Operative, safety and economic outcomes of endoscopic submucosal dissection in older adults: a retrospective real-world analysis

Abstract

Background

Endoscopic submucosal dissection (ESD) is a cornerstone organ-preserving treatment for colorectal neoplasia; however, the impact of advanced age on procedural efficiency, safety, and economic burden remains controversial. This study aimed to comprehensively evaluate these outcomes in older adults within a real-world clinical setting.

Methods

We retrospectively reviewed 255 patients who underwent colorectal ESD between January 2022 and December 2023. Patients were divided into elderly (≥ 65 years, n = 119) and non-elderly (< 65 years, n = 136) groups. Operative outcomes (resection and wound-management speeds), safety (delayed bleeding and perforation), and economic metrics (hospital stay and costs) were evaluated. Multivariable regression, age-stratified models, and 1:1 propensity score matching (PSM) were utilized to rigorously adjust for baseline confounders.

Results

Elderly patients presented with significantly larger and morphologically more complex lesions. Nevertheless, multivariable analyses demonstrated no significant independent association between age and resection speed or wound-management speed, which was fully supported by subsequent PSM and restricted cubic spline (RCS) analyses. Complication rates were extremely low (2.0% for bleeding, 0.8% for perforation) with no descriptive differences between groups. Although initial regression models suggested a slight age-related increase in hospital costs and length of stay, these economic disparities were attenuated and lost significance after applying PSM to balance clinical drivers such as ASA physical status and anesthesia methods (matched hospital costs: 11.39k vs. 10.57k RMB, p = 0.249; identical median stay: 4 vs. 4 days).

Conclusions

Appropriately selected elderly patients achieve operative and safety outcomes comparable to younger individuals during colorectal ESD. The initial economic variances observed are likely driven by underlying comorbidities and anesthetic requirements rather than chronological age itself. Clinical decisions should prioritize comprehensive, patient-centered assessments over strict age cutoffs.

Background

Endoscopic submucosal dissection (ESD) is a cornerstone treatment for colorectal neoplasia [1], prized for its high curative rates and organ-preserving benefits [2–4]. However, evidence regarding its applicability in older adults remains inconsistent. Several studies have reported that ESD achieves comparable technical success and safety outcomes in elderly and non-elderly patients, suggesting that age alone should not be considered a contraindication [5–8]. In contrast, other investigations indicate that elderly patients may experience reduced therapeutic benefit and higher risks, especially for the post-operative complications, such as bleeding, intraoperative perforation and prolonged hospital stays [9–13]. Adding to this complexity, published data on the economic implications of ESD in the elderly population are scarce. While ESD is generally considered cost-effective in broader patient groups [1, 14, 15], few studies have specifically evaluated healthcare utilization, hospital costs, or procedure-related efficiency in older adults. More importantly, these studies primarily compare ESD with alternative treatment modalities, rather than evaluating the costs of ESD across different age groups, leaving a significant gap in the real-world assessment of its value in this rapidly growing elderly population.

To address these uncertainties, our study evaluated the impact of age on multiple clinically relevant outcomes of colorectal ESD using a real-world cohort and multi-model statistical analysis to ensure robust comparability between elderly and non-elderly patients. Beyond standard safety and therapeutic endpoints, we incorporated operator-related outcomes (such as resection efficiency and wound-management performance) and economic outcomes (including hospital days and total costs), enabling a more comprehensive assessment of how age influences both clinical and healthcare-resource dimensions of ESD. By integrating technical, patient-centered, and economic perspectives, our analysis provides a deeper understanding of the true benefits and risks of ESD in older adults and offers practical evidence to support clinical decision-making in this increasingly common scenario.

Methods

Study design and participants

This study included 255 adult patients from Min-hang District Central Hospital from January, 2022 to December, 2023. The inclusion criteria were as follows: (1) clinical diagnosis of colorectal neoplastic lesions; (2) receiving ESD treatment. Patients with uncertain or incomplete information were excluded.

