Robotic-assisted versus laparoscopic surgery for colorectal cancer in high-risk patients: a systematic review and meta-analysis

Abstract

Background

Evidence of superiority of robotic-assisted surgery for colorectal resections remains limited. This systematic review and meta-analysis aims to compare robotic-assisted and laparoscopic surgical techniques in high-risk patients undergoing resections for colorectal cancer.

Methods

Systematic searches were performed using Pubmed, Embase and Cochrane library databases from inception until December 2024. Randomised and non-randomised studies reporting outcomes of robotic-assisted or laparoscopic resections in the following high-risk categories were included: obesity, male gender, the elderly, low rectal cancer, neoadjuvant chemoradiotherapy and previous abdominal surgery. Comparative meta-analyses for all sufficiently reported outcomes were completed. Risk of bias was assessed using the ROBINS-I and RoB 2 tools for non-randomised and randomised studies, respectively.

Results

48 studies, including a total of 34,846 patients were eligible for inclusion and 32 studies were utilised in the comparative meta-analyses. Conversion to open rates were significantly lower for robotic-assisted surgery in patients with obesity, male patients and patients with low rectal tumours (obese OR 0.41 [CI 0.32–0.51], p < 0.00001); male gender (OR 0.28 [CI 0.22–0.34], p < 0.00001); low tumours OR 0.10 [CI 0.02–0.58], p = 0.01). Length of stay was significantly reduced for robotic-assisted surgery in patients with obesity (SMD 0.25 [CI − 0.41 to − 0.09], p = 0.002). Operative time was significantly longer in all subgroups (obesity SMD 0.57 [CI 0.31–0.83], p < 0.0001; male gender SMD 0.77 [CI 0.17–1.37], p = 0.01; elderly SMD 0.50 [CI 0.18–0.83], p = 0.002; low rectal tumours SMD 0.48 [CI 0.12–0.84], p = 0.008; neoadjuvant chemoradiotherapy SMD 0.72 [CI 0.34–1.09], p = 0.0002; previous surgery SMD 1.55 [CI 0.05–3.06], p = 0.04). When calculable, blood loss, length of stay, complication rate and lymph node yield were comparable in all subgroups.

Conclusions

This review provides further evidence of non-inferiority of robotic-assisted surgery for colorectal cancer and demonstrates conversion rates are superior in specific, technically challenging operations.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10151-025-03141-3.

Introduction

Robotic-assisted surgery (RAS) has significantly impacted colorectal surgical management over the past two decades and has become established as a safe and viable alternative to laparoscopic surgery. RAS provides technical advantages including articulation of instruments, three-dimensional vision, remote operating and improved ergonomics [1, 2]. Its benefits over laparoscopic surgery have been widely documented in urological and gynaecological surgery [3–5]. In colorectal surgery, however, there is more limited evidence. The largest and most robust randomised control trial to date, the ROLARR trial, compared RAS with conventional laparoscopic surgery in patients undergoing rectal cancer resections [6]. This trial utilised conversion to open rates as the primary outcome and demonstrated the non-inferiority of RAS but no benefit. Subgroup analyses hinted at possible lower conversion rates for patients with obesity or male patients but the trial was not powered to detect significant differences in these patient groups. Considering the significant cost burdens [7] and barriers to training [8] associated with RAS, further evidence is required to clarify the benefits of this approach.

RAS may provide the most benefit in more technically challenging or high-risk operations. The following patient groups are independently associated with technical challenges or inferior outcomes in colorectal cancer resections and, as such, make up the subgroups of interest for this study: male gender [9–12], obesity [13], low rectal tumours [9, 11], neoadjuvant chemoradiotherapy [9, 11, 12], the elderly [10, 14] and patients with previous abdominal surgery [15, 16]. In male patients, the narrower anatomical pelvis provides more limited space for dissection during total mesorectal excision (TME). Similarly, limited space and poor visualisation are problems encountered in both low rectal tumours and patients with obesity. Neoadjuvant radiotherapy is associated with tissue oedema, fibrosis and scarring, and adhesion formation from previous abdominal surgery can alter dissection planes and is associated with a higher risk of iatrogenic injury. Elderly patients do not provide specific anatomical challenges but are at higher risk of significant comorbidities which impact perioperative physiology, tissue healing and recovery.

