Socioeconomic disparities in survival of patients with superficial colorectal cancer

Abstract

Background

Limited data are available regarding the impact of socioeconomic disparities on the survival of patients with superficial colorectal cancer (SCRC).

Methods

We used population-based data from the U.S. Surveillance, Epidemiology, and End Results (SEER) program registry to conduct a retrospective cohort study to examine the impact of gender, race, geographic location, and socioeconomic factors on survival outcomes in SCRC patients. Additionally, we explored how socioeconomic factors influenced survival in SCRC patients when stratified by age at diagnosis. To assess the influence of socioeconomic factors on overall survival (OS) and cancer-specific survival (CSS), we employed the Cox proportional hazards model, adjusting for relevant clinical characteristics.

Results

50,671 SCRC patients were included in the final analysis. Compared to patients aged 20–39, older age groups showed progressively higher hazards of CSS, with adjusted hazard ratios (AHR) of 1.21 (95%CI 0.85–1.72) for the 40–49 age group, 1.45 (95%CI 1.05-2.00) for the 50–59 age group, 1.97 (95%CI 1.43–2.70) for the 60–69 age group, and 4.25 (95%CI 3.11–5.82) for those aged 70 or older. Statistically significant increases in CSS hazard were observed in the 50–59, 60–69 and 70 years or older age groups. The AHR for OS among patients aged 40–49, 50–59, 60–69, and 70 or older was 1.54 (95%CI 1.18–2.01), 2.24 (95%CI 1.75–2.86), 4.07 (95%CI 3.20–5.18), and 10.98 (95%CI 8.65–13.94), respectively. Female patients had higher 5-year CSS (by 0.80% points) and OS (by 2.50% points) rates compared to male patients. Black patients showed comparable CSS to White patients. However, their OS was significantly lower (AHR = 1.08, 95% CI 1.02–1.14), corresponding to a 1.70-percentage-point reduction in the 5-year OS rate. Furthermore, among patients aged 50–59, the hazard of death was elevated for those who were unmarried (AHR = 1.90, 95% CI 1.68–2.15) or had low income (AHR = 2.27, 95% CI: 1.62–3.18) alone, but it was markedly higher for those possessing both risk factors (AHR = 3.63, 95% CI 2.17–6.10). Rural residents also demonstrated a survival advantage, particularly in the 40–49 age group.

Conclusions

This study found significant socioeconomic disparities in SCRC survival, with older age, male gender, Black race, unmarried status, and lower income associated with poorer outcomes. Moreover, we identified a critical intersectional pattern in which marital status and income were jointly correlated with inferior survival outcomes. Potential strategies including expanded screening coverage, optimized medical care accessibility and culturally adapted health policies are proposed as future research directions, which may help alleviate clinical disparities and facilitate health equity in SCRC management.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-026-05726-2.

Introduction

Superficial colorectal cancer (SCRC), characterized by tumor invasion limited to the mucosa or submucosa, represents a critical window for early intervention in colorectal cancer (CRC) management. Epidemiological studies indicate that SCRC accounts for approximately 15–25% of all newly diagnosed CRC cases [1, 2]. When detected and treated early, the 5-year cancer-specific survival rate exceeds 95%, which underscores the potential for curative treatment and the importance of screening and accessible care [3, 4].

Despite this generally favorable prognosis, emerging evidence suggests that not all patient groups benefit equally from advances in early detection and minimally invasive treatment. Significant disparities in CRC outcomes have been documented across socioeconomic and demographic groups. In the United States, differences in incidence and survival are observed by age, gender, race, and geographic region [5]. Notably, Black patients had a higher hazard of mortality compared to White patients, even when diagnosed at early stages [6]. Additionally, males experience a 40% higher incidence and mortality rate compared to females [7]. Socioeconomic status (SES) further exacerbates these disparities. Studies consistently show that individuals from lower SES backgrounds face delayed diagnoses, reduced access to advanced endoscopic techniques, and poorer survival outcomes [8–10].

However, most studies examining these disparities have focused on CRC collectively or on advanced-stage disease. It remains unclear to what extent socioeconomic factors influence outcomes in early-stage SCRC. This is a critical question for a highly curable disease, where such disparities would be unexpected and would represent a critical inequity. Although gender and age are known to influence incidence, their interaction with socioeconomic variables in relation to survival in SCRC populations is poorly characterized. Moreover, while biological mechanisms may contribute to outcome differences, systemic barriers related to access, care continuity, and social support likely play crucial roles.

The Surveillance, Epidemiology, and End Results (SEER) database provides a platform for analyzing the impact of patient- and disease-related factors on clinical outcomes. In this study, we aimed to investigate the association between socioeconomic factors and survival outcomes in a cohort of patients with SCRC. Crucially, we examined whether known disparities persist after adjustment for tumor characteristics and whether these inequities vary across age groups. By focusing on a population with a high expected survival rate, we sought to identify disparities that may reflect systemic inequities beyond disease biology, thereby providing a stronger rationale for targeted interventions to achieve equity in early cancer care.

