Comparison of the Learning Curve and Outcomes of Robotic Assisted Pediatric Pyeloplasty

Transcription

1 Comparison of the Learning Curve and Outcomes of Robotic Assisted Pediatric Pyeloplasty Mathew D. Sorensen,* Catherine Delostrinos, Michael H. Johnson, Richard W. Grady and Thomas S. Lendvay From the Division of Pediatric Urology, Seattle Children s Hospital (RWG, TSL) and Department of Urology (MDS, RWG, TSL), University of Washington School of Medicine (CD, MHJ), Seattle, Washington Purpose: We compared the learning curve and outcomes in children undergoing robotic assisted laparoscopic pyeloplasty during the initiation of a robotic surgery program compared to the benchmark of open pyeloplasty. Materials and Methods: The records of our first consecutive 33 children undergoing robotic assisted laparoscopic pyeloplasty from 2006 to 2009 were retrospectively reviewed and compared to those of age and gender matched children who underwent open repair done by senior faculty surgeons before the initiation of our robotic surgery program. We compared operative time, complications, postoperative pain, length of stay and surgical success for 2 surgeons who adopted the robotic approach at an academic teaching institution. Results: We found no significant differences in length of stay, pain score or surgical success at a median followup of 16 months. The number of complications was similar and they tended to be early and technical in the robotic assisted laparoscopic pyeloplasty group. Overall average operative time was 90 minutes longer (38%) for robotic assisted laparoscopic pyeloplasty (p 0.004). When evaluated chronologically, there was evidence of a learning curve. After 15 to 20 robotic cases overall operative times for robotic assisted laparoscopic cases was consistently within 1 SD of our average open pyeloplasty time with no significant difference in overall operative time (p 0.23). Of the decrease in overall operative time 70% was due to decreased pyeloplasty time rather than peripheral time. Conclusions: There was similar safety and efficacy with robotic assisted laparoscopic pyeloplasty, although complications tended to be technical and early in our initial experience. Operative time decreased with experience and after 15 to 20 cases it was similar to that of open pyeloplasty with similar outcomes and surgical success. Abbreviations and Acronyms MAG3 mercaptoacetyltriglycine RAL robotic assisted laparoscopic Study received Seattle Children s Hospital institutional review board approval. * Correspondence; University of Washington School of Medicine, 1959 Northeast Pacific St., Box , Seattle, Washington (telephone: ; FAX: ; mathews@u.washington.edu). Financial interest and/or other relationship with Spi Surgical. Key Words: kidney, laparoscopy, robotics, learning, outcome and process assessment (health care) THE learning curve has become an increasingly important focus of evaluating RAL procedures. While the urological literature includes reports of adult procedure learning curves, no studies have adequately described the learning curve for RAL pediatric urological reconstructive procedures. Learning curve is a frequently used and often poorly defined term. On the way to achieving mastery the curve represents the initial challenges in competence, and the change in technical proficiency and efficiency with increasing experience. 1,2 In adults increasing surgeon experience decreases operative time for robotic prostatectomy within the first /11/ /0 Vol. 185, , June 2011 THE JOURNAL OF UROLOGY Printed in U.S.A by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RESEARCH, INC. DOI: /j.juro

2 2518 LEARNING CURVE AND OUTCOMES OF ROBOTIC ASSISTED PEDIATRIC PYELOPLASTY 50 cases, although the learning curve likely extends beyond 150 to 200 cases in terms of surgical margins. 3 8 For practitioners familiar with robotic procedures RAL radical cystectomy has an estimated learning curve of only 20 cases and reports have reassured us that, although operative time and blood loss were initially higher, these initial cases could be done without observed compromises in oncological parameters. 9 Other procedures, such as RAL partial nephrectomy and RAL hysterectomy, appear to have a learning curve of 15 to 30 cases but the complication rate may be higher in these initial cases As a group at a center where RAL pyeloplasty is now regularly performed, we evaluated the learning curve and outcomes of these initial cases. We defined the learning curve as time needed to achieve similar outcomes of success, operative time and complications for RAL pyeloplasty compared to the outcomes of open pyeloplasty performed by senior faculty. We focused primarily on safety and outcomes as markers of proficiency since prior studies demonstrated inferior outcomes during the initiation and establishment of such a new technology. 