Safety of long vs. short fasting after ESD/STER for esophageal tumors ≤ 2 cm: a propensity score-matching study Authors: Kang Li, Foqiang Liao, Taiyu Chen, Zhenhua Zhu, Xiaolin Pan, Shunhua Long, Guohua Li, Xiaojiang Zhou, Youxiang Chen, Yin Zhu, Xu Shu Published: 2025 ABSTRACT BackgroundEndoscopic submucosal dissection (ESD) and submucosal tunneling endoscopic resection (STER) have become the safe and effective techniques for early-stage esophageal cancer and small-diameter submucosal tumors of the esophagus. Yet, there remains insufficient research on the appropriate fasting period following ESD or STER in patients with early esophageal neoplasms. This study aims to assess the impact of different fasting durations—long versus short—on postoperative complication rates and hospital stay durations.MethodsPatients with esophageal tumors less than 2 cm in diameter who underwent ESD or STER between 2014 and 2020 in the Department of Gastroenterology at the First Affiliated Hospital of Nanchang University were analyzed.ResultsIn intention to treat (ITT) analysis, 266 patients were part of the long fasting group, and 93 patients were part of the short fasting group. After matching, propensity score matching analysis (PSMA) resulted in 88 matched pairs between short and long fasting group, covariates were well balanced across both groups. Stratified analysis of esophageal ESD or STER indicate that the incidence of Postoperative discomfort and complications did not differ significantly between the two groups(all P > 0.05). And the length of hospital stay remained significantly longer in the long fasting group compared to the short fasting group with esophageal ESD (7.63 ± 2.56 days vs. 6.52 ± 2.17 days, P = 0.01) or esophageal STER (7.86 ± 2.07 days vs. 6.67 ± 1.90 days, P = 0.037). Supplementary per protocol (PP) analysis also showed similar trends in the incidence of Postoperative adverse events or complications and the hospitalization duration.ConclusionsFor small esophageal tumors (< 2 cm), short fasting after ESD or STER shows similar complication rates to long fasting but significantly shortens hospitalization, Improving its efficiency in postoperative care.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12876-025-04548-w. Introduction In recent years, the incidence of esophageal cancer has risen, making it the sixth most common cancer globally. Statistics indicate that approximately 544,000 individuals succumbed to esophageal cancer globally in 2020 [1]. Additionally, the detection of other esophageal tumors, such as, leiomyomas, gastrointestinal stromal tumors (GISTs), lipomas, and polyps, has increased with the advancement of endoscopic ultrasound (EUS) technology [2]. Some studies have suggested that endoscopic submucosal dissection (ESD) is considered as the safe and efficient treatment for early-stage esophageal cancer and submucosal benign tumors [3, 4]. However, the significant anatomical differences between the esophagus and the stomach or colon, coupled with the limited surgical field and operational space during endoscopy, contribute to the technical complexity of performing ESD in the esophagus [5, 6]. The emergence of endoscopic submucosal tunnel resection (STER) has solved some technical problems to some extent. Through the refinement of ESD techniques, STER aids in the accurate pathological evaluation and comprehensive handling of submucosal tumors from the muscularis propria. Various research has indicated that STER offers benefits such as shorter surgery times and reduced expenses compared to thoracoscopic resection [7, 8]. The optimal timing for early postoperative feeding following endoscopic procedures remains a subject of considerable debate in clinical practice. Current guidelines predominantly recommend a 48 to 72 h fasting period post-procedure. This recommendation is based on evidence showing that most post-endoscopic bleeding events occur within the first 72 h after diagnostic or therapeutic interventions [9, 10]. However, emerging evidence from randomized controlled trials (RCTs) on feeding regimens after endoscopic hemostasis challenges this conventional approach. Notably, trials involving both esophageal and non-esophageal variceal hemorrhage