Documente Academic
Documente Profesional
Documente Cultură
DOI 10.1007/s11701-015-0513-4
ORIGINAL ARTICLE
Received: 6 March 2015 / Accepted: 28 April 2015 / Published online: 16 May 2015
Springer-Verlag London 2015
Abstract The objective of this randomized, controlled sutured for 10 min using the da Vinci Surgical System.
trial was to assess whether voluntary participation in a Performances were videotaped, anonymized, and subse-
proctored, proficiency-based, virtual reality robotic sutur- quently graded independently by three robotic surgeons. 27
ing curriculum using the da Vinci Skills SimulatorTM participants were randomized. 23 of the 27 completed both
improves robotic suturing performance. Residents and at- the pre- and post-test, 13 in the training group and 10 in the
tending surgeons were randomized to participation or non- control group. Mean training time in the intervention group
participation during a 5 week training curriculum. Robotic was 238 136 min (SD) over the 5 weeks. The primary
suturing skills were evaluated before and after training outcome (improvement in GOALS? score) and the sec-
using an inanimate vaginal cuff model, which participants ondary outcomes (improvement in GEARS, total knots,
satisfactory knots, and the virtual reality suture sponge 1
task) were significantly greater in the training group than
This manuscript was presented at the Society of Gynecologic
the control group in unadjusted analysis. After adjusting for
Oncology 45th Annual Meeting on Womens Cancer, Tampa, FL,
USA, March 23, 2014 and at the Society of Gynecologic Oncology of lower baseline scores in the training group, improvement in
Canada 35th Annual General Meeting. June 13, 2014. Niagara Falls, the suture sponge 1 task remained significantly greater in
ON, Canada. the training group and a trend was demonstrated to greater
improvement in the training group for the GOALS? score,
Condensation: In this randomized controlled trial, learners
randomized to participation in a virtual reality robotic suturing GEARS score, total knots, and satisfactory knots.
curriculum showed a trend towards greater improvement when scored
on their ability to suture an inanimate vaginal cuff model using the da Keywords Randomized controlled trial Patient
Vinci Surgical System. simulation Computer simulation Graduate medical
ClinicalTrials.gov: NCT01811095, www.clinicaltrials.gov. education Robotics Operative surgical procedures
123
180 J Robotic Surg (2015) 9:179186
123
J Robotic Surg (2015) 9:179186 181
Fig. 1 The da Vinci Surgical System set-up with training arms and the inanimate vaginal cuff model
inanimate model with the actual da Vinci Surgical System achieve the completed score (equivalent to attaining all
and (2) completion of the suture sponge 1 task using the da component scores C60 % and overall score green
Vinci Skills SimulatorTM. For suturing the inanimate checkmark) non-consecutively on 2 occasions for the
model, we used the model of the vaginal cuff described by following tasks: camera targeting 1, camera targeting 2,
Finan [19], constructed from a beer huggie to represent suture sponge 1, suture sponge 2, suture sponge 3 and to
the vaginal cuff, a block of wood for positioning, and complete the horizontal suturing defect (HSD) task ten
balloons to represent the bladder and rectum, and to distend times. Participants in the training group set their own hours
the vaginal cuff. Prior to the pre-test, participants were for training. Given the voluntary nature of the study, there
given a verbal overview of how to use the da Vinci sur- was no obligation to train and participants were free to try
gical system. They were then instructed to suture closed the other simulation exercises outside the curriculum, although
vaginal cuff model using figure of eight stitches with participants were made aware of, and encouraged to
square knots composed of a double throw and then three achieve all the goals of the proficiency curriculum. Par-
single throws. This task was recorded using the camera of ticipants were informed that, on average, it was estimated
the da Vinci Surgical System. The post-test was identical that it would take 5 h to attain all the goals of the profi-
to the pre-test. The vaginal cuff model along with the set- ciency curriculum and that the suggested training schedule
up used on the testing day is shown in Fig. 1. Supple- was 1 h per week for 5 weeks. The study coordinator (DK),
mentary file 1 shows a demonstration of the task of su- at the time experienced in robotic virtual reality simulation
turing the vaginal cuff model. but not actual robotic surgery, was present for all training
sessions and offered suggestions for improvement.
