadjusted RR (95% CI) Adjusted RR (95% CI) n (%) Laparoscopic Hysterectomy With Power Morcellation (Ref Without Power Morcellation) n (%) (Ref) Unadjusted RR (95% CI) Adjusted RR (95% CI) Operative 307 (3.6) 594 (2.6) 1.40 ( ) 1.27 ( ) k 58 (2.8) 356 (2.7) 1.00 ( ) 0.84 ( ) Medical 283 (3.3) 345 (1.5) 2.24 ( ) 2.41 ( ) 22 (1.0) 171 (1.3) 0.79 ( ) 0.84 ( ) Perioperative 497 (5.8) 379 (1.7) 3.68 ( ) 3.18 ( ) 35 (1.7) 223 (1.7) 0.97 ( ) 0.88 ( ) Ref, reference; RR, relative risk, CI confidence interval. * Includes laparoscopic and vaginal cases. Includes robotic-assisted cases. Adjusted relative risk estimated using logistic regression analysis controlling for age, race, body mass index, Charlson Index, indication, uterine weight, any procedures (yes or no), and surgeon volume. P,.001. k P,.01. Permanente facilities and are included in the bivariate and multivariate analyses (Tables 2 4). The uterine weight was less than 500 g for the majority of hysterectomies (79%). The majority of women with large uteri of greater than 1,000 g (69%) received abdominal hysterectomies (Table 2). As one would expect, the majority of women with prolapse or incontinence received vaginal hysterectomies (77%). Although operative time was shorter for abdominal hysterectomies (2.0 hours, interquartile range ) compared with minimally invasive hysterectomies (2.4 hours, interquartile range , P,.001), mean postoperative stay and estimated blood loss were greater for abdominal hysterectomies (48.5 hours, interquartile range and 200 ml, interquartile range , respectively) compared with minimally invasive hysterectomies (9.2 hours, interquartile range and 100 ml, interquartile range , respectively, P,.001). The associations among age, race, parity, and BMI were statistically significant in the multivariable analysis of characteristics associated with receiving abdominal compared with minimally invasive hysterectomy; however, there were clinically meaningful associations with uterine weight and indication for surgery (Table 3). Women with uterine weight of greater than 501 1,000 g (adjusted RR 3.73, 95% CI ) and greater than 1,000 g (adjusted RR 11.62, 95% CI ) and indication of ovarian neoplasm (adjusted RR 7.82, 95% CI ) were significantly more likely to receive an abdominal hysterectomy. Consistent with declining abdominal hysterectomy numbers, abdominal hysterectomies were less likely to be associated with surgeons with high hysterectomy volume (greater than 20 cases per year) (adjusted RR 0.24, 95% CI ). Among women undergoing laparoscopic hysterectomy (nonrobotic and robotic), supracervical hysterectomy was the factor with the strongest association with receiving power morcellation (adjusted RR 43.89, 95% CI ). Women undergoing laparoscopy with uterine weight of 501 1,000 g and greater than 1,000 g were also more likely to receive power morcellation (adjusted RR, 3.90, 95% CI and adjusted RR, 5.74, 95% CI , respectively). As one might expect, women with a precancer diagnosis were less likely to receive power morcellation (adjusted RR 0.27, 95% CI ). Overall complication rates were low; however, the highest rates of complications were seen in abdominal hysterectomies (Table 4). After controlling for age, race, BMI, Charlson Index, surgical indication, uterine weight, any procedures, and surgeon volume, women with abdominal hysterectomies were statistically significantly more likely to have operative and medical complications than women receiving minimally invasive hysterectomies (adjusted RR 1.27, 95% CI , P,.01 and 2.41, 95% CI , respectively, P,.001) and there was a clinically meaningful difference in perioperative complications (3.18, 95% CI , P,.001) (Table 4). There were no differences in adjusted risk of complications for laparoscopic hysterectomies with power morcellation compared with without. DISCUSSION The use of minimally invasive hysterectomy for benign conditions in this large integrated health care system increased from 39.8% in 2008 to 93.1% in Previous reports of hysterectomy trends have not revealed such high rates. Desai et al 16 reported VOL. 129, NO. 6, JUNE 2017 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates 1003

