OBJECTIVE: To compare the surgical outcomes of robot-assisted laparoscopic myomectomy (robot-assisted), standard laparoscopic myomectomy (laparoscopic), and open myomectomy (abdominal).
METHODS: Myomectomy patients were identified from the case records of the Cleveland Clinic and stratified into three groups. Operative and immediate postoperative outcomes were compared. Data analysis was performed using analysis of variance, Kruskal-Wallis analysis of ranks, χ2, and Fisher exact tests where appropriate.
RESULTS: From a total of 575 myomectomies, 393 (68.3%) were abdominal, 93 (16.2%) were laparoscopic, and 89 (15.5%) were robot-assisted. The three groups were comparable regarding the size, number, and location. Significantly heavier myomas were removed in the robot-assisted group (223 [85.25, 391.50] g) compared with the laparoscopic group (96.65 [49.50, 227.25] g, P<.001) and were lower than in the abdominal group (263 [ 90.50, 449.00] g, P=.002). Higher blood loss was reported in the abdominal group compared with the other two groups, with a median (interquartile range) of blood loss in milliliters of 100 (50, 212.50), 200 (100, 437.50) and 150 (100, 200) in the laparoscopic, abdominal, and robot-assisted groups, respectively. The actual surgical time in minutes was 126 (95, 177) in the abdominal group, 155 (98, 200) in the laparoscopic group, and 181 (151, 265) in robot-assisted group (P<.001). Patients in the abdominal group had a higher median length of hospital stay of 3 (2, 3) days, compared with 1 (0, 1) day in the laparoscopic group and 1 (1, 1) days in the robot-assisted group (P<.001).
CONCLUSION: Robotic-assisted myomectomy is associated with decreased blood loss and length of hospital stay compared with traditional laparoscopy and to open myomectomy. Robotic technology could improve the utilization of the laparoscopic approach for the surgical management of symptomatic myomas.
LEVEL OF EVIDENCE: II
Robot-assisted surgery by appropriately trained surgeons may allow more myomectomies to be performed by a laparoscopic approach.
From Obstetrics, Gynecology and Women's Health Institute, Cleveland Clinic, Cleveland, Ohio; Cleveland Clinic Florida, Weston, Florida; Quantitative Health Sciences, Cleveland Clinic Foundation, Cleveland, Ohio; the Department of Obstetrics and Gynecology, Benha University, Benha, Egypt; and the Department of Obstetrics and Gynecology, Assiut University, Assiut, Egypt.
Winner of the Society of Reproductive Surgeon's in Training Award at the 66th annual meeting of the American Society for Reproductive Medicine, October 23–27, 2010, Denver, Colorado.
Corresponding author: Tommaso Falcone, MD, Department of Obstetrics and Gynecology, A81, The Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH; e-mail: email@example.com.
Financial Disclosure The authors did not report any potential conflicts of interest.
Myomas are the most common tumor in women during the reproductive age.1 Symptomatic myomas are often associated with excessive menstruation, abnormal uterine bleeding, pressure symptoms, infertility, or all of these.2 Since it was first described by Victor Bonney in 1931, myomectomy remains the gold standard surgical treatment for women desiring future fertility and uterine conservation.3,4 Since its introduction by Kurt Semm in 1979,5 laparoscopic myomectomy brought surgical benefits including smaller scars, decreased postoperative pain, less need for postoperative analgesics, less blood loss, and shorter convalescence period followed by rapid return to the normal activity.3,6–8
These benefits were also substantiated by the only prospective randomized controlled trial published with pregnancy as the primary outcome.8 Despite the fact that it is becoming increasingly popular, laparoscopic myomectomy remains underutilized because of its inherent challenges and limitations. These challenges include dissection of the myoma from its bed using the correct plane and multilayer closure of the myoma bed. These limitations could affect the strength of the scar and the subsequent risk of uterine rupture. The risk of uterine rupture in pregnancy after a laparoscopic myomectomy has been reported to be 1%.9 Given this particular risk, laparoscopic myomectomy is currently performed only by expert laparoscopists.3,10–12
Since its very initial use in Reproductive Surgery,13 the surgical robot has been shown to overcome the difficulties encountered in conventional operative laparoscopy. The 3-dimensional vision, improved ergonomics, wide range of movements, absence of the fulcrum effect along with the improved instrument dexterity eliminated most of the limitations of traditional laparoscopy. Since its approval by the Food and Drug Administration, the da Vinci robot has been used successfully in many gynecologic surgical procedures including but not limited to hysterectomy, myomectomy, sacrocolpopexy, and tubal reanastomosis.14–16
Earlier reports on robotic myomectomy denoted several advantages. First, it enabled precise dissection of the myoma using the endowrist instruments. Second, it helped to access myomas in difficult anatomic locations. In addition, it simplified the complex task of suturing the myoma bed precisely with proper approximation of the edges with adequate multilayering. Few publications have reviewed the surgical outcome of robotic myomectomy. These include a case series of 35 patients,17 a report of full-term uncomplicated pregnancy following robotically assisted myomectomy,18 and a case of a large myoma that had been successfully removed with robotic assistance.19 More recently, the surgical outcome of robotic myomectomy was compared with the standard laparoscopic procedure20,21 and to the open approach independently.22 In addition, the effect of body mass index (BMI, calculated as weight (kg)/[height (m)]2) on surgical outcomes of 77 patients who had undergone robotic myomectomy were reported in 2009.23 Our objective is to compare the surgical outcome of open myomectomy (abdominal), laparoscopic myomectomy (laparoscopic), and robot-assisted myomectomy (robot-assisted).
