Stereopsis is the perception of depth that arises from comparison of disparities in the images projected to two laterally separated eyes [1]. Most surgeons (except those who lack depth perception) use this visual effect in open surgery. Conventional two-dimensional (2D) minimally invasive surgery (MIS) using single-channel endoscopes and 2D screens is akin to monocular viewing.

Contemporary 3D MIS systems capture separate images using dual-channel laparoscopes consisting of either two separate rod lenses or two separate chips at the end of the scope to provide two vertically separated images. This produces different fixed distance perspectives of the operative field and simulates binocular imaging as if the viewer were positioned at the tip of the laparoscope [2]. In most modern commercially available 3D systems, users wear lightweight glasses that polarize alternate horizontal rows of pixels corresponding to the right- and left-eye images.

The first randomized controlled trial (RCT) investigating three-dimensional imaging was reported 20 years ago [3]. This used a now outdated system as technological developments have led to improved system quality and overcome many of the technical and resolution challenges seen in early platforms. Several studies have been published in recent years, both in experimental and clinical settings, suggesting 3D systems present a number of potential benefits to surgeons, trainees and patients [4]. Therefore, the European Association of Endoscopic Surgery (EAES) sponsored this consensus development conference on the use of 3D laparoscopy MIS systems. The aim of this consensus was to draw a number of statements based on the available evidence and develop recommendations for the MIS surgical community.

Materials and methods

The scope of this consensus on the use of 3D systems in laparoscopic surgery consisted of three main parts: general topics, organ-specific data and ongoing trials. The coordinating team (AA, YM and NV) formulated a list of questions related to each topic to be specifically addressed, which guided literature searches (Table 1) conducted in cooperation with a medical information specialist of the University of Torino. An initial literature search was conducted in order to identify any additional topics of interest. All searches were performed in both PubMed and Embase electronic libraries on 22 September 2017 with no limitation regarding year of publication or language. Search strings are displayed in Table 2. Study inclusion criteria were (a) trials on 3D technology in the selected topic (Table 2); (b) RCTs, prospective and retrospective observational comparative studies. Case reports or series of less than 10 patients were excluded.

Table 1 List of questions regarding the use of 3D to be addressed
Full size table
Table 2 Literature searches
Full size table

Endpoints

The two primary endpoints were the impact of 3D on operative time and complications (both intra- and post-operative). Eligible organ-specific studies were also merged into a single dataset for operative time assessment in addition to separate subgroup analyses. Complications were analysed together and also underwent planned subgroup analysis where only those that appeared directly related to surgery were included. RCTs and prospective studies were used for the analysis of all outcomes but complications also included data retrieved from retrospective studies.

Research team

The coordinators invited 13 surgeons and engineering members of the EAES executive and technology committee with recognized expertise on the topic of 3D vision to join the panel of experts. Each was asked to nominate at least one young surgical researcher to participate. An international research team of 14 young surgical researchers was formed to review and evaluate the existing literature on the use of 3D technology in laparoscopy. Young researchers were mentored by a senior expert. The final list of topics was approved by the experts and subsequently divided among the teams (Table 3).

Table 3 Topics and distribution among experts and young researches
Full size table

All search hits were screened by topic and reviewed by two team members for eligibility, based on title and abstract. If considered eligible, full-text articles were reviewed and summarized. In cases of disagreement, the coordinators acted as referee and made the final decision. Standardized data extraction forms were used across all topics. A PRISMA chart was completed for each literature search according to recommendations [5]. The methodological quality of included RCT was assessed using the Cochrane risk of bias score [6]. All included studies were evaluated with the GRADE system [7,8,9]. GRADE is a systematic and explicit approach to judging quality of evidence and strength of recommendations. GRADE specifically assesses methodological flaws, consistency of results across different studies, generalizability of research results and treatment effectiveness. The original Centre for Evidence-Based Medicine levels of evidence (LoE) system was used [9].

The highest levels of identified evidence were assessed first. If there was a systematic review of sufficient quality, it was used to answer the research question with a statement. When data were considered sufficient, consensus statements were prepared by each team and scored with a grade of recommendation (GoR).