Whilst there are an increasing number of observational studies comparing robotic surgery to laparoscopic surgery in high-risk patients, randomised control trials are lacking. Therefore, the highest quality evidence will be provided by systematic review and meta-analysis of observational studies. Suwa et al. conducted a meta-analysis comparing outcomes in patients with and without obesity undergoing robotic colorectal surgery [17]. They concluded obesity is associated with a significantly longer operating time and an increased risk of conversion to open, in keeping with findings from laparoscopic surgery. No previous meta-analyses compare RAS to laparoscopic approaches in high-risk groups.

This review aims to perform a direct comparison of robotic-assisted versus laparoscopic surgery in high-risk patients to ascertain any benefit of RAS in addition to proving non-inferiority.

Methods

This systematic review has been reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [18]. A prospective review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) database (registration no. CRD42022326555).

Search strategy

A systematic search was performed using Pubmed, the Cochrane Library and EMBASE databases. The following search algorithm was used for each database: (Robotic) AND (colorectal surgery OR rectal surgery) AND (high risk OR obese OR obesity OR male OR gender OR age OR elderly OR location OR tumour location OR low tumour OR chemoradiation OR radiation OR previous surgery OR abdominal surgery) AND (outcomes). Results were filtered to human studies published in the English language. The final searches took place no later than December 2024.

Eligibility criteria

Inclusion

All randomised and non-randomised studies which report original data on any outcome of robotic colorectal surgery in high-risk patients are included. The high-risk categories we identified are obesity, male gender, age > 65 years, low tumour location, previous chemoradiation and previous abdominal surgery. These categories were chosen after careful inspection of current data concerning the risk factors for complications in colorectal surgery [10, 19].

Exclusion

All studies published in languages other than English were excluded from this review. Case reports were excluded and studies with no comparison to either laparoscopic surgery or a corresponding lower-risk category were excluded.

Study selection and data extraction

Each article identified by the initial search was manually and independently screened, using the study title and abstract in reference to the eligibility criteria. Screened articles were then included in a full-text review to confirm final eligibility.

The primary outcome was conversion to open rate. Secondary outcomes included overall complication rate (all Clavien Dindo grades), estimated blood loss, length of stay, operative time, readmission rate and the lymph node yield.

Data was manually and independently extracted onto a prospectively designed database. Corresponding authors were contacted if necessary to provide clarity or further detail during data extraction. All non-randomised studies underwent bias analysis using the ROBINS-I tool [20] and the RoB 2 tool was used for randomised trials [21].

Statistical analysis

Meta-analyses are presented as Forest plots. For the pooling of dichotomous data, the Mantel–Haenszel method was used and results are presented as an odds ratio (OR) with a corresponding 95% confidence interval (CI). For continuous data, the inverse variance (IV) method was used and results are presented as a standardised mean difference (SMD) with 95% CI. Analysis was completed using a random-effects model. When original studies reported outcomes using a median and a range or interquartile range (IQR), mean and standard deviation (SD) was estimated using previously published methods [22, 23]. Review Manager v5.4 (RevMan) software (Cochrane Collaboration, Oxford, UK) was used for comparative meta-analyses. Pooled weighted meta-analyses (PWM) were calculated using a random-effects model and either the metaprop or metan commands in Stata v14 (Stata Corp) software for dichotomous and continuous data, respectively. Pooled weighted meta-analyses are presented with corresponding 95% confidence intervals.

Results

Study selection

A total of 1841 records were identified from database searches, Fig. 1. Title and abstract screening excluded 1329 irrelevant records. Full-text screening of 211 papers was completed and 162 studies were excluded for not meeting the eligibility criteria. 48 papers were eligible for inclusion in this review. In order to have comparable homogenous data, the 32 studies [24–55] which performed a direct case-matched comparison of laparoscopic and robotic-assisted surgery in high-risk patients were included in comparative meta-analyses. The remaining 16 studies [6, 56–70] comparing high-risk to lower-risk patients undergoing robotic-assisted surgery were included in pooled summary meta-analyses but not in the comparative meta-analyses.