PATIENTS AND METHODS
This study was approved by the Institutional Review Board (IRB) of the Cleveland Clinic Foundation. A total number of 575 myomectomies performed between January 1995 and December 2009 were included in this study. These patients were stratified into three groups according to the surgical route. Group I included 393 patients who underwent open abdominal myomectomy between 1995 and 2009. Group II included of 93 patients who underwent laparoscopic myomectomy between 2003 and 2009. Group III included a series of 89 patients who underwent robotically assisted myomectomy between January 2008 and December 2009 using the da Vinci surgical system. Patients' data were retrieved from our electronic medical recording system as well as the patients' paper charts. Demographic data including age, height, weight, BMI, parity, history of previous abdominal surgery, and indication of the current myomectomy were collected. Preoperative characteristics of the myomas including the number, location, depth of infiltration, the maximum diameter and the size as evaluated by pelvic ultrasound or magnetic resonance imaging or both were included. The weight of the myomas was evaluated postoperatively as part of the histopathologic evaluation. Surgical outcome measures included variables describing the events related to the surgical procedure including the actual surgical time in minutes, estimated blood loss, hemoglobin drop, intraoperative and postoperative complications, and the length of hospital stay were collected. Postoperative hemoglobin was measured two or three times over the postoperative period in our institution depending on the patient's condition. We did include only the measurement performed around 6 am on postoperative day 1 in all cases for standardization purposes.
The robotic surgical technique was similar to what is published elsewhere (see the video online at http://links.lww.com/AOG/A218). Briefly, the patient was placed in the low dorsal lithitomy position with the arms tucked by their sides. A Foley catheter was inserted and the bladder and a uterine manipulator were placed. The 12-mm camera port was placed in the midline at the umbilicus or 8–10 cm above the upper border of the uterus. The two 8-mm side ports were inserted 8 cm lateral and at 15 degrees below the camera port. A third 8-mm robotic port was occasionally used. Another 10-mm port was also placed. This port was used by the assistant to introduce or remove the sutures, manipulate the myoma with a tenaculum, for morcellation, and for suction-irrigation. After inserting all trocars, the patient was placed in a steep Trendelenburg and the surgical cart with four robotic arms was brought between the patient's legs and each trocar was attached to the assigned robotic arm.
At first the pelvic cavity was scanned with evaluation of the targeted myomas regarding the number, location, and size and to exclude other pelvic pathology. Then diluted vasopressin (20 units in 100–200 mL of saline) was injected into the myoma to ensure adequate hemostasis using a 16-gauge spinal needle. Subsequently, the uterine incision was made over the prominent bulge of the myoma using monopolar cautery or a harmonic scalpel. Occasionally the myoma was excised with conventional laparoscopy before docking the robot. Then the robotic endowrist instruments were used to dissect the myoma out of its bed. Countertraction was provided through the accessory port using a corkscrew or a tenaculum. When the myoma was freed completely all around, the vascular pedicle was coagulated and the myoma was removed from the bed. Subsequently, multilayered closure of the bed—typically between two and four layers—using interrupted and continuous polyglactin 910 (Vicryl) or polydioxanone suture (PDS II, Ethicon) was performed with adequate hemostasis. The uterine serosa was closed securely using a continuous running polydioxanone suture. In general, either polyglactin 910 or polydioxanone 0 was used for the muscularis layer and 2-0 polydioxanone for the serosa layer. After closure of the myoma bed, an electric tissue morcellator was introduced through the accessory port and specimens were extracted.