6,7,10 12 MATERIALS AND METHODS We systematically evaluated the first 33 children undergoing RAL pyeloplasty at our institution as we initiated our RAL surgery program in April Cases were performed during a 34-month period by 2 pediatric urologists using the da Vinci Standard surgical system. Each surgeon had prior laparoscopic experience (approximately 20 cases yearly) but minimal laparoscopic pyeloplasty experience. Neither surgeon had performed any robotic procedures. The last case in this study represents the 106th RAL case performed at our institution and the 67th performed by these 2 urological surgeons. Children who underwent RAL pyeloplasty were compared to children who underwent open pyeloplasty. Since patient age is associated with differences in operative time, complications and postoperative pain, RAL and open cases were matched based on gender and frequency matched for age (0 to 5, 6 to 9 and 10 years or greater). 13,14 To decrease case selection bias open cases were only eligible for inclusion if they were performed before the time when the attending surgeon was performing RAL surgery. Open cases were done by senior faculty with an average of more than 10 years of experience and are thought to represent surgical competence and likely mastery. The learning curve was defined by comparing surgical success and complications. Also, overall operative time in RAL pyeloplasty cases was evaluated chronologically and compared to the average 1 SD overall operative time for open pyeloplasty. Operative time was further evaluated based on the time from skin incision to dressing to perform the pyeloplasty portion of the procedure as well as peripheral time (in room, anesthesia, initial positioning, cystoscopy, repositioning and out of room time). Secondarily we analyzed postoperative pain, length of stay and blood loss. Postoperative pain scores were regularly assessed by nursing staff using a standard 0 to 10 scale. 15 Complications, and their nature and timing in the robotic experience were assessed. RAL and open cases were required to have had at least 90 days of followup, defined as time from surgery to the most recent clinic visit. Success in the 2 groups was defined postoperatively as improved drainage on MAG3 renography and/or improved hydronephrosis combined with renal parenchymal growth on ultrasound or computerized tomography and resolution of renal colic crisis in patients who presented with pain symptoms. Robotic cases were performed using a 12 mm camera port and 2, 8 mm instrument ports. A 14 gauge angiocatheter was used to assist with antegrade ureteral stent placement, tissue retraction, smoke evacuation and suction. 16 Of the cases 54% were done in transmesenteric fashion, when possible. A urethral catheter was left in place overnight and a ureteral stent was left in place for 6 weeks. Open cases were performed through a subcostal or flank incision. A urethral catheter was typically removed the morning after surgery, followed by removal of a percutaneous Penrose drain in the flank. Ureteral stents were placed in 12 patients (36%) and removed 6 weeks later. In each group pain was managed by regional anesthesia or narcotic boluses, followed by oral analgesia with or without ketorolac. Outcomes were compared using conditional logistic regression with robust SEs with a priori adjustment for all matching factors. Chi-square analysis was done for binary variables and Student s t test was performed for continuous variables with significance considered at p Analysis was performed using Stata, version 10. Approval for this study was obtained from the Seattle Children s Hospital institutional review board. RESULTS Patient age and gender were well matched with similar presentations and imaging (table 1). The intraoperative approach to urinary diversion and postoperative imaging differed between the groups (table 1). In the 2 groups 97% success was achieved, that is in 32 of the 33 patients each. In cases of RAL and open failure endoscopic dilation was done. Followup duration was similar between the groups. As a group at a tertiary referral center, we tend to follow remote patients for approximately 12 to 18 months and then return them to the care of the primary care provider and local urologist. During the study period overall average operative time in RAL cases was 38% longer than in open cases with an absolute difference of 90 minutes (p 0.004, table 2). This longer time was due primarily to longer pyeloplasty time (skin incision to dressing, p 0.002) rather than peripheral time (in room, anesthesia, setup, cystoscopy and out of room time, p 0.12). When analyzed chronologically and compared to the overall operative time for the average open pyeloplasty done by senior surgical faculty, a learning curve was identified for the surgeon with the greatest experience