cohorts demonstrated non-inferiority of early enteral nutrition initiation (< 24 h post-procedure) compared to delayed feeding protocols (>24 or >48 h) in terms of rebleeding rates, hospitalization duration, and nutritional outcomes [11, 12]. Furthermore, a randomized controlled trial (RCT) on postoperative management after gastric endoscopic submucosal dissection (ESD) found that accelerated feeding protocols (within 24 h post-procedure) did not increase complication rates. Additionally, they were associated with higher patient satisfaction, greater cost-effectiveness (mean difference $395.4), and reduced hospital stay length (mean difference 0.9 days) [13]. These cumulative findings suggest that early postoperative feeding may be a safe and potentially superior alternative to traditional prolonged fasting in selected patient populations undergoing therapeutic endoscopy. The current evidence base regarding postoperative fasting protocols following endoscopic interventions for esophageal pathologies remains notably underdeveloped. Therefore, we design a retrospective study to systematically evaluate the relationship between postoperative fasting duration and clinical outcomes after feeding in ESD or STER for esophageal neoplasms (≤ 2 cm). Method and materials Participants and study design We conducted a retrospective analysis utilizing data from the electronic medical records system at the First Affiliated Hospital of Nanchang University. 641 patients with esophageal tumors who underwent ESD or STER from January 2014 to January 2020 were identified. ITT analysis was performed in the initial cohort, exclusion criteria included: (1) multiple tumors; (2) tumor diameter > 2 cm; (3) referral surgeries; (4) incomplete records of fasting duration; and incomplete demographic data; Based on whether the postoperative fasting duration was ≤ 1 day, patients were classified into short fasting and long fasting groups. Every case included was logged at the Human Genetic Resource Center of the First Affiliated Hospital of Nanchang University. Approval for this study was granted by the ethics committee at the First Affiliated Hospital of Nanchang University, with all patients giving their informed consent. Data collection and variable definitions The following variables were systematically documented: demographic characteristics (age, sex), comorbidities (hypertension, diabetes), tumor features (location, size, pathological type), procedural details (technique [ESD/STER], operation time, en bloc resection status, intraoperative adverse events), and postoperative outcomes. Tumor location was categorized according to the Japanese Classification of Esophageal Carcinoma (15th edition). Histological classification adhered to the WHO Classification of Tumors of the Digestive System (5th edition). Intraoperative bleeding was defined as endoscopically confirmed hemorrhage requiring therapeutic intervention (Forrest Ia–IIb). Perforation was identified by direct visualization during procedure or radiographic evidence of pneumomediastinum on postoperative computed tomography. Postoperative outcomes included bleeding (clinically evident hemorrhage with endoscopic confirmation), perforation (clinically suspected cases with radiographic confirmation), fever (body temperature > 38.0 °C), nausea/vomiting (documented episodes requiring medical attention), and thoracoabdominal pain (patient-reported symptoms prompting clinical evaluation). The length of hospital stay was calculated from the procedure date to discharge. Endoscopic procedures and postoperative management All endoscopic procedures were performed by a specialized team of board-certified endoscopists who had completed the standardized institutional learning curve for ESD/STER and maintained a high procedural volume. This ensured a consistent and high level of expertise across the entire study period. Standardized technique protocols were implemented using a single-channel endoscope (GIF-Q260J; Olympus, Japan). For ESD procedures: (1) Lesion demarcation via white-light endoscopy with narrow-band imaging (NBI) enhancement; (2) Submucosal cushion creation using adrenaline-indigo carmine solution (1:100,000); (3) Circumferential incision with Dual Knife