Conventional laparoscopy was performed with a technique similar to the robotic one except suturing was done with laparoscopic needle holders. Open myomectomy was performed by standard reported techniques including the use of dilute vasopressin and tourniquets for the uterine and ovarian vessels.24
Following visual examination for normality, symmetrically distributed numerical variables were summarized with means and standard deviations while other variables were summarized with medians and ranges. Frequencies and percentages were used to summarize categorical measures. For numerical variables, univariable comparisons are conducted with either analysis of variance (for normally distributed data) or Kruskal-Wallis analysis of ranks (for non-normal data). χ2 and Fisher exact tests were employed for categorical data. Dunn's test has been employed for pair-wise comparisons following the Kruskal-Wallis test.
To model the length of stay, the data were categorized into groups of 0–1 day, 2–3 days, or 4 or more days. Categorization was done because the residuals for a linear model with the number of days did not meet the assumptions of normality nor constant variance; the length of stay is modeled using the proportional odds logistic regression.25
Hemoglobin drop was modeled using linear regression. Blood loss was subjected to a logarithmic transformation to rectify violations of normality assumptions; the transformation exhibited noticeable improvement on the normal QQ plot. In both models, surgical approach was the primary factor and was considered with BMI, age, previous myomectomy, weight of the removed myoma, and the number of myomas removed. Variables were eliminated using a stepwise selection procedure based on the AIC criterion. Statistical analysis was conducted using R 2.9.1 software. Differences were considered significant at P<.05.
All patients included in this study requested fertility preserving surgical management of their symptomatic myomas. From a total of 575 myomectomies performed during the study period, 393 (68.3%) patients underwent open myomectomy, 93 (16.2%) patients underwent conventional laparoscopic myomectomy, and 89 (15.5%) patients underwent robotic myomectomy using the da Vinci surgical system. Similarly, the median (interquartile range [IQR]) weights were 75.57 (62.85, 90.72), 64.86 (59.10, 76.66), and 68.04 (57.65, 82.56) kg in the open, laparoscopic, and robotic groups, respectively. Patients in the open group were significantly heavier than patients in the laparoscopic and robotic groups (P<.001; Table 1). All three groups were comparable regarding the patients' heights,(0.18). Due to the significant differences in the patients' weight, the median BMI (IQR) was significantly higher in the open group at 27.61 (23.43, 32.81) compared with 24.10 (22, 28.01) in the laparoscopic and robotic groups at 25.15 (22.14, 29.44) (P<.001). No significant differences were seen among the three groups regarding their parity (P=.13) (Table 1).
Overall, prior history of myomectomies was encountered equally in the three groups with values of 9 out of 393 (2.3%), 6 out of 93 (6.4%), and 4 out of 89 (4.5%) in the open, laparoscopic, and robotic groups, respectively (P=.086). Significantly higher numbers of previous operative laparoscopies (P=.017), tubal ligations (P=.04) and cesarean deliveries (P=.047) were performed in the open group compared with the other two groups (Table 1).
The anatomic, pathologic, and surgical parameters features of the myomas of the three groups were evaluated. The median (IQR) of the maximum diameter of the removed myoma in centimeters as measured by preoperative ultrasound or magnetic resonance imaging (MRI) examination was 7.50 (5.05, 10.20) cm in the open group, 6.70 (4.20, 10) cm in the laparoscopic group, and 7.70 (5.40, 10.50) cm in the robotic group (Fig. 1 and Table 2). Significantly larger myomas were removed in the open and robotic groups compared with the laparoscopic group (P=.036).
The median (IQR) weight of the removed myomas in grams as measured postoperatively during the pathologic examination showed significantly heavier myomas in the open group (263 [90.5, 449.00] g, than in the laparoscopic group (96.65 [49.50, 227.25] g and in the robotic group (223 [85.25, 391.50] g with P<.001; Fig. 2). The weight of the removed myomas was comparable between both robotic and open groups while being significantly lower in the laparoscopic group. In addition, significantly heavier myomas were removed in the robotic group when compared with the laparoscopic group (P<.001).