Training intervention
Outcomes
The training group was assigned to a proficiency-based
robotic surgery simulation curriculum which took place The primary outcome was skill in performing the task of
over a 5 week period between pre- and post-test sessions, suturing the vaginal cuff model. This was assessed by three
and was completed in addition to usual clinical duties or blinded raters (two gynecologic oncologists and one gyne-
residency training. We used the da Vinci Skills Simulator cologic oncology fellow, all experienced in robotic surgery)
tasks and the goals of the training curriculum were to using the GOALS? score. This score is composed of the five
123
182 J Robotic Surg (2015) 9:179186
domains of the GOALS score developed for assessing skill in Assessed for eligibility
laparoscopy which include autonomy, efficiency, tissue (n = 30)
Allocation
training (n=12)
GOALS? includes 7 domains, 6 of which form the GEARS Participated in proficiency-
based curriculum but did not Engaged in some training
score [21]. Thus, our data allowed us to calculate the GEARS achieve all goals (n = 8) with da Vinci Skills Simulator
outside of study (n=1)
score as a secondary outcome of the study. Completed proficiency-based
The secondary outcomes were the number total knots and curriculum (n = 5)
123
J Robotic Surg (2015) 9:179186 183
and 10 in the control group. Table 1 shows the baseline score (p = 0.002), the GEARS score (p = 0.002), total
characteristics of the participants. Despite a rigorous ran- knots (p \ 0.001), satisfactory knots (p \ 0.001), and the
domization protocol, there were imbalances between groups suture sponge 1 score (p \ 0.001) (Table 2). The control
likely due to the small size of the trial and the lack of group also showed statistically significant improvements in
stratification by baseline scores. The training group, on av- the GEARS score (p = 0.041), total knots (p \ 0.001) and
erage, tended to be older, contained more attending physi- satisfactory knots (p = 0.004). Unadjusted analysis com-
cians, had less robotic surgery experience, less robotic paring the magnitude of the improvement in the training
simulation experience, and significantly lower pretest scores group to that of the control group showed statistically
for both the virtual reality simulator task (suture sponge 1) significantly greater improvement in the training group for
and the inanimate model task (GOALS? score). the GOALS? score, the GEARS score, total knots, satis-
The mean total time spent training with the da Vinci factory knots, and the suture sponge 1 score (Table 3).
Skills Simulator over the 5 week study period was 238 min At baseline, the training group had significantly lower
among the training group participants (range 15600 min). GOALS? scores (p = 0.020) and GEARS scores (p = 0.023)
Five of the 13 study participants achieved all the training than the control group. However, at the post-test, no statisti-
goals of the proficiency curriculum, with a mean time of cally significant differences were found between groups in the
272 min (range 175600 min). The three shortest training GOALS? scores (p = 0.455) or the GEARS scores
times were among participants who did not complete the (p = 0.598), suggesting that participation in the curriculum
curriculum. One participant in the control group performed allowed the training group to catch up to the control group.
approximately 3 h of robotic simulation training with the Training group participants with baseline GOALS?
da Vinci Skills SimulatorTM during the study period scores less than 14 out of 35 tended to have greater im-
outside of the study protocol. One participant in the provements (mean 8.5, n = 8) than those with scores
training group and two participants in the control group greater than 14 (mean 3.0, n = 5) with a p value of 0.067.