J (KD-655Q; Olympus) in EndoCut Q mode (Effect 3, Duration 2, Interval 3); (4) Submucosal dissection using hybrid knife in Swift Coagulation mode (Effect 4, 40 W)(Fig. 1). For STER: (1) Distal mucosal incision (3–5 cm from lesion); (2) Submucosal tunneling creation with insulated-tip knife; (3) En bloc tumor resection under direct visualization(Fig. 2). Following recovery from anesthesia, patients were transferred to the ward for postoperative care. All patients received standardized postoperative management, including fluid resuscitation, mucosal protectants, and proton pump inhibitors (40 mg twice daily of esomeprazole or pantoprazole). A structured fasting protocol was implemented based on senior physicians’ recommendations. Oral intake was gradually resumed following a stepwise approach: clear liquids initially, followed by semi-solid foods, and finally advancing to regular solid diet. Oral feeding was initiated with a clear liquid diet, consisting of water, clear broth, and diluted juice. Patients began with small volumes of 50–100 mL per serving, offered 3–4 times per day, under close monitoring of tolerance. If no discomfort, nausea, or vomiting occurred over approximately 24 h, the diet was advanced to semi-solid items such as porridge, yogurt, and pureed foods. Subsequent progression to a regular solid diet—starting with soft, easily digestible foods—was guided by clinical evaluation by the attending physician and nursing staff. Tolerance was defined as the absence of significant adverse symptoms including thoracoabdominal pain, nausea, vomiting, or distension. Any recurrence of symptoms resulted in either delaying diet advancement or reverting to the prior dietary stage. All patients remained stable during their hospital stay with no mortality or transfers to other facilities. Fig. 1Procedures of esophageal leiomyoma STER. a Identification of lesions under endoscopy; (b) Submucosal injection; (c). Establish a submucosal tunnel; (d) Enter the tunnel and gradually peel off; (e) Exposure of tumor; (f) Complete removal of tumor body; (g) Titanium clip closes tunnel opening; (h) Measure the resected tumor Procedures of esophageal leiomyoma STER. a Identification of lesions under endoscopy; (b) Submucosal injection; (c). Establish a submucosal tunnel; (d) Enter the tunnel and gradually peel off; (e) Exposure of tumor; (f) Complete removal of tumor body; (g) Titanium clip closes tunnel opening; (h) Measure the resected tumor Fig. 2Procedures of esophageal cancer ESD. (a) Identification of endoscopic lesions; (b) Mark the scope of resection of the lesion; (c) Submucosal injectionl; (d) Stepwise dissection of lesions; (e) Titanium clip to close the wound; (f) Measure the resected tumor Procedures of esophageal cancer ESD. (a) Identification of endoscopic lesions; (b) Mark the scope of resection of the lesion; (c) Submucosal injectionl; (d) Stepwise dissection of lesions; (e) Titanium clip to close the wound; (f) Measure the resected tumor Statistical methods Statistical analyses were performed using R software (version 4.2.0). Continuous variables were summarized as mean ± standard deviation (SD) or median with interquartile range [IQR], depending on distribution normality assessed by the Shapiro–Wilk test. Group comparisons utilized independent samples t-test or Mann–Whitney U test for continuous variables, and chi-square or Fisher’s exact test for categorical variables, expressed as frequencies and percentages. Propensity score matching (PSM) was employed to mitigate confounding, implemented via the “MatchIt” package. Propensity scores were derived from a multivariable logistic model incorporating age, gender, tumor characteristics (size, location, pathology), comorbidities, procedural parameters (operation time, intraoperative bleeding, perforation, wound closure), and medication use (antithrombotics, antibiotics). A 1:1 nearest-neighbor matching was performed with a caliper of 0.2 SD of the logit propensity score. Covariate balance was evaluated using standardized mean differences (SMD < 0.1). Stratified analyses were conducted within the matched cohort to assess outcomes separately for ESD and STER procedures. A per-protocol (PP) sensitivity analysis excluded patients