3 LEARNING CURVE AND OUTCOMES OF ROBOTIC ASSISTED PEDIATRIC PYELOPLASTY 2519 Table 1. Demographics and surgical variables in 33 patients with RAL and 33 with open pyeloplasty RAL Open p Value* Age (yrs): No No No. 10 or greater Mean SD No. gender (%): 1.0 M 10 (30) 10 (30) F 23 (70) 23 (70) Mean SD wt (kg) Mean SD body mass index (kg/m 2 ) Mean SD American Society of Anesthesiologists score No. presenting symptoms (%): 0.20 Pain/Dietl s crisis 19 (58) 19 (58) Prenatal hydronephrosis 8 (24) 2 (6) Stones/urinary tract infection 4 (12) 5 (15) Other 2 (6) 7 (21) No. preop imaging: 0.12 MAG Magnetic resonance urogram 4 0 Computerized tomography/ultrasound 5 2 No. complex: Pyelolithotomy 3 2 Repeat pyeloplasty 2 1 Ruptured ureteropelvic junction 1 Horseshoe kidney 1 Stent: Antegrade Retrograde 6 2 Totals 32 (97) 12 (36) Mean duration (wks) No. drain (%): Penrose 30 (91) Closed suction 1 (3) % Postop imaging: 0.05 MAG Ultrasound Computerized tomography 3 3 % Success Followup (mos): Mean SD Median (IQR) 16 (11 23) 12 (5 35) 0.55 * Conditional logistic regression with robust SE adjusted for age category and gender. (see figure). Overall average operative time in the initial 10 RAL pyeloplasty cases was about 6 hours for each surgeon. These initial cases included the 3 longest and most difficult cases, including repeat pyeloplasty in 1 and concurrent large pyelolithotomy in 2. After 15 to 20 cases there was no statistically significant difference in overall operative time for RAL and open pyeloplasty (p 0.23). Overall operative time was also consistently within 1 SD of that of our average open pyeloplasty. Peripheral and the pyeloplasty time decreased with experience but 70% of the decrease was due to pyeloplasty surgical time. The number of complications were similar between the groups (p 0.45). However, complications in the RAL pyeloplasty group tended to be early and technical (see figure). There were 3 complications (9%) in the open group, including a patient who was rehospitalized twice for pyelonephritis and another who was rehospitalized for dehydration, persistent nausea, vomiting and flank pain, which ultimately resolved spontaneously. A third complication was technical in a child in whom asymptomatic stent migration was managed conservatively. There were 5 complications (15%) in the RAL group. One boy had gross hematuria with clot obstruction of the ureteral stent, which was managed conservatively. Another 4 patients in the RAL group had technical complications due to urine extravasation from the anastomotic closure. Two patients were treated conservatively with prolonged urinary drainage, 1 underwent percutaneous nephrostomy tube placement and 1 underwent replacement of a malpositioned stent. After our first year of performing RAL pyeloplasty we began performing intraoperative ultrasound using the S-Nerve ultrasound sys-

4 2520 LEARNING CURVE AND OUTCOMES OF ROBOTIC ASSISTED PEDIATRIC PYELOPLASTY Table 2. Operative and perioperative comparison of patients with RAL vs open pyeloplasty RAL Open Difference p Value* Mean SD time (mins): Peripheral Pyeloplasty Overall operative Mean SD estimated blood loss (ml) Mean SD length of stay (days) No. complications: Intraop Postop (%) 5 (15) 3 (9) 0.45 Technical postop 4 1 Other postop 1 2 * Conditional logistic regression with robust SE adjusted for age category and gender. tem to confirm distal stent position. After this no further episodes of stent malposition occurred. All patients with a postoperative complication ultimately had a successful outcome. Estimated blood loss, length of stay and intraoperative complications were similar in the 2 groups. When analyzed annually, we found a decrease in length of stay with each year of experience, that is a mean of 3.5, 2.0 and 1.2 days for each successive study year, respectively (p 0.01). There were no differences in postoperative pain scores, which may be partly explained by differences in the approach to postoperative pain control. At our institution there were 5 mechanical failures in our initial 100 robotic procedures (5%) during a 12-month period. Three of these failures developed in patients who underwent robotic pyeloplasty. Two cases were completed using conventional laparoscopy and the other was converted to conventional laparoscopy for 1.5 hours while the robot was repaired and then completed robotically. No patient in the RAL cohort required open conversion. Ultimately the 2 robotic arms were replaced. DISCUSSION Our experience demonstrates the realities of the 50% longer overall operative time in our initial robotic cases compared to open operative time in cases done by senior faculty. These RAL cases were performed by surgeons with prior laparoscopic experience and no prior robotic experience except basic pig and dry laboratory experience. Operative time decreased considerably with increasing experience, such that after 15 to Learning curve comparison of overall operative time for RAL vs open pyeloplasty per surgeon. After 15 to 20 cases overall operative times for RAL was consistently within 1 SD of average open pyeloplasty time with no significant difference (p 0.23). Five complications (arrows) tended to be technical and early in learning curve. Gray box represents average overall operative time for open pyeloplasty 1 SD.

