The search strategy retrieved 1,148 citations by interrogating EMBASE with the MeSH words 'microsurgery' and 'education', 'microsurgery' and 'training', and 'microsurgery' and 'curriculum'. These were limited respectively to 9 RCTs and 7 abstracts, 12 RCTs and 4 abstracts, and 0 RCTs and 0 abstracts. The strategy retrieved 1,460 citations by interrogating MEDLINE using Multi-Field Search-"All Fields", meshing 'microsurgery' and 'education' and 'microsurgery' and 'training', with a total of 21 RCTs and 10 abstracts. The strategy retrieved 2,277 citations by interrogating PubMed in a similar way to EMBASE, with a total of 36 RCTs and 14 abstracts. 5 RCT, met our inclusion criteria and were included in this review [4
Grober et al. [6
]-The Educational Impact of Bench Model Fidelity on the Acquisition of Technical Skill: The Use of Clinically Relevant Outcome Measures
This double-blind, randomised controlled trial evaluated whether the acquisition of skill and clinically relevant outcomes are impacted by bench model fidelity. Fifty junior surgical residents were voluntarily recruited for the trial. Trainees with prior experience of >5 microsurgical cases as the primary surgeon (i.e., performing greater than 80% of the procedure) were excluded. The trainees participated in a one-day microsurgical training course. Initially all trainees received an instructional video demonstrating basic microsurgical skills followed by a baseline assessment pre-test drill. Trainees were randomized to one of three groups; those receiving training on high fidelity models (n=21), those receiving training on low fidelity models (n=19), and those receiving didactic training (n=10). High fidelity (HF) model training was performed anastamosing the anaesthestised rat vas deferens. Low fidelity (LF) model training was performed anastamosing silicone tubing, and the didactic group was identified as the control group.
Video recordings of assessments were scored by blinded experts using validated global rating scores. The final product was assessed in a blinded fashion for patency, suture precision and overall quality. The anastamosed vas deferens remained in the living rat for 30 days and was re-evaluated 30-days post anastomosis and tested for: patency, the presence of any sperm granuloma (suggesting leak), and the presence of sperm on microscopy from the abdominal end of the anastomosis to assess functional patency.
The method of randomization was not described, nor the method of concealment from the expert investigators. Therefore randomisation was graded as (B) unclear. Several questions remained unanswered, i.e., How were the candidates randomized? Was a power calculation performed? How were the candidates entering the trial blinded? Participants were blinded to the intervention as were the experts that were assessing the video recordings. Skill assessment, one of the principal outcomes in this study, was conducted. Handling of withdrawals and losses-There were no withdrawals or losses and all of the 50 participants enrolled in this clinical trial were accounted for and therefore this criterion was graded as (A) yes. As most of the criteria were met, this study was assessed as at low risk of bias (plausible bias unlikely to seriously alter the results) ().
The primary outcomes included: global rating & checklist score, suture precision placement & quality, and anastamosis completion time. Blinded assessors assessed the outcomes using a validated global rating scoresheet. Differences between the pre- and post-est scores were significantly greater in those that received the hands-on model training compared to those who received didactic training alone (P=0.004). The pre- and post-scores were not significantly different between the HF and LF groups. The post test scores were not significantly different in all three groups. Overall suture quality and precision did not differ in all three groups. Anasatomosis completion times were significantly faster in those who received hands on training.
The secondary outcomes included: anastamotic patency, granuloma at the anastamotic site presence of sperm on microscopy, and trainee preference. Seventy-two percent of the rats survived and were re-explored. Of these 50% of the 'didactic' mice survived, 79% of the LF mice survived, and 76% of the HF mice survived. There was significantly higher delayed anastamotic patency among bench-model trainees (P=0.039). There was no significant difference in granuloma formation or the rates of sperm presence on microscopy between the 3 groups. 90% of trainees preferred the HF model. There were no adverse effects reported.
Grober et al. [5
]-Laboratory Based Training in urological microsurgery with bench model simulators: a randomized controlled trial evaluating the durability of technical skill
This double-blind, randomised controlled study evaluated the durability and retention of skill acquisition with clinically relevant outcomes. Fifty junior surgical residents were voluntarily recruited, and randomized to either a high fidelity model, low fidelity model or didactic teaching, as in the previous study. Four months after focused teaching, 18 trainees were invited back (13 bench model trainees, and 5 didactic trainees), and re-assessed on a high fidelity, live animal model.
Blinded experts using validated global rating scores assessed video recordings. During the post-test assessment, all participants were blindly assessed on a live rat model of vas deferens anastamosis. This was then assessed for anastamotic patency.