who developed significant symptoms or complications during fasting. Results from both the PP and stratified analyses aligned with the primary findings, supporting their robustness. All tests were two-sided, with P < 0.05 considered significant. Reporting followed STROBE guidelines. Results Study population A total of 641 patients who underwent ESD or STER for esophageal tumors were initially screened. After applying exclusion criteria for multiple tumors, tumor size > 2 cm, referral surgery, and incomplete data, 359 patients were analyzed according to the intention-to-treat principle. They were categorized by their assigned postoperative fasting duration (short vs. long) for all primary analyses. From this ITT cohort, 105 patients who developed significant symptoms during fasting were further excluded, resulting in 254 patients eligible for the per-protocol (PP) analysis. Propensity score matching was subsequently applied to both the ITT and PP cohorts, each yielding 88 and 71 well-balanced matched pairs for outcome comparison.(more details in Fig. 3). Fig. 3Flow chart of patients enrolled in the study ‘ Flow chart of patients enrolled in the study ‘ Baseline characteristics Baseline characteristics of patients in ITT analysis were summarized in Table 1. Patients in the long fasting group were significantly older (56.34 ± 11.63 vs. 51.20 ± 11.04 years, P < 0.001), with larger median tumor size (12 mm vs. 10 mm, P = 0.016). Analyses of data on pathology revealed a higher proportion of leiomyoma in the short fasting group (69.89% vs. 46.24%, P < 0.001) and more malignant lesions in the long fasting group (19.17% vs. 3.23%, P < 0.001). No significant differences were observed in gender distribution or comorbidities (all P > 0.05). While surgical approach distribution was similar between groups (P = 0.736), the long fasting group experienced higher rates of intraoperative bleeding (36.47% vs. 25.81%, P = 0.061) and perforation (7.52% vs. 3.23%, P = 0.146). The short fasting group demonstrated higher ESD completing wound closure rates (80.65% vs. 64.29%, P = 0.012) and shorter operation times (27.22 ± 15.53 min vs. 39.99 ± 23.67 min, P < 0.001), while en bloc resection rates remained comparable between groups (90.32% vs. 92.48%, P = 0.511). In the use of drugs, the long fasting group had a higher rate of postoperative antibiotic use than short fasting(19.17% vs. 3.23%, P = 0.023), and there was no significant difference in the use of antithrombotic drugs between the two groups(1.08% vs. 1.88%, P = 0.959).The baseline characteristics of patients in PP analysis can be detailed in Supplementary Table 3. Among the enrolled patients, the details of antithrombotic and antibiotic medication are summarized in Table 2. All patients on antithrombotic drugs (n = 6) discontinued therapy perioperatively, with a mean discontinuation period of 6.83 ± 1.73 days. Medications included single or dual antiplatelet agents as well as oral anticoagulants. For antibiotics (n = 48), third-generation cephalosporins were the most frequently administered agents for prophylaxis or treatment, while other classes such as second-generation cephalosporins, quinolones, and combination regimens were also used. Table 1Baseline characteristics of patients in the ITT analysis cohort before and after matchingBefore matchingAfter matchingShort fasting group (n = 93)Long fasting group (n = 266)P valueShort fasting group (n = 88)Long fasting group (n = 88)P valueAge (years, mean ± SD)51.20 ± 11.0456.34 ± 11.63< 0.00151.30 ± 11.1753.25 ± 12.110.270 Gender [n(%)] 0.4170.340 Female34 (36.56)85 (31.95)32 (36.36)26 (29.55) Male59 (63.44)181 (68.05)56 (63.64)62 (70.45) Comorbidity [n(%)] Diabetes2 (2.15)5 (1.88)1.002 (2.27)1 (1.14)1.00 Hypertension13 (13.98)51 (19.17)0.26013 (14.77)14 (15.91)0.830 History of abdominal surgery10 (10.75)29 (10.90)0.96810 (11.36)7 (7.95)0.440 Lesion location 0.0470.520 Upper 1/31 (1.08)16 (6.02)1 (1.14)0 (0.00) Middle 1/367 (72.04)160 (60.15)62 (70.45)58 (65.91) Lower 1/325 (26.88)90 (33.83)25 (28.41)30 (34.09) Pathology type[n(%)] < 0.0010.520 Leiomyoma65 (69.89)123 (46.24)62 (70.45)56 (63.64) Squamous carcinoma3 (3.23)51 (19.17)3 (3.41)7 (7.95) Dysplasia4 (4.30)45 (16.92)4 (4.55)3 (3.41) Others21 (22.58)47 (17.67)19 (21.59)22 (25.00) Pathology [n (%)] < 0.0010.190 Benign90 (96.77)215 (80.83)85 (96.59)81 (92.05) Malignant3 (3.23)51 (19.17)3 (3.41)7 (7.95) Operation method [n(%)] 0.7360.510 ESD69 (74.19)202 (75.94)64 (72.73)60 (68.18) STER24 (25.81)64 (24.06)24 (27.27)28 (31.82)Tumor size [mm, median (IQR)] All procedures10.00(7.00–15.00)12.00(10.00–16.00)0.01610.00(7.00–15.00)12.00(8.00–15.00)0.393 ESD10.00(6.00–15.00)12.00(8.00–15.00)0.01212.00(10.00–16.25.00.25)15.00(10.00–18.00)0.541 STER12.00(10.00–16.25.00.25)15.00(10.00–18.00)0.54712.00(10.00–16.25.00.25)15.00(10.00–18.5.00.5)0.471 Intraoperative bleeding 24 (25.81)97 (36.47)0.061023 (26.14)26 (29.55)0.610 Intraoperative perforation [n(%)] 3 (3.23)20 (7.52)0.1463 (3.41)5 (5.68)0.720Operation time [min, median (IQR)]27.22 ± 15.5339.99 ± 23.67< 0.00127.76 ± 15.5030.90 ± 17.140.205Fasting durations[day, median (IQR)]1.00(1.00–1.00)2.00(2.00–3.00)< 0.0011.00(1.00–1.00)2.00(2.00–3.00)/ En bloc resection 84 (90.32)246 (92.48)0.51179 (89.77)78 (88.64) Closured of ESD wound [n (%)] 0.01201.00 complete closure75 (80.65)171 (64.29)72 (81.82)71 (80.68) partially closure5 (5.38)20 (7.52)3 (3.41)3 (3.41) no closure13 (13.98)75 (28.20)13 (14.77)14 (15.91) Drug use [n(%)] Antibiotics6 (6.45)42 (15.79)0.02306 (6.82)9 (10.23)0.420 Antithrombotic Drugs1 (1.08)5 (1.88)0.9591 (1.14)1 (1.14)1.00Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic resection, SD Standard deviation, IQR Interquartile range Table 1Baseline characteristics of patients in the ITT analysis cohort before and after matchingBefore matchingAfter matchingShort fasting group (n = 93)Long fasting group (n = 266)P valueShort fasting group (n = 88)Long fasting group (n = 88)P valueAge (years, mean ± SD)51.20 ± 11.0456.34 ± 11.63< 0.00151.30 ± 11.1753.25 ± 12.110.270 Gender [n(%)] 0.4170.340 Female34 (36.56)85 (31.95)32 (36.36)26 (29.55) Male59 (63.44)181 (68.05)56 (63.64)62 (70.45) Comorbidity [n(%)] Diabetes2 (2.15)5 (1.88)1.002 (2.27)1 (1.14)1.00 Hypertension13 (13.98)51 (19.17)0.26013 (14.77)14 (15.91)0.830 History of abdominal surgery10 (10.75)29 (10.90)0.96810 (11.36)7 (7.95)0.440 Lesion location 0.0470.520 Upper 1/31 (1.08)16 (6.02)1 (1.14)0 (0.00) Middle 1/367 (72.04)160 (60.15)62 (70.45)58 (65.91) Lower 1/325 (26.88)90 (33.83)25 (28.41)30 (34.09) Pathology type[n(%)] < 0.0010.520 Leiomyoma65 (69.89)123 (46.24)62 (70.45)56 (63.64) Squamous carcinoma3 (3.23)51 (19.17)3 (3.41)7 (7.95) Dysplasia4 (4.30)45 (16.92)4 (4.55)3 (3.41) Others21 (22.58)47 (17.67)19 (21.59)22 (25.00) Pathology [n (%)] < 0.0010.190 Benign90 (96.77)215 (80.83)85 (96.59)81 (92.05) Malignant3 (3.23)51 (19.17)3 (3.41)7 (7.95) Operation method [n(%)] 0.7360.510 ESD69 (74.19)202 (75.94)64 (72.73)60 (68.18) STER24 (25.81)64 (24.06)24 (27.27)28 (31.82)Tumor size [mm, median (IQR)] All procedures10.00(7.00–15.00)12.00(10.00–16.00)0.01610.00(7.00–15.00)12.00(8.00–15.00)0.393 ESD10.00(6.00–15.00)12.00(8.00–15.00)0.01212.00(10.00–16.25.00.25)15.00(10.00–18.00)0.541 STER12.00(10.00–16.25.00.25)15.00(10.00–18.00)0.54712.00(10.00–16.25.00.25)15.00(10.00–18.5.00.5)0.471 Intraoperative bleeding 24 (25.81)97 (36.47)0.061023 (26.14)26 (29.55)0.610 Intraoperative perforation [n(%)] 3 (3.23)20 (7.52)0.1463 (3.41)5 (5.68)0.720Operation time [min, median (IQR)]27.22 ± 15.5339.99 ± 23.67< 0.00127.76 ± 15.5030.90 ± 17.140.205Fasting durations[day, median (IQR)]1.00(1.00–1.00)2.00(2.00–3.00)< 0.0011.00(1.00–1.00)2.00(2.00–3.00)/ En bloc resection 84 (90.32)246 (92.48)0.51179 (89.77)78 (88.64) Closured of ESD wound [n (%)] 0.01201.00 complete closure75 (80.65)171 (64.29)72 (81.82)71 (80.68) partially closure5 (5.38)20 (7.52)3 (3.41)3 (3.41) no closure13 (13.98)75 (28.20)13 (14.77)14 (15.91) Drug use [n(%)] Antibiotics6 (6.45)42 (15.79)0.02306 (6.82)9 (10.23)0.420 Antithrombotic Drugs1 (1.08)5 (1.88)0.9591 (1.14)1 (1.14)1.00Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic resection, SD Standard deviation, IQR Interquartile range Baseline characteristics of patients in the ITT analysis cohort before and after matching 10.00 (7.00–15.00) 12.00 (8.00–15.00) 10.00 (6.00–15.00) 12.00 (8.00–15.00) 12.00 (10.00–16.25.00.25) 15.00 (10.00–18.00) 12.00 (10.00–16.25.00.25) 15.00 (10.00–18.00) 12.00 (10.00–16.25.00.25) 