Consensus development process

A face-to-face consensus meeting was held in London on 20 January 2018 to present all findings and drafted consensus statements and recommendations, which were finalized during two further virtual meetings. A modified Delphi method was used, as anonymity was not applicable in our situation [10,11,12]. All statements and recommendations were shared with the proposed LoE and subjected to voting for agreement or disagreement. In case of 100% consensus, the statements and recommendations were accepted. Where there was lack of consensus, the research team responsible for that topic presented the underlying evidence and rationale for their statement. After discussion, further voting rounds were conducted until an agreement was reached.

All finalized recommendations and statements with LoE and GoR were presented at a dedicated session during the 26th EAES congress London 2018. Fifty delegates attended and voted on each statement in two aspects: (a) Do you agree with the above-mentioned recommendation? (b) Will this recommendation change your practice?

Statistical analyses

All analyses were performed by a specialist biostatistician (RP), according to the original treatment allocation (intention-to-treat analysis). For binary outcome data, the relative risk (RR) and 95% confidence interval (CI) were estimated using the Mantel–Haenszel method; a RR < 1 favoured 3D vision. For continuous outcome data, the mean differences (MD) and 95% CI were estimated using inverse variance weighting with negative MD values favouring 3D vision. When means and/or standard deviation were not reported they were estimated from reported medians, ranges and sample size as described by Hozo [13]. A fixed-effects model was used in all meta-analyses, repeating the same analyses using a random-effects model as described by DerSimonian and Laird [14].

Risk of bias assessment

Publication bias was assessed by generating a funnel plot and performing the rank correlation test of funnel plot asymmetry. Heterogeneity was assessed by the I2 measure of inconsistency. Potential sources of heterogeneity were explored by different sensitivity analyses: comparing fixed- versus random-effects models (incorporating heterogeneity by using the random-effect method); checking the results of cumulative (sequentially including studies by date of publication) and influential meta-analyses (calculating pooled estimates by omitting one study at a time) and performing subgroup analyses. All analyses were performed with the R package (v3.5.0 Package Meta, The R Foundation for Statistical Computing, Vienna-A. http://www.R-project.org) [15] and Review Manager (RevMan 5.3; The Cochrane Collaboration).

Results

The literature searches yielded 9967 hits. In total, 138 articles were included and reviewed in detail to define 13 consensus statements and two recommendations. Nearly all studies presented in this review used dual-channel 3D laparoscopes and HD passive polarizing systems.

General topics

The impact of 3D vision on basic laparoscopy

Statement: 3D vision improves outcomes for junior trainees performing standardized box trainers tasks using properly setup 3D HD systems and passive polarized glasses (LoE: high).

The evidence behind this statement was derived from 14 studies, of which half were RCTs, comparing outcomes of basic experiments using either 2D or 3D vision systems [16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]. Most RCTs were well designed. Primarily due to insufficient reporting, selection bias could not be excluded. Only three studies were considered to have a high randomisation bias risk with three considered to be at low risk of bias for participant blinding and 29% for outcome assessment blinding.

Ten papers focused on specific tasks with all but one demonstrating a significant reduction in errors. In five papers, this was assessed on novices. Seven papers (three studying novices) assessed task completion time with all demonstrating a reduction in operative time with 3D.

The primary endpoints of included articles varied from task completion time, quality of task performance, enacted errors and subjective questionnaire assessment of comfort and headache. Some studies also used completely different imaging systems for 2D versus 3D [20, 30]. Few studies standardized specific conditions of 3D setup (monitor height, monitor distance and viewing angle) [27, 29] or tested participants for their stereovision abilities which could impact outcomes [28, 29].

Impact of 3D vision on training

Statement: The use of 3D imaging systems improves laparoscopic box trainer task completion time and error rate but this benefit has not been studied in clinical practice (LoE moderate).

Evidence from 72 primary studies in 19 countries across four continents supported this statement [3, 19,20,21, 24, 28,29,30, 32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96]. Publication dates varied from 1996 to 2017 suggesting that a spectrum of 3D systems were used although 57% of the publications were reported in 2014 onwards. Studies included 2452 participants: 1367 (55.8%) were laparoscopically naïve, primarily medical students, with 644 trainees (26.3% [486 junior (26.3%), senior 186 (7.6%)]) and 404 expert surgeons studied. Primary endpoints were operative time (95.8%), enacted errors (62.5%), task-specific score (22.2%), instrument path length (13.9%), repetitions performed (5.6%) and instrument movement speed (4.2%).