The method of randomization was not described nor the method of concealment from the expert investigators therefore randomisation was graded as (B) unclear. Several questions remained unanswered i.e., how were the candidates re-selected for retention testing, it states "voluntarily returned", however this is unclear. Was a power calculation performed? How were the candidates entering the initial trial blinded? How did the interim clinical opportunities impact on outcomes. Only the assessors were blinded for this study. Skill assessment, one of the principal outcomes in this study, was conducted. There were no withdrawals or losses-all of the 18 participants enrolled in this clinical trial were accounted for and therefore this criterion was graded as (A) yes. This study was rated as moderate risk (plausible bias that raised some doubt about the results) of bias because of the possible confounding factors; most were selected from the bench model training group (13/18), and only 5/18 from the didactic training group ().
The primary outcomes included: global rating & checklist score, microsurgical drill: looking at dexterity, visuo-spatial awareness and skill, and a retention test. Blinded assessors assessed the outcomes using a validated global rating score-sheet. Patency testing was carried out on the live model using methylene blue dye injection.
Eight out of the 18 subjects had been exposed to microsurgical clinical opportunities in the interim but there was no significant difference with the frequency of microsurgery practice between the two groups however there was a significantly positive correlation between the number of interim practice opportunities and retention test/global rating scores (P=0.02). Global rating scores still remained significantly higher in those that had initially received hands-on bench model training compared with those that received didactic training (P=0.02) and therefore indicates that simulation training supercedes theoretical training. Anastamotic patency. Anastamotic patency rates were also significantly higher in those that received bench model training.
Moulton et al. [7
]-Teaching Surgical Skills: What kind of practice makes perfect?
This single-center, single-blinded, randomised controlled trial evaluated the effect of mass training vs. distributed training on microsurgical skill acquisition and the transferability of the acquired skills to life-like models. 38 postgraduate year one [PGY1], PGY2 & PGY3 surgical residents volunteered for the trial. These trainees participated in a one-day (4-session) mass training microsurgical course or a 4-week distributed (one session/week) microsurgical training course.
Trainees were stratified according to their post-graduate year, and randomized to one of two experimental groups: mass training (n=19), and distributed training (n=19). The mass training group received 4-training sessions in one day, and the distributed group received the same 4 training sessions over 4 weeks. In the first session participants watched a video on the principles of microsurgery, and practiced suturing on a Penrose drain. In the second session, they practiced microvascular anastomosis on a 2 mm poly-vinyl chloride artery simulation model. In the third and fourth sessions participants practiced microvascular anastomoses using the arteries of a turkey thigh. Each group received an equal overall training time. The microsurgical drill test was used in the pre-test during the first training session. The post-test was also a microsurgical drill carried out at the end of the fourth training sessions. The retention and transferability test was carried out one month after the training was completed, and this required that the trainees carry out a microsurgical drill test and an in-vivo infra-renal anastomosis on a live rat model respectively.
Blinded experts using previously validated global rating scores, checklists and end product evaluation methods assessed the pre-test, post-test and retention test (performed one month after the training). This was performed by two blinded experts assessing the video recordings independently. Computer based evaluations were also carried out for time to completion and motion analysis assessments. Clinically relevant outcome measures included the patency of the infra-renal anastomosis, narrowing of the anastomosis, bleeding and completion of the anastomosis.
The method of randomization was clearly described and considered adequate (A). Several questions remained unanswered i.e. Was a power calculation performed? How did they account for sub-speciality experience? Participants were blinded to the intervention, as were the experts that were assessing the video recordings. Skill assessment, one of the principal outcomes in this study, was conducted. There were no withdrawals or losses, and all of the 38 participants enrolled in this clinical trial were accounted for-therefore this criterion was graded as (A) yes. As most of the criteria were met, this study was assessed as at low risk of bias (plausible bias unlikely to seriously alter the results ().
The primary outcomes included: global rating & checklist score of pre-test, post-test and retention test, computer based evaluation of time to completion and motion efficiency, transferability to an in-vivo rat model of infra-renal anastomosis. Blinded independent experts assessed video recordings of the pre-test, post-test and retention test. There was no significant difference between the two groups at pre-testing, nor at immediate post-testing. Time to completion and the number of hand movements were not significantly different between the mass and distributed groups either, both at pre-test and post-test. Both groups showed significant improvement between the pre-test and post-test when using the global rating score, time to completion and motion analysis, but only the distributed group showed significant improvement when utilizing the validated checklist and end-product evaluation. The retention drill revealed the distributed group performed better in the computer based assessments but not expert-based outcomes. Nonetheless, the distributed group out-performed the mass training group in all expert-based outcome measures during the transfer, the clinically relevant outcome, but not the computer-based evaluations. Inter-rater reliability was also assessed using Cronbach's alpha and varied between 0.67 and 0.89 on all expert based outcome measures. There were no adverse effects to report.
