Incidence of colorectal cancer and advanced polyps by post-polypectomy surveillance interval: Real-world evidence from Thailand

Abstract

BACKGROUND

Globally, colorectal cancer (CRC) is the third most common cancer and the second leading cause of cancer-related mortality. Given the limited colonoscopy capacity in low- and middle-income countries, it is essential to identify surveillance intervals that maintain clinically acceptable CRC detection rates while using resources efficiently.

AIM

To examine the incidence of CRC and advanced polyps across different post-polypectomy surveillance intervals.

METHODS

Medical records of post-polypectomy patients at Siriraj Hospital (June 2011 to October 2023) were retrospectively reviewed. Patients were classified into low-risk and high-risk groups based on baseline colonoscopy findings, and the incidence of CRC and advanced polyps was calculated and compared across early, usual, and late surveillance intervals. Odds ratios were estimated using logistic regression, and survival analysis was performed.

RESULTS

Among 2875 included patients, 41.63% were low-risk. CRC incidence was 0.70% (2.04 cases per 1000 patient-years), whereas the incidence of CRC and advanced polyps was 12.52% (36.67 cases per 1000 patient-years). The incidence proportions of CRC and advanced polyps for early, usual, and late intervals in the high-risk group were 15.39%, 17.17%, and 17.42%, respectively. These proportions in the low-risk group were 5.63%, 7.19%, and 9.63%, respectively. The odds ratio for CRC and advanced polyps in late vs early intervals was 1.79 (95% confidence interval: 1.12-2.85; P = 0.015) in the low-risk group. Event-free survival for CRC and advanced polyps fell below 80% when the surveillance interval exceeded 8 years in the low-risk group.

CONCLUSION

Tailoring surveillance intervals improves outcomes and resource allocation. High-risk patients require surveillance every 2-3 years; low-risk patients may safely extend intervals to 7-8 years unless new symptoms arise.

Key Words: Advanced colorectal polyp; Colonoscopy; Colonoscopy surveillance; Colorectal cancer; Post-polypectomy

Core Tip: This 12-year retrospective cohort study provided real-world evidence to optimize post-polypectomy surveillance in resource-limited settings. Analyzing 2875 Thai patients, we found that high-risk patients require surveillance every 2-3 years to minimize colorectal cancer risk. While, low-risk patients may safely extend surveillance intervals to 7-8 years unless new symptoms arise. Tailoring surveillance by risk group offers a viable strategy to maximize endoscopic capacity without compromising patient safety, providing a practical framework for low- and middle-income countries to improve colorectal cancer prevention outcomes.

INTRODUCTION

Background/rationale

Globally, colorectal cancer (CRC), encompassing cancers of the colon and rectum, ranks as the third most common cancer and the second leading cause of cancer-related mortality[1]. According to World Health Organization data from 2022, incidence rates were 17.7 per 100000 men and 14.5 per 100000 women[2]. Colonoscopy is considered the gold standard for CRC screening and for surveillance after polypectomy, because it allows both lesion detection and immediate polyp removal, thereby reducing cancer incidence and mortality.

Certain high-risk groups, including individuals with a personal history of CRC or advanced adenomas, require more frequent surveillance because of their elevated risk of recurrence or progression to cancer[3]. Surveillance intervals depend on prior polyp characteristics (e.g., size, number, histology, and dysplasia grade) but also vary considerably across countries and guidelines, reflecting differences in risk stratification, healthcare resources, and cost-effectiveness. Unnecessarily frequent surveillance colonoscopies are costly, demand substantial endoscopic resources, and increase the risk of complications, whereas excessively extended intervals may lead to higher CRC incidence and decreased patient compliance.

Under the 2020 United States Multi-Society Task Force on Colorectal Cancer guidelines, patients with ≥ 10 polyps on a single examination should undergo surveillance at 1 year[4]. Those with high-risk features - ≥ 3 polyps, a polyp ≥ 10 mm, tubulovillous/villous/traditional serrated adenoma histology, or high-grade dysplasia - should undergo surveillance at 3 years. Patients without these findings may extend the interval to 7 years[4]. Thailand’s National Cancer Institute guidelines follow the same risk stratification but recommend a 5-year interval for patients without high-risk findings[5]. The 2020 European Society of Gastrointestinal Endoscopy guidelines recommend surveillance at 3 years exclusively for patients with ≥ 1 completely removed adenoma ≥ 10 mm, high-grade dysplasia, or ≥ 5 adenomas[6].