The vast majority of included studies were single centre and utilized crossover designs where participants completed the same tasks in 2D and 3D. 68 studies (94.4%) were performed in box trainer simulators with three animal experiments (two ex-vivo, one live [91]). Only two studies included operative room performance with both using laparoscopic cholecystectomies [3, 80]. Participants performed an average of three tasks (IQR 2–4, range 1–10). Only 36% of studies used previously validated tasks mainly taken from the Fundamentals of Laparoscopic Surgery (FLS) tasks, McGill Inanimate System for Training and Evaluation of Laparoscopic Skills (MISTELS) or European training in basic laparoscopic urological skills (E-BLUS) systems. The selection of the tasks within each simulation was not fully explained and only three studies assessed all tasks from their chosen program.

Of the 33 identified RCTs, primarily due to insufficient reporting, selection bias could not be excluded [6]. Only 30% of the studies were considered to have a low randomization bias risk, with 15% of studies maintaining allocation concealment. Blinding of participants to imaging modality is challenging although one group had students wear 3D glasses before entering the testing room. Only two studies were assessed as low risk of outcome assessment bias [47, 80]. One reviewed deidentified 2D videos of performance and one used automated tracking technology to record instrument metrics. All other assessments were performed by directly observing performance and therefore not blinded to allocation. The use of laboratory-based studies meant attrition bias was low but inadequate reporting meant that selective reporting and other bias could not be fully assessed. Overall compliance with the CONSORT statement was low across all included trials.

Pooling all 145 endpoints from included studies, 3D was significantly better in 90 (62.1%). 3D imaging was associated with a significant reduction in task completion time in 44 of 69 studies (63.8%). In the 45 studies using errors as the primary outcome, 28 (62.2%) observed a significant reduction when using 3D. Task-specific scores were significantly higher in the 3D arm in 56.3% of the 16 studies. Instrument path length, repetitions needed and instrument movement speeds were significantly improved by 3D in 50%, 50% and 100% of studies, respectively. It is noteworthy that both clinical studies did not show any time or error count differences between 2D and 3D modalities. Across all studies and endpoints, 2D was seen to be significantly quicker in two studies only (1.4% of all endpoints).

Impact of 3D vision on cognitive load

Statement: 3D laparoscopy does not introduce a higher cognitive workload and may result in decreased experienced cognitive workload provided that the viewing setup is optimal (LoE: moderate).

The systematic search identified 1684 hits with seven eligible for inclusion [3, 27, 52, 55, 68, 91, 97]. Three studies were randomized trials [3, 27, 52]. Only one study was of high quality according to Cochrane risk of bias tool [27].

Buia et al. stated that the perceived cognitive workload in 3D laparoscopy was not higher. This was shared by Lin et al. and Feng et al. [27, 52, 91]. Sakata and colleagues found cognitive workload using 3D laparoscopy was lower provided the viewing setup was optimal [98]. Benefits other than lowering cognitive workload were reported by Foo et al. and Kong et al.; however, they reported no conclusion on cognitive workload [55, 97]. The only adverse effects of 3D vision were reported by Hanna et al. [3] although this used an outdated 3D system [3].

Pitfalls of 3D vision in laparoscopy

Statement: 3D systems may increase visual fatigue, discomfort and headaches when setup is not optimal (LoE: moderate).

Out of 481 hits found, only 3 articles were included.

3D laparoscopic systems provide an improved stereoscopic vision which facilitates tasks performance, especially in inexperienced subjects [99]. Nevertheless, signs of visual discomfort, such as headache and visual fatigue, dry eyes or double vision, are reported in all studies [1, 99, 100]. Interestingly Zhou et al. [100] found out that whereas individuals experienced the above-mentioned discomfort symptoms, objective visual functional parameters (distance and near exophoria and esophoria, fusion range, accommodative convergence/accommodation, and tear film breakup time) did not worsen during 3D laparoscopy. Sakata et al. [1] suggested that looking at the screen from eccentric positions causes variable degrees of double vision, whereas an optimal position results when the centre of the screen is aligned with the eyes of the viewer.

