Surgical outcomes after robot-assisted versus laparoscopic pancreatoduodenectomy: multicentre propensity-matched comparison from the Italian Group of Minimally Invasive Pancreatic Surgery Authors: Giuseppe Quero, Roberta Menghi, Niccolò Napoli, Allegra Ripolli, Giovanni Ferrari, Michele Mazzola, Elisa Bannone, Giovanni Butturini, Massimo Giuseppe Viola, Marco Garatti, Giovanni Tebala, Fabrizio Di Benedetto, Luca Moraldi, Carlo Molino, Sergio Alfieri, Umberto Bracale, Roberto Troisi, Raffaele Dalla Valle, Marco Vivarelli, Elio Jovine, Riccardo Memeo, Gianluca Garulli, Luca Ottaviani, Stefano Berti, Andrea Coratti, Giorgio Ercolani, Roberto Salvia, Luca Morelli, Riccardo Casadei, Roberto Coppola, Alberto Brolese, Vicenzo Tondolo, Alessandro Zerbi, Ugo Boggi, the IGOMIPS Registry Study Group Published: 2026 ABSTRACT AbstractBackgroundMinimally invasive approaches to pancreatoduodenectomy (PD) are increasingly being adopted worldwide. However, direct comparisons between laparoscopic PD (LPD) and robot-assisted PD (RPD) remain limited and inconclusive. This study aimed to compare perioperative outcomes between LPD and RPD within a national multicentre cohort.MethodsThis retrospective multicentre study included patients who underwent LPD or RPD at participating Italian centres within the Italian Group of Minimally Invasive Pancreas Surgery registry between October 2019 and December 2024. Propensity score matching was performed in a 1 : 1 ratio based on age, body mass index, American Society of Anaesthesiologists score, tumour type (pancreatic ductal adenocarcinoma versus other histologies), receipt of neoadjuvant chemotherapy, and requirement for vascular and/or organ resection. Primary outcomes were estimated blood loss and operative time. Secondary outcomes included conversion rate, postoperative morbidity, pancreas-related complications, reoperation, length of hospital stay, histopathological features, and deaths.ResultsAmong 774 minimally invasive PDs, 163 (21.1%) were LPD and 611 (78.9%) were RPD. After propensity score matching , 161 LPDs were compared with 161 RPDs. RPD was associated with a longer operative time (median 506 (interquartile range (i.q.r.) 427.5–600) versus 500 (411.5–551) minutes; P = 0.030) but lower estimated blood loss (median 200 (i.q.r. 100–337.5) ml versus 250 (150–500) ml; P < 0.001). On multivariable analysis, surgical approach was not an independent predictor of operative time, whereas RPD was independently associated with lower estimated blood loss (β −0.372, 95% confidence interval −0.641 to −0.102; P = 0.007). Conversion to open surgery was significantly more frequent after LPD (24 of 161; 14.9%) than RPD (11 of 161; 6.8%) (P = 0.023). Postoperative outcomes—including major morbidity, pancreas-specific complications, reoperation, length of hospital stay, and in-hospital or 90-day deaths—were comparable between groups. Histopathological findings were also similar.ConclusionRPD was associated with lower estimated blood loss and a reduced conversion rate compared with LPD. However, these differences were modest in absolute terms and did not translate into clinically meaningful differences in postoperative outcomes, which remained comparable between the two approaches.Keywords: intraoperative outcomes, postoperative outcomes, pancreas-related complications, minimally invasive surgery, pancreatic surgery Introduction Pancreatoduodenectomy (PD) is widely recognized as one of the most complex abdominal surgical procedures and remains associated with substantial postoperative morbidity and death1,2. Historically, PD has been performed through an open approach. However, the adoption of minimally invasive surgery has improved perioperative outcomes across several surgical procedures, including reduced blood loss, shorter length of hospital stay (LOS), and lower complication rates3. These advantages have stimulated growing interest in extending laparoscopic and robotic techniques to technically demanding procedures such as PD. The first laparoscopic PD (LPD) was performed by Gagner and Pomp4 in 1994. Since then, numerous studies5–10 have evaluated the safety and feasibility of LPD compared with open PD, although with inconsistent results. Whereas several retrospective series and three randomized clinical trials suggested favourable outcomes with LPD5–7,9,10, the Dutch multicentre LEOPARD-2 trial8 was prematurely terminated because of significantly higher postoperative complication rates in the laparoscopic arm. These conflicting findings may partly reflect the inherent technical limitations of conventional laparoscopy, including restricted degrees of freedom, the fulcrum effect, and limited depth perception associated with two-dimensional vision11. The robotic platform was developed to overcome these limitations while preserving the advantages of minimally invasive surgery by providing three-dimensional visualization, enhanced dexterity with seven degrees of freedom, tremor filtration, and improved ergonomics11. Robotic PD (RPD), first performed in 2001 and reported by Giulianotti et al.12 in 2003, has been increasingly adopted, supported by encouraging evidence from single- and multicentre studies. Nevertheless, direct comparisons between LPD and RPD remain limited. Most available comparative studies are retrospective, single-centre, and based on relatively small cohorts, resulting in inconclusive evidence13–17. Some authors have reported improved perioperative outcomes after RPD, including lower blood loss, fewer postoperative complications, shorter LOS, and lower readmission rates14. More recently, a nationwide-matched analysis from Japan18 further supported these observations by demonstrating improved short-term outcomes with RPD, including reduced blood loss and lower conversion rates. Conversely, other studies have reported higher rates of postoperative pancreatic fistula (POPF), surgical site infection, and delayed gastric emptying (DGE) after RPD than after LPD13. Given the evolving role of minimally invasive surgery in PD, a direct comparison between LPD and RPD is warranted to define better the relative contribution of the surgical platform to perioperative outcomes. The Italian Group of Minimally Invasive Pancreas Surgery (IGoMIPS), established in 