In a United Kingdom study, Cross et al[7] reported that high-risk patients had 3-year incidence rates of 15.3% for advanced adenomas, 10.0% for high-risk lesions, and 1.5% for CRC. High-risk features included ≥ 2 polyps with ≥ 1 measuring ≥ 10 mm, severe dysplasia, or ≥ 5 polyps. By contrast, low-risk patients (1-2 polyps < 10 mm) had lower rates of advanced adenomas (7.8%), high-risk lesions (3.7%), and CRC (1.1%) over the same interval[7]. Overall, current evidence supports post-polypectomy surveillance colonoscopy even for low-risk lesions[8-10]. However, whether surveillance at 5- or 10-year intervals is sufficient to effectively reduce cancer risk in these low-risk groups remains unclear.

Current Thai colonoscopy surveillance guidelines are largely adapted from international protocols. Given the limited colonoscopy capacity in low- and middle-income countries[11], it is essential to identify surveillance intervals that maintain clinically acceptable CRC detection rates while ensuring efficient use of resources. However, evidence on polyp recurrence rates, high-risk lesion detection, and CRC incidence after polypectomy in the Thai population remains limited.

Objectives

To address this gap, we aimed to examine the incidence of CRC and advanced polyps, evaluate the effect of surveillance intervals on detection rates, and propose optimal intervals for Thailand’s healthcare context.

MATERIALS AND METHODS

Study design and setting

This retrospective cohort study reviewed colonoscopy and pathology records of post-polypectomy patients who underwent surveillance colonoscopy at the Faculty of Medicine Siriraj Hospital, Mahidol University, between June 1, 2011, and October 31, 2023. The study was approved by the Ethics Committee of the Faculty of Medicine Siriraj Hospital (No. Si 104/2024) before data collection.

Patient identification

Patients undergoing post-polypectomy colonoscopy surveillance were identified from colonoscopy databases using International Classification of Diseases, Tenth Revision (ICD-10) and Ninth Revision (ICD-9) codes as well as internal procedural codes. ICD-10 search codes were K51.4 (inflammatory colonic polyps), K62 (rectal polyps and other anorectal diseases), K63 (other intestinal diseases), and D12 (benign colorectal neoplasms). ICD-9 procedural codes were 45.42 (endoscopic polypectomy of the large intestine) and 45.43 (endoscopic destruction of other large-intestinal lesions). The internal procedural code for colonoscopy (Siriraj itemized code GI38) was also included. All data were reviewed by a research assistant and verified by an internal medicine resident and a gastroenterologist.

Eligibility criteria

Adult patients (aged ≥ 18 years) who underwent ≥ 2 colonoscopy visits, with the first colonoscopy detecting and removing ≥ 1 polyp, were included. Patients were excluded if they had any of the following at baseline: Malignant colonic polyps or CRC, a history of malignant colonic polyps or CRC, familial adenomatous polyposis, Lynch syndrome, juvenile or hamartomatous polyps, Peutz-Jeghers syndrome, inflammatory bowel disease or colitis, inadequate bowel preparation, incomplete or unknown colonoscopy completeness, or missing data. Patients who underwent post-polypectomy surveillance colonoscopy ≤ 30 days after baseline and had no subsequent follow-up were also excluded because the short interval suggested a post-polypectomy complication requiring early re-colonoscopy. All consecutive patients meeting the eligibility criteria were included in the analysis.

Baseline visit and surveillance interval definitions

The baseline visit was defined as the first colonoscopy with confirmed polyp detection and complete removal. The surveillance interval was defined as the time between the baseline visit and the next colonoscopy examination.

Consolidation of multiple procedures

When patients underwent multiple colonoscopy procedures or related surgery within 30 days of either the baseline or follow-up visit, results were consolidated into a single visit using hierarchical rules. For example, if endoscopic mucosal resection was performed after incomplete polypectomy or colectomy was performed because of high-risk pathologic findings, the worst outcomes in terms of polyp size, number, and histology were used for the analysis.

Risk classification

Post-polypectomy patients were classified into low-risk and high-risk groups based on pathologic findings at the baseline visit. High-risk patients were those with ≥ 10 polyps on a single examination or any of the following baseline features. These features included 3-9 polyps, or ≥ 1 polyp with size ≥ 10 mm, high-grade dysplasia, or tubulovillous/villous/traditional serrated adenoma histology. Low-risk patients were those with 1-2 polyps on a single examination without any of these risk features.