Anastakis et al. [4
]-Assessment of technical skills transfer from the bench training model to the human model
This double-blind, randomised controlled trial evaluated whether technical skills learned by bench models are transferrable to a human cadaver model. Twenty-tree surgical PGY1 residents were recruited for this study and randomized. These trainees participated in a 3-day microsurgical training course. They were assigned to one of three groups: text only, bench model training, or cadaver model training, with 2 of each of the 6 procedures taught using one of the three modalities thus serving as their own control. Each course was a four-hour session, apart from the didactic training. One week following the intervention, the delegates were invited to carry out the procedure on a human cadaver model in an operating room environment to assess transferability. Two examiners, none of whom were previous instructors, were asked to independently evaluate the candidates, and were blinded to the instructional modality used. They were assessed using validated checklists & global rating scores.
The method of randomization was not described, nor the method of concealment from the expert investigators-therefore randomisation was graded as (B) unclear. Several questions remained unanswered i.e., What was the inter-rater reliability? Was a power calculation performed? How were the candidates entering the trial blinded? How was experience accounted for when randomsing the subjects? Participants were blinded to the study model and intervention, and the experts were blinded to the teaching delivery modality as they were assessing them carrying out the live cadaveric procedures. There were no withdrawals or losses, and all of the 23 participants enrolled in this clinical trial were accounted for-therefore this criterion was graded as (A) yes.
The study was graded as moderate to high risk bias. Initially the results analysis did not show any significant differences between the groups for both checklist scores and global rating scores. The team subsequently re-analysed the date to control for the wide variation in subject skill and procedure difficulty, and it was only after repeated measures of analysis of variance that a significant effect on training modality on transferability of skill was accounted for ().
The primary outcomes included: global rating score, checklist score, transferability. Blinded expert examiners evaluated the delegates' performance in the pre- and post-test microsurgical drill, as well as the transferability study where the candidates had to carry out the procedures on live cadaveric models. After controlling for variance and skill difficulty, the cadaveric and bench model forms of training had a marginally significant impact, only 7% to 10% increase, on the residents' ability to perform each procedure on the human cadaver, when compared to the manual reading group. There were no adverse effects to report.
Price et al. [8
]-A randomized evaluation of simulation training on performance of vascular anastomosis on a high fidelity in-vivo model: The role of deliberate practice
This single-center, single-blinded, randomised controlled trial evaluated the effect of independent and deliberate simulator practice, during non-clinical time, on the performance of an end-to-end anastomosis in an in-vivo model. 39 PGY1 & PGY2 surgical trainees were stratified and randomized to an expert guided tutorial on a procedural trainer group or to an expert guided tutorial on a procedural trainer combined with self directed practice on the same procedural trainer. The distribution of PGY1 & PGY2 were similar between the two groups and the distribution of subspecialties was similar. Trainees were randomized to one of two groups; those receiving training by an expert guided tutorial on a procedural trainer (n=18) and those receiving training by an expert guided tutorial on a procedural trainer followed by 10 sessions of independent self directed training (n=21). The pre-test was taken at the end of the expert-guided tutorial.
Blinded experts used validated Objective Structured Assessment of Technical Skill (OSATS) scores. Initially, the candidates were given an expert guided tutorial and trained on performing end-to-end anastomoses. The 4th anastomosis carried out was considered the pre-test sample and scored. It is unclear if this was done blind. Two weeks after the initial tutorial, the candidates were invited to perform an in-vivo carotid anastomosis on a porcine model, that was blindly assessed by 2 independent expert cardiac surgeons.
The method of randomization was clearly described and considered adequate (A). Several questions remained unanswered i.e., Was a power calculation performed? Participants were blinded to the intervention, as were the experts assessing the video recordings. Skill assessment, one of the principal outcomes in this study, was conducted. 3 candidates from each study group were lost to follow-up.
As most of the criteria were met, this study was assessed as at low risk of bias (plausible bias unlikely to seriously alter the results ().
The primary outcome was the OSAT score. Blinded independent experts assessed the anastomoses in the in-vivo porcine carotid artery model using OSAT scores. The animal models were used to create a high fidelity simulation model. The group that undertook independent self-directed training after the expert tutorial scored significantly higher OSAT scores. This was also the case when the PGY1 subgroup and PGY2 subgroup were considered independently. When the OSAT domains were considered independently, a statistically significant improvement was seen with the independent practice group in 4 out of the 7 domains. The subgroup analysis supports the robustness of the overall findings.
The secondary outcomes included: time to completion, and end-product evaluation. End product scores were significantly higher, and time to completion was significantly shortened, in the trainees randomized to self-directed practice. Inter-rater reliability was also assessed, and was high between the expert observers (intra-class correlation coefficient=0.8). There were no adverse effects to report.