2019, was created to collect nationwide data prospectively from minimally invasive pancreatic centres and monitor related outcomes. Within this framework, the present multicentre observational cohort study aimed to compare intraoperative outcomes between LPD and RPD using data from the IGoMIPS registry. To minimize potential selection bias, propensity score matching (PSM) was performed using covariables expected to influence perioperative outcomes. Methods Study design and data collection The IGoMIPS registry is a prospective registry approved by the Ethics Committee (Humanitas Research Hospital; Prot. No. Ce Humanitas ex D.M. 8/2/2013–213/19)19 and registered (registration number: NCT06381362 (http://www.clinicaltrials.gov)). The present study was conducted and reported in accordance with the STROBE guidelines for observational studies. All patients who underwent elective minimally invasive PD (LPD or RPD) for any indication were retrospectively retrieved from the IGoMIPS registry and included in the present analysis. The registry prospectively includes all minimally invasive pancreatic procedures performed across 42 Italian pancreatic referral centres from October 2019 onwards, with postoperative follow-up up to 90 days after surgery. Hybrid or hand-assisted procedures were excluded to ensure procedural consistency. Surgical approaches (laparoscopic, robotic, and hybrid) were defined according to the Brescia guidelines for minimally invasive pancreatic surgery20. Patients scheduled for PD who ultimately underwent a different surgical procedure were excluded. To ensure the methodological robustness of the PSM analysis, only patients with complete data for the selected matching covariables were included. Patients with missing data for the primary or predefined secondary outcomes, or for clinically relevant perioperative variables required for adjusted analyses, were also excluded. Baseline characteristics included age, sex, body mass index (BMI, kg/m2), American Society of Anesthesiologists (ASA) score, history of abdominal surgery, neoadjuvant therapy (NAT), and lesion type. Surgical procedures were performed according to the standard techniques adopted at each participating centre. Operative variables included operative time, estimated blood loss (EBL), associated vascular and/or organ resection, pancreatic texture (soft versus firm), main pancreatic duct diameter (≤ 3 mm versus > 3 mm), type of PD (pylorus-preserving or classical Whipple), intraoperative pancreatic stent placement, and type of reconstruction (pancreatojejunostomy or pancreatogastrostomy). Postoperative management followed institutional protocols. Postoperative variables included complications graded according to the Clavien–Dindo classification, POPF, post-pancreatectomy haemorrhage (PPH), DGE, bile leak, gastrojejunostomy leak, chyle leak, intra-abdominal collection, surgical site infection, sepsis, reoperation, LOS, readmission, in-hospital deaths, and 90-day deaths. POPF, PPH, and DGE were defined and graded according to the International Study Group of Pancreatic Surgery1,21,22. Oncological outcomes included histopathological diagnosis, tumour diameter, number of harvested and metastatic lymph nodes, resection margin status (R) according to the Royal College of Pathologists23, and tumour stage according to the American Joint Committee on Cancer/Union for International Cancer Control 8th edition24. Study outcomes The primary outcome was the comparison of intraoperative outcomes—EBL and operative time—between RPD and LPD. These endpoints were selected to reflect intraoperative performance and to allow a standardized comparison between surgical approaches across multiple centres. Secondary outcomes included conversion to open surgery, postoperative morbidity, pancreas-related complications, postoperative transfusion, reoperation, LOS, histopathological features, and deaths. To account for centre-level clustering and temporal trends related to the progressive adoption of RPD, sensitivity analyses were performed for the primary outcome (EBL) and key secondary outcomes (conversion to open surgery), with additional exploratory analyses for failure-to-rescue, using mixed-effects regression models including centre as a random intercept, and calendar era (2019–2020, 2021–2022, and 2023–2024) as a fixed-effect covariable. Statistical analysis To reduce selection bias between the RPD and LPD cohorts, PSM was performed in a 1 : 1 ratio. Propensity scores were estimated using a multivariable logistic regression model including age, BMI, ASA score, tumour histology (pancreatic ductal adenocarcinoma versus other histologies), NAT, and associated vascular and/or organ resection as covariables, selected based on their potential influence on the primary outcomes. Patients were matched using nearest-neighbour matching without replacement and a caliper width of 0.2 standard deviations of the logit of the propensity score. Balance between groups was assessed using standardized mean differences (MDs), with values < 0.1 considered indicative of adequate balance. Continuous variables are reported as median (interquartile range (i.q.r.)), whereas categorical variables are reported as number and percentages. Univariable analysis was performed using Student’s t tests or Mann–Whitney U tests for continuous variables, and Fisher’s tests or χ2 tests for categorical variables, as appropriate. For all analyses, a two-sided P value < 0.05 was considered statistically significant. Univariable and multivariable analyses were performed to explore variables associated with operative time and EBL. Both outcomes were analysed as continuous variables. Operative time was analysed using linear regression, whereas EBL was log-transformed to account for its right-skewed distribution and improve model fit. Given the exploratory aim of the analysis, both baseline and intraoperative variables were included. Results are reported as unstandardized regression coefficients (β) with corresponding 95% confidence intervals (c.i.) and P values. A two-sided P value < 0.05 was considered statistically significant. All analyses were performed using IBM SPSS Statistics® version 25 (IBM Corp., Armonk, NY, United States). To explore the potential impact of missing