Surveillance interval classification

In accordance with international guidelines, recommended post-polypectomy surveillance intervals are 3 years for the high-risk group and 5 years for the low-risk group. However, in clinical practice, many patients do not undergo surveillance within the recommended timeframe. Therefore, surveillance intervals were categorized into 3 groups: Early, usual, and late. Usual intervals were defined as surveillance performed within the recommended period, with an acceptable deviation of ± 3 months. Early intervals were defined as surveillance performed > 3 months before the recommended time, and late intervals as surveillance performed > 3 months after the recommended time.

Outcomes

The primary outcomes were the incidence proportions and rates of CRC and advanced polyps (defined as ≥ 10 polyps, polyp size ≥ 10 mm, high-grade dysplasia, tubulovillous/villous/traditional serrated adenoma, or CRC). The secondary outcomes were the differences in incidence proportions across the 3 surveillance interval groups (i.e., early, usual, and late).

Optimal surveillance intervals for CRC and advanced polyps were determined using predefined event-free survival thresholds of > 99% for CRC and > 80% for the composite outcome of CRC and advanced polyps. The > 99% CRC-free survival threshold was selected in accordance with international post-polypectomy surveillance guidelines that aim to maintain the incidence of interval CRC below 1%, reflecting a conservative approach to patient safety[6,12]. In addition, the > 80% event-free survival threshold for the composite outcome was chosen as a pragmatic criterion to balance the prevention of clinically significant neoplasia with the burden of excessive surveillance colonoscopy[13].

Statistical analysis

Descriptive statistics were used to summarize baseline characteristics. Continuous variables were reported as mean (SD), and categorical variables as n (%). Differences among subgroups were assessed using the independent t test or one-way analysis of variance for continuous variables and the Fisher exact test for categorical variables.

Incidence proportions (%) and rates (cases per 1000 patient-years) of CRC and advanced polyps were estimated. Survival analysis was performed to evaluate differences in event-free survival across surveillance interval groups, and Kaplan-Meier curves were used to illustrate event-free survival distributions. Univariable logistic regression was used to estimate odds ratios (ORs) with 95% confidence interval (CI) for the incidence of CRC and advanced polyps, using the early or usual surveillance group as the reference.

All tests were 2-tailed, with P < 0.05 considered statistically significant. All analyses were performed using Stata, release 17.0.

Sample size calculation

The sample size was calculated using a 2-independent-proportions formula to account for differences in incidence proportions between risk groups. Based on a previous study[7], the incidence proportion of CRC and advanced polyps among low-risk patients who underwent early or usual colonoscopy surveillance (≤ 5 years) was 12.5%. The corresponding proportion among those who underwent late surveillance (> 5 years) was 17.9%. Among high-risk patients who underwent early or usual surveillance (≤ 3 years), the incidence proportion of CRC was 1.1%, compared with 3.2% among those who underwent late surveillance (> 3 years).

With an α level of 0.05 and power of 0.80, the required sample size was 1460 participants in the low-risk group and 1682 in the high-risk group, for a total of 3142. However, the study enrolled all consecutive eligible participants during the data collection period to maximize statistical power for investigating the impact of different surveillance intervals on CRC and advanced polyp detection rates.

RESULTS

Patient selection

A total of 11332 patients underwent colonoscopy during the data collection period. After applying the exclusion criteria, 2875 post-polypectomy patients were included: 1197 (41.63%) classified as low-risk and 1678 (58.37%) as high-risk (Figure 1).

Figure 1 Patient selection and allocation flowchart based on the Strengthening the Reporting of Observational Studies in Epidemiology Guidelines of 11332 patients assessed for eligibility, 8457 were excluded (see exclusion criteria listed in the figure). The remaining 2875 post-polypectomy patients were allocated to a low-risk group (n = 1197; recommended 5-year surveillance interval) or a high-risk group (n = 1678; recommended 3-year surveillance interval). Each risk group was further subdivided by actual surveillance timing into early, usual, and late interval subgroups. CRC: Colorectal cancer.

Baseline characteristics

Overall, more than half (52.38%) of patients were female. The mean (SD) age at baseline colonoscopy was 63.6 (9.7) years and was higher in the high-risk group than in the low-risk group. In the high-risk group, most patients had 3-4 polyps (37.9%), and the most common polyp sizes were < 10 mm and 10-19 mm (39.5%). Tubular adenoma was the most common histologic type across both risk groups: 74.6% in the high-risk group and 100% in the low-risk group. Traditional serrated adenoma was the second most common type, found in 22.6% of the high-risk group. High-grade dysplasia was observed in 19.7% of high-risk patients (Table 1).