Fig. 1.

PRISMA diagram

Study characteristics

The included studies comprise 42 retrospective observational studies, four prospective cohort studies and two randomised control trials. The eligible studies include a total of 34,846 patients. A summary of study characteristics and their inclusion criteria is listed in Table 1.

Table 1.

Study characteristics

Author and year	Type of study	Site(s) of resection	Inclusion criteria	Total number	Laparoscopic	Robotic
Obesity
Hellan 2014	Retrospective	Rectal	BMI > 30	126	0	126
Keller 2015	Retrospective	Rectal + colonic	BMI > 30	68	0	68
Lagares-Garcia 2015	Prospective	Rectal + colonic	BMI > 30	34	0	34
Cardinali 2016	Retrospective	Colonic	BMI > 30	15	8	7
Gorgun 2016	Retrospective	Rectal	BMI > 30	56	27	29
Shiomi 2016	Retrospective	Rectal	VFA > 130 cm2	82	30	52
Ackerman 2017	Retrospective PSM	Rectal	BMI > 30	130	66	64
Baukloh 2017	Prospective	Rectal	BMI > 30	57	0	57
Bayraktar 2017	Retrospective	Rectal	BMI > 30	30	0	30
Jayne 2017	RCT	Rectal	BMI > 30	107	0	107
Schootman 2017	Retrospective PSM	Colonic	BMI > 30	4404	3589	815
Harr 2017	Retrospective	Rectal + colonic	BMI > 30	108	0	108
Pai 2017	Prospective	Rectal	BMI > 30	33	0	33
Duchalais 2018	Retrospective	Rectal	BMI > 30	58	0	58
Panteleimonitis 2018	Retrospective PSM	Rectal	BMI > 30	124	61	63
Peacock 2020	Retrospective	Rectal	BMI > 30	161	0	161
Ozben 2021	Retrospective	Colonic	BMI > 30	42	0	42
Hannan 2022	Retrospective	Rectal + colonic	BMI > 30	34	0	34
Juang 2023	Retrospective PSM	Rectal + colonic	BMI > 40	1296	648	648
Albayati 2022	Retrospective	Rectal	BMI > 30	2385	1824	561
Glencer 2022	Retrospective	Rectal	BMI > 30	6722	4711	2011
Zhao 2024	Retrospective PSM	Rectal	VFA > 100 cm2	242	121	121
Male
Ackerman 2017	Retrospective PSM	Rectal	Male	614	311	303
Baukloh 2017	Prospective	Rectal	Male	229	0	229
Jayne 2017	RCT	Rectal	Male	317	0	317
Serin 2015	Retrospective	Rectal	Male	79	65	14
Schootman 2017	Retrospective PSM	Colonic	Male	6489	5259	1230
Esen 2018	Retrospective	Rectal	Male	83	42	41
Elderly
Fernandez 2013	Retrospective	Rectal	≥ 65 years	72	59	13
Ceccarelli 2017	Retrospective	Rectal + colonic	≥ 65 years	38	0	38
Oldani 2017	Retrospective	Rectal + colonic	≥ 70 years	9	0	9
De’Angelis 2018	Retrospective PSM	Rectal + colonic	≥ 70 years	160	102	58
Palomba 2021	Retrospective	Rectal + colonic	≥ 65 years	83	51	32
Srinath 2022	Retrospective	Colonic	≥ 80 years	171	86	85
Ali 2023	Retrospective	Rectal	≥ 70 years	108	63	45
Lu 2024	Retrospective	Colonic	≥ 80 years	7550	7061	489
Low tumour location
Park 2010	Retrospective	Rectal	1–8 cm from AV	123	82	41
Baek 2013	Retrospective	Rectal	Mean 5.5 cm from AV	84	37	47
Park 2013	Retrospective	Rectal	Mean 3.5 cm from AV	80	40	40
Kuo 2014	Retrospective	Rectal	2–6 cm from AV	64	28	36
Yoo 2015	Retrospective	Rectal	< 5 cm from AV	70	26	44
Feng 2022	RCT	Rectal	< 5 cm from AV	347	173	174
Chemoradiotherapy
Saklani 2013	Retrospective	Rectal	Chemoradiotherapy	138	64	74
Serin 2015	Retrospective	Rectal	Chemoradiotherapy	79	65	14
Kim 2016	Retrospective	Rectal	Chemoradiotherapy	99	66	33
Huang 2017	Retrospective	Rectal	Chemoradiotherapy	78	38	40
Chen 2021	Retrospective PSM	Rectal	Chemoradiotherapy	171	95	76
Fiorenzo 2021	Prospective	Rectal	Chemoradiotherapy	102	64	38
Morohashi 2022	Retrospective	Rectal	Chemoradiotherapy	55	0	55
Yamanashi 2022	Retrospective PSM	Rectal	Chemoradiotherapy	60	30	30
Previous abdominal surgery
Park 2017	Retrospective	Rectal + colonic	Previous open or laparoscopic procedure that invaded peritoneal space	850	612	238
Huang 2020	Retrospective PSM	Rectal	Previous abdominal siurgery	32	0	32
Milone 2021	Retrospective PSM	All GI resections	Previous open abdominal surgery	98	62	36
Total				34,846	25,666	9180