On imaging, myomas were similarly distributed in the anterior wall, posterior wall, and fundus in all three groups (Table 2). Similarly, cervical myomas were encountered equally during the three surgical approaches. Interestingly, significantly more broad ligament myomas (P=.006) were successfully removed robotically compared with the two other approaches. Similarly, more multiple corporeal myomas were encountered in the robotic group (P=.041). Overall, significantly higher numbers of myomas were removed in the open and robotic groups compared with the laparoscopic group (P<.001). Of interest, the largest number of submucous myomas were removed robotically (P=.02). A summary of the ultrasound data and the anatomic and surgical features describing the removed myomas are detailed in Table 2.
The intraoperative and immediate postoperative outcomes were compared across the three groups and are presented in Table 3. The actual surgical time was significantly less with the open group than in the two other groups (P<.001; Fig. 3), where the median and IQR were 126 (95, 177), 155 (98, 200), and 181 (151, 265) minutes for the open, laparoscopic, and robotic groups, respectively. The actual surgical time is defined as the time from the incision to the closure. It reflects the absolute time of the surgical procedure. Only when the open group was compared with the robotic group were the actual surgical times significantly higher in the robotic group (P<.001). There were no significant differences upon comparing the open with the laparoscopic group or the laparoscopic with the robotic group.
Overall, a significantly higher blood loss was reported in the open group compared with the other two groups with the median (IQR) of the blood loss of 200 (100, 437.50) mL, 150 (100, 200) mL, and 100 (50, 212.50) mL in the open, laparoscopic, and robotic groups, respectively (P<.001; Fig. 4). On comparing the different groups, a significantly higher blood loss was reported in the open group compared with the laparoscopic group (P<.001) and in the open compared with the robotic group (P<.001), while there was no significant difference blood loss between the robotic and laparoscopic groups (P=.818). On translating the estimated blood loss to a postoperative drop in the patient's hemoglobin concentration, the median (IQR) of hemoglobin drop was 2 (1.40, 2.90), 1.55 (1.20, 2.40), and 1.30 (0.80, 2.28) g/dL in the open, laparoscopic, and robotic groups, respectively (Fig. 5). The hemoglobin drop was significantly lower in the robotic group compared with the other two groups (P<.001). A significantly lower drop in hemoglobin was detected in the robotic group compared with the open group (P<.001). On the other side the two minimally invasive approaches (laparoscopic and robotic) were comparable regarding the postoperative drop in hemoglobin (P=.431).
The need for blood transfusion was reported in 27 cases in the entire cohort, giving a blood transfusion rate of 4.7%. Out of these, 25 cases (6.4%) were reported in the open group and 2 cases (2.2%) in the robotic group. No patients in the laparoscopic group required blood transfusion. Three postoperative complications were also reported within the total number of cases. One case of wound separation was reported in the open group that required readmission and intervention for wound closure. Two complications were reported in the laparoscopic group including bowel injury that required conversion to laparotomy and the second was postoperative pyrexia with no major effect on the patient's health. No postoperative complications were reported in the robotic group. Finally, the median (IQR) length of patients' hospital stay was 3 (2, 3) days in the open group, 1 (0, 1) days in the laparoscopic group, and 1 (1, 1) days in the robotic group. Patients in the open group had a significantly longer median length of hospital stay compared with the laparoscopic and robotic groups (P<.001).
The purpose of robotic technology is to convert open cases to minimally invasive ones. However, if procedures can be performed by conventional laparoscopy without the robot then it would not be cost-effective to use this added technology. Myomectomy is a particularly challenging procedure because of the extensive suturing required. Myomectomy is mostly performed abdominally due to the inherent limitations of the conventional laparoscopic instruments for either dissection of the myoma or suturing of the myoma bed.
The current study compares the outcomes of the three different approaches of myomectomy. This study suggests that the introduction of robotic technology allowed cases that typically would have been done by laparotomy to be done by laparoscopy. In this study conventional laparoscopy was used for smaller myomas only while robotic laparoscopy allowed surgery on cases similar to the open ones. The robotic assistance enabled us to remove larger myomas and myomas at difficult locations as those with submucous extension and broad ligament myomas. The robotic assistance was associated with less blood loss, less drop in the postoperative hemoglobin, and shorter hospital stay. However, the length of stay is often due to surgical tradition without clear objective criteria. Nonetheless, the length of stay in our study is typical of laparotomy cases reported in the literature.