Table 1 Baseline patient and polyp characteristics by risk group.

	High-risk (3-year interval)		Low-risk (5-year interval)		P value1
	N = 1678	%	N = 1197	%
Sex					< 0.001
Women	822	49	684	57.1
Men	856	51	513	42.9
Age in years at baseline
mean (SD)	64.8 (9.5)	N/A	62.0 (9.7)	N/A	< 0.001
< 50	92	5.5	117	9.8	< 0.001
50-75	1328	76.1	971	81.1
> 75	258	15.4	109	9.1
Number of polyps					N/A
1-2	556	33.1	1197	100
3-4	636	37.9	0	0
5-10	273	16.3	0	0
> 10	213	12.7	0	0
Size of polyps in mm					N/A
< 10	637	39.5	1046	100
10-19	636	39.5	0	0
≥ 20	339	21	0	0
Adenoma histology					N/A
Tubular	1252	74.6	1197	100
Tubulovillous	2	0.1	0	0
Villous	45	2.7	0	0
Traditional serrated	379	22.6	0	0
Adenoma dysplasia grade					N/A
No/low-to-intermediate	1348	80.3	1196	100
High	330	19.7	0	0

¹Continuous variables were compared using the independent t test; categorical variables were compared using the Fisher exact test. Data are presented as n (%) unless otherwise specified.

N: Total number of patients; N/A: Not applicable.

In the low-risk group, sex distribution did not differ across surveillance intervals; however, patients in the early surveillance group had a higher baseline age (Table 2).

Table 2 Baseline characteristics of low-risk patients by surveillance interval, N = 1197.

	Early		Usual		Late		P value1
	n = 622	%	n = 139	%	n = 436	%
Sex							0.255
Women	350	56.3	73	52.5	261	59.9
Men	272	43.7	66	47.5	175	40.1
Age in years at baseline
mean (SD)	62.5 (10.5)		61.5 (7.9)		61.6 (9.1)		< 0.001
< 50	60	9.7	10	7.2	47	10.8	0.033
50-75	492	79.1	123	88.5	356	81.6
> 75	70	11.2	6	4.3	33	7.6
Number of polyps							N/A
1-2	622	100	139	100	436	100
Size of polyps in mm							N/A
< 10	622	100	139	100	436	100
Adenoma histology							N/A
Tubular	622	100	139	100	436	100
Adenoma dysplasia grade							N/A
No/low-to-intermediate	622	100	139	100	436	100

¹Continuous variables were compared using one-way analysis of variance; categorical variables were compared using the Fisher exact test. Data are presented as n (%) unless otherwise specified. All low-risk patients had 1-2 polyps < 10 mm, tubular adenoma histology, and no/low-to-intermediate dysplasia; therefore, polyp characteristics are uniform across intervals.

N: Total number of patients; n: Number of patients per subgroup; N/A: Not applicable.

Among the high-risk group, patients who underwent usual-interval surveillance were the oldest. A higher proportion of patients with larger polyps, traditional serrated adenoma histology, and high-grade dysplasia was observed in the early surveillance group (Table 3).

Table 3 Baseline characteristics of high-risk patients by surveillance interval, N = 1678.

	Early		Usual		Late		P value1
	n = 849	%	n = 201	%	n = 551	%
Sex							0.337
Women	431	50.8	95	47.3	296	47.1
Men	418	49.2	106	52.7	332	52.9
Age in years at baseline
mean (SD)	64.9 (10.2)		65.6 (8.7)		64.3 (8.7)		< 0.001
< 50	52	5.6	6	3	34	6.2	0.001
50-75	708	76.2	162	81.8	458	83.1
> 75	169	18.2	30	15.2	59	10.7
Number of polyps
1-2	373	40.1	54	27.3	129	23.4	< 0.001
3-4	301	32.4	74	37.3	261	47.4
5-9	140	15.1	34	17.2	99	18
≥ 10	115	12.4	36	18.2	62	11.2
Size of polyps in mm							< 0.001
< 10	247	27.9	82	42.3	308	57.9
10-19	361	40.7	90	46.4	185	34.8
≥ 20	278	31.4	22	11.3	39	7.3
Adenoma histology							< 0.001
Tubular	617	66.4	164	82.8	471	85.5
Tubulovillous	1	0.1	0	0	1	0.2
Villous	31	3.4	1	0.5	13	2.3
Traditional serrated	280	30.1	33	16.7	66	12
Adenoma dysplasia grade							< 0.001
No/low-to-intermediate	645	69.4	184	92.9	519	94.2
High	284	30.6	14	7.1	32	5.8

¹Continuous variables were compared using 1-way analysis of variance; categorical variables were compared using the Fisher exact test. Data are presented as n (%) unless otherwise specified.