PSM propensity score matching, RCT randomised control trial, BMI body mass index, VFA visceral fat area, AV anal verge, GI gastrointestinal

The included studies report on both rectal and colonic resections. 23 studies utilised in the comparative meta-analyses reported on surgery undertaken for rectal cancer only, and nine studies reported on segmental colectomies alone or both colonic and rectal resections. Given this confounding variable, a sensitivity analysis utilising only rectal resections has been completed for each meta-analysis. The results of sensitivity analyses are reported in each section and accompanying forest plots are presented in Fig. S1.

Only one randomised control trial (RCT) was used in the comparative meta-analyses for low rectal tumours. A further sensitivity analysis was undertaken without the RCT to ensure outcomes remained the same, which they did. The forest plots for these analyses are presented in Fig. S2.

Bias and quality analysis

All non-randomised studies selected for inclusion were screened for risk of bias with the ROBINS-I tool, Fig. 2. The two randomised control trials were assessed using the RoB 2 tool and found to be at low risk of bias. Overall, the included studies showed a moderate to high risk of bias, largely to due to the presence of confounding bias. Visual assessment of the funnel plots for each meta-analysis demonstrated no evidence of significant reporting bias.

Fig. 2.

Risk of bias analysis. Diagrammatic representation of the risk of bias assessed using the ROBINS-I 2 tool from the Cochrane group, presented using the robvis tool

Obesity

Obesity was defined as a BMI greater than 30 or visceral fat area (VFA) > 100 cm² in all studies in the meta-analyses. Ten case-matched studies reported on robotic-assisted versus laparoscopic surgery in patients with obesity and were included in the meta-analyses [24, 27, 31, 37, 42, 44, 48–50, 55]. These studies included a total of 4371 robotic-assisted cases and 11,085 laparoscopic. Forest plots of select meta-analyses are presented in Fig. 3. All meta-analyses not presented in Fig. 3 can be found in Fig. S3.

Fig. 3.

Forest plots. Forest plots representing the comparative meta-analyses of select outcomes. M-H Mantel–Haenszel, CI confidence interval, SD standard deviation

A meta-analysis including the seven studies which reported operative time found significantly longer operations using robotic-assisted surgery (SMD 0.57 [CI 0.31–0.83], p < 0.0001). Nine studies reported conversion-to-open rates, and meta-analysis demonstrated a significantly lower conversion rate in robotic-assisted surgery (OR 0.41 [CI 0.32–0.51], p < 0.00001). Six studies reported on length of stay, and meta-analysis suggested a reduced length of stay in patients undergoing robotic-assisted surgery (SMD 0.25 [CI − 0.41 to − 0.09], p = 0.002). Blood loss (SMD 0.23 [CI − 0.51 to 0.05], p = 0.11), overall complication rate (OR 0.96 [CI 0.81–1.14], p = 0.68), and lymph node yield (SMD 0.11 [CI − 0.07 to 0.28], p = 0.24) were comparable between groups.

