This study aims to evaluate the outcomes of the initial 58 cases of hiatal hernia surgery performed using a robotic approach, assess the safety of implementing this technique, and secondly, analyze the learning curve of the surgical team.

Regarding the descriptive analysis of baseline characteristics (Table 1), it is noteworthy that nearly 50% of the patients had a BMI greater than 30 kg/m2. Additionally, more than 70% of the patients had a paraesophageal hernia. Both conditions are directly associated with hernia recurrence [10, 11].

Regarding the morbidity associated with the robotic technique, previous studies have demonstrated its safety and effectiveness in antireflux procedures, with postoperative complication rates at 30 days ranging from 15 to 23% and mortality rates of 0% to 2.5% [12, 13]. In the current series, there was only one major complication in the PS group (Table 2), which occurred in the first patient operated on. This event is likely attributable to the initial lack of experience of the surgical team.

In contrast, RS, which often involves more extensive and challenging dissection, has historically been associated with higher morbidity and conversion rates [14]. In laparoscopic approach, morbidity and conversion rates reach up to 20–30% and 12%, respectively [9, 15]. Robotic surgery appears to have improved upon these outcomes. Mertens et al. reported a postoperative complication rate of 10.6%, with 2.6% being major complications [16]. In our series (Table 2), two patients in the RS group experienced intraoperative events (accidental opening of the pleura and gastric perforation) that were successfully managed during the procedure. Due to the more complex dissection involved, such events are more frequent in RS [17]. There was only one readmission in RS group due to symptomatic pleural effusion managed conservatively. No statistical differences were found between groups (p = 0.702). Thus, in the RS group, there were no major complications associated with the procedure, aligning these results more closely with series such as that of Sowards et al., which reported even lower rates of postoperative morbidity (1% for PS and 0% for RS) [18].

The length of hospital stay has also shown variability in different series with the introduction of robotic approaches [19, 20] ranging from 2 to 7 days [21]. In our series, hospital stay averaged 2 days, slightly longer in RS compared to PS (3 vs 2 days), with statistical significance (P = 0.005), but similar to other reported series [21]. Elmously et al. have described even shorter hospital stays and have attributed this to the implementation of early discharge programs rather than the type of surgical approach.

With respect to symptom resolution (Table 2), the overall series achieved a satisfaction rate of over 85%. Although the evaluation method is subjective, it is commonly used by other authors who report similar data, comparable to outcomes seen with conventional laparoscopic surgery [22]. However, there are limited studies presenting satisfaction results alongside quality of life, reflux, or dysphagia assessments. Nonetheless, satisfaction rates remain consistently high at 82% [20, 23]. Thus, there was no statistically significant difference observed between PS and RS groups (p = 0.31).

Hiatal hernia surgery represents a critical procedure in benign robotic surgery of the upper gastrointestinal tract, demanding a range of intricate skills such as tissue manipulation, hiatus dissection, and intracorporeal suturing. The mastery of these skills has positioned this procedure as a stepping stone to more complex esophagogastric oncologic surgeries.

The adoption of new surgical technologies inevitably entails a learning curve. One of the crucial insights from learning curve analyses sought by surgeons is determining the requisite case experience needed to overcome this curve.

To evaluate the learning curve in our study, we focused on surgical time due to the low morbidity observed, employing the CUSUM method for its ability to visually depict the evolution of learning and identify critical turning points [24]. This method aids in pinpointing when the learning phase has been surpassed, potentially shortening the curve for more complex procedures such as oncologic surgery [25].

In our series, the average surgical time was 121 min, comparable to studies by Morino et al. (131 min) [26] Nakadi et al. (137 min) [27] and lower than others [17, 28]. However, many authors have reported longer surgical times compared to conventional approaches [27, 29], often attributed to robot setup time, though this is not universally confirmed [30].

Our analysis identified three distinct phases based on the inflection points of the CUSUM curve (Fig. 2). Phase 1 exhibits the expected learning curve slope, representing the training phase (operations 1 to 14). Phase 2 shows a plateau, signifying the beginning of the improvement phase (operations 15 to 25), where the surgeon begins to demonstrate increased proficiency with accumulated experience. Phase 3 displays a decline in the plateau, indicating the expertise period (operations 26 onwards), consistent with typical learning curve patterns [31]. The minor fluctuations observed within each phase of our study were attributed to varying complexities of hernia cases, particularly the introduction of RS starting from the 10th case (marked by red spots). Additionally, the upward trend seen from the 43rd case onwards was associated with increased case complexity, including two RS cases, and the involvement of two new surgeons in the procedures.

Recent research on the learning curve for robotic hiatal hernia repair and fundoplication indicated mastery was achieved after 85 cases, with a learning curve observed over 40 cases. This study noted decreases in average surgical time, blood loss, and hospital stay across different learning phases, although these improvements did not correlate with reduced morbidity [32]. Similarly, Cundy et al. identified three distinct phases in their learning curve for pediatric patients, with training concluding around the 37th case and mastery after the 48th case [8].

In contrast, another study on robotic foregut surgery reported a longer learning curve extending up to 86 cases, encompassing various surgical procedures [33]. These studies, conducted without proctored teams, achieved mastery much later compared to our findings. Conversely, the studies employing a proctored pathway have demonstrated potential reductions in the learning curve [7], though several factors influence this process, such as prior experience in laparoscopic techniques, background in esophagogastric surgery, surgeon skill, and procedural complexity.

We have demonstrated that the learning curve may be as short as 14 cases and a proctored pathway involving simulation-based training, a multi-day wet lab course, followed by robotic procedures overseen by robotic upper GI experts, may be the way to reduce or nearly eliminate this learning curve.

This study has limitations. It is a retrospective analysis providing data from a single institution and a single experienced surgeon. It includes different types of surgery (primary and revisional), fundoplication (partial and complete), and other modifications of the technique such as the placement of mesh in some cases. These variables could alter the evolution of the surgical time and, consequently, the learning curve. Therefore, extrapolating the results to other centers must be approached with caution due to the mix of patient cases and surgeon-related variables. This methodology for evaluating the learning curve might be more suitable for self-evaluation of results rather than establishing a cut-off point for strategic decisions.

Additionally, the study is limited by the lack of objective evidence during follow-up, which affects the relative validity of the satisfaction results.