N: Total number of patients; n: Number of patients per subgroup.

Primary outcomes

The overall incidence proportion of CRC was 0.70% (95%CI: 0.04%-10.7%), with an incidence rate of 2.04 cases per 1000 patient-years (95%CI: 1.24-3.14). The combined incidence proportion of CRC and advanced polyps was 12.52% (95%CI: 11.3%-13.8%), with an incidence rate of 36.67 cases per 1000 patient-years (95%CI: 33.04-40.58).

Secondary outcomes: CRC alone

Among high-risk patients, CRC incidence proportions were 0.54% (95%CI: 0.17%-1.25%) for early, 0.51% (95%CI: 0.01%-2.78%) for usual, and 1.45% (95%CI: 0.63%-2.84%) for late surveillance intervals. Among low-risk patients, the corresponding values were 0.16% (95%CI: 0.004%-0.89%), 0%, and 1.15% (95%CI: 0.37%-2.66%), respectively. For the high-risk group, the ORs for CRC were 0.94 (95%CI: 0.11-8.07; P = 0.954) for usual vs early and 2.72 (95%CI: 0.89-8.36; P = 0.080) for late vs early. For the low-risk group, the OR for CRC in the late vs early interval was 7.20 (95%CI: 0.84-61.88; P = 0.072; Table 4).

Table 4 Incidence of colorectal cancer by post polypectomy surveillance interval and risk group.

Interval	n (N = 2875)	Follow-up time (patient-years)	CRC cases	Incidence rate (cases per 1000 patient-years)	Incidence proportion (%)	OR1	95%CI1	P value1
High-risk (3-year interval) group
Total	1678	4555.09	14	3.07	0.83
Early	929	1131.04	5	4.42	0.54	Reference
Usual	198	590.94	1	1.69	0.51	0.94	0.11-8.07	0.954
Late	551	2833.11	8	2.82	1.45	2.72	0.89-8.36	0.080
Low-risk (5-year interval) group
Total	1197	5261.89	6	1.14	0.50
Early	622	1601.17	1	0.62	0.16	Reference
Usual	139	687.37	0	0.00	0	N/A	N/A	N/A
Late	436	2973.35	5	1.68	1.15	7.20	0.84-61.88	0.072

¹Univariable logistic regression. The early surveillance group served as the reference category. Follow-up time is measured in patient-years.

CI: Confidence interval; CRC: Colorectal cancer; N: Total number of patients; n: Number of patients per subgroup; N/A: Not applicable; OR: Odds ratio.

Secondary outcomes: CRC and advanced polyps combined

Among high-risk patients, incidence proportions of CRC and advanced polyps for early, usual, and late surveillance intervals were 15.39% (95%CI: 13.13%-17.88%), 17.17% (95%CI: 12.19%-23.16%), and 17.42% (95%CI: 14.34%-20.85%), respectively. Among low-risk patients, the corresponding values were 5.63% (95%CI: 3.95%-7.74%), 7.19% (95%CI: 3.50%-12.83%), and 9.63% (95%CI: 7.03%-12.80%), respectively.

In the high-risk group, the OR for CRC and advanced polyps was 1.14 (95%CI: 0.76-1.72; P = 0.532) for usual vs early. The corresponding ORs were 1.16 (95%CI: 0.87-1.54; P = 0.305) for late vs early and 1.02 (95%CI: 0.66-1.56; P = 0.936) for late vs usual. In the low-risk group, the OR was 1.30 (95%CI: 0.63-2.69; P = 0.480) for usual vs early. The ORs were 1.79 (95%CI: 1.12-2.85; P = 0.015) for late vs early and 1.38 (95%CI: 0.67-2.82; P = 0.384) for late vs usual (Table 5).

Table 5 Incidence of colorectal cancer and advanced polyps by post polypectomy surveillance interval and risk group.