9 a benign minimally invasive hysterectomy rate of 47.1% in 2012 and Wright et al 2 reported a rate of 45.8% in Both of these were national inpatient samples, which may have underestimated rates. However, Harris et al 6 reported a rate of 75.9% in 2015 in a Michigan state inpatient and outpatient sample. The high rate in our study is likely the result of the implementation of the minimally invasive hysterectomy initiative aimed at quality improvement. Kaiser Permanente Southern California also reported a large increase in minimally invasive hysterectomy from 38% in 2005 to 78% in 2010 after regional efforts to increase minimally invasive hysterectomy. 17 Although some aspects of the strategy to increase minimally invasive hysterectomy at Kaiser Permanente such as surgeon training and proctoring could be implemented across different practice settings, it is not clear whether such high minimally invasive hysterectomy rates can be achieved in other practice settings. In contrast to other reports in the literature demonstrating a decrease in minimally invasive hysterectomy after the FDA safety communications, our high rate of minimally invasive hysterectomy persisted after the cessation in use of power morcellation in ,7 From 2008 to 2015, the number of laparoscopic hysterectomies without power morcellation exceeded the number with power morcellation, suggesting that the increase in minimally invasive hysterectomy was only weakly associated with morcellation technique. This study demonstrates the minimally invasive hysterectomy rate is primarily associated with uterine size. Large uterine size (greater than 1,000 g) was the factor with the strongest association with receipt of abdominal hysterectomy compared with minimally invasive hysterectomy. In addition, large uterine size was also associated with receipt of power morcellation during laparoscopic hysterectomy, although the association was not as strong. The only other factor that was strongly associated with power morcellation was supracervical hysterectomy. Laparoscopic supracervical hysterectomy numbers were already decreasing before the FDA safety communications, possibly as a result of growing comfort with laparoscopic surgical techniques. After the cessation of use of power morcellation in 2014, the number of laparoscopic supracervical hysterectomies continued to decrease. This suggests that maintenance of the high minimally invasive hysterectomy rate was probably the result of a combination of factors including less reliance on laparoscopic supracervical hysterectomies and adoption of alternate surgical techniques for removal of large uteri without power morcellation but via manual morcellation. The high minimally invasive hysterectomy rate at Kaiser Permanente was achieved with limited use of robotic technology. National data demonstrate rapid uptake of robotic-assisted surgery despite questions about the cost-effectiveness of the technology, with robotic-assisted hysterectomy accounting for 12.6% of laparoscopic hysterectomies in 2012 compared with 8.2% in 2010 in the National Inpatient Sample. 16,18 The difference in robotic hysterectomy uptake seen in our sample is most likely the result of several factors including limited availability of the robotic platform (only four robots across the region) and the minimally invasive hysterectomy initiative at Kaiser Permanente that limited use of the robotic platform. There was a slight decrease in the proportion of vaginal hysterectomies from 2008 to The quality improvement initiative at Kaiser Permanente Northern California had a primary focus on laparoscopic techniques. It is possible that a higher proportion of vaginal hysterectomies could be obtained with increased focus on training and patient selection. Evidence from a decade ago demonstrates that vaginal hysterectomy is associated with better outcomes and fewer complications than laparoscopic or abdominal hysterectomy. 