A pooled weighted mean (PWM) of conversion to open rate was 7.6% [CI 5.2–10.1%] and was calculated utilising 21 studies reporting outcomes of robotic-assisted surgery in patients with obesity [6, 24, 27, 31, 37, 42, 44, 48–50, 56–66]. Table 2 presents all calculable PWMs for each outcome in each subgroup.

Table 2.

Pooled weighted means

	Mean (%)	95% CI (%)		Mean (mins)	95% CI (mins)
Conversion to open rate			Operative time
Obesity	7.6	5.2–10.1	Obesity	289	251–327
Male gender	9.1	5.4–12.7	Male gender	271	117–425
Elderly	5.7	1.3–10.2	Elderly	264	56–462
Low rectal tumours	NC	NC	Low rectal tumours	318	223–414
Neoadjuvant chemoradiotherapy	5.6	0.7–11.8	Neoadjuvant chemoradiotherapy	348	256–441
Previous abdominal surgery	0.5	0–1.3	Previous abdominal surgery	270	224–317

	Mean (%)	95% CI (%)		Mean (days)	95% CI (days)
Perioperative complication rate			Length of stay
Obesity	28	21.7–34.3	Obesity	6.5	4.2–8.8
Male gender	19	5.4–32.7	Male gender	5.2	2.6–7.8
Elderly	38.3	24.2–52.5	Elderly	9.8	5.4–14.2
Low rectal tumours	22.0	14.3–29.7	Low rectal tumours	11.2	7.7–14.8
Neoadjuvant chemoradiotherapy	26.8	17.1–36.4	Neoadjuvant chemoradiotherapy	10.3	6.7–13.9
Previous abdominal surgery	15.1	0–32.7	Previous abdominal surgery	8.2	3.6–12.9

CI confidence interval, NC not calculable

On sensitivity analysis, looking at rectal resections alone, all outcomes remain the same.

Male gender

Four studies examined male gender as a high-risk factor and included a total of 934 robotic-assisted cases and 3762 laparoscopic cases [24, 29, 42, 43]. All but one of the studies looked at resections for rectal cancer alone, with the fourth study including all types of colectomies. There were no significant differences between the two groups with regards to complication rate (OR 1.09 [CI 0.91–1.30], p = 0.36) or length of stay (SMD 0.19 [CI − 0.51–0.13], p = 0.25). Operative time was again significantly longer in the robotic-assisted group (SMD 0.77 [CI 0.17–1.37], p = 0.01). However, conversion to open was significantly lower in the robotic-assisted group (OR 0.28 [CI 0.22–0.34], p < 0.00001), Fig. 3.

PWM conversion to open rate from five studies of robotic-assisted surgery in male patients [6, 24, 42, 43, 60] was 9.1% [CI 5.4–12.7%].

Sensitivity analysis of rectal resections alone did not alter the primary outcome but did alter operative time to be equivalent between the two groups (SMD 0.75 [CI − 0.70 to 2.19] p = 0.31) and length of stay then favoured RAS (SMD − 0.40 [CI − 0.75 to 0.06] p = 0.02).

Elderly

Six studies were eligible for inclusion in the comparative meta-analyses. The authors reported outcomes of robotic-assisted versus laparoscopic surgery in elderly patients, including a total of 722 robotic-assisted cases and 7422 laparoscopic [28, 30, 36, 45, 51, 54]. Definitions of elderly varied between studies, but all patients had a pooled mean age of 84.3 ± 2.3 years.

Operative time was significantly longer in the robotic-assisted group (SMD 0.50 [CI 0.18–0.83], p = 0.002). There was no difference between the two groups with regards to conversion to open (OR 0.28 [CI 0.07–1.11], p = 0.07), complication rate (OR 1.02 [CI 0.59–1.78], p = 0.94), length of stay (SMD 3.20 [CI − 10.45 to 4.06], p = 0.39), estimated blood loss (SMD 0.04 [CI − 0.29 to 0.22], p = 0.77) or lymph node yield (SMD 0.34 [CI − 0.78 to 0.10], p = 0.13).

PWM conversion to open rate from six studies of elderly patients undergoing robotic-assisted colorectal resections [28, 30, 36, 51, 67, 68] was 5.7% [CI 1.3–10.2%].