data, baseline characteristics of included and excluded patients were compared in the overall study population. Sensitivity analyses were performed using mixed-effects regression models including centre as a random intercept and calendar era as a fixed-effect covariable. A linear mixed-effects model was applied for the continuous outcome of EBL, whereas mixed-effects logistic regression models were used for the binary outcomes of conversion to open surgery and failure-to-rescue after major postoperative complications. Results Before matching Clinico-demographic characteristics Between October 2019 and December 2024, a total of 1002 planned minimally invasive PDs were recorded in the IGoMIPS registry. After applying the predefined inclusion and exclusion criteria, 774 patients (77.2%) were included in the final analysis. Among these, 163 (21.1%) underwent LPD and 611 (78.9%) underwent RPD (Fig. 1). To explore the potential impact of selection bias related to missing data, baseline characteristics of included and excluded patients were compared (Table S1). Baseline characteristics were comparable across all evaluated variables. Over the study period, a clear temporal shift in surgical approach was observed, with a progressive decline in the laparoscopic approach and a concomitant increase in the robotic approach, which became predominant in the most recent semesters (Fig. 2). Before PSM, patients in the LPD cohort were significantly older than those in the RPD group (median 71 (i.q.r. 62–77) versus 68 (59–75) years; P = 0.005). All other clinico-demographic characteristics, including BMI, ASA score, previous abdominal surgery, and receipt of NAT, were comparable between the two groups (Table 1). Patient selection flowchart After applying predefined inclusion and exclusion criteria to 1002 planned minimally invasive PDs in the Italian Group of Minimally Invasive Pancreas Surgery registry, 774 patients were included in the final cohort. Distribution of LPD and RPD before and after PSM is reported. PD, pancreatoduodenectomy; LPD, laparoscopic PD; RPD, robotic PD, PSM, propensity score matching, BMI, body mass index, ASA, American Society of Anesthesiologists; EBL, estimated blood loss; POPF, postoperative pancreatic fistula; DGE, delayed gastric emptying. Line graph depicting the temporal distribution of laparoscopic and robotic pancreatoduodenectomy across consecutive semesters S1: January–June; S2: July–December, from October 2019 to December 2024. Table 1Clinico-demographic characteristics of the unmatched and matched cohortLPDRPD P Unmatched cohort (n = 774) (n = 163)(n = 611) Sex Male95 (58.3%)333 (54.5%)0.383 Female68 (41.7%)278 (45.5%) Age (years), median (i.q.r.)71 (62–77)68 (59–75)0.005 Body mass index, median (i.q.r.)24.5 (22.3–27.2)24.2 (22.1–26.7)0.370 ASA score 1–2100 (61.3%)363 (59.4%)0.090 3–463 (38.7%)248 (40.6%) Previous abdominal surgery68 (42.0%)210 (34.4%)0.072 NAT12 (7.4%)39 (6.4%)0.653 Lesion type PDAC67 (41.1%)259 (42.4%)0.764 Other96 (58.9%)352 (57.6%) Matched cohort (n = 322) (n = 161)(n = 161) Sex Male94 (58.4%)93 (57.8%)0.910 Female67 (41.6%)68 (42.2%) Age (years), median (i.q.r.)71 (62.0–76.5%)69 (60.0–75%)0.143 Body mass index, median (i.q.r.)24.5 (22.3–27.2)23.7 (21.6–26.8)0.192 ASA score 1–299 (61.5%)99 (61.5%)0.180 3–462 (38.5%)62 (38.5%) Previous abdominal surgery68 (42.5%)63 (39.1%)0.543 NAT12 (7.5%)9 (5.6%)0.502 Lesion type PDAC66 (41.0%)66 (41.0%)1 Other95 (59.0%)95 (59.0%)Values are n (%) unless otherwise stated. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; i.q.r., interquartile range; ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma. Clinico-demographic characteristics of the unmatched and matched cohort Values are n (%) unless otherwise stated. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; i.q.r., interquartile range; ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma. Intraoperative outcomes (Table 2) Table 2Intraoperative outcomes for LPD and RPD before and after matchingLPDRPD P Unmatched cohort (n = 774) (n = 163)(n = 611) Operative time (min), median (i.q.r.)500 (407–552)495 (420–580)0.100 Pancreatic texture Firm98 (60.1%)407 (66.6%)0.121 Soft65 (39.9%)204 (33.4%) Pancreatic duct diameter (mm), median (i.q.r.)3 (2–5)4 (2–5)0.070 Vascular resection10 (6.1%)37 (6.1%)0.971 Associated organ resection5 (3.1%)14 (2.3%)0.560 EBL (ml), median (i.q.r.)250 (150–500)200 (100–300)< 0.001 Pancreatic stent91 (56.5%)428 (72.5%)< 0.001 Whipple procedure140 (86.4%)164 (26.8%)< 0.001 Pancreatic anastomosis P–J155 (95.1%)577 (91.1%)0.130 P–G7 (4.3%)16 (2.5%) Other017 (2.8%) Conversion24 (14.7%)38 (6.2%)< 0.001 Matched cohort (n = 322) (n = 161)(n = 161) Operative time (min), median (i.q.r.)500 (411.5–551)506 (427.5–600)0.030 Pancreatic texture Firm97 (60.2%)87 (54.0%)0.262 Soft64 (39.8%)74 (46.0%) Pancreatic duct diameter (mm), median (i.q.r.)3 (2–5)4 (2–5)0.221 Vascular resection10 (6.2%)9 (5.6%)0.812 Associated organ resection5 (3.1%)5 (3.1%)1 EBL (ml), median (i.q.r.)250 (150–500)200 (100–337.5)< 0.001 Pancreatic stent90 (56.6%)112 (73.2%)0.002 Whipple procedure138 (86.3%)49 (30.6%)< 0.001 Pancreatic anastomosis P–J154 (95.7%)152 (94.4%)0.021 P–G7 (4.3%)3 (1.9%) Other06 (3.7%) Conversion24 (14.9%)11 (6.8%)0.023Values are n (%) unless otherwise stated. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; min, minutes; i.q.r., interquartile range; EBL, estimated blood loss. Intraoperative outcomes for LPD and RPD before and after matching Values are n (%) unless otherwise stated. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; min, minutes; i.q.r., interquartile range; EBL, estimated blood loss. RPD was associated with lower EBL (median 200 (i.q.r. 100–300) ml versus 250 (150–500) ml; P < 0.001), a higher rate of pylorus-preserving procedures (73.2% (447) versus 14.1% (23); P < 0.001) and more frequent pancreatic stent placement (428 (72.5%) versus 91 (56.5%); P < 0.001). No significant differences were observed in operative time (median 500 (i.q.r. 407–552) and 495 (420–580) minutes (min) for LPD and RPD, respectively; P = 0.100) or type of pancreatic anastomosis, with pancreatojejunostomy representing the most common reconstruction in both cohorts (155 (95.1%) and 577 (91.1%) in the LPD and RPD groups, respectively; P = 0.130). Similarly, an equivalent number of vascular resections were performed in both cohorts (10 (6.1%) versus 37 (6.1%); P = 0.971). Conversely, conversion to open surgery was significantly more frequent in the LPD group (24 (14.7%)) compared with the RPD group (38 (6.2%)) (P < 0.001). Postoperative outcomes Rates of major complications and pancreas-related complications, including POPF, DGE, and PPH, were similar between the two cohorts. However, postoperative fluid collections were more frequent after RPD than LPD (191 (31.3%) versus 36 (22.1%); P = 0.021). Overall, 104 patients (13.4%) required reoperation, with no difference between LPD and RPD cohorts (22 (13.5%) and 82 (13.5%), respectively; P = 0.994). Likewise, the choice of minimally invasive versus open approach for reoperation did not differ (P = 0.903). Postoperative transfusion rates were comparable between groups (P = 0.591). Comparable rates of in-hospital (3.7% (6) and 3.4% (21) for LPD and RPD, respectively; P = 0.883) and 90-day deaths (6.7% (11) and 5.1% (31) for LPD and RPD, respectively; P = 0.404) were documented in both groups. Oncological outcomes, including tumour diameter, number of harvested lymph nodes, resection margin status, and pathological stage, did not differ between the two approaches. Detailed postoperative and histopathological outcomes are reported in Table 3. Table 3Postoperative and histopathological outcomes for LPD and RPD before and after matchingUnmatched cohort(n = 774)Matched cohort(n = 322)LPD (n = 163)RPD (n = 611) P LPD (n = 161)RPD (n = 161) P Clavien–Dindo I–II54 (52.9%)221 (58.6%)0.30152 (52.0%)64 (65.3%)0.560 III–IV48 (47.1%)156 (41.4%)48 (48.0%)34 (34.7%)DGE32 (19.8%)123 (20.2%)0.90132 (20.0%)24 (14.9%)0.231 DGE grade A14 (8.6%)30 (4.9%)0.23114 (43.8%)5 (20.8%)0.202 B14 (8.6%)61 (10.0%)14 (43.8%)15 (62.5%) C4 (2.5%)31 (5.1%)4 (12.5%)4 (16.7%)POPF28 (17.2%)138 (21.8%)0.20228 (17.4%)35 (21.7%)0.322 POPF grade B24 (42.1%)106 (46.3%)0.19024 (42.1%)27 (45.0%)0.421 C4 (7.0%)32 (14.0%)4 (7.0%)8 (13.3%)PPH29 (17.8%)103 (16.9%)0.77128 (17.4%)18 (11.2%)0.110 PPH grade A6 (3.7%)16 (2.6%)0.0705 (17.9%)4 (22.2%)0.894 B11 (6.7%)39 (6.4%)11 (39.3%)6 (33.3%) C12 (7.4%)48 (7.9%)12 (42.9%)8 (44.4%)Fluid collection36 (22.1%)191 (31.3%)0.02136 (22.4%)48 (29.8)0.125Bile leak9 (5.5%)50 (8.2%)0.2559 (5.6%)9 (5.6)1.000Chyle leak7 (4.3%)25 (4.1%)0.9117 (4.3%)6 (3.7)0.782Sepsis20 (12.3%)108 (17.7%)0.09020 (12.4%)24 (14.9)0.510Postoperative transfusion49 (30.1%)197 (32.2%)0.59149 (30.4%)43 (26.7%)0.463Reoperation22 (13.5%)82 (13.5%)0.99422 (13.7%)16 (10.0%)0.311 Reoperation approach Minimally invasive9 (5.5%)28 (4.6%)0.9039 (40.9%)1 (6.2%)0.040 Open13 (8.0%)54 (8.5%)13 (59.1%)15 (93.8%)Gastrojejunostomy leakage5 (3.1%)16 (2.6%)0.7505 (3.1%)2 (1.2%)0.251Surgical site infection10 (6.1%)25 (4.1%)0.27110 (6.2%)4 (2.5%)0.100Length of hospital stay (days), median (i.q.r.)12 (8–22)14 (9–27)0.07011 (8–21)15 (8–25.5)0.060Readmission26 (16.0%)93 (15.5%)0.88126 (16.1%)20 (12.6%)0.363Discharge with abdominal drainage11 (6.7%)67 (11.0%)0.36311 (6.8%)18 (11.2%)0.394Postoperative deaths6 (3.7%)21 (3.4%)0.8836 (3.7%)3 (1.9%)0.31090-day deaths11 (6.7%)31 (5.1%)0.40411 (6.8%)6 (3.7%)0.212 Histopathological features Tumour diameter (mm), median (i.q.r.)25 (17.5–32)25 (18–32)0.45124.5 (17.2–32)25 (18–33.5)0.430 Harvested lymph nodes22 (16.5–30)24 (17–34)0.11021.5 (16.2–30)25 (17–34)0.285 Positive lymph nodes0 (0–2)1 (0–3.2)0.0300 (0–2)1 (0–4)0.031 Positive resection margin21 (14.5%)83 (15.7%)0.71121 (14.6%)29 (20.3%)0.201 T category‡ 118 (26.9%)54 (20.8%)0.16218 (27.3%)14 (21.2%)0.492 244 (65.6%)166 (64.1%)43 (65.1%)42 (63.7%) 34 (6.0%)38 (14.7%)4 (6.1%)9 (13.6%) 41 (1.5%)1 (0.4%)1 (1.5%)1 (1.5%) N category‡ 024 (35.8%)70 (27.0%)0.25124 (36.4%)12 (18.2%)0.060 124 (35.8%)104 (40.2%)23 (34.8%)30 (45.4%) 219 (28.4%)85 (32.8%)19 (28.8%)24 (36.4%)Values are n (%) unless otherwise stated. ‡Tumour staging was evaluated only for pancreatic ductal adenocarcinoma lesions. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; DGE, delayed gastric emptying; POPF, postoperative pancreatic fistula; PPH, post-pancreatectomy haemorrhage; i.q.r., interquartile range. Postoperative and histopathological outcomes for LPD and RPD before and after matching Values are n (%) unless otherwise stated. ‡Tumour staging was evaluated only for pancreatic ductal adenocarcinoma lesions. LPD, laparoscopic pancreatoduodenectomy; RPD, robotic pancreatoduodenectomy; DGE, delayed gastric emptying; POPF, postoperative pancreatic fistula; PPH, post-pancreatectomy haemorrhage; i.q.r., interquartile range. PSM PSM resulted in 161 matched pairs. After matching, covariable balance improved substantially, with all MDs < 0.1, indicating adequate balance between groups. Detailed balance diagnostics are reported in Table S2 and Fig. S1. After matching After PSM, 161 patients who underwent RPD were compared with 161 who underwent LPD. Clinico-demographic characteristics were comparable between the two approaches (Table 1). Intraoperative outcomes RPD was associated with a longer operative time (median 506 (i.q.r. 427.5–600) minutes) than LPD (500 (411.5–551) minutes) (P = 0.030). Conversely, EBL was higher in the LPD cohort (median 250 (i.q.r. 150–500) ml) compared with the RPD cohort (200 (100–337.5) ml) (P < 0.001). No differences were observed in the rate of vascular resection (6.2% (10) for LPD and 5.6% (9) for RPD; P = 0.812) (Table 2). Pylorus-preserving PD was performed more frequently in the RPD group (69.5% (112)), whereas LPD patients more commonly underwent a classical Whipple procedure (138 (86.3%)) (P < 0.001). Although pancreatojejunostomy was the predominant reconstruction in both groups (154 (95.7%) and 152 (94.4%) in the LPD and RPD cohorts, respectively), pancreatogastrostomy was more frequently performed after LPD (7 (4.3%) versus 3 (1.9%); P = 0.021). Likewise, pancreatic stent placement was more common in the RPD group (112 (69.5%) versus 90 (56.6%); P = 0.002). Conversion to open surgery remained significantly more frequent in the LPD cohort (24 (4.9%)) compared with RPD (11 (6.8%); P = 0.023). Postoperative outcomes Postoperative outcomes were comparable between the approaches (Table 3). No differences were observed in the rates of Clavien–Dindo grade ≥ 3 complications or pancreas-related complications. Reoperation was required in 38 patients (11.8%), with no significant difference between cohorts (22 (13.7%) and 16 (10.0%) in the LPD and RPD groups, respectively; P = 0.311). However, reoperation via an open approach was more frequent after RPD (15 of 16 (93.8%)) than after LPD (13 of 22 (59.1%)) (P = 0.040). Similarly, no significant differences in postoperative transfusion rates were observed between the two groups (P = 0.463). Postoperative in-hospital deaths did not differ significantly, although it was higher in the LPD cohort (6 (3.7%)) compared with the RPD group (3 (1.9%)) (P = 0.310). Similarly, 90-day deaths were comparable between the two approaches (11 (6.8%) and 6 (3.7%) for the LPD and RPD groups, respectively; P = 0.212). Oncological outcomes, including tumour diameter (P = 0.430), number of harvested lymph nodes (P = 0.285), resection margin status (P = 0.201), and pathological staging (P = 0.492), were comparable between the two approaches (Table 3). Multivariable analysis of prognostic factors for operative time and EBL On univariable analysis, female sex (P < 0.001), ASA score ≥ 3 (P < 0.001), and type of surgical procedure (pylorus-preserving versus Whipple; P = 0.002) were significantly associated with operative time. The robotic approach was also associated with operative time at univariable analysis (P = 0.041). On multivariable analysis, female sex (β = −30.7 min, 95% c.i. −53.4 to −8.0; P = 0.008), ASA score (β = −44.6 min, −67.7 to −21.4; P < 0.001), and type of surgical procedure (β = −78.8 min, −106.2 to −51.3; P < 0.001) remained independently associated with operative time. In contrast, the effect of surgical approach was attenuated and was no longer statistically significant after adjustment (β = 12 min, −22.3 to 33.7; P = 0.140) (Table 4). Table 4Prognostic factors analysis for operative timeUnivariable regression analysisMultivariable regression analysisβ (min) P β (min) P Age (years)−0.7 (−1.90, 0.50)0.241 Female sex (versus male)−43.5 (−67.80, −19.16)< 0.001− 30.7 (−53.40, −8.00)0.008Body mass index (kg/m2)2.06 (−0.89, 5.03)0.170ASA score (≥ 3 versus < 3)−53.92 (−78.36, −29.48)< 0.001−44.60 (−67.70, −21.40)< 0.001Previous abdominal surgery−0.55 (−25.53, 24.42)0.962NAT6.72 (−42.84, 56.28)0.791Tumour type (PDAC versus others)−9.72 (−34.58, 15.13)0.440Surgical approach (RPD versus LPD)28.10 (2.00, 54.70)0.04112.00 (− 22.30, 33.70)0.140Associated organ resection45.98 (−24.38, 116.36)0.204Vascular resection11.10 (−40.80, 63.00)0.671Pancreatic texture (firm versus soft)13.34 (−11.34, 38.00)0.280Pancreatic duct diameter (mm)−4.00 (−8.70, 0.61)0.090Pylorus-preserving versus Whipple−38.40 (−62.90, −13.85)0.002−78.80 (−106.20, −51.30)< 0.001Conversion−21.84 (−42.75, 19.94)0.123Tumour diameter (mm)0.08 (−0.90, 1.07)0.861Results are expressed as unstandardized regression coefficients (β). Values in parentheses are 95% confidence intervals. Multivariable analyses were exploratory, and results should be interpreted as associational. min, minutes; ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma; RPD, robotic pancreatoduodenectomy; LPD, laparoscopic pancreatoduodenectomy. Prognostic factors analysis for operative time Results are expressed as unstandardized regression coefficients (β). Values in parentheses are 95% confidence intervals. Multivariable analyses were exploratory, and results should be interpreted as associational. min, minutes; ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma; RPD, robotic pancreatoduodenectomy; LPD, laparoscopic pancreatoduodenectomy. On univariable analysis, age (P = 0.030), female sex (P < 0.001), surgical approach (P < 0.001), type of surgical procedure (P = 0.003), and conversion to open surgery (P = 0.010) were significantly associated with EBL. On multivariable analysis, surgical approach remained independently associated with EBL, with robotic procedures associated with lower blood loss (β −0.372, 95% c.i. −0.641 to −0.102; P = 0.007). Female sex was also independently associated with lower EBL (β −0.456, −0.684 to −0.227; P < 0.001). In contrast, the effects of age, type of surgical procedure, and conversion to open surgery were no longer statistically significant after adjustment (Table 5). Table 5Prognostic factors analysis for EBLUnivariable regression analysisMultivariable regression analysisβ P Β P Age (years)0.012 (0.001, 0.023)0.0300.009 (−0.002, 0.019)0.101 Female sex (versus male)−0.467 (−0.7, −0.234)< 0.001− 0.456 (−0.684, −0.227)< 0.001Body mass index (kg/m2)0.027 (−0.001, 0.056)0.061ASA score (≥ 3 versus < 3)−0.113 (−0.355, 0.130)0.360Previous abdominal surgery−0.002 (−0.240, 0.230)0.990NAT0.310 (−0.157, 0.785)0.191Tumour type (PDAC versus others)−0.017 (−0.256, 0.221)0.883Surgical approach (RPD versus LPD)−0.450 (−0.687, −0.228)< 0.001−0.372 (−0.641, −0.102)0.007Associated organ resection0.280 (−0.384, 0.958)0.401Vascular resection0.449 (−0.043, 0.942)0.072Pancreatic texture (firm versus soft)−0.066 (−0.303, 0.17)0.581Pancreatic duct diameter (mm)0.038 (−0.008, 0.084)0.100Pylorus-preserving versus Whipple−0.355 (−0.591, −0.119)0.003−0.05 (−0.325, 0.226)0.720Conversion0.480 (0.109, 0.851)0.0100.348 (−0.092, 0.808)0.231Tumour diameter (mm)−0.003 (−0.013, 0.006)0.494Values in parentheses are 95% confidence intervals. EBL was log-transformed due to right-skewed distribution. Regression coefficients (β) represent relative changes in estimated blood loss. ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma; RPD, robotic pancreatoduodenectomy; LPD, laparoscopic pancreatoduodenectomy. Prognostic factors analysis for EBL Values in parentheses are 95% confidence intervals. EBL was log-transformed due to right-skewed distribution. Regression coefficients (β) represent relative changes in estimated blood loss. ASA, American Society of Anesthesiologists; NAT, neoadjuvant therapy; PDAC, pancreatic ductal adenocarcinoma; RPD, robotic pancreatoduodenectomy; LPD, laparoscopic pancreatoduodenectomy. Sensitivity analyses Sensitivity analyses confirmed the robustness of the main findings (Tables S3–S5). After adjustment for centre-level clustering and calendar era, surgical approach remained independently associated with EBL and conversion to open surgery, with LPD showing a higher likelihood of conversion. Conversely, no significant association was observed between surgical approach and failure-to-rescue. Calendar era was not significantly associated with the investigated outcomes. These findings support the stability of the observed associations after accounting for institutional variability and temporal trends. Discussion This national multicentre PSM analysis provides a comparative assessment of RPD and LPD performed in high-volume centres with established expertise in minimally invasive pancreatic surgery. The main findings indicate that RPD is associated with significantly lower intraoperative blood loss and a reduced conversion rate compared with LPD. Importantly, these advantages did not translate into significant differences in postoperative outcomes. Postoperative morbidity, pancreas-specific complications, and oncological outcomes were comparable between the two approaches. The lower EBL observed after RPD is consistent with previous comparative studies15,25–27 reporting improved intraoperative outcomes with the robotic approach . This may be partly explained by the technical advantages of the robotic platform, including three-dimensional visualization, enhanced dexterity, tremor filtration, and improved instrument stability, which may facilitate precise dissection and haemostasis. Similarly, the lower conversion rate observed in the RPD group may reflect the ability of the robotic platform to manage technically demanding intraoperative scenarios better, as previously suggested in comparative studies15,17,28. These findings are also in line with more recent evidence, including a recent nationwide matched analysis from Japan18, which reported reduced blood loss and lower conversion rates after RPD compared with laparoscopic and open approaches. Conversely, other authors17,28 have reported comparable EBL between RPD and LPD but a reduced need for intraoperative transfusions in the robotic cohort. Taken together, the available evidence suggests that the intraoperative benefits of RPD are plausible, but their magnitude may vary according to patient selection, institutional expertise, and procedural standardization. In this context, the role of the learning curve should also be considered when interpreting these intraoperative findings. Both LPD and RPD are technically demanding procedures, and surgeon experience is known to influence operative performance, conversion rates, and perioperative outcomes. Several studies14,26,28–31 have shown that increasing experience with RPD is associated with progressive improvements in operative efficiency and short-term outcomes. In the present study, the progressive adoption of RPD over time may partly reflect increasing institutional and team proficiency with the robotic platform. Therefore, some of the observed advantages of RPD, particularly lower blood loss and fewer conversions, may be related not only to the intrinsic technical features of the robotic system but also to procedural maturity and accumulated experience. However, surgeon-specific experience metrics were not available in the registry, and this aspect could not be directly analysed. Notably, sensitivity analyses accounting for calendar era did not show a significant temporal effect on the investigated outcomes, supporting the robustness of the main findings. Operative time remains a debated issue in minimally invasive pancreatic surgery. Previous comparative studies and meta-analyses5,15,17,26,28,32,33 have reported heterogeneous results, with some showing similar operative times between LPD and RPD and others suggesting longer procedures during the early phase of robotic adoption. In the present study, operative time was slightly longer in the robotic cohort, but surgical approach was not an independent determinant after adjustment. This suggests that factors other than the platform itself may play a more relevant role. Differences in procedural strategy may have contributed to this finding. In particular, RPD was more frequently performed as a pylorus-preserving procedure, which often requires hand-sewn gastrointestinal reconstruction, whereas LPD more commonly involved antrectomy and stapled reconstruction. Operative time should therefore be interpreted in the context of technical heterogeneity rather than as a direct measure of platform performance alone. Despite differences in intraoperative outcomes, postoperative results were similar between groups. Rates of major complications and pancreas-specific complications, including POPF, DGE, and PPH, did not differ significantly between RPD and LPD. This is consistent with recent systematic reviews and network meta-analyses11,15–17,34 showing comparable postoperative morbidity between the two minimally invasive approaches. The similarity in pancreas-related complications may be explained by the balanced distribution of established risk factors, including pancreatic texture, duct diameter, and underlying pathology. LOS, readmission, postoperative transfusion, reoperation, and deaths were also comparable, suggesting that the modest intraoperative advantages observed with RPD did not translate into measurable short-term clinical benefits in this cohort. The management of postoperative complications requiring reintervention deserves specific consideration. In this study, reinterventions after LPD were more frequently performed using a minimally invasive approach, whereas those after RPD were more often managed through laparotomy. This difference may reflect logistical limitations, as robotic platforms are not always readily available in urgent or emergency settings, unlike conventional laparoscopy. Although evidence on robotic surgery in emergency settings remains limited35, differences in reintervention modality may influence the effectiveness of surgical rescue following severe complications and could partially affect downstream outcomes. In this context, it is noteworthy that 90-day deaths, although not statistically different, were nearly doubled in the LPD cohort. This numerical disparity may be influenced by multiple factors, including patient selection and perioperative management. However, the present study was not designed to assess the relationship between reintervention