Cost-effectiveness

No prospective study or RCT directly investigating the cost of the 3D laparoscopy could be found. Hence, no statement made relating to the cost-effectiveness of 3D systems compared to 2D systems can be made.

Pooled evidence on potential clinical benefits from 3D imaging

Operative time

Statement: 3D laparoscopy shortens the operative time across all analysed surgical specialities (general surgery, urology and gynaecology) (LoE: high).

We included 18 primary studies, reporting data about operative time, from nine countries across three continents [3, 26, 101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116] (Fig. 1). All but one study was published after 2013 suggesting a relatively limited variability of modern 3D systems were used. Studies included 1729 individuals: 487 (25.5%) solid organ operations; 1289 (74.5%) hollow organ procedures; 875 (50.6%) cases contained laparoscopic suturing while 794 papers (45.9%) did not specify the type of surgery. 647 procedures were general surgical with others consisting of urology and gynaecology procedures. The seven identified RCTs were assessed for bias using the Cochrane risk tool [6] (Table 4). Primarily due to insufficient reporting, selection bias could not be excluded. All were deemed to have a low randomization bias risk with 57% of studies maintaining allocation concealment. Blinding of participants to imaging modality was not possible. Overall compliance with the CONSORT statement was low.

Fig. 1
figure 1

Reproduced with permission from Moher et al. [117]

PRISMA 2009 flow diagram—22 September 2018: operative time.

Full size image
Table 4 Risk of bias (Cochrane risk tool) for RCTs selected for operative time assessment
Full size table

Pooling data derived from the 18 included studies, and 3D significantly shortened operative time. A mean difference (MD) of 11 min (8% [95%CI 1.72–20.29]) in favour of 3D with a high heterogeneity was seen (I2 96%, Fig. 2). Similarly, operative time in procedures including laparoscopic suturing a MD of 15 min was seen (11% [95%CI 2.70–27.2], I2 87%, Fig. 3). This effect size was reduced in procedures not including laparoscopic suturing with a MD of 6 min (5% [95%CI 0.11–11.79], I2 80%, Fig. 4) and absent in hollow organ procedures with a MD of 2.7 min (3.8% [95 CI% 2.91–8.33], I2 80%, p = 0.34). The analysis of the operative time in procedures performed on solid organs showed a MD of 21.7 min (14% [95 CI% 6.45–36.94], I2 88%, Fig. 5). Finally, the analysis of the operative time in procedures performed by general surgeons shows a MD of 7.44 min (4% [95 CI% 0.66–14.23], I2 85%, Fig. 6).

Fig. 2
figure 2

ABC

Full size image
Fig. 3
figure 3

DEF

Full size image
Fig. 4
figure 4

HIJ

Full size image
Fig. 5
figure 5

KLM

Full size image
Fig. 6
figure 6

NOP

Full size image

Recommendation: We recommend the use of 3D vision in laparoscopy to reduce the operative time (GoR: Low).

The recommendation was voted by 38 delegates with 36 (95%) agreeing and 44% stating this recommendation was likely to change their practice. 8 (19%) were already using 3D, while 15 (37%) disagreed that this recommendation would influence their practice.

Complications

Statement: The pooling of data from different settings seems to suggest a lowering in the overall rate of complications after surgical procedures involving suturing in 3D laparoscopy (LoE: low).

We included 18 primary prospective and retrospective studies from eight countries [26, 101,102,103, 105, 106, 108,109,110,111,112, 114, 118,119,120,121,122,123]. The flowchart is described in Fig. 7 with risk of bias reported in Table 5. All studies were published after 2013 suggesting a relatively limited variability of 3D systems used. Studies included 1733 individuals. 12 prospective studies included 1039 patients and six retrospective included 694 patients. 10 papers containing 713 patients regarded procedures including laparoscopic suturing with nine general surgical papers of 958 patients and all others consisting of urology and gynaecology procedures. Not all reported complications were considered for the analysis, as some appeared unrelated to the surgical procedure (Table 6).

Fig. 7
figure 7

Reproduced with permission from Moher et al. [117]

PRISMA 2009 flow diagram—22 September 2018: complications.