Collected patient data included demographics, lesion location, lesion size, lesion morphology, pathological pattern, duration of resection, resection pattern, postoperative complications, hospital stay, hospital costs. Enrolled patients older than 65 years were placed in the elderly group, and patients under 65 years of age were placed in the non-elderly group.

Surgery

All enrolled patients discontinued antiplatelet agents, anticoagulants, and any Chinese herbal medicines that may increase the risk of postoperative bleeding one week before surgery. During the procedure, the margins of the elevated lesion were marked with electrocautery. A mixture of normal saline, epinephrine, and methylene blue was injected submucosally at multiple points around the periphery of the lesion to achieve sufficient lifting. An incision was made in the mucosa using a Dual Knife (Olympus, Japan), followed by a step-by-step submucosal dissection. Throughout the dissection, repeated submucosal injections were administered to maintain the appropriate plane within the submucosal layer. Visible vessels on the wound surface were coagulated as needed, and hemostatic clips were applied for further hemostasis when necessary. Carbon dioxide insufflation was used to reduce abdominal discomfort caused by distension. Perforations of the intestinal wall identified during the procedure were typically closed using metal clips. All procedures were performed by senior certified endoscopists.

Outcomes

The observed outcomes were categorized into operative, safety, and economic measures. Operative outcomes included resection speed and wound management speed, calculated as the lesion area (length × width) divided by the respective procedure time [16, 17]. Safety outcomes consisted of postoperative delayed bleeding and intraoperative perforation. Economic outcomes included the length of hospital stay and total costs.

Sample size calculation

The sample size was estimated using published data on resection speed during colorectal ESD. Prior studies reported mean resection speeds of approximately 5–6 cm²/h with standard deviations around 2.4–3.0 cm²/h [16, 17]. Assuming a detectable between-group difference of 0.7 cm²/h, a standard deviation of 2.5 cm²/h, a two-sided α of 0.05, and a power of 80%, the required sample size was calculated using a two-independent-sample comparison of means. This yielded an estimated minimum sample size of 200 patients (approximately 100 per group). Our final cohort included 255 patients, exceeding this requirement.

Sensitivity and age-stratified analyses

To validate our primary conclusions, we performed a sensitivity analysis using a higher age threshold of ≥ 75 years. Furthermore, to explore the impact of advancing age on ESD outcomes, we stratified the entire cohort into three age categories: <65, 65–74, and ≥ 75 years.

Statistical analysis

Categorical variables were presented as frequencies and percentages, and compared between groups using the Pearson chi-squared test or Fisher’s exact test, as appropriate. Quantitative data were expressed as medians with interquartile ranges (Q1, Q3) and analyzed using the Mann-Whitney U test. Variables with a P-value < 0.25 in the univariate analyses were regarded as potential confounding factors and subsequently included in the multivariable models. Linear regression models were used for primary outcomes.

Due to the right-skewed distribution of hospital costs and the discrete count nature of the length of stay, ordinary linear regression was deemed inappropriate. Instead, multivariable analyses were conducted using a generalized linear model (GLM) with a Gamma distribution and a log-link function to evaluate hospital costs. Furthermore, a negative binomial regression model was employed to analyze the length of stay. To confirm the adequacy and goodness-of-fit of these models, residual diagnostics were performed by plotting standardized deviance residuals against the linear predictors. Finally, box plots and line graphs were utilized to visualize the comparative tendencies between the two groups. Additionally, the same statistical analyses were applied to the sensitivity and age-stratified analyses.

Moreover, to validate the conclusions, propensity score matching (PSM) was performed using the R package ‘MatchIt.’ The propensity scores were derived from a multivariable logistic regression model. Patients were matched 1:1 based on the nearest-neighbor method with a caliper width of 0.1 and without replacement. We evaluated the difference between the matched groups using Wilcoxon signed-rank test for continuous outcomes and McNemar’s test for categorical outcomes. All statistical analyses were performed with SPSS software (version 31, SPSS Inc., Chicago, IL, USA), and R (version 4.5.2; R Foundation for Statistical Computing, Vienna, Austria). A p < 0.05 was assumed to indicate statistically significant differences for all tests.