3,19 We demonstrated increased morbidity for abdominal hysterectomies; however, comparative analysis of laparoscopic and vaginal hysterectomies is beyond the scope of this study. Future studies comparing surgical characteristics (eg, operative time and postoperative stay), complications, and costs of vaginal and laparoscopic hysterectomies are needed given the increase in experience with laparoscopic surgery in the last decade. This study s strengths include a large sample size and use of comprehensive electronic health records. We document trends and associations; however, we cannot infer causation for changes in practice over time. Interpretation of hysterectomy incidence may be limited because we did not account for potential loss of insurance in the denominator. Our study is also limited in that we did not examine nonpower morcellation or prior surgeries because this information is not reliably documented. Data may not be generalizable to other practice settings; however, this study demonstrates that minimally invasive hysterectomy rates can remain high without power morcellation. REFERENCES 1. Updated laparoscopic uterine power morcellation in hysterectomy and myomectomy: FDA safety communication. Available 1004 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates OBSTETRICS & GYNECOLOGY

10 at: ucm htm. Retrieved June 1, Wright JD, Herzog TJ, Tsui J, Ananth CV, Lewin SN, Lu YS, et al. Nationwide trends in the performance of inpatient hysterectomy in the United States. Obstet Gynecol 2013;122: Nieboer TE, Johnson N, Lethaby A, Tavender E, Curr E, Garry R, et al. Surgical approach to hysterectomy for benign gynaecological disease. The Cochrane Database Systematic Reviews 2009, Issue 3. Art. No.: CD Walsh CA, Walsh SR, Tang TY, Slack M. Total abdominal hysterectomy versus total laparoscopic hysterectomy for benign disease: a meta-analysis. Eur J Obstet Gynecol Reprod Biol 2009;144: Barron KI, Richard T, Robinson PS, Lamvu G. Association of the U.S. Food and Drug Administration morcellation warning with rates of minimally invasive hysterectomy and myomectomy. Obstet Gynecol 2015;126: Harris JA, Swenson CW, Uppal S, Kamdar N, Mahnert N, As-Sanie S, et al. Practice patterns and postoperative complications before and after US Food and Drug Administration safety communication on power morcellation. Am J Obstet Gynecol 2016;214:98.e Wright JD, Chen L, Burke WM, Hou JY, Tergas AI, Ananth CV, et al. Trends in use and outcomes of women undergoing hysterectomy with electric power morcellation. JAMA 2016; 316: Rogo-Gupta LJ, Lewin SN, Kim JH, Burke WM, Sun X, Herzog TJ, et al. The effect of surgeon volume on outcomes and resource use for vaginal hysterectomy. Obstet Gynecol 2010;116: Boyd LR, Novetsky AP, Curtin JP. Effect of surgical volume on route of hysterectomy and short-term morbidity. Obstet Gynecol 2010;116: Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol 1994;47: Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42: Twijnstra AR, Blikkendaal MD, van Zwet EW, van Kesteren PJ, de Kroon CD, Jansen FW. Predictors of successful surgical outcome in laparoscopic hysterectomy. Obstet Gynecol 2012;119: Driessen SR, Sandberg EM, la Chapelle CF, Twijnstra AR, Rhemrev JP, Jansen FW. Case-mix variables and predictors for outcomes of laparoscopic hysterectomy: a systematic review. J Minim Invasive Gynecol 2016;23: Brummer TH, Jalkanen J, Fraser J, Heikkinen AM, Kauko M, Mäkinen J, et al. FINHYST, a prospective study of 5279 hysterectomies: complications and their risk factors. Hum Reprod 2011;26: Grimes DA. Epidemiologic research with administrative databases: red herrings, false alarms and pseudo-epidemics. Hum Reprod 2015;30: Desai VB, Xu X. An update on inpatient hysterectomy routes in the United States. Am J Obstet Gynecol 2015;213: Andryjowicz E, Wray T. Regional expansion of minimally invasive surgery for hysterectomy: implementation and methodology in a large multispecialty group. Perm J 2011;15: Swenson CW, Kamdar NS, Harris JA, Uppal S, Campbell DA Jr, Morgan DM. Comparison of robotic and other minimally invasive routes of hysterectomy for benign indications. Am J Obstet Gynecol 2016;215:650.e Choosing the route of hysterectomy for benign disease. ACOG Committee Opinion No American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114: Build Your Reference List at The Green Journal s website now makes research easier. The full-text view of each article features an Article Tools box. Clicking on the link Export to Citation Manager allows you to export the citation information to EndNote or ProCite. rev 2/2017 VOL. 129, NO. 6, JUNE 2017 Zaritsky et al Power Morcellation and Minimally Invasive Hysterectomy Rates 1005