This was a single-center study using prospectively collected data from 125 consecutive patients who underwent SPDG between November 2011 and December 2015 at Seoul National University Bundang Hospital in Korea. All operations were performed by 2 surgeons, who had experience of more than 2,000 cases of LDG (surgeon A) and 200 cases of LDG (surgeon B) before starting SPDG. SPDG was conducted only for EGC confirmed in the preoperative examination and was excluded if there was evidence of advanced gastric cancer or metastasis to other organs. Previous abdominal surgery itself was not a contraindication to SPDG, except for cases where severe adhesion was expected due to multiple surgical procedures. There was no restriction to patient body mass index (BMI). The surgical procedure of SPDG has already been reported in a previous study [5]. The protocol of this retrospective cohort study was approved by the institutional review board of Seoul National University Bundang Hospital, Seongnam, Korea, an academic hospital affiliated with Seoul National University, College of Medicine (approval No. L-20171194).

Baseline characteristics, including patients' age, BMI, comorbidity, and previous operation, were recorded. Before surgery, the patients' conditions were assessed using the American Society of Anesthesiologists (ASA) physical status classification system.

The following variables were analyzed for evaluating the quality of the operations: operative time, estimated blood loss (EBL), number of retrieved lymph nodes (LNs), duration of postoperative hospital stay, and complication rate. The postoperative complications included the 30-day morbidity and mortality. The complications were graded using the Clavien-Dindo classification [7]; those under Clavien-Dindo grade I were not regarded as an event, whereas those with grade II or higher (requiring medical or surgical interventions) were regarded as an event.

There are several statistical methods to evaluate the learning curve, and among them, the moving average method was used. Two sequential variables, i.e., operative time and EBL, were used to define the learning curve. The complication rate was not included, since it was too low to perform reliable statistical analyses [89].

Statistical analyses were performed using the SPSS software (version 18.0). All values were expressed as means±standard deviations. The Student's t-test and Mann-Whitney U test were used for continuous variables and the χ² test for categorical variables. A P-value of <0.05 was considered statistically significant.

The general characteristics of the patients in each group are shown in Table 1. SPDG was performed successfully in all 125 cases without any serious intraoperative or postoperative complications. Surgeon A performed 68 (female-to-male sex ratio, 91.1%:8.82%) cases, and surgeon B performed 57 (female-to-male sex ratio, 61.4%:38.5%) cases. No significant differences were observed in age, BMI, ASA score, and history of abdominal operation, except for sex distribution. Surgeon A's patient group had a relatively higher proportion of female patients than surgeon B's patient group.

For analyzing the learning curve of each surgeon, cases were extracted from 6 (surgeon A) and 5 (surgeon B) sequential groups. The mean operative time in each group was calculated. The operative time was shorter in the beginning but increased after the middle group with surgeon A (Fig. 1). In contrast, surgeon B showed a steady state after 30 cases of SPDG (Fig. 2). The “training phase” was defined as the state before proficiency, while the “developed phase” was defined as the state after proficiency, indicating the period that the surgeon had become adept at SPDG.

The surgical quality of the 2 phases for each surgeon was compared (Tables 2 and 3). With surgeon B, the first 3 groups were in the training phase, and the following 2 groups were in the developed phase. In the comparison of the operative time according to each phase, it significantly decreased in the developed phase (157.8±38.4 minutes vs. 118.1±34.5 minutes, P=0.003). With surgeon A, as it was difficult to differentiate between the training phase and the developed phase using only the operative time, the entire phase was divided into half to compare the surgical quality. Only EBL (50.8±55.8 mL vs. 17.9±39.2 mL, P=0.002) decreased in the later phases, while the other variables did not show significant differences. The operative time was similar between the 2 phases (160.8±51.6 minutes vs. 173.3±35.2 minutes, P=0.6).

The early complication rates after surgery are shown in Table 4. The complication rate for surgeon A was 7.35%, and that for surgeon B was 5.26%. Complications under Clavien-Dindo grade IV and higher occurred in 4 out of 5 cases with surgeon A; these were not observed with surgeon B.

To examine the favorable factors for SPDG, those influencing the operative time were analyzed, and a sub-analysis was conducted for sex and BMI. The operative time between sexes in the entire patient group was similar (male-to-female sex ratio, 173.6±62.1 minutes: 164.2±52.1 minutes; P=0.20), showing no significant difference between the 2 groups. In the analysis of BMI using 25 kg/m² as the reference value, which is a standard value for obesity in Korea, the operative time was significantly shorter in the group with a BMI of <25 kg/m² (159.3±41.7 minutes vs. 194.25±81.1 minutes, P=0.001). Additional analysis of BMI according to sex was also conducted. The mean BMI was 22.5±3.31 kg/m² in the female patients and 22.5±2.60 kg/m² in the male patients, showing similar distributions in both groups. The operative time between the female patients with BMIs of <30 and ≥30 kg/m² showed a significant difference (156.6±41.1 minutes vs. 196.6±20.5 minutes, P=0.049).

No significant differences in the postoperative morbidities were observed between the 2 groups (Table 4). The early (<30-day) complication rates for surgeons A and B were 7.35% and 5.26%, respectively (P=0.24). The detailed information on the complications is summarized in Table 4. No in-hospital mortality was observed in either group. To assess the severity of postoperative morbidities, early complication cases were graded using the Clavien-Dindo classification. In surgeon B's patient group, complications under grade III and higher did not occur; in surgeon A's patient group, complications under grade III and higher occurred (5.88%).