Result

Demographics and clinical characteristics

A total of 255 patients who underwent colorectal ESD were included (119 elderly, 136 non-elderly). The characteristics were summarized in the Table 1. The overall median age was 63 years (IQR, 52–70), with the elderly group being significantly older (70 vs. 53 years). Sex distribution (58.0% male) and lesion location were comparable between groups. However, lesion size was significantly larger in the elderly group (2.8 vs. 2.5 cm², p = 0.027). Morphology differed markedly (p < 0.001): LST-G (18.5% vs. 8.1%) and LST-NG (22.7% vs. 12.5%) were more common in the elderly, while submucosal-type lesions predominated in the non-elderly (25.7% vs. 9.2%). Pathological findings also varied (p = 0.001), with more intramucosal carcinomas in the elderly (41.2% vs. 24.3%) and more neuroendocrine tumors in the non-elderly (23.5% vs. 8.4%).

Table 1.

Baseline characteristics of the study population

	Total (N = 255)	Elderly group (N = 119)	Non-elderly group (N = 136)	P value
Age	63 (52,70)	70 (67,76)	53 (44,60)	NA
Gender (male%)	58.0% (148)	54.6% (65)	61.0% (83)	0.301
Lesion location				0.217
Left hemicolon	45.5% (116)	44.5% (53)	46.3% (63)
Right hemicolon	23.9% (61)	28.6% (34)	19.9% (27)
Rectum	30.6% (78)	26.9% (32)	33.8% (46)
Lesion size (cm2)	2.7 (1.5,4.0)	2.8 (1.8,5.0)	2.5 (1.4,3.7)	0.027
Lesion morphology				< 0.001
LST-G	12.9% (33)	18.5% (22)	8.1% (11)
LST-NG	17.3% (44)	22.7% (27)	12.5% (17)
Protruding type	51.8% (132)	49.6% (59)	53.7% (73)
Submucosal type	18.0% (46)	9.2% (11)	25.7% (35)
Anesthesia method				0.029
Intravenous anesthesia	92.2% (235)	88.2% (105)	95.6% (130)
General anesthesia	7.8% (20)	11.8% (14)	4.4% (6)
ASA grade				0.099
Class-1	59.2% (151)	53.8% (64)	64.0% (87)
Class-2	40.8% (104)	46.2% (55)	36.0% (49)
Complete resection (%)	98.4% (251)	99.2% (118)	97.8% (133)	0.711
Curative resection (%)	96.5% (246)	97.5% (116)	95.6% (130)	0.634
Pathological pattern				0.001
Adenoma	47.8% (122)	48.7% (58)	47.1% (64)
Intramucosal carcinoma	32.2% (82)	41.2% (49)	24.3% (33)
Submucosal carcinoma	1.6% (4)	0.8% (1)	2.2% (3)
Neuroendocrine tumors	16.5% (42)	8.4% (10)	23.5% (32)
Leiomyoma	2.0% (5)	0.8% (1)	2.9% (4)
Outcomes
Duration of resection (h)	0.47 (0.28,0.78)	0.48 (0.33,0.75)	0.47 (0.27,0.85)	0.599
Duration of wound management (h)	0.22 (0.15,0.32)	0.23 (0.17,0.32)	0.20 (0.13,0.32)	0.139
Postoperative delayed bleeding	2.0% (5)	1.7% (2)	2.2% (3)	1.000
Intraoperative perforation	0.8% (2)	0.8% (1)	0.7% (1)	1.000
Hospital days	4 (3,5)	4 (4,5)	4 (3,5)	< 0.001
Hospital costs (thousand RMB)	10.6 (9.2,12.5)	11.4 (9.7,12.8)	10.2 (9.03,12.0)	0.001

High and comparable rates of complete resection (98.4%) and curative resection (96.5%) were achieved. Resection duration, wound management time, and complication rates (delayed bleeding 2.0%, perforation 0.8%) were similar between groups. However, elderly patients had a significantly longer hospital stay (4 vs. 4 days, p < 0.001) and higher costs (median 11.4k vs. 10.2k RMB, p = 0.001).