assisted at robotic surgery during the study period, in- A multivariate analysis was then conducted (Table 4),
cluding one participant in the control group who had the controlling for baseline scores with a regression model
opportunity to suture the vaginal cuff in live surgery twice containing pre-score and group allocation as independent
during the study period. variables and improvement in score as the dependent
The training group showed statistically significant im- variable. In this analysis, group allocation (training or
provements between pre-test and post-test in the GOALS? control) was no longer a statistically significant predictor of
123
184 J Robotic Surg (2015) 9:179186
GOALS? score (max 35) Training group 15.1 7.3331.0 21.4 1229.7 0.002*
Control group 20.6 13.330.3 22.8 16.731.3 0.052
GEARS score (max 30) Training group 12.7 6.3326.7 18.4 9.6726.3 0.002*
Control group 17.4 1126 19.4 13.327 0.041*
Total knots Training group 1.65 05.6 3.38 1.935.73 \0.001*
Control group 2.45 0.934.87 3.42 1.936.4 \0.001*
Satisfactory knots Training group 1.04 05.07 2.73 1.134.4 \0.001*
Control group 2.02 0.934.2 2.87 15.6 0.004*
Suture sponge 1 score (max 100) Training group 49.0 385 90.7 77100 \0.001*
Control group 74.5 4494 77 6394 0.599
* Statistically significant at a p value \0.05
change in GOALS? score (p = 0.26), change in GEARS cuff, extent of capturing essential anatomy, extent of cap-
score (p = 0.20), change in total knots (p = 0.12), or turing essential surgical challenges, utility for training, and
change in satisfactory knots (p = 0.26). Group allocation utility for assessment. However, ratings for proximity of
did remain a highly statistically significant predictor of the bowel and bladder, appearance of the bowel, and ap-
change in suture sponge 1 score, even after controlling for pearance of the bladder were lower: median ratings 3, 2.5,
baseline score (p = 0.001). and 3.5, respectively, suggesting that this aspect of the
Face and content validity of the vaginal cuff model were model was less realistic.
supported by surveys completed by 13 obstetrics and gy- Comparison of the scores provided by the three
necology participants, one gynecologic oncologist, and one evaluators demonstrated high internal consistency for the
gynecologic oncology fellow. Their input showed median GEARS score, GOALS?, efficiency, total knots, and sat-
ratings of four out of a maximum five for the domains of: isfactory knots (Kendalls coefficient of concordance 0.85,
shape of the vaginal cuff, resistance to suture of the vaginal 0.84, 0.75, 0.93, 0.83, respectively). Statistically significant
123
J Robotic Surg (2015) 9:179186 185
correlations were found between the GOALS? score and participants with lower baseline GOALS? scores. This
the suture sponge 1 score on the same day, providing could have important implications for development of future
evidence of concurrent validity (Spearmans q 0.510, training curricula which are more appropriately tailored
p value \0.001). Suture sponge 1 pre-test score was sig- based on the participants pre-training skill level.
nificantly correlated with GOALS? score at the post-test The high inter-observer agreement for the GEARS score
(Spearmans q 0.451, p value 0.031), providing evidence of provides further evidence regarding the validity of this
predictive validity, with or without training. Strong corre- scale by an independent group. Furthermore, the high inter-
lations were found between the GOALS? score and total observer agreement for total knots and satisfactory knots
knots (Spearmans q 0.875, p value \0.001), and between and their correlation with the GEARS score suggest these
GOALS? score and satisfactory knots (Spearmans individual measurements may provide simpler substitutes
q 0.853, p value \0.001). No statistically significant dif- for the GEARS scale in the limited setting of assessing
ferences were found in injury to bowel or bladder between suturing ability. Finally, our study provides some valida-
groups or within groups at pre- versus post-test. tion of the vaginal cuff model as a training tool with face
and content validity.