On sensitivity analysis of rectal resections alone, all outcomes remain the same with the exception of lymph node yield which was greater in laparoscopic surgery (OR 0.55 [CI − 0.94 to − 0.16], p = 0.006).

Low rectal tumour

Six studies looked at low rectal tumours, including 382 robotic-assisted cases and 386 laparoscopic [25, 34, 38, 40, 46, 52]. The pooled mean distance of the tumour from the anal verge was 4 ± 1.5 cm. Of the two studies that reported on conversion to open rates, these were found to be significantly lower in the robotic-assisted group (OR 0.10 [CI 0.02–0.58], p = 0.01).

Operative time was significantly longer in the robotic-assisted group (SMD 0.48 [CI 0.12–0.84], p = 0.008), Fig. 3. There was no difference between the two groups with regards to complication rate (OR 0.94 [CI 0.58–1.52], p = 0.81), estimated blood loss (SMD − 0.37 [CI − 0.89 to 0.15], p = 0.16), length of stay (SMD − 0.34 [CI − 0.87 to 0.18], p = 0.20) or lymph node yield SMD − 0.07 [CI − 0.34 to 0.19], p = 0.59).

When a robotic-assisted technique was used, only one patient from a pooled total of 338 from five studies required conversion to open for a low rectal cancer [25, 34, 38, 40, 52].

Neoadjuvant radiotherapy

Seven studies examined patients who had received long-course neoadjuvant chemoradiotherapy for low rectal cancer, with a total number of 305 robotic-assisted cases and 422 laparoscopic [26, 32, 33, 41, 43, 47, 53], Fig. 3. Operative time was significantly longer in the robotic-assisted group (SMD 0.72 [CI 0.34–1.09], p = 0.0002). There was no difference between the two groups with regards to conversion to open (OR 0.47 [CI 0.09–2.43], p = 0.37), complication rate (OR 1.10 [CI 0.66–1.81], p = 0.72), estimated blood loss (SMD 0.17 [CI − 0.74 to 0.40], p = 0.56), length of stay (SMD 0.07 [CI − 0.22 to 0.08], p = 0.36) or lymph node yield (SMD 0.18 [CI − 0.28 to 0.64], p = 0.44).

After neoadjuvant radiotherapy, PWM conversion to open rate from six studies [6, 33, 41, 43, 47, 53] was 5.6% [CI 0.7–11.8%].

All studies reported on rectal cancers only and, therefore, sensitivity analyses were not completed.

Previous abdominal surgery

Two studies looked at patients who had previous abdominal surgery, with a total number of 274 robotic-assisted cases and 674 laparoscopic [35, 39]. There was no significant difference between the two groups with regards to conversion to open (OR 0.42 [CI 0.08–2.21], p = 0.30), complication rate (OR 0.89 [CI 0.35–2.27], p = 0.80) or length of stay (SMD 0.06 [CI − 0.14 to 0.27], p = 0.54). Again, operative time was significantly longer for RAS (SMD 1.55 [CI 0.05–3.06], p = 0.04), Fig. 3.

The same two studies were the only ones to report conversion to open rates after previous surgery and PWM was 0.5% [CI 0–1.3%].

Both studies included colonic and rectal resections and so sensitivity analysis was not possible. Furthermore, both studies included a wide range of previous surgeries, including some minimally invasive operations, and only one utilised an adhesion severity scoring system, limiting the robustness of primary data and meta-analysis.

Discussion

This review has demonstrated significantly lower conversion to open rates in male patients, patients with obesity and patients with low rectal tumours undergoing robotic-assisted surgery for colorectal cancer when compared to laparoscopic surgery. It has also demonstrated shorter length of stay in patients with obesity undergoing robotic-assisted surgery, and male patients when looking at rectal resections alone.

As the prevalence of RAS has increased, so has the evidence demonstrating equivalent outcomes for robotic-assisted and laparoscopic colorectal surgery. The exception to this has been conversion rates. Observational studies and meta-analyses of RAS have consistently shown superiority in conversion to open rates [71–73]. However, before this review, obesity was the only high-risk subgroup explored with level 1 or 2 evidence, suggesting lower conversion to open rates as well as decreased length of stay [61, 74]. This review confirms these findings. There is strong evidence depicting the relationship between obesity and colorectal cancer [75–77] and, as the prevalence of obesity continues to increase worldwide [78], so will the proportion of these patients in a colorectal surgeon’s workload.