After propensity score matching, a total of 88 pairs were selected. All baseline variables (including gender, lesion location, anesthesia method, ASA grade, lesion morphology, complete resection, curative resection, pathological patterns and lesion size) achieved adequate balance between the elderly and non-elderly groups, with standardized mean differences reduced to below 0.20 and most below 0.10. Variance ratios and distributional metrics also fell within acceptable ranges, indicating that the matched cohorts were well balanced and suitable for subsequent comparative analyses. Characteristics after PSM are summarized in Table 5.

Table 5.

The characteristics of total patients after PSM

	Elderly group (N = 88)	Non-elderly group (N = 88)	SMD	P value
Age	70 (67,76)	54 (48,60)	/	/
Gender (male%)	54.5% (48)	56.8% (50)	0.0456	/
Lesion location				/
Left hemicolon	50.0% (44)	45.5% (40)	0.0915	/
Right hemicolon	26.1% (23)	28.4% (25)	-0.0503	/
Rectum	23.9% (21)	26.1% (23)	-0.0513	/
Lesion size (cm2)	2.75 (1.80,5.00)	2.84 (1.80,4.50)	-0.0512	/
Lesion morphology				/
LST-G	12.5% (11)	12.5% (11)	0.0000	/
LST-NG	19.3% (17)	18.2% (16)	0.0271	/
Protruding type	55.7% (49)	56.8% (50)	-0.0227	/
Submucosal type	12.5% (11)	12.5% (11)	0.0000	/
Anesthesia method				/
Intravenous anesthesia	89.8% (79)	93.2% (82)	0.0459	/
General anesthesia	10.2% (9)	6.8% (6)	0.0459	/
ASA grade				/
Class-1	58.0% (51)	56.8% (50)	0.0661	/
Class-2	42.0% (37)	43.2% (38)	0.0661	/
Complete resection (%)	98.9% (87)	100% (88)	-0.1245	/
Curative resection (%)	97.7% (86)	97.7% (86)	0.0000	/
Pathological pattern				/
Adenoma	50.0% (44)	53.4% (47)	-0.0682	/
Intramucosal carcinoma	36.4% (32)	34.1% (30)	0.0462	/
Submucosal carcinoma	1.1% (1)	0%	0.1245	/
Neuroendocrine tumors	11.4% (10)	11.4% (10)	0.0000	/
Leiomyoma	1.1% (1)	1.1% (1)	0.0000	/
Outcomes
Duration of resection(h)	0.48 (0.32,0.76)	0.52 (0.28,0.85)	/	0.671
Duration of wound management(h)	0.25 (0.17,0.32)	0.21 (0.15,0.32)	/	0.984
Resection speed (cm2/h)	6.28 (4.09,9.06)	5.38 (3.26,9.15)	/	0.335
Wound management speed (cm2/h)	12.00 (6.97,20.65)	13.50 (7.20,21.57)	/	0.876
Postoperative delayed bleeding	2.3% (2)	3.4% (3)	/	1.000
Intraoperative perforation	1.1% (1)	1.1% (1)	/	1.000
Hospital days	4 (4,5)	4 (3,4)	/	0.010
Hospital costs (thousand RMB)	11.39 (9.66,12.83)	10.57 (9.03,12.15)	/	0.249

Elderly VS. non-elderly: operative outcomes

Multivariable linear regression demonstrated that age was not associated with resection speed (multivariate p = 0.430) or wound-management speed (multivariate p = 0.937) (Table 2; Fig. 1A and B), indicating no significant operational disadvantage for elderly patients after adjustment for lesion characteristics, ASA grade, and anesthesia methods. Furthermore, these results remained robust in both the sensitivity analysis using a ≥ 75 years cutoff (resection speed: p = 0.287; wound-management speed: p = 0.097) and the age-stratified trend analysis (p = 0.979 and p = 0.381, respectively) (Table 4). Ultimately, these findings were fully supported by the propensity score matching (PSM) analysis, where resection speed (6.28 vs. 5.38 cm²/h, p = 0.335) and wound-management speed (12.00 vs. 13.50 cm²/h, p = 0.876) remained highly comparable between the elderly and non-elderly cohorts (Table 5). Together, the multivariable regression, age-stratified, and PSM analyses consistently suggest that age does not independently influence technical performance during colorectal ESD. Furthermore, RCS analyses revealed no significant non-linear association between age and operative outcomes (Fig. 2A and B).