Standard hemostatic techniques such as the use of vasopressin, electrocautery, and suturing were enhanced by the robotic surgical system's improved 3-dimensional vision system and “wristed” instrumentation, which made suturing more ergonomic and more precise. In our cohort 2 patients in the robotic group required blood transfusions compared with 25 in the laparotomy group. Despite the fact that the operative times were significantly higher for the robotic group, this was offset by the significantly shorter hospital stay, which was one third that of the laparotomy group.
There is increasing interest in robotically assisted myomectomy. The feasibility of this approach was shown in a series of 35 cases.17 Since then, subsequent articles compared robotically assisted myomectomy to either open technique22 or laparoscopic approach.20,21 Our findings were strengthened by the fact that 89 cases underwent robotic myomectomy which represent a large series. Also this number of cases has been compared with large cohorts of laparoscopic myomectomy and open cases performed at the same institution. In our work the weight of the removed myoma appears not to be a limitation against using the surgical robot to perform myomectomy as it was comparable between both robotic and open groups while being significantly higher than the laparoscopic group. The significantly lower blood loss encountered in the robotic group was similar to the results of Advincula and associates.22 However, Nezhat et al reported no difference in the estimated blood loss between robotic and laparoscopic cases.21 Bedient et al compared robotic (N=40) and laparoscopic myomectomy and found no significant differences between the two procedures regarding the short-term surgical outcome measures, which is very much consistent with our finding.20
Although both robotic and laparoscopic techniques are minimally invasive approaches, easier maneuverability was encountered with the use of the robotic endowrist instruments, helping with both dissection and suturing. Similar to what has been previously published, the surgical time remains one of the challenges of the robotic myomectomy. This time will influence cost, which could be offset by decreased length of stay if compared with laparotomy cases. A formal cost analysis was not performed in this study because of the limitations of our institutional cost database.
One of the major limitations of robotic surgery is the absence of haptic feedback. This could lead to two important challenges intraoperatively. The first is finding and treating all myomas—even small ones that are not obvious on preoperative imaging—an important issue for avoiding re-intervention over time. The second is to determine the strength needed for suturing without breaking the suture. It is especially the case with early cases and its effect will decrease but not be entirely eradicated with experience. Visual cues from the 3-dimensional image may alleviate this limitation. Excellent preoperative imaging of the myoma location will also compensate for this limitation.
All the laparoscopic and the robotic procedures were performed by experienced surgeons with a minimum of 10 years of practice. The majority of the patients in the open myomectomy group were performed by fellowship-trained surgeons as well, with a subgroup of patient procedures performed by generalists. We did not analyze the relationship between the surgeon's experience and surgical outcomes in this work. However, the robotic surgeons had a minimum of 5 years of experience. It is unlikely that the experience of these surgeons influenced the outcome in this report.
This study has several limitations. It is a retrospective study and the decision to perform a procedure with one technique or another was not rigidly defined but based on experience. In addition, it is limited by the lack of long-term outcomes including subsequent pregnancy and uterine rupture. The data suggest that some open myomectomy cases can be converted to minimally invasive surgery with the use of the robot by appropriately trained surgeons who maintain robotic and laparoscopic skills consistently.
1. Marshall LM, Spiegelman D, Barbieri RL, Goldman MB, Manson JE, Colditz GA, et al. Variation in the incidence of uterine leiomyoma among premenopausal women by age and race. Obstet Gynecol 1997;90:967–73.
2. Stewart EA. Uterine fibroids. Lancet 2001;357:293–8.
3. Falcone T, Bedaiwy MA. Minimally invasive management of uterine fibroids. Curr Opin Obstet Gynecol 2002 Aug;14:401–7.
4. Goldberg J, Pereira L. Pregnancy outcomes following treatment for fibroids: uterine fibroid embolization versus laparoscopic myomectomy. Curr Opin Obstet Gynecol 2006;18:402–6.
5. Semm K. New methods of pelviscopy (gynecologic laparoscopy) for myomectomy, ovariectomy, tubectomy and adnectomy. Endoscopy 1979;11:85–93.