Full size image
Table 5 Risk of bias (Cochrane risk tool) for RCTs selected for complications assessment
Full size table
Table 6 List of complications considered for the analysis
Full size table

Overall compliance with the CONSORT statement was low across all five included RCTs, which were also assessed using the Cochrane risk tool [6]. Due to insufficient reporting, selection bias could not be excluded. All were deemed to have a low randomisation bias with 60% of studies also maintaining allocation concealment. Blinding of surgeons to imaging modality was not possible.

Using data pooled from the 18 included studies, a significant overall reduction complications was observed (RR 0.75, [95 CI% 0.60–0.94], I2 0%). However, no significant difference was observed when considering only general surgical procedures (RR 0.78, 95CI 0.60–1.02, I2 0%). When considering procedures including laparoscopic suturing, 3D showed a significant reduction in complications (RR 0.57 [95 CI% 0.35–0.90], I2 0%, Fig. 8). Omitting one study each time, the RR varied from 0.54 to 0.61 but without any statistically significant variation for the I2, demonstrating that no trial was a potential source of inconsistency. This was also confirmed in a subgroup analysis. When including only RCTs and prospective studies, the reduction in complications increased (RR 0.50 [95 CI% 0.25–0.97], I2 0, Fig. 9).

Fig. 8
figure 8

STU

Full size image
Fig. 9
figure 9

XYZ

Full size image

Recommendation: Future research is recommended to specifically investigate the potential benefit of the use of 3D laparoscopy system on complication rate (GoR: high).

This recommendation was voted by 45 delegates with 39 (87%) agreeing. 30 (72%) stated this recommendation was likely to change their practice while 6 members disagreed (14%) and 6 members (14%) reported already using 3D systems.

Organ specific

Cholecystectomy

Statements: There is no evidence that 3D vision is superior or inferior to 2D in laparoscopic cholecystectomy in terms of intra- and post-operative complications (LoE: moderate).

Less experienced surgeons could benefit from 3D imaging resulting in shorter operative time in laparoscopic cholecystectomy (LoE: low).

We analysed 402 hits. Six studies met inclusion criteria containing 309 elective patients (four prospective studies [80, 116, 125, 126] and two RCTs [3, 115]). Publication dates varied from 1998 to 2017, although 84% were reported from 2013 onwards.

All studies used operative time as their primary endpoint. Three studies demonstrated a reduction in operative time in favour of 3D [115, 116, 126], but in two studies, this was reported only in novices while no difference was noticed in the expert group. In all studies, the rate of conversion to open cholecystectomy and intra- and post-operative complication rate were not different in 3D. Four studies analysed the error rates during laparoscopic procedures with no significant difference seen [3, 80, 115, 126].

Colorectal surgery and appendectomy

Statement: 3D visualization shortens operative time in right colectomy compared to 2D (LoE: moderate).

We analysed a total of 552 hits of which six papers met the inclusion criteria [122,123,124, 127,128,129], with one RCT [124]. The procedures analysed were all resectional surgery but heterogeneous. Meta-analysis of these trials highlights a significant reduction in operative time (− 13.4 min; CI95% − 26.05, − 0.83, p < 0.01) but no significant difference in complications or lymph node yield. The RCT was performed by Curro et al. [124] and includes 40 right colectomies, 40 sigmoidectomies and 40 anterior resections, showing no difference in terms of complications or operative time.

For the topic of appendectomy, a total of 75 articles were identified; however, no article met the inclusion criteria.

Upper GI and bariatrics

Statements: 3D systems shorten operative time in hiatal hernia repair and mini gastric by-pass procedures (LoE: moderate).

There are no significant advantages in 3D for gastrectomy and sleeve gastrectomy (LoE: moderate).

Literature search identified 656 abstracts. After screening, six studies were included [120, 121, 130], of which three were RCTs [112,113,114]. The surgical procedures were hiatal hernia repair, bariatric surgery and two gastric cancer and two oesophagectomy studies. Operative time was investigated by all three RCTs with two showing a statistically significant advantage for 3D. There was significant time reduction in the 3D group for both bariatric surgery and hiatal hernia repair compared to 2D, but not for gastric cancer or oesophageal cancer surgery. Surgical complications were reported by all six studies with no significant differences seen.