In conclusion, participation in a well-defined virtual
Discussion reality training curriculum results in a trend of improved
ability to perform robotic suturing as assessed using an
Our results are in agreement with those of two other ran- inanimate task. However, statistical significance was not
domized controlled trials assessing virtual reality simula- demonstrated for the primary outcome, improvement in the
tion using the da Vinci Skills Simulator which suggest: a GOALS? score. Although our trial has limitations, taken
trend to improve in ability to perform inanimate tasks in the context of two other randomized controlled trials
robotically, less than perfect translation from virtual reality with similar results [10, 11], it suggests the need to develop
to inanimate tasks, and greatest benefit for participants with improved virtual reality exercises and supports inanimate
minimal to no prior robotic simulation or robotic surgery models as a worthwhile complement to virtual reality.
experience [10, 11]. Rather than reflecting inherent
limitations of the da Vinci Skills Simulator, which remains
a platform with great flexibility and potential, this may Conflict of interest Authors Kiely, Gotlieb, Lau, Zeng, Samoue-
rather indicate limitations of the current virtual reality lian, Ramanakumar, Zakrzewski, Brin, Fraser, Korsieporn, Drudi, and
exercises. Nevertheless, the gap in translation between Press declare that they have no conflict of interest.
virtual reality and inanimate model tasks provides support
Informed consent Informed consent was obtained from all study
for the development of inanimate robotic surgical skill participants.
curricula [23] and advanced inanimate robotic surgical
simulations [24, 25] as a possible intermediate step be- Compliance with ethical standards All procedures performed in
tween virtual reality and live surgery on patients. this study were in accordance with the ethical standards of the McGill
Faculty of Medicine Institutional Review Board and with the 1964
The main limitations of our study include baseline im- Helsinki declaration and its later amendments.
balances in pre-test scores between groups and recruitment
which did not attain our target sample size. In addition, there
were some training group participants with short training References
times who did not complete the curriculum as well some
contamination of the control group by live robotic surgery 1. Reznick RK, MacRae H (2006) Teaching surgical skillschan-
experience or simulation training (off protocol) during the ges in the wind. N Engl J Med 355:26642669
2. Liss MA, McDougall EM (2013) Robotic surgical simulation.
study period. All of these weaken the strength of our ultimate
Cancer J 19:124129
conclusion of no significant difference between groups in our 3. Hung AJ, Zehnder P, Patil MB, Cai J, Ng CK, Aron M et al
primary outcome, improvement in the GOALS? score. (2011) Face, content and construct validity of a novel robotic
The strengths of our study are rigorous outcome assess- surgery simulator. J Urol 186:10191024
4. Alzahrani THR, Alkhayal A, Delisle J, Drudi L, Gotlieb W,
ment with blinded video review as well as the possibility of
Fraser S, Bergman S, Bladou F, Andonian S, Anidjar M (2013)
comparing virtual reality and inanimate model scores at Validation of the da Vinci Surgical Skill Simulator across three
baseline and conclusion. These aspects allowed us to surgical disciplines: a pilot study. Can Urol Assoc J 7:e520e529
establish that virtual reality training translated well into 5. Lee JY, Mucksavage P, Kerbl DC, Huynh VB, Etafy M, McDougall
EM (2012) Validation study of a virtual reality robotic simulator
improved performance of the virtual reality task (highly
role as an assessment tool? J Urol 187:9981002
significant after adjustment) but not so well into performance 6. Hung AJ, Jayaratna IS, Teruya K, Desai MM, Gill IS, Goh AC
of the inanimate task (not significant after adjustment). In (2013) Comparative assessment of three standardized robotic