Table 2.

The operative outcomes between the two groups

	p value	β coefficient	Multivariate p value	Multivariate β coefficient
Resection speed
Age	0.009	0.092 (0.02,0.16)	0.430	0.66 (-0.98, 2.30)
Lesion size	< 0.001	0.76 (0.58,0.94)	< 0.001	0.65 (0.46, 0.84)
Lesion morphology
LST-G	< 0.001	9.89 (6.79,12.98)	0.811	6.99 (-5.03.6.43)
LST-NG	< 0.001	5.50 (2.64,8.36)	0.282	-2.87 (-8.10,2.36)
Protruding type	< 0.001	5.19 (2.87,7.51)	0.343	-2.35 (-7.20,2.51)
Submucosal type	Ref	Ref	Ref	Ref
Pathological pattern	< 0.001	-2.03 (-2.79,-1.28)	0.041	-1.50 (-2.95, -0.06)
Lesion location
Left hemicolon	< 0.001	3.53 (1.48,5.59)	0.083	2.10 (-0.28,4.49)
Right hemicolon	< 0.001	5.31 (2.90,7.71)	0.455	1.06 (-1.73,3.85)
Rectum	Ref	Ref	Ref	Ref
Anesthesia method	0.644	-0.80 (-4.20,2.60)	0.585	-0.82 (-3.76, 2.12)
ASA grade	0.803	-0.24 (-2.10,1.62)	0.870	-0.14 (-1.75, 1.48)
Wound management speed
Age	0.338	1.93 (-2.02,5.88)	0.937	-0.13 (-3.38, 3.12)
Lesion size	< 0.001	2.14 (1.79,2.49)	< 0.001	2.06 (1.69, 2.43)
Lesion morphology
LST-G	< 0.001	16.39 (9.61,23.17)	0.340	5.54 (-5.84,16.91)
LST-NG	< 0.001	13.90 (7.64,20.17)	0.442	4.07 (-6.32,14.46)
Protruding type	< 0.001	12.88 (7.79,17.97)	0.317	4.92 (-4.71,14.55)
Submucosal type	Ref	Ref	Ref	Ref
Pathological pattern	< 0.001	-3.61 (-5.27,-1.95)	0.875	-0.23 (-2.64,3.10)
Lesion location
Left hemicolon	0.001	7.49 (3.00,11.98)	0.108	3.88 (-0.86,8.61)
Right hemicolon	< 0.001	9.28 (4.05,14.52)	0.916	0.30 (-5.24,5.84)
Rectum	Ref	Ref	Ref	Ref
Anesthesia method	0.648	-1.71 (-9.05,5.63)	0.470	-2.15 (-7.98,3.68)
ASA grade	0.527	-1.30 (-5.31,2.72)	0.731	-0.56 (-3.78,2.65)

Fig. 1.

A The association between resection speed and age in both groups; (B) the association between wound management speed and age in both groups

Table 4.

The sensitivity analysis between age and outcomes

	P value		Multivariate p value	OR/β coefficient
Outcomes (≥ 75 year-old subgroup)
Resection speed	0.897	0.17 (-2.45,2.80)	0.287	-1.21 (-3.47, 1.06)
Wound management speed	0.715	-1.06 (-6.73,4.61)	0.097	-3.80 (-8.28, 0.68)
Hospital days	0.574	0.11 (-0.28,0.50)	0.715	0.07 (-0.32, 0.47)
Hospital costs	0.043	0.08 (0.01,0.17)	0.245	0.04 (-0.03, 0.11)
Outcomes (age-stratified)
Resection speed	0.198	0.83 (-0.43,2.09)	0.979	0.02 (-1.11, 1.14)
Wound management speed	0.627	0.68 (-2.06,3.41)	0.381	-1.00 (-3.23, 1.24)
Hospital days	0.351	0.09 (-0.10,0.28)	0.514	0.07 (-0.13, 0.27)
Hospital costs	0.004	0.058 (0.02,0.10)	0.032	0.04 (0.01, 0.07)

Fig. 2.