15.00 (10.00–18.5.00.5) 1.00 (1.00–1.00) 2.00 (2.00–3.00) 1.00 (1.00–1.00) 2.00 (2.00–3.00) Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic resection, SD Standard deviation, IQR Interquartile range Table 2Details of medication(Antithrombotic drugs and Antibiotics)Antithrombotic Drugs Total(n = 6)Single-antiplatelet agents Aspirin2 Clopidogrel1 Dual-antiplatelet agents Aspirin + Clopidogrel2 Anticoagulation agents Rivaroxaban1 Discontinue of antithrombotic Drugs 6 The number of days off medication 6.83 ± 1.73 Antibiotics Total(n = 48) Third-generation cephalosporins 27 Second-generation cephalosporins 10 Quinolones 4 Others Third-generation cephalosporins + Metronidazole3 Latamoxef Sodium2 Cefminox Sodium1 Imipenem and Cilastatin Sodium1 Table 2Details of medication(Antithrombotic drugs and Antibiotics)Antithrombotic Drugs Total(n = 6)Single-antiplatelet agents Aspirin2 Clopidogrel1 Dual-antiplatelet agents Aspirin + Clopidogrel2 Anticoagulation agents Rivaroxaban1 Discontinue of antithrombotic Drugs 6 The number of days off medication 6.83 ± 1.73 Antibiotics Total(n = 48) Third-generation cephalosporins 27 Second-generation cephalosporins 10 Quinolones 4 Others Third-generation cephalosporins + Metronidazole3 Latamoxef Sodium2 Cefminox Sodium1 Imipenem and Cilastatin Sodium1 Details of medication(Antithrombotic drugs and Antibiotics) Propensity score matching analysis Since there were many variables with statistical differences in the initial cohort and bias due to confounding factors, which also affected our comparison of long versus short fasting after gastric ESD or STER for esophageal tumors, we used the nearest neighbor method for 1:1 PSM analysis. At the same time, we use scatter plots and histograms to evaluate the matching effect. All variables exhibit a favorable equilibrium in the baseline traits among the pairs (Table 1), and both scatterplot (Fig. 4) and histogram (Fig. 5) show good matching effects. Fig. 4Scatter plot of the tendency distribution before and after propensity score matching Scatter plot of the tendency distribution before and after propensity score matching Fig. 5Histogram of the tendency distribution before and after propensity score matching Histogram of the tendency distribution before and after propensity score matching Outcomes of stratified analysis of esophageal ESD or STER in patients in the initial vs. matched cohort Given the procedural differences between esophageal ESD and STER, outcomes were analyzed separately (Tables 3 and 4). Before matching, the short fasting group had a significantly lower incidence of fever in the ESD cohort (11.59% vs. 27.72%, P = 0.006), with a similar trend in the STER cohort (4.17% vs. 23.44%, P = 0.076). Hospitalization was consistently shorter in the short fasting group for both ESD (P < 0.001) and STER (P = 0.006), with no significant differences in other adverse events and postoperative complications. After matching, differences in fever and other symptoms were no longer significant in either cohort (all P > 0.05). However, the short fasting group still showed significantly shorter hospitalization for both ESD (P = 0.01) and STER (P = 0.037). Subgroup analysis of SMTs and malignant tumors after ESD (Supplementary Tables 1–2) showed no significant differences in adverse events and postoperative complications(P > 0.05), but hospitalization duration still differed significantly for SMTs before and after matching. In the supplementary PP analysis, after excluding complications during fasting, no significant differences were found in adverse events for either procedure (all P > 0.05), while hospitalization length remained significantly shorter in the short fasting group across procedures and matching status (all P < 0.05); (Supplementary Tables 4–5). Table 3Outcome of patients in the ITT analysis cohort of esophageal ESD or STER before matchingEsophageal ESDEsophageal STERShort fasting group (n = 69)Long fasting group (n = 202)P valueShort fasting group (n = 24)Long fasting group (n = 64)P valuePostoperative adverse events [n (%)] Fever8 (11.59)56 (27.72)0.0061 (4.17)15 (23.44)0.076 Nausea or vomiting2 (2.90)4 (1.98)1.003 (12.50)11 (17.19)0.559 Thoracoabdominal pain5 (7.25)20 (9.90)0.5113 (12.50)11 (17.19)0.835Postoperative bleeding [n (%)]0 (0.00)1 (0.50)1.00001.00Postoperative perforation [n(%)]00/00/Duration of hospitalization days, (mean ± SD)6.39 ± 2.158.48 ± 2.61< 0.0016.67 ± 1.908.22 ± 2.410.006Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Table 3Outcome of patients in the ITT analysis cohort of esophageal ESD or STER before matchingEsophageal ESDEsophageal STERShort fasting group (n = 69)Long fasting group (n = 202)P valueShort fasting group (n = 24)Long fasting group (n = 64)P valuePostoperative adverse events [n (%)] Fever8 (11.59)56 (27.72)0.0061 (4.17)15 (23.44)0.076 Nausea or vomiting2 (2.90)4 (1.98)1.003 (12.50)11 (17.19)0.559 Thoracoabdominal pain5 (7.25)20 (9.90)0.5113 (12.50)11 (17.19)0.835Postoperative bleeding [n (%)]0 (0.00)1 (0.50)1.00001.00Postoperative perforation [n(%)]00/00/Duration of hospitalization days, (mean ± SD)6.39 ± 2.158.48 ± 2.61< 0.0016.67 ± 1.908.22 ± 2.410.006Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Outcome of patients in the ITT analysis cohort of esophageal ESD or STER before matching Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Table 4Outcome of patients in the ITT analysis cohort of esophageal ESD or STER after matchingEsophageal ESDEsophageal STERShort fasting group (n = 64)Long fasting group (n = 60)P valueShort fasting group (n = 24)Long fasting group (n = 28)P valuePostoperative adverse events [n (%)] Fever8 (12.50)14 (23.33)0.1151 (4.17)6 (21.43)0.158 Nausea or vomiting2 (3.12)1 (1.67)0.2890 (0.00)1 (3.57)0.882 Thoracoabdominal pain4 (6.25)7 (11.67)1.003 (12.50)5 (17.86)1.00Postoperative bleeding [n (%)]00/00/Postoperative perforation [n(%)]00/00/Duration of hospitalization days, (mean ± SD)6.52 ± 2.177.63 ± 2.560.0106.67 ± 1.907.86 ± 2.070.037Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Table 4Outcome of patients in the ITT analysis cohort of esophageal ESD or STER after matchingEsophageal ESDEsophageal STERShort fasting group (n = 64)Long fasting group (n = 60)P valueShort fasting group (n = 24)Long fasting group (n = 28)P valuePostoperative adverse events [n (%)] Fever8 (12.50)14 (23.33)0.1151 (4.17)6 (21.43)0.158 Nausea or vomiting2 (3.12)1 (1.67)0.2890 (0.00)1 (3.57)0.882 Thoracoabdominal pain4 (6.25)7 (11.67)1.003 (12.50)5 (17.86)1.00Postoperative bleeding [n (%)]00/00/Postoperative perforation [n(%)]00/00/Duration of hospitalization days, (mean ± SD)6.52 ± 2.177.63 ± 2.560.0106.67 ± 1.907.86 ± 2.070.037Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Outcome of patients in the ITT analysis cohort of esophageal ESD or STER after matching Abbreviations: ESD Endoscopic submucosal dissection, STER Submucosal tunneling endoscopic, SD Standard deviation, ITT Intention-to-Treat Discussion Postoperative nutritional management is a critical determinant of surgical outcomes and patient recovery, regardless of whether the approach is open or endoscopic [14]. However, the optimal timing for transitioning from parenteral to enteral or oral nutrition after surgery remains uncertain [15]. Recent evidence indicates that enteral nutrition effectively replaces parenteral support in esophagogastric malignancies, offering clinical, economic, and metabolic benefits [16–18]. These benefits are thought to arise from early enteral nutrition, which preserves intestinal barrier integrity, reduces bacterial translocation, modulates immunity, and attenuates systemic inflammation [19–21]. Advances in minimally invasive surgery have redirected clinical attention to early oral feeding protocols. Sun HB et al. reported in a randomized trial that early oral intake after McKeown minimally invasive esophagectomy accelerates gastrointestinal recovery without increasing complications [22]. The same group later observed that reduced inflammatory markers in the early-feeding cohort likely mediated the attenuated stress response [23]. Collectively, these findings support the clinical benefits of prompt oral nutrition after minimally invasive surgery. Notably, evidence on post-ESD nutritional protocols—especially for esophageal procedures—remains limited. Current literature remains sparse in evaluating feeding strategies post-esophageal (ESD) or submucosal tunneling endoscopic resection (STER). Whether early feeding