addition, the curriculum appeared to be most beneficial for surgery training methods. BJU Int 112:864871
123
186 J Robotic Surg (2015) 9:179186
7. Abboudi H, Khan MS, Aboumarzouk O, Guru KA, Challacombe 16. Rosenthal ME, Ritter EM, Goova MT, Castellvi AO, Tesfay ST,
B, Dasgupta P et al (2013) Current status of validation for robotic Pimentel EA et al (2010) Proficiency-based fundamentals of la-
surgery simulatorsa systematic review. BJU Int 111:194205 paroscopic surgery skills training results in durable performance
8. Fried GM, Feldman LS, Vassiliou MC, Fraser SA, Stanbridge D, improvement and a uniform certification pass rate. Surg Endosc
Ghitulescu G et al (2004) Proving the value of simulation in 24:24532457
laparoscopic surgery. Ann Surg 240:518525 (discussion 258) 17. Scott DJ, Ritter EM, Tesfay ST, Pimentel EA, Nagji A, Fried GM
9. Carter FJ, Schijven MP, Aggarwal R, Grantcharov T, Francis NK, (2008) Certification pass rate of 100% for fundamentals of la-
Hanna GB et al (2005) Consensus guidelines for validation of paroscopic surgery skills after proficiency-based training. Surg
virtual reality surgical simulators. Surg Endosc 19:15231532 Endosc 22:18871893
10. Hung AJ, Patil MB, Zehnder P, Cai J, Ng CK, Aron M et al 18. Schulz KF, Grimes DA (2002) Allocation concealment in ran-
(2012) Concurrent and predictive validation of a novel robotic domised trials: defending against deciphering. Lancet 359:614618
surgery simulator: a prospective, randomized study. J Urol 19. Finan MA, Rocconi CM (2010) A novel method for training
187:630637 residents in robotic hysterectomy. J Robot Surg 4:3339
11. Vaccaro CM, Crisp CC, Fellner AN, Jackson C, Kleeman SD, 20. Vassiliou MC, Feldman LS, Andrew CG, Bergman S, Leffondre
Pavelka J (2013) Robotic virtual reality simulation plus standard K, Stanbridge D et al (2005) A global assessment tool for
robotic orientation versus standard robotic orientation alone: a evaluation of intraoperative laparoscopic skills. Am J Surg
randomized controlled trial. Female Pelvic Med Reconstr Surg 190:107113
19:266270 21. Goh AC, Goldfarb DW, Sander JC, Miles BJ, Dunkin BJ (2012)
12. Sroka G, Feldman LS, Vassiliou MC, Kaneva PA, Fayez R, Fried Global evaluative assessment of robotic skills: validation of a
GM (2010) Fundamentals of laparoscopic surgery simulator clinical assessment tool to measure robotic surgical skills. J Urol
training to proficiency improves laparoscopic performance in the 187:247252
operating rooma randomized controlled trial. Am J Surg 22. PASS: Power analysis and sample size software. http://www.
199:115120 ncss.com/software/pass/. Accessed 6 Feb 2015
13. Dulan G, Rege RV, Hogg DC, Gilberg-Fisher KM, Arain NA, 23. Arain NA, Dulan G, Hogg DC, Rege RV, Powers CE, Tesfay ST
Tesfay ST et al (2012) Proficiency-based training for robotic et al (2012) Comprehensive proficiency-based inanimate training
surgery: construct validity, workload, and expert levels for nine for robotic surgery: reliability, feasibility, and educational bene-
inanimate exercises. Surg Endosc 26:15161521 fit. Surg Endosc 26:27402745
14. Dulan G, Rege RV, Hogg DC, Gilberg-Fisher KK, Tesfay ST, 24. Marecik SJ, Prasad LM, Park JJ, Pearl RK, Evenhouse RJ, Shah
Scott DJ (2012) Content and face validity of a comprehensive A et al (2008) A lifelike patient simulator for teaching robotic
robotic skills training program for general surgery, urology, and colorectal surgery: how to acquire skills for robotic rectal dis-
gynecology. Am J Surg 203:535539 section. Surg Endosc 22:18761881
15. Palter VN, Graafland M, Schijven MP, Grantcharov TP (2012) 25. Kiely DJ, Gotlieb WH, Jardon K, Lau S, Press JZ (2015) Ad-
Designing a proficiency-based, content validated virtual reality vancing surgical simulation in gynecologic oncology: robotic
curriculum for laparoscopic colorectal surgery: a Delphi ap- dissection of a novel pelvic lymphadenectomy model. Simul
proach. Surgery 151:391397 Healthc 10:3842
123