Restricted cubic spline (RCS) analyses of the association between age and operative outcomes. A Predicted resection speed. B Predicted wound management speed. Solid lines represent the estimated values, and shaded areas indicate 95% confidence intervals. Non-linear P values are presented in the top right corners

Elderly VS. Non-elderly: Safety outcomes

In the total cohort, the overall incidences of postoperative delayed bleeding (non-elderly vs. elderly: 2.2% [95% CI: 0.8%–6.3%] vs. 1.7% [95% CI: 0.5%–5.9%]; p = 1.000) and intraoperative perforation (0.7% [95% CI: 0.1%–4.0%] vs. 0.8% [95% CI: 0.1%–4.6%]; p = 1.000) were extremely low. These rare events precluded a meaningful multivariable regression analysis. After matching, the observed complication rates remained consistently low in both groups (Table 5). We acknowledge that with only 5 delayed bleeding events and 2 perforations, this study is underpowered to statistically demonstrate safety equivalence, and the risk of a Type II error cannot be excluded. Nevertheless, while the event counts are insufficient for definitive non-inferiority claims, the absolutely low frequencies of these complications provide valuable preliminary reference. These descriptive findings suggest a generally favorable and acceptable safety profile in both age groups, indicating no obvious catastrophic safety signals attributable to older age, though larger-scale studies are required for further validation.

Elderly VS. Non-elderly: economic outcomes

Regarding hospital stay, multivariable regression, sensitivity (≥ 75 years), and age-stratified analyses consistently showed no independent association with advanced age (all p > 0.05) (Tables 3 and 4). Although the propensity score-matched (PSM) cohort revealed a statistical difference (p = 0.010), the identical median stay (4 vs. 4 days) and marginal distributional shift (interquartile range: 4–5 vs. 3–4 days) suggest minimal clinical relevance (Table 5; Fig. 3A).

Table 3.

The economic outcomes between the two groups

	p value	β coefficient	Multivariate p value	Multivariateβ coefficient
Hospital days
Age	0.337	0.13 (-0.14,0.41)	0.492	0.10 (-0.19, 0.39)
Lesion size	0.229	0.02 (-0.01,0.05)	0.459	0.01 (-0.02, 0.04)
Lesion morphology
LST-G	0.203	0.32 (-0.17,0.81)	0.486	0.36 (-0.66,1.38)
LST-NG	0.505	0.16 (-0.30,0.62)	0.666	0.20 (-0.71,1.12)
Protruding type	0.624	0.09 (-0.28,0.47)	0.646	0.20 (-0.65,10.6)
Submucosal type	Ref	Ref	Ref	Ref
Pathological pattern	0.614	-0.03 (-0.15,0.09)	0.657	0.06 (-0.20, 0.31)
Lesion location
Left hemicolon	0.983	0.01 (-0.32,0.32)	0.806	-0.05 (-0.47,0.37)
Right hemicolon	0.335	0.18 (-0.19,0.55)	0.891	0.04 (-0.46,0.53)
Rectum	Ref	Ref	Ref	Ref
Anesthesia method	0.920	-0.03 (-0.48,0.53)	0.960	0.01 (-0.51, 0.53)
ASA grade	0.712	-0.05 (-0.33,0.23)	0.666	-0.06 (-0.35,0.22)
Hospital costs
Age	0.005	0.08 (0.02,0.14)	0.023	0.06 (0.01, 0.11)
Lesion size	< 0.001	0.02 (0.02,0.03)	< 0.001	0.02 (0.01, 0.02)
Lesion morphology
LST-G	< 0.001	0.24 (0.14,0.34)	0.024	0.21 (0.03,0.39)
LST-NG	0.002	0.15 (0.06,0.24)	0.069	0.15 (-0.01,0.31)
Protruding type	0.010	0.10 (0.02,0.17)	0.023	0.18 (0.02,0.33)
Submucosal type	Ref	Ref	Ref	Ref
Pathological pattern	0.034	-0.03 (-0.05,-0.01)	0.036	0.05 (0.01, 0.09)
Lesion location
Left hemicolon	0.991	0.01 (-0.06,0.06)	0.267	-0.04 (-0.12,0.03)
Right hemicolon	< 0.001	0.18 (0.11,0.26)	0.050	0.09 (0.01,0.18)
Rectum	Ref	Ref	Ref	Ref
Anesthesia method	0.473	-0.04 (-0.15,0.07)	0.159	-0.07 (-0.16, 0.03)
ASA grade	0.981	0.01 (-0.06,0.01)	0.760	0.01 (-0.04,0.06)