strategy and deaths, and no specific analysis was performed to explore this association. Therefore, any potential link between these findings should be considered hypothesis-generating only and warrants further investigation in dedicated studies. The higher utilization of internal stents observed in the RPD group represents an additional noteworthy finding. This difference may reflect surgeon preference or reconstructive strategy when performing robotic pancreatoenteric anastomoses, rather than an intrinsic requirement of the robotic approach. Importantly, increased stent use did not translate into differences in fistula rates, suggesting that this variable did not adversely affect clinical outcomes but should nevertheless be acknowledged as part of the technical heterogeneity inherent to multicentre analyses. Regarding oncological adequacy, RPD and LPD yielded comparable histopathological outcomes. No relevant differences were observed in tumour diameter, lymph node harvest, resection margin status, or pathological stage. These findings are consistent with previous studies11,17,27,29 suggesting that minimally invasive PD can achieve oncological results comparable to conventional approaches when performed in experienced centres. Although some reports13 have suggested potential oncological advantages of RPD, particularly in terms of margin status and lymph node retrieval, the present study does not support a clear superiority of one minimally invasive platform over the other from an oncological standpoint. This study has several limitations. First, its retrospective design represents the main methodological limitation. Although PSM was used to reduce baseline imbalance, residual confounding cannot be excluded. In addition, some variables known to influence postoperative outcomes, such as pancreatic texture and duct diameter, were not included in the matching model because the primary focus was on intraoperative outcomes. Second, the multicentre nature of the registry introduces variability in operative technique, perioperative management, and postoperative follow-up. Although all participating institutions were pancreatic referral centres with expertise in minimally invasive surgery, differences in surgeon experience and centre volume may still have influenced outcomes. Third, surgeon-specific learning curve data were unavailable, limiting the ability to assess directly the impact of individual experience. Fourth, selection bias related to allocation to minimally invasive surgery remains possible, as suggested by the relatively low proportion of patients receiving NAT and the limited number of vascular resections. Finally, the absence of long-term follow-up prevents assessment of oncological survival outcomes. Despite these limitations, this study provides robust real-world evidence from a national registry including exclusively high-volume centres with substantial experience in minimally invasive pancreatic surgery. The findings support the safety and feasibility of both RPD and LPD and suggest that the robotic platform may offer specific intraoperative advantages, particularly in terms of blood loss and conversion, without compromising postoperative or histopathological outcomes. In conclusion, RPD was associated with lower EBL and reduced conversion rates compared with LPD. However, these differences were modest and did not translate into clinically meaningful improvements in postoperative outcomes. Both approaches appear safe and feasible in experienced centres. Further prospective studies, ideally incorporating surgeon-specific learning curve metrics and long-term oncological follow-up, are warranted to define better the relative role of robotic and laparoscopic approaches in minimally invasive PD. Collaborators Members of the IGOMIPS Registry Study Group: A. Gemini (Azienda Ospedaliera S. Maria, Terni, Italy), L. Tirloni (AOU Careggi, Firenze, Italy), A. Maffongelli (‘Antonio Cardarelli’ Hospital, Naples, Italy), A. Giani (ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy), V. P. Barbieri (G. Panico Hospital, Tricase, Lecce, Italy), R. Ballarin (University of Modena and Reggio Emilia, Modena, Italy), A. Giardino (Pederzoli Hospital, Peschiera del Garda, Italy), M. Cesare Giglio (University of Naples Federico II, Naples, Italy), M. Giuffrida (Azienda Ospedaliero-Universitaria di Parma, Parma, Italy), R. Rossi (AUO delle Marche, Ancona, Italy), A. Manzoni (Fondazione Poliambulanza Istituto Ospedaliero, Brescia, Italy), G. Zimmitti (Fondazione Poliambulanza Istituto Ospedaliero, Brescia, Italy), L. Mastrangelo (IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy), F. Valentina (‘F. Miulli’ General Hospital, Acquaviva delle Fonti, Bari, Italy), L. Veneroni (Ospedale Infermi, Rimini, Italy; Azienda Unica della Romagna, Rimini, Italy), G. L. Baiocchi (ASST Cremona, Cremona, Italy), M. P. F. Dorma (Misericordia Hospital, Grosseto, Italy), G. Capretti (IRCCS Humanitas Research Hospital, Milan, Italy), A. Mereu (Chirurgia Generale ed Oncologica Ospedale Michele e Pietro Ferrero, Verduno, Italy), P. Di Gioia (Hospital Pierantoni-Morgagni, Forlì, Italy), C. Ricci (University of Bologna, Bologna, Italy), S. Marcucci (Azienda Sanitaria Universitaria Integrata del Trentino, Trento, Italy), S. Tramontano (San Giovanni and Ruggi Hospital Trust, Salerno, Italy), D. Caputo (University Campus Bio-Medico di Roma, Rome, Italy; Fondazione Policlinico Universitario Campus Bio-Medico, 00128 Rome, Italy), C. Fiorillo (Fondazione Policlinico Universitario Agostino Gemelli IRCCS di Roma, Rome, Italy), A. Comandatore (University of Pisa, Pisa, Italy), A. Esposito (University of Verona, Verona, Italy). Supplementary Material Contributor Information Giuseppe Quero, Digestive Surgery Unit, Fondazione Policlinico Universitario Agostino Gemelli IRCCS di Roma, Rome, Italy; Dipartimento di Medicina e Chiurgia Traslazionale, Università Cattolica del Sacro Cuore di Roma, Rome, Italy. Roberta Menghi, Digestive Surgery Unit, Fondazione Policlinico Universitario Agostino Gemelli IRCCS di Roma, Rome, Italy; Dipartimento di Medicina e Chiurgia Traslazionale, Università Cattolica del Sacro Cuore di Roma, Rome, Italy. Niccolò Napoli, Division of General and Transplant Surgery, University of