Interval	n (N = 2875)	Follow-up time (patient-years)	CRC and advanced polyp cases	Incidence rate (cases per 1000 person-years)	Incidence proportion (%)	OR1	95%CI1	P value1
High-risk (3-year interval) group
Total	1678	4555.09	273	59.93	16.27
Early	929	1131.04	143	126.43	15.39	Reference
Usual	198	590.94	34	57.54	17.17	1.14	0.76-1.72	0.532
Late	551	2833.11	96	33.89	17.42	1.16	0.87-1.54	0.305
Late vs usual						1.02	0.66-1.56	0.936
Low-risk (5-year interval) group
Total	1197	5261.89	87	16.53	7.27
Early	622	1601.17	35	21.86	5.63	Reference
Usual	139	687.37	10	14.55	7.19	1.30	0.63-2.69	0.480
Late	436	2973.35	42	14.13	9.63	1.79	1.12-2.85	0.015
Late vs usual						1.38	0.67-2.82	0.384

¹Univariable logistic regression. The early surveillance group served as the reference category unless otherwise specified (late vs usual comparison is also shown). Follow-up time is measured in patient-years.

CI: Confidence interval; CRC: Colorectal cancer; N: Total number of patients; n: Number of patients per subgroup; OR: Odds ratio.

Event-free survival for CRC

Based on guideline-recommended intervals (3 years for the high-risk group, 5 years for the low-risk group), the CRC survival rate was 99.55% (95%CI: 98.93%-99.81%) in the high-risk group. In the low-risk group, the rate was 99.90% (95%CI: 99.32%-99.99%).

Proposed optimal surveillance intervals for CRC

Using a CRC survival threshold of > 99%, optimal surveillance intervals were determined from Thai patient data. For the high-risk group, surveillance at 2-3 years yielded a CRC survival rate of 99.55% (95%CI: 98.93%-99.81%). Extending the interval to 3-4 years reduced the rate to 98.80% (95%CI: 97.61%-99.40%). For the low-risk group, surveillance at 5-6 years yielded a CRC survival rate of 99.90% (95%CI: 99.32%-99.99%). Extending the interval to 6-7 years reduced the rate to 97.95% (95%CI: 94.84%-99.19%; Figure 2A).

Figure 2 Kaplan-Meier estimates by risk group. A: Kaplan-Meier estimates of colorectal cancer-free survival by risk group. Kaplan-Meier survival curves illustrate the probability of remaining free of colorectal cancer (CRC) over time (years) for the high-risk group (blue line; recommended 3-year surveillance interval) and the low-risk group (red line; recommended 5-year surveillance interval). Both groups maintained CRC-free survival rates above 99% within guideline-recommended intervals. The high-risk group showed a gradual decline beyond 7 years, whereas the low-risk group remained relatively stable throughout the follow-up period; B: Kaplan-Meier estimates of event-free survival for CRC and advanced polyps by risk group. Kaplan-Meier survival curves illustrate the probability of remaining free of both CRC and advanced polyps over time (years) for the high-risk group (blue line; recommended 3-year surveillance interval) and the low-risk group (red line; recommended 5-year surveillance interval). Event-free survival declined more steeply in the high-risk group, falling below 80% at approximately 3-4 years. The low-risk group maintained event-free survival above 80% for up to 7-8 years, with a more gradual decline thereafter. The marked separation between curves supports risk-stratified surveillance interval recommendations.

Event-free survival for CRC and advanced polyps

Based on the same guideline-recommended intervals, the event-free survival rate for CRC and advanced polyps was 86.85% (95%CI: 84.74%-88.68%) at 2-3 years in the high-risk group. In the low-risk group, the rate was 94.43% (95%CI: 92.50%-95.87%) at 4-5 years.

Proposed optimal surveillance intervals for CRC and advanced polyps

Using a survival threshold of > 80% for CRC and advanced polyps, optimal intervals were similarly derived. In the high-risk group, surveillance at 2-3 years yielded a survival rate of 86.85% (95%CI: 84.74%-88.68%). Extending the interval to 3-4 years reduced the rate to 79.44% (95%CI: 76.58%-82.00%), falling below the threshold. In the low-risk group, the interval could potentially be extended to 7-8 years, at which the survival rate was 80.38% (95%CI: 74.71%-84.92%). Extending the interval further to 8-9 years reduced the rate to 76.84% (95%CI: 69.87%-82.40%; Figure 2B).

DISCUSSION

This study investigated the incidence of CRC and advanced polyps and their association with different post-polypectomy surveillance intervals. It aimed to determine optimal surveillance timing within the Thai healthcare context, which is representative of the challenges faced by low- and middle-income countries.