5 LEARNING CURVE AND OUTCOMES OF ROBOTIC ASSISTED PEDIATRIC PYELOPLASTY RAL pyeloplasty cases overall operative time was not significantly different from that of open pyeloplasty (p 0.23). Of this decrease in RAL cases 70% was due to improved proficiency with the pyeloplasty portion of the procedure (skin incision to dressing) rather than the peripheral time (in room, anesthesia, positioning, cystoscopy, repositioning and out of room time), although each contributed to the decrease in overall operative time. In the early stages of our robotic pyeloplasty experience we observed outcomes similar to those in our open pyeloplasty cohort. The number of technical complications was higher in the robotic group and tended to occur early in our experience (see figure), providing further evidence of the learning curve. After year 1 we began performing intraoperative ultrasound to identify the bladder curl of the ureteral stent and no further episodes of stent malposition occurred. There were no significant differences in overall blood loss, length of stay or pain score, although there was a decrease in length of stay with each year of experience. We expected length of stay to be similar since we applied more conservative postoperative observation in our initial patients. The success rate was 97% in our open and RAL groups with more than a year of average followup. All patients underwent postoperative imaging, although those in the RAL pyeloplasty group were more rigorously imaged by renograms. During year 1 as we conservatively approached this new technique, almost all patients underwent postoperative MAG3 renography except the 2 who underwent concurrent pyelolithotomy. Our initial mechanical failure rate is higher than the 1% to 2% cited in the literature and highlights the importance of maintaining standard laparoscopy skills. 17 No mechanical failure resulted in a significant increase in operative time, estimated blood loss or length of hospitalization. Excluding these mechanical failures did not alter our conclusions for any of the outcomes analyzed and, thus, they were included in all analyses. Our institution has since upgraded to the da Vinci Si system and we have not experienced a mechanical failure in 35 cases using the new robot. Defining the learning curve for reconstructive procedures represents a unique challenge. The factors available for evaluation of proficiency are somewhat limited, especially compared to oncological cases, which have several variables, such as margin status, number of lymph nodes, continence, potency and overall/cancer specific survival, in addition to the variables that we evaluated, such as surgical success, complications, pain and blood loss. We acknowledge that it likely takes 15 to 20 cases to achieve rudimentary proficiency when we evaluated the somewhat crude outcomes of safety, efficacy and operative time. Also, although operative time is likely to continue to decrease with greater experience, achieving true technical mastery of a pediatric reconstructive procedure likely takes far more than the 33 cases that we evaluated. 4,6,7 These findings have important implications for practitioners and institutions electing to perform these types of robotic pediatric reconstructive procedures. Initial operative time is longer and depending on the case volume it may take several years for operative time to approach that of established open procedures. These initial cases were performed at the beginning of establishing our robotic surgery program and, thus, surgical staff and surgeons were acclimating to the new technology. We believe that there is a synergistic effect, in that an experienced robotics surgery staff may accelerate a novice surgeon learning curve and vice versa. We compared an open procedure performed by experienced senior faculty to a newly adopted robotic surgical technique. Although this sets a high standard, we believe that it offers a realistic comparison of the options that would have otherwise been open to families presenting for surgery at our institution. Furthermore, there were differences in surgical and postoperative management between the RAL and open groups with the RAL group favoring a ureteral stent while most cases had a Penrose drain and sometimes a ureteral stent (97% vs 91% and 36%, respectively). Thus, a second anesthesia was required in almost every robotics case, a factor that must be discussed when obtaining consent for this procedure from the families. The RAL group also tended to undergo more imaging, including nuclear renography and ultrasound, while the open group tended to be followed with serial ultrasound only, reflecting practitioner bias. We thought that it was important to match patients with RAL and open surgery by age and gender since age is associated with differences in operative time, complications and pain in children who undergo laparoscopic procedures. 13,14,18 This study has limitations. It is a retrospective study done at a tertiary academic referral center and is subject to the inherent limitations of this study design. The degree of resident and/or fellow participation in either group was not assessed and could have biased surgical time. This study includes data primarily from 1 surgeon. However, we found it reassuring that the second surgeon appeared to have experienced a similar learning curve in the initial 9 cases (p 0.61). This study includes our first 33 RAL pyeloplasty cases, which may have limited our power to detect small differences between the groups. Despite our efforts to minimize case selection bias by selecting controls from a time when the attending surgeons were not performing any RAL procedures, patients were not randomized. In our initial robotic experience we attempted several complex cases, which likely impacted outcomes. We suggest that complex cases should be reserved for after the initial learning curve has been traversed. Comparing outcomes and operative times for our initial experience with a new technology vs a well estab-