Fig. 3.

A The distribution of hospital days in each group; (B) the association between hospital costs and age in both groups

For hospital costs, initial multivariable and age-stratified models indicated a slight age-related increase (p = 0.023 and p = 0.047, respectively; Tables 3 and 5). However, this association lost significance in both the ≥ 75 years sensitivity analysis (p = 0.287) and the PSM cohort (11.39k vs. 10.57k RMB, p = 0.249). Taken together, these findings indicate that elevated costs in older adults are likely driven by underlying comorbidities and anesthesia requirements rather than chronological age itself (Fig. 3B).

Discussion

In this retrospective real-world study, we comprehensively evaluated the effect of age on the outcomes of colorectal ESD, integrating technical, safety, and economic dimensions. Existing literature remains divided regarding the suitability of ESD for older adults, with some studies suggesting comparable outcomes across age groups, while others report increased risks and reduced therapeutic benefit in elderly patients [5–13]. Our findings add clarity to this inconsistent landscape by demonstrating that elderly patients, when appropriately evaluated and selected, can undergo ESD with outcomes similar to those of younger individuals.

A key innovation of this study lies in the inclusion of operator-related outcomes, specifically resection speed and wound-management speed, which directly reflect procedural difficulty and operator burden. These metrics have rarely been used in previous age-related ESD analyses. Both multivariable regression and propensity score matching showed that elderly patients did not experience slower resection or reduced technical efficiency. These consistent findings indicate that chronological age alone does not increase the procedural challenge of colorectal ESD, even though elderly patients in our cohort presented with larger and more complex lesions. This result underscores the procedural feasibility of ESD in the elderly population. Procedural difficulty may be influenced by lesion size (β coefficient = 0.65 [0.46, 0.84], adjusted p < 0.001), which is consistent with previous reports [18–20].

Safety outcomes presented a generally favorable profile, although the statistical limitations associated with rare events must be acknowledged. Intraoperative perforation and postoperative delayed bleeding are both severe complications after ESD [21, 22]. In our study, the incidence of delayed bleeding in both the elderly and non-elderly groups was lower than that reported in the general population. This may be attributed to the fact that all ESD procedures for elderly patients were performed by highly experienced endoscopists with high annual case volumes, and meticulous hemostasis followed by prophylactic clip closure was routinely applied to the post-resection defect [23]. Moreover, all patients underwent preoperative evaluation including complete blood counts and coagulation profiles, and medications known to increase bleeding risk, such as antiplatelet agents [24] and anticoagulants [25], were appropriately withheld before the procedure. While both unadjusted and propensity-matched comparisons showed equally low complication rates in elderly and non-elderly patients, the absolute rarity of these events (only 5 delayed bleeds and 2 perforations in the total cohort) precluded meaningful multivariable regression modeling. As a result, our study is underpowered to statistically demonstrate definitive safety equivalence, and the risk of a Type II error cannot be excluded. Nevertheless, the descriptive absence of catastrophic safety signals in the elderly cohort provides valuable preliminary reference. It supports the notion that, under strict perioperative management, procedural safety is likely more strongly determined by lesion characteristics and operator expertise than by chronological age itself [18, 19, 26, 27].