Furthermore, the results demonstrate the advantage of robotic-assisted surgery in male patients and patients with low rectal tumours. Conversion rates are significantly lower with RAS, confirming in practice the benefits of improved instrument articulation in narrow spaces. When just rectal resections were examined, length of stay was shorter for RAS in male patients.

The pooled weighted means show that male patients and patients with obesity have the highest conversion to open rates among all the subgroups analysed. Conversion to open surgery in any population is associated with increased morbidity and mortality in addition to adverse long-term oncological outcomes [79, 80]. Suwa et al. [17] found higher conversion rates and significantly longer operative times in patients with obesity undergoing robotic colorectal cancer resections compared to patients without obesity. Similarly, Baukloh et al. [60] conducted a multicentre prospective study in 348 patients undergoing robotic surgery due to rectal cancer and found male patients had significantly higher rates of major complications and anastomotic leakage. Therefore, RAS should be considered the modality of choice for male patients and/or patients with obesity, especially those with low rectal tumours, to decrease conversion rates and improve outcomes in these high-risk subgroups.

In elderly patients, this review finds no evidence of superior outcomes for RAS when compared to laparoscopic surgery but does find significantly longer operating times. Therefore, it is questionable whether elderly patients not belonging to any other high-risk subgroup will benefit from RAS. This is even more important considering this subgroup is at greater risk from lengthier surgery [81, 82].

Longer surgery for any patient is an independent risk factor for postoperative complications and increased length of stay [83, 84]. This finding is likely due to the setup and docking required for robotic-assisted surgery and the established learning curve for surgeons and theatre staff. However, once RAS becomes more established in individual centres, operative time decreases and is on par with a laparoscopic approach. Furthermore, evidence is emerging to suggest the learning curve for RAS may be faster than the learning curve for laparoscopy, and implementation of robotic training programmes can be associated with reduced operating times and fewer complications [85, 86].

The limitations of this meta-analysis lie largely in the primary evidence. Out of the 32 studies included in the comparative meta-analyses, all but one are non-randomised observational studies with significant bias due to confounding. The majority of sample sizes were small. Furthermore, the ‘previous abdominal surgery’ subgroup analyses are at higher risk of error given the inherent biases in the primary studies’ methodologies. The studies included span over a decade, from 2010 to 2024. Within that time, robotic-assisted surgery has progressed significantly, drawing into question the comparability of older and newer studies. Finally, the source studies are at an overall moderate to high risk of bias which can have an effect on the reliability of data syntheses.

Conclusions

This systematic review and meta-analysis demonstrates equivalent or superior outcomes of RAS when compared to laparoscopic surgery in high-risk colorectal patients. Conversion to open rates are significantly lower for robotic-assisted surgery in male patients, patients with obesity and patients with low rectal tumours, when compared to laparoscopy. Length of stay is also significantly reduced in patients with obesity undergoing all resection types, and in male patients undergoing rectal resections alone. These findings suggest robotic-assisted surgery should be considered as the preferred modality in these cohorts. However, operative times remain significantly longer in robotic-assisted surgery, which must be weighed against the cost implications and learning curve.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

OR

Odds ratio

PWM

Pooled weighted means

RAS

Robotic-assisted surgery

SD

Standard deviation

SMD

Standardised mean difference

TME

Total mesenteric excision

Author contributions

SG was responsible for initial searches, study screening, data extraction and quality analysis. SGP and SM were responsible for data extraction. JW and KA were responsible for data analysis. SA and KA conceived the review. JW and SG completed the first draft of the manuscript and all co-authors critically reviewed, revised the manuscript and provided final approval.

Funding

No funding was sought or required for the preparation of this manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Conflict on interest

Drs. Sukhpreet Gahunia, James Wyatt, Simon G. Powell, Shareef Mahdi, Shakil Ahmed and Kiran Altaf have no competing interests or financial ties to disclose.