6 2522 LEARNING CURVE AND OUTCOMES OF ROBOTIC ASSISTED PEDIATRIC PYELOPLASTY lished open procedure likely influenced our evaluation but we believe that this comparison is honest and real for those considering adoption of this technology. Although prior studies have demonstrated similar outcomes, it was not possible to evaluate traditional laparoscopic pyeloplasty since our institutional experience was too small to allow any meaningful comparisons. 19 CONCLUSIONS The initial learning curve for RAL pyeloplasty appears to be 15 to 20 cases. True mastery of the procedure extends beyond this initial evaluation but it is important to note that during this learning curve we observed similar safety and efficacy outcomes with the caveat of a few early technical complications. This reality should be brought up during preoperative patient counseling. ACKNOWLEDGMENTS Dr. Martin Koyle reviewed the manuscript, Jeff Bice assisted with data acquisition and Carrie Wachob assisted with figures. REFERENCES 1. Wright TP: Factors affecting the cost of airplanes. J Aeronaut Sci 1936; 3: Perspectives on Experience. Boston: Boston Consulting Group 1972; p Menon M, Shrivastava A, Tewari A et al: Laparoscopic and robot assisted radical prostatectomy: establishment of a structured program and preliminary analysis of outcomes. J Urol 2002; 168: Patel VR, Tully AS, Holmes R et al: Robotic radical prostatectomy in the community setting the learning curve and beyond: initial 200 cases. J Urol 2005; 174: Ahlering TE, Woo D, Eichel L et al: Robot-assisted versus open radical prostatectomy: a comparison of one surgeon s outcomes. Urology 2004; 63: Doumerc N, Yuen C, Savdie R et al: Should experienced open prostatic surgeons convert to robotic surgery? The real learning curve for one surgeon over 3 years. BJU Int 2010; 106: Herrell SD and Smith JA Jr: Robotic-assisted laparoscopic prostatectomy: what is the learning curve? Urology 2005; 66: 105. EDITORIAL COMMENT 8. Ahlering TE, Skarecky D, Lee D et al: Successful transfer of open surgical skills to a laparoscopic environment using a robotic interface: initial experience with laparoscopic radical prostatectomy. J Urol 2003; 170: Pruthi RS, Smith A and Wallen EM: Evaluating the learning curve for robot-assisted laparoscopic radical cystectomy. J Endourol 2008; 22: Haseebuddin M, Benway BM, Cabello JM et al: Robot-assisted partial nephrectomy: evaluation of learning curve for an experienced renal surgeon. J Endourol 2010; 24: Bell MC, Torgerson JL and Kreaden U: The first 100 da Vinci hysterectomies: an analysis of the learning curve for a single surgeon. S D Med 2009; 62: Seamon LG, Fowler JM, Richardson DL et al: A detailed analysis of the learning curve: robotic hysterectomy and pelvic-aortic lymphadenectomy for endometrial cancer. Gynecol Oncol 2009; 114: Lee RS, Retik AB, Borer JG et al: Pediatric robot assisted laparoscopic dismembered pyeloplasty: comparison with a cohort of open surgery. J Urol 2006; 175: Casale P, Patel RP and Kolon TF: Nerve sparing robotic extravesical ureteral reimplantation. J Urol 2008; 179: Wong DL and Baker CM: Pain in children: comparison of assessment scales. Pediatr Nurs 1988; 14: Hotaling JM, Shear S and Lendvay TS: 14-gauge angiocatheter: the assist port. J Laparoendosc Adv Surg Tech A 2009; 19: Kim WT, Ham WS, Jeong W et al: Failure and malfunction of da Vinci Surgical systems during various robotic surgeries: experience from six departments at a single institute. Urology 2009; 74: Yee DS, Shanberg AM, Duel BP et al: Initial comparison of robotic-assisted laparoscopic versus open pyeloplasty in children. Urology 2006; 67: Mufarrij PW, Woods M, Shah OD et al: Robotic dismembered pyeloplasty: a 6-year, multi-institutional experience. J Urol 2008; 180: While there are several challenges to this study, the effort to quantitate and, thereby, begin to understand surgical learning is admirable. The concept of the learning curve emerged in the early aircraft industry and was described by Wright (reference 1 in article). It was based on the hypothesis that if an operation (not necessarily surgical) were repeated, it would take less time to perform. The learning curve was a representation of the number of repetitions needed to decrease the time spent. To measure this Wright described the learning percent as the difference in time taken after the number of operations is doubled. In this report it was about 67% (100% minus 33%, that is 6 hours for the initial 10 cases and about 4 hours for the second 10 or 2 hours/6 hours 33%). This is good compared to some industries. This value provides an assessment of the learning curve that may be compared among providers and systems. Craig A. Peters Department of Pediatric Surgery Children s National Medical Center Washington, D.C.