Beyond technical and safety endpoints, our study incorporated economic outcomes, providing a broader assessment of ESD’s real-world value in older adults. Initially, our multivariable regression and age-stratified analyses suggested that advanced age was associated with slightly higher hospitalization costs. However, after employing propensity score matching to rigorously balance baseline characteristics—most notably ASA physical status and anesthesia methods—this cost difference was attenuated and no longer statistically significant. This represents a critical real-world clinical insight: the elevated healthcare costs often observed in elderly patients are likely driven by their underlying comorbidities, poorer baseline physical status, and the subsequent need for more intensive perioperative management (e.g., general anesthesia or advanced monitoring), rather than chronological age itself. Similarly, while PSM revealed a statistically significant difference in hospital stay, the median duration was identical (4 days) between the two groups. This marginal distributional variance lacks substantial clinical significance. Together, these findings indicate that age alone does not intrinsically impose a greater economic burden when baseline comorbidities are well-controlled.

Taken together, the findings of this study support a crucial clinical implication: the decision to perform ESD should not rely on chronological age alone. Our integrated evaluation, spanning operator performance, procedural safety, and economic burden, demonstrates that elderly patients do not experience significant inherent disadvantages across these domains. Instead, a more appropriate framework for decision-making should comprehensively evaluate functional status, comorbidities, nutritional condition, frailty, and lesion characteristics. By optimizing perioperative management for underlying diseases, the economic and safety outcomes of elderly patients can be aligned with those of younger individuals. This patient-centered approach may prevent the unnecessary exclusion of older adults who stand to benefit from ESD and promote more equitable access to organ-preserving endoscopic therapy.

Several limitations of the present study should be acknowledged. First, due to the retrospective design, we lacked granular data regarding detailed frailty assessments, specific nutritional indices, and standardized comorbidity scores, meaning the potential for residual confounding cannot be completely eliminated despite rigorous propensity score matching. Second, while our findings support the feasibility of ESD in older adults, it is crucial to acknowledge that advanced age, particularly when combined with severe comorbidities and antithrombotic therapy, can inherently elevate procedural risks. In our cohort, bleeding risks were minimized as all patients strictly discontinued anticoagulants and antiplatelets prior to the procedure. Furthermore, although we successfully adjusted for key variables such as ASA physical status and anesthesia methods, our retrospective design precluded the inclusion of several critical elderly-specific confounders. The absence of data on the Charlson Comorbidity Index, specific baseline laboratory values, and detailed frailty or nutritional indices introduces the potential for selection bias and residual confounding. Consequently, our conclusions must be interpreted with caution, as the relatively robust and highly selected nature of the elderly patients in our cohort may have contributed to the favorable outcomes observed. Therefore, future prospective, large-scale multicenter studies incorporating comprehensive geriatric assessments and focusing on octogenarians are warranted to further validate these preliminary findings.

Conclusions

In summary, our integrated analysis demonstrates that elderly patients achieve operative and safety outcomes comparable to those of younger individuals undergoing colorectal ESD. While initial unadjusted models suggested a slight age-related increase in healthcare costs, propensity-matched analyses revealed that this economic variance is likely driven by underlying comorbidities and anesthesia requirements rather than chronological age itself. Therefore, age alone should not preclude consideration of ESD. Appropriately selected older adults, particularly those with well-managed baseline conditions, can benefit from this organ-preserving technique without incurring excess procedural risk or inherently higher costs. These findings support the broader but judicious use of ESD in the growing elderly population, emphasizing the need for comprehensive patient-centered assessments over strict age cutoffs.

Abbreviations

ESD

Endoscopic submucosal dissection

PSM

Propensity score matching

LST-G

Laterally spreading tumor with granular type

LST-NG

Laterally spreading tumor with non- granular type

Authors’ contributions

LF, XHZ contributed to the study design and manuscript revision; XDZ contributed to data collection, analysis, and drafting of the manuscript. JHT contributed to collect data and edit manuscript, YLZ contributed to collect data.

Funding

None.

Data availability

The datasets are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Shanghai Minhang District Central Hospital (No.2026-035-01K). Due to the retrospective nature of the study and the use of de-identified data, the requirement for informed consent was formally waived by the Ethics Committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.