benefits patients after esophageal ESD or STER as it does after minimally invasive surgery or gastrointestinal ESD remains unclear [13, 22]. To address this clinical uncertainty, we conducted a retrospective cohort analysis of patients undergoing esophageal ESD and STER at our institution. Through propensity score matching, we systematically compared the safety profiles of extended versus abbreviated postoperative fasting regimens. Baseline characteristics differed between fasting groups. Patients assigned to prolonged fasting were older (mean age 56.34 ± 11.63 vs. 51.20 ± 11.04 years; P < 0.001), had larger lesions (median 12 mm [IQR 10–16] vs. 10 mm [IQR 7–15]), and showed a higher prevalence of malignancy (19.17% vs. 3.23%; P < 0.001). These differences may reflect resection strategy: malignant lesions often require wider margins, increasing complication risk and prompting longer fasting. A Korean multicentre study showed a 2.5-fold increase in complications when ESD targets lesions ≥ 3 cm [24]. Furthermore, the long fasting group exhibited a significantly higher incidence of intraprocedural hemorrhage (36.47%vs.25.81%, P = 0.061) and perforation (7.52% vs. 3.23%, P = 0.146). Endoscopists may prolong fasting to mitigate perceived risks of delayed complications and to allow mucosal healing. In a retrospective study, for patients with intraoperative gastric perforation, He et al. used a 48-hour window before starting to stare at a liquid diet, which resulted in no increase in postoperative complications [25]. After 1:1 propensity-score matching, baseline variables were balanced (all SMD < 0.1). No major complications were recorded after oral feeding in either group. Especially in supplementary PP analysis, thoracoabdominal pain and fever occurred infrequently in both groups. This low incidence may reflect our focus on events limited to the oral feeding period. Second, the low and non-significant morbidity suggests that fasting < 24 h does not impair wound healing or anastomotic integrity.Of note, Previous data indicate that fasting >24 h after major intestinal tumour resection correlates with prolonged ICU stay and increased infection risk [26]. Finally, the long fasting group still exhibited significantly prolonged median hospitalization duration after matching, a finding consistent with prospective investigations in gastric ESD populations, in which short-term fasting may contribute to shorter duration of disease [13]. This study has several limitations. First, its retrospective and single-center design inherently introduces risks of unmeasured confounding and selection bias. The application of ‘ITT’ and ‘PP’ terminologies to our non-randomized study may introduce bias. Our ‘PP’ analysis—a post-hoc sensitivity analysis excluding patients with complications—risks selection bias by potentially removing outcomes caused by the exposure itself. Furthermore, residual confounding from unmeasured factors remains possible despite propensity score matching. Second, it is a single-center retrospective analysis, which does not allow for precise documentation of patients’ specific dietary intake after fasting; it can only roughly distinguish between liquid and solid foods. Third, the sample size is relatively small, and the data on postoperative complications are limited, which may affect the statistical power of the results. Furthermore, as a retrospective study, it lacked systematic data on objective nutritional parameters (e.g., serum albumin, inflammatory markers) and nutrition-related complications (e.g., postoperative hypoglycemia, (venous thromboembolism) VTE) due to their non-routine collection in clinically stable patients during the short postoperative stay. Therefore, we could not assess the potential nutritional and metabolic differences between the fasting groups. Consequently, future large-scale, multicenter randomized controlled trials are requisite to further substantiate the relevant findings. Conclusion For small esophageal tumors (< 2 cm), early oral feeding after ESD or STER maintains comparable safety profiles to conventional delayed feeding protocols. Additionally, it shortens hospital stay. Supplementary Information Supplementary Material 1. Supplementary Material 1.