6. Mais V, Ajossa S, Guerriero S, Mascia M, Solla E, Melis GB. Laparoscopic versus abdominal myomectomy: a prospective, randomized trial to evaluate benefits in early outcome. Am J Obstet Gynecol 1996;174:654–8.
7. Malzoni M, Rotond M, Perone C, Labriola D, Ammaturo F, Izzo A, et al. Fertility after laparoscopic myomectomy of large uterine myomas: operative technique and preliminary results. Eur J Gynaecol Oncol 2003;24:79–82.
8. Seracchioli R, Rossi S, Govoni F, Rossi E, Venturoli S, Bulletti C, et al. Fertility and obstetric outcome after laparoscopic myomectomy of large myomata: a randomized comparison with abdominal myomectomy. Hum Reprod 2000;15:2663–8.
9. Dubuisson JB, Fauconnier A, Deffarges JV, Norgaard C, Kreiker G, Chapron C. Pregnancy outcome and deliveries following laparoscopic myomectomy. Hum Reprod 2000;15:869–73.
10. Al-Mahrizi S, Tulandi T. Treatment of uterine fibroids for abnormal uterine bleeding: myomectomy and uterine artery embolization. Best Pract Res Clin Obstet Gynaecol 2007;21:995–1005.
11. Hurst BS, Matthews ML, Marshburn PB. Laparoscopic myomectomy for symptomatic uterine myomas. Fertil Steril 2005;83:1–23.
12. Luciano AA. Myomectomy. Clin Obstet Gynecol 2009;52:362–71.
13. Falcone T, Goldberg J, Garcia-Ruiz A, Margossian H, Stevens L. Full robotic assistance for laparoscopic tubal anastomosis: a case report. J Laparoendosc Adv Surg Tech A 1999;9:107–13.
14. Falcone T, Goldberg JM. Robotics in gynecology. Surg Clin North Am 2003;83:1483–9, xii.
15. Falcone T, Goldberg JM, Margossian H, Stevens L. Robotic-assisted laparoscopic microsurgical tubal anastomosis: a human pilot study. Fertil Steril 2000;73:1040–2.
16. Di Marco DS, Chow GK, Gettman MT, Elliott DS. Robotic-assisted laparoscopic sacrocolpopexy for treatment of vaginal vault prolapse. Urology 2004;63:373–6.
17. Advincula AP, Song A, Burke W, Reynolds RK. Preliminary experience with robot-assisted laparoscopic myomectomy. J Am Assoc Gynecol Laparosc 2004;11:511–8.
18. Bocca S, Stadtmauer L, Oehninger S. Uncomplicated full term pregnancy after da Vinci-assisted laparoscopic myomectomy. Reprod Biomed Online 2007;14:246–9.
19. Mao SP, Lai HC, Chang FW, Yu MH, Chang CC. Laparoscopy-assisted robotic myomectomy using the da Vinci system. Taiwan J Obstet Gynecol 2007;46:174–6.
20. Bedient CE, Magrina JF, Noble BN, Kho RM. Comparison of robotic and laparoscopic myomectomy. Am J Obstet Gynecol 2009;201:566.e1–5.
21. Nezhat C, Lavie O, Hsu S, Watson J, Barnett O, Lemyre M. Robotic-assisted laparoscopic myomectomy compared with standard laparoscopic myomectomy–a retrospective matched control study. Fertil Steril 2009;91:556–9.
22. Advincula AP, Xu X, Goudeau S 4th, Ransom SB. Robot-assisted laparoscopic myomectomy versus abdominal myomectomy: a comparison of short-term surgical outcomes and immediate costs. J Minim Invasive Gynecol 2007;14:698–705.
23. George A, Eisenstein D, Wegienka G. Analysis of the impact of body mass index on the surgical outcomes after robot-assisted laparoscopic myomectomy. J Minim Invasive Gynecol 2009;16:730–3.
24. Silva BA, Falcone T, Bradley L, Goldberg JM, Mascha E, Lindsey R, et al. Case-control study of laparoscopic versus abdominal myomectomy. J Laparoendosc Adv Surg Tech A 2000;10:191–7.
25. Venables WN, Ripley BD. Modern APPLIED STATISTICS with S. 4th ed. New York (NY): Springer; 2002.
Figure. No caption available.
Supplemental Digital Content
© 2011 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.