Liver, pancreas, spleen and adrenal surgery

Literature search identified 666 hits for liver, 320 for pancreas, 190 for spleen and 152 for adrenal gland. Only one single prospective comparative study was found, focusing on different kinds of anatomical liver resections. The study [111] shows a significant reduction of blood loss (1255 ml versus 654 ml) and complications (33% vs. 14%) in favour of 3D. No prospective study or RCT dealing directly with the pancreas, spleen and adrenal surgery could be found. Hence, no statement made relating to the use of 3D systems in liver, pancreas, spleen or adrenal surgery could be made.

Abdominal wall

No prospective study or RCT dealing directly with 3D laparoscopic abdominal wall or hernia surgery was found. Hence, no statement could be made.

Gynaecology

Statements: 3D laparoscopy could be of benefit in terms of operative time in more complex procedures (LoE: moderate).

The literature search identified 1273 hits. Five articles met the inclusion criteria [107,108,109,110, 131] containing three RCTs [107, 109, 131] and a total of 693 patients: 371 in 2D group and 322 in 3D group. All procedures required suturing in the reconstructive phase. Two studies reported a shorter operative time in the 3D group [108, 110], but overall no statistically significant difference was observed (MD 2.12, [CI95% − 24.87 to 29.11], I2 81%, p = 0.88). No difference in terms of blood loss was seen. Meta-analysis of complications showed no differences. One study reported a significantly shorter hospital stay in favour of 3D surgery [108].

Urology

Statements: 3D laparoscopy significantly reduces the operative time but not perioperative complications in prostatectomy and renal lodge surgeries (LoE: high).

The literature search identified 1653 hits. Seven studies met inclusion criteria [26, 103,104,105,106, 132, 133] with four RCTs [26, 104,105,106]. A total of 460 patients were included: 224 in 3D and 236 2D. The urological procedures included 122 donor nephrectomies, 121 radical prostatectomies, 93 partial nephrectomies, 54 simple nephrectomies, 40 pyeloplasties, 21 radical nephrectomies, six radical cystectomies and three other laparoscopic surgeries. The operative time was significantly shorter in two of the four RCTs [104, 105], but in meta-analysis a statistically significant was seen (MD − 25.6 min, [95 CI% − 46.45 to − 4.75], I2 88%, p = 0.02). Blood loss was significantly lower in two of the three RCTs reporting it [26, 104]. Two of the three RCTs demonstrated shorter suturing time in 3D [26, 104]. When considering complications in suturing cases, a total of 11/179 and 18/223 for the 3D and 2D group were recorded, respectively. Meta-analysis did not show any difference in complications when including all studies or radical prostatectomy alone.

Ongoing trials

At the time of writing, searching registries of privately and publicly funded clinical studies for the terms “3D” and “laparoscopy”, we found 18 ongoing RCTs [134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151] registered on ClinicalTrials.gov and ISRCTN registries: 12 in Europe, three in the United States and three in Asia. Five have completed recruiting, while three are in setup. Four deal with cholecystectomy, three colorectal surgery, two gastrectomy, two hernia repair, six gynaecology surgery and one radical prostatectomy. In 10 studies the primary outcome is operative time, three error count, two complications, two patient reported outcomes (pain assessment and quality of life) and one lymph-node yield. Common secondary outcomes are conversion rate, assessment of fatigue, blood loss, readmission, mortality, oncologic safety and specimen quality, nerve sparing and functionality.

Discussion

Over the past two decades, the rapid advancement in 3D imaging technology appears to have successfully overcome previous barriers to its wide clinical use. It was important for the EAES to sponsor this consensus conference in order to define recommendations, based on available evidence to provide guidance to clinicians regarding the impact of 3D laparoscopy on their practice. This project has generated consensus statements and recommendations, demonstrating that 3D laparoscopy has some advantages in reducing operating times, cognitive load and possibly complications.