Pisa, Pisa, Italy. Allegra Ripolli, Division of General and Transplant Surgery, University of Pisa, Pisa, Italy. Giovanni Ferrari, Department of Minimally Invasive and Oncologic Surgery, ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy. Michele Mazzola, Department of Minimally Invasive and Oncologic Surgery, ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy. Elisa Bannone, Unit of HPB Surgery, Pederzoli Hospital, Peschiera del Garda, Italy. Giovanni Butturini, Unit of HPB Surgery, Pederzoli Hospital, Peschiera del Garda, Italy. Massimo Giuseppe Viola, General Surgery Department, G. Panico Hospital, Tricase, Lecce, Italy. Marco Garatti, Department of General Surgery, Fondazione Poliambulanza Istituto Ospedaliero, Brescia, Italy. Giovanni Tebala, Digestive and Emergency Surgery Unit, Azienda Ospedaliera S. Maria, Terni, Italy. Fabrizio Di Benedetto, HPB Surgery and Liver Transplant Unit, University of Modena and Reggio Emilia, Modena, Italy. Luca Moraldi, HPB Surgery Unit, AOU Careggi Firenze, Firenze, Italy. Carlo Molino, Department of Oncological Surgery Team 1, ‘'Antonio Cardarelli'’ Hospital, Naples, Italy. Sergio Alfieri, Digestive Surgery Unit, Fondazione Policlinico Universitario Agostino Gemelli IRCCS di Roma, Rome, Italy; Dipartimento di Medicina e Chiurgia Traslazionale, Università Cattolica del Sacro Cuore di Roma, Rome, Italy. Umberto Bracale, Chirurgia Generale e d’Urgenza, AOU San Giovanni e Ruggi d’Aragona, Salerno, Italy. Roberto Troisi, Department of Clinical Medicine and Surgery, Hepato-Pancreato-Biliary Surgery Unit, University of Naples Federico II, Naples, Italy. Raffaele Dalla Valle, Programma di Chirurgia Oncologica ad Indirizzo Epato-Bilio-Pancreatico, Azienda Ospedaliero-Universitaria di Parma, Parma, Italy. Marco Vivarelli, Hepatobiliary, Pancreatic and Transplantation Surgery, AUO delle Marche, Ancona, Italy. Elio Jovine, Department of Sperimental Surgery, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy. Riccardo Memeo, Unit of Hepato-Biliary and Pancreatic Surgery, ‘F. Miulli’ General Hospital, Acquaviva delle Fonti, Bari, Italy; Department of Medicine and Surgery, LUM University, Casamassima, Italy. Gianluca Garulli, Chirurgia Generale d’ Urgenza Laparoscopica e Robotica, Eras Certified Centre, Ospedale Infermi Rimini, Azienda Unica della Romagna, Rimini, Italy. Luca Ottaviani, ASST Cremona, General Surgery Unit, Cremona, Italy. Stefano Berti, Department of General Surgery, Chirurgia Generale ed Oncologica Ospedale Michele e Pietro Ferrero, Verduno, Italy. Andrea Coratti, Department of General Surgery, Emergency Surgery and Gastroenterological Diseases, Misericordia Hospital, Grosseto, Italy. Giorgio Ercolani, Department of Medical and Surgical Sciences (DIMEC), University of Bologna, Hospital Pierantoni-Morgagni Forlì, Forlì, Italy. Roberto Salvia, Pancreatic Surgery Unit, Department of Engineering for Innovation Medicine (DIMI), University of Verona, Verona, Italy. Luca Morelli, Division of General Surgery, University of Pisa, Pisa, Italy. Riccardo Casadei, Division of Pancreatic Surgery, IRCCS, AOUBO, Department of Internal Medicine and Surgery (DIMEC), Alma Mater Studiorum, University of Bologna, Bologna, Italy. Roberto Coppola, Operative Research Unit of General Surgery, Fondazione Policlinico Universitario Campus Bio-Medico, Rome, Italy. Alberto Brolese, Department of General Surgery & Hepato-Pancreato-Biliary (HPB) Unit—ASUIT, Azienda Sanitaria Universitaria Integrata del Trentino, Trento, Italy. Vicenzo Tondolo, Dipartimento di Medicina e Chiurgia Traslazionale, Università Cattolica del Sacro Cuore di Roma, Rome, Italy; General Surgery Unit, Fatebenefratelli Isola Tiberina—Gemelli Isola, Rome, Italy. Alessandro Zerbi, Pancreatic Surgery, IRCCS Humanitas Research Hospital, Milan, Italy. Ugo Boggi, Division of General and Transplant Surgery, University of Pisa, Pisa, Italy. the IGOMIPS Registry Study Group: A Gemini, L Tirloni, A Maffongelli, A Giani, V P Barbieri, R Ballarin, A Giardino, M Cesare Giglio, M Giuffrida, R Rossi, A Manzoni, G Zimmitti, L Mastrangelo, F Valentina, L Veneroni, G L Baiocchi, M P F Dorma, G Capretti, A Mereu, P Di Gioia, C Ricci, S Marcucci, S Tramontano, D Caputo, C Fiorillo, A Comandatore, and A Esposito Funding The authors have no funding to declare. Author contributions Giuseppe Quero (Conceptualization, Writing—original draft), Roberta Menghi (Data curation), Niccolò Napoli (Formal analysis), Allegra Ripolli (Formal analysis), Giovanni Ferrari (Investigation, Methodology), Michele Mazzola (Investigation), Elisa Bannone (Conceptualization), Giovanni Butturini (Validation, Visualization), Massimo Giuseppe Viola (Formal analysis, Validation), Marco Garatti (Validation, Visualization), Giovanni Tebala (Methodology), Fabrizio Di Benedetto (Investigation, Methodology), Luca Moraldi (Formal analysis, Validation), Carlo Molino (Validation), Sergio Alfieri (Writing—review & editing), Umberto Bracale (Visualization), Roberto Troisi (Supervision), Raffaele Dalla Valle (Methodology, Supervision), Marco Vivarelli (Supervision, Validation), Elio Jovine (Validation, Visualization), Riccardo Memeo (Supervision), Gianluca Garulli (Methodology, Validation), Luca Ottaviani (Data curation), Stefano Berti (Methodology, Visualization), Andrea Coratti (Validation, Visualization), Giorgio Ercolani (Data curation, Supervision), Roberto Salvia (Validation, Visualization), Luca Morelli (Data curation, Investigation), Riccardo Casadei (Visualization), Roberto Coppola (Supervision, Validation), Alberto Brolese (Methodology), Vincenzo Tondolo (Visualization, Writing—review & editing), Alessandro Zerbi (Validation, Visualization), and Ugo Boggi (Supervision, Validation, Writing—review & editing) Disclosure The authors declare no conflict of interest. Supplementary material Supplementary material is available at BJS Open online. Data availability The data supporting the findings of this study are available upon reasonable request from the corresponding author: atgiuseppe.quero@policlinicogemelli.it. References Associated Data Data Citations Supplementary Materials Data Availability Statement The data supporting the findings of this study are available upon reasonable request from the corresponding author: atgiuseppe.quero@policlinicogemelli.it.