Among 2875 patients followed over 12 years, the overall CRC incidence was 0.70% (95%CI: 0.04%-10.7%), and the combined incidence of CRC and advanced polyps was 12.52% (95%CI: 11.3%-13.8%). When early, usual, and late surveillance intervals were compared, CRC incidence did not differ across intervals in either risk group. However, in the low-risk group, delayed surveillance was associated with a higher risk of CRC and advanced polyps, whereas no such difference was observed in the high-risk group.

Optimal surveillance intervals were evaluated against current guideline recommendations (3 years for high-risk, 5 years for low-risk)[4,6]. To maintain a CRC survival rate exceeding 99%, surveillance within 3 years for the high-risk group and within 6 years for the low-risk group was required. For the composite outcome of CRC and advanced polyps, a target survival rate of > 80% was applied. This required surveillance within 3 years for the high-risk group and within 8 years for the low-risk group.

We compared our findings with several studies evaluating the incidence of CRC and advanced polyps across risk groups and surveillance intervals. A meta-analysis of > 800000 participants from 14 studies reported a CRC incidence of 0.63% in high-risk patients at 3 years, and 0.37% and 0.67% in low-risk patients at 5 and 10 years, respectively[14]. Three retrospective cohort studies from England reported a CRC incidence of 0.5%-1.5% (1.4-3.6 per 1000 person-years) and an advanced adenoma incidence of 7.8%-15.3%[7,15,16]. Another meta-analysis of > 1000000 participants from 13 studies found a CRC incidence of 0.52 per 1000 person-years[17]. Cohort studies from various countries also reported a wide range of CRC incidence: Poland, 0.46 per 1000 person-years[18]; Sweden, 1.6%-8.6%[19]; Japan, 0.34-0.46 per 1000 person-years[20]; and South Korea, 0.45-0.54 per 1000 person-years[21]. A cohort study from Australia found an advanced polyp incidence of 19% at 5 years[22].

In comparison, our study found an overall CRC incidence of 0.70% (2.04 cases per 1000 patient-years) and a combined incidence of CRC and advanced polyps of 12.52% (36.67 cases per 1000 patient-years). These findings are consistent with global evidence, suggesting that the natural history of CRC in Thailand is comparable to that reported in multinational data from various countries.

However, in resource-limited settings, adopting surveillance intervals based on data from disparate populations may not be appropriate for the Thai context. Moreover, the classification of patients into low-risk and high-risk groups varies in definition and terminology across countries and time periods[4,6,23-26].

Among high-risk patients, CRC incidence gradually increased from early to late surveillance intervals. The incidence of CRC and advanced polyps also rose with longer intervals, although no statistically significant difference in ORs was observed. Despite the lack of statistical significance, these trends suggest a potential increase in risk with delayed surveillance. Based on our findings, surveillance in the high-risk group should be conducted within guideline-recommended intervals. Early-to-usual surveillance intervals should be prioritized to minimize the risk of missed CRC and advanced polyps across all risk groups. Extended intervals may increase the risk of CRC and advanced polyps; however, in the low-risk group, modest postponements could be acceptable when resources are limited.

In the low-risk group, CRC-free survival rates remained high for up to 6 years, and the combined survival free of CRC and advanced polyps remained stable for up to 8 years. These findings suggest that extending the surveillance interval to 7-8 years may be a safe and viable strategy for low-risk individuals in this population.

Consistent with our findings, a retrospective English cohort study reported that relative CRC risk increased from 1.54% to 4.44% when high-risk patients extended surveillance from 3 years to 4 years, compared with < 18 months[7]. A retrospective cohort study in Japan similarly found a higher CRC incidence among patients with advanced adenomas at baseline, with a hazard ratio of 3.49 for intervals > 3 years[27]. Conversely, among low-risk patients, no statistically significant difference in CRC risk was observed over surveillance intervals of up to 6 years.

Together with existing evidence, our findings support the feasibility of extended surveillance for low-risk patients, offering a practical solution for resource-constrained healthcare systems. Such an approach may help reduce unnecessary procedures, procedure-related burden, and healthcare resource utilization. Our results are generally consistent with international recommendations, including those from the United States Multi-Society Task Force on Colorectal Cancer[4] and the European Society of Gastrointestinal Endoscopy[6], in which high-risk patients are advised to undergo surveillance colonoscopy at 3 years, whereas low-risk patients may undergo surveillance at longer intervals ranging from 5 years to 10 years[28,29]. Nevertheless, surveillance strategies should remain individualized, and earlier colonoscopy should be considered when new symptoms, additional risk factors, or concerns regarding baseline examination quality are present.