This consensus follows the recent health technology objective assessment report [4] by the EAES-affiliated Italian Society of Endoscopic Surgery and new technologies (SICE). An expert consensus was now felt appropriate as a combination of structured review of the evidence with leading multi-disciplinary technological and surgical expertise to generate statements and recommendations for clinical practice. We used a modified Delphi approach to reach consensus on the statements and grade of evidence adhering to the PRISMA, Cochrane risk of bias and GRADE principles. Agreement was reached on 13 proposed consensus statements (Table 7). Paucity of available evidence limited us to two recommendations underpinned by the meta-analyses.

Table 7 List of statements and recommendations
Full size table

This project also involved pooling the evidence from multiple studies to allow meaningful conclusions by increasing the sample size. Meta-analysis of the included clinical studies across all specialities showed 3D systems reduced operative time by a mean of 11 min representing 8% of case time. This effect size was larger in procedures involving laparoscopic suturing. Box trainer task time was also shorter with 3D.

We found a large body of evidence investigating the role of 3D on the performance on box training. Results of the laboratory experimental trials showed 3D was associated with significantly better performance than 2D in the majority of technical tasks. Consideration for the learning effect when repeating identical tasks with a different imaging modality should be made as performance could be higher irrespective of imaging used. While affected by the risk of bias and methodological flaws, our findings suggest that 3D technology improves laparoscopic box trainer simulator task performance. This could speed time to competency with fewer enacted error events in laparoscopically naïve or junior participants. There are no data, however, on whether these benefits translate to operating room performance or patient outcomes. A Cochrane review concluded that the benefits of box training have not been shown to translate to real-world performance [152]. As improved operating performance represents the goal of all minimal access training, further dedicated translational and longitudinal studies are clearly indicated.

The major output of this consensus is represented by a clear advantage of 3D in the overall rate of complications, in particular after surgical procedures involving suturing. The data should be considered cautiously due to the variety of procedures; however, this is likely to increase the generalizability of the results. Nevertheless, this difference was even more evident when only complications strictly related to vision, such as intraoperative injuries, leaks and fistulas, were considered. When including only RCTs and prospective studies, the benefit of the use of 3D increased showing a halving of complications, while maintaining no heterogeneity. This could be explained by the increasing task complexity with laparoscopic suturing. It appears that presence of stereoscopy seems to be crucial due to a more accurate appreciation of depth perception during this technique. This is consistent with the other studies which demonstrated superiority with 3D suturing technique [153]. Nevertheless, this benefit of 3D was observed as either a secondary endpoint in the vast majority of the studies or in small cohorts. Further robust research is required to confirm these findings and specifically investigate the true benefit of this technique on clinical outcomes.

Limitations

Our findings are in line with the available 2D/3D systematic reviews on this topic [75, 154, 155] which reported similar methodological concerns. In identified studies including a number of RCTs, significant heterogeneity was observed which limited the number of recommendations made. There is a clear need for further randomized studies that use validated and reproducible tasks or standardization of intervention delivery. Wherever possible equipment, viewing distance, table height and ergonomics should also be standardized. Compliance with the CONSORT statement, A-priori sample size calculations, homogenous participant groups, surgical experience, stereopsis visual assessment, validated blinded assessment methods and robust randomisation tools are additional considerations that would strengthen results and add to our understanding. Although we have identified that 3D imaging can assist surgical efficiency, no study attempted to assess quality of surgery or cost-effectiveness of this technology. Limited by study reporting, we could not investigate the effective of varying stereopsis abilities on outcome data.

It is well acknowledged that there is approximately 1–30% percent of the general population, and 9.7% of surgeons are stereo blind and would not be expected to benefit from 3D [156, 157]. This may limit the generalizability of our proposed recommendations and testing for stereo blindness is strongly advocated before clinical practice or research involving 3D systems.

Future research

This exercise has highlighted areas of equipoise as well as clinically important gaps in the literature and can serve as a guide for future clinical studies. Currently, there are 18 ongoing clinical trials that are related to this topic and already registered. Health economic analysis is required to evaluate the cost-effectiveness of 3D systems and the observed reduction in operative time and complication data. Dedicated research assessing the impact of 3D systems on clinical outcomes is indicated.

Conclusion

3D laparoscopic systems reduce procedure time in both the operating room and box trainer settings. 3D vision may also reduce perioperative complications particularly in procedures involving laparoscopic suturing.