Notably, CRC diagnosed in patients who underwent early surveillance might be categorized as interval CRC; however, we chose to use the broader term - incident CRC - rather than interval CRC to encompass CRC detected during both scheduled and unscheduled surveillance.

This large retrospective cohort study addresses a critical evidence gap by establishing optimal post-polypectomy surveillance intervals for the Thai healthcare system and similar middle-income settings. The study also features a follow-up period of > 12 years, which we consider sufficient to evaluate CRC incidence given the typical adenoma-to-carcinoma development timeframe of approximately 10 years[30]. We conducted a risk-stratified analysis encompassing all variations in surveillance intervals, enabling examination of CRC and advanced polyp incidence across intervals reflecting real-world settings where patients may not adhere to recommended schedules. Our study further enhances clinical relevance by reporting not only CRC incidence but also the incidence of advanced polyps, which are precancerous lesions that may progress to CRC.

The comprehensive review of electronic hospital records allowed for accurate outcome assessment and the recommendation of surveillance intervals relevant to the Thai population. Only a small amount of missing data was observed, enhancing the reliability of our findings.

Our findings offer practical, policy-relevant guidance that is applicable to current clinical practice and provides direction for determining optimal surveillance intervals in resource-limited settings.

Several limitations should be considered. First, this retrospective cohort design inherently involves differences in population characteristics and the potential for unmeasured confounders and bias. However, because long-term follow-up is needed to assess the effect of surveillance intervals on CRC incidence, alternative designs such as randomized controlled trials or prospective cohort studies may be impractical. The retrospective design also precluded the collection of all potential confounders, such as comorbidities, bowel preparation quality, and colonoscopy quality, limiting the analysis to univariable logistic regression. Nevertheless, the study was conducted at the largest university hospital in Thailand, where colonoscopy procedures were performed under standardized protocols throughout the study period. All procedures were carried out either by experienced endoscopists or by gastroenterology fellows under the direct supervision of expert staff, which should minimize variability in colonoscopy quality over the 12-year study period. However, because formal quality measures were not consistently available, their potential influence on surveillance outcomes could not be fully assessed. Future studies should consider including prior comorbidities and other known confounding factors to establish more comprehensive findings.

Second, reliance on ICD-10 and ICD-9 codes may have resulted in incomplete case identification because of coding omissions or absent procedural documentation. Because our database did not capture colonoscopies performed at outside institutions, the findings may not fully reflect real-world clinical practice. Future multicenter studies with integrated registries are needed to improve data completeness and generalizability.

Third, selection and information bias are possible in this retrospective cohort. High-risk patients may have been prioritized for early surveillance on the basis of risk stratification tools. In addition, as this was a single-center study, we were unable to retrieve data from other institutions, which may have led to incomplete ascertainment of incident CRC or advanced polyps detected during repeat colonoscopies performed elsewhere. However, we conducted a comprehensive review of all available medical records and found no documentation indicating that any included patients underwent colonoscopy at other institutions. The proportion of incomplete or missing data was modest, but it may nonetheless have led to risk misclassification and subsequent deviations in observed clinical outcomes.

Finally, although all consecutive patients were included over a 12-year period, the low-risk group did not meet the projected sample size required to detect statistically significant differences in CRC and advanced polyp incidence. Because the observed incidence was lower than the estimates used in our initial sample size calculations, the study may be underpowered for certain outcome comparisons. Indeed, a post hoc power analysis of the low-risk group yielded a power of 63.1% (at an α level of 0.05), falling below the conventional 80% threshold. Therefore, the lack of statistically significant differences should be attributed to this limited sample size rather than interpreted as evidence of no difference. Despite this limited power, our findings suggest that low-risk patients may safely extend surveillance intervals to 7-8 years, given that the survival rate free of CRC and advanced polyps remained above the 80% threshold. Although our results provide valuable point estimates, larger multicenter cohorts are recommended to achieve the statistical power necessary for definitive subgroup analyses.

CONCLUSION

In low- and middle-income settings, surveillance strategies should be tailored by risk group to optimize limited colonoscopy capacity. Based on our findings, a surveillance colonoscopy interval of 2-3 years is recommended for the high-risk group. Low-risk patients can tolerate extended surveillance intervals of up to 7-8 years unless new symptoms arise.

ACKNOWLEDGEMENTS

We express our gratitude to the Division of Ambulatory Medicine, Department of Medicine, Faculty of Medicine Siriraj Hospital, Mahidol University, for their support. The authors also thank David Park for English-language editing.