Methods

Data source

The Surveillance, Epidemiology, and End Results (SEER) program systematically collects comprehensive cancer incidence data through a network of population-based cancer registries. In this retrospective cohort study, we analyzed data exclusively from the SEER 17-registry database (November 2023 submission), representing a nationally representative sample covering 45.9% of the U.S. population [11]. The included registries encompass diverse geographic regions (San Francisco-Oakland, Connecticut, Detroit, Hawaii, Iowa, New Mexico, Seattle-Puget Sound, Utah, Atlanta, Alaska Natives, San Jose-Monterey, Los Angeles, Greater California, Kentucky, Louisiana, New Jersey, and Greater Georgia), ensuring broad socioeconomic and demographic representation. Our study focused on patients diagnosed with SCRC between 2007 and 2021, with all treatment patterns, survival outcomes, and socioeconomic variables captured within the U.S. healthcare context. The SEER database contained de-identified patient information, which was exempt from Institutional Review Board (IRB) approval requirements and did not necessitate obtaining informed consent.

Study population

To avoid confounding by advanced disease and to align with our aim of investigating the association between socioeconomic factors and survival outcomes in early-stage SCRC, we focused on a cohort of SCRC patients with Stage 0 (Tis N0 M0) or Stage I (T1 N0 M0) disease. This corresponded to tumors confined to the mucosa or submucosa without lymph node involvement or distant metastasis. Stage assignment prioritized pathological staging when available, using AJCC 6th (2007–2009), 7th (2010–2017), and 8th (2018 onward) edition T/N/M variables, respectively. T1N0M0 was defined as T1, N0, and M0 according to the corresponding AJCC edition. As the AJCC T variables do not uniformly include a ‘Tis’ code across editions, TisN0M0 cases were identified via CS_EXTENSION in situ codes combined with AJCC N0 and M0. CS staging variables were supplemented when original AJCC staging data were missing. Clinical stage from CS variables was adopted for patients without surgical resection, and a sensitivity analysis excluding this population was performed to confirm result robustness.

For patient inclusion in the study, we considered those who were diagnosed with SCRC between 2007 and 2021 and were aged 20 or older. Additionally, SCRC had to be the first primary tumor diagnosed in these patients. Furthermore, the diagnoses needed to be histologically confirmed, which was determined by specific International Classification of Diseases for Oncology, Third Edition (ICD-O-3) site and primary site codes (C18.0-C18.9, C19.9, and C20.9). Regarding patient exclusion, we did not include individuals with multiple primary tumors. Patients were also excluded if information on their age, race, income bracket, and area of residence was unknown. Moreover, those with missing data on the presence of another malignancy, treatment details, follow-up information, or mortality variables were not part of the study. A flow diagram for patient selection was shown in Fig. 1.

Fig. 1.

Flow diagram for selection of the study patients

Study outcomes and covariates

The primary outcomes of interest were the five-year Overall Survival (OS) and Cancer-Specific Survival (CSS) rates, with Hazard Ratios (HRs) adjusted to evaluate the influence of socioeconomic factors on survival and identify existing disparities. OS was defined as the interval from diagnosis to death from any cause. The primary outcome of this study was CSS, defined as the time from diagnosis to cancer-related death. Non-cancer-related deaths were right-censored in the analysis. OS was considered a secondary outcome. We examined the following variables: age group, gender, race, marital status, household income, and residential area. Age was categorized into brackets (20–39, 40–49, 50–59, 60–69, and 70 years or older). Notably, this stratification was informed by distinct clinical and methodological considerations across the age spectrum. Specifically, the 20–39 age group represented a distinct subset of Early-Onset CRC (EOCRC), often characterized by diagnostic delays and a more aggressive disease biology. The 40–49-year group represented the newly screening-eligible cohort, allowing assessment of early screening impact. In contrast, the older strata (50–59, 60–69, and ≥ 70 years) accounted for substantial shifts in comorbidity burden and treatment tolerance. Crucially, this granular approach enabled us to model the non-linear epidemiology of CRC and to precisely examine how socioeconomic factors differentially impact survival across these clinically distinct phases of life. Race and ethnicity were classified according to the SEER recode as White, Black, Asian/Pacific Islander, and American Indian/Alaska Native. Marital status encompassed married individuals and those who were unmarried (including separated, divorced, never married, or widowed). The median household income variable, as obtained from the SEER database, was reported in inflation-adjusted dollars to ensure comparability across the study period (2007–2021). These adjustments were performed by SEER using the American Community Survey (ACS) data and were standardized to the year of 2022 [12]. Income was analyzed in increments of ten thousand dollars, ranging from below US$40,000 to over US$70,000. Residential areas were classified as urban or rural, with urban areas further subdivided based on population size and rural areas categorized by their proximity to urban centers. It’s important to note that all the socioeconomic data used in this study were exclusively sourced from the U.S. SEER registry data. These data offer a reflection of the economic conditions within the U.S. healthcare system.

Statistical analysis

The data for this study were extracted through the SEER*Stat software (version 8.4.5) and subsequently analyzed using R statistical software (version 3.6.1). Categorical variables were presented in terms of frequencies and percentages. Survival rates were estimated using the Kaplan-Meier method, with 5-year OS and CSS rates calculated for each subgroup. Between-group differences were assessed for statistical significance using log-rank tests, and these rates were reported alongside 95% confidence intervals (CIs). Adjusted hazard ratios (AHR) were derived from Cox proportional hazards regression models to evaluate the impact of socioeconomic and clinical factors on survival. We assessed the proportional hazards (PH) assumption by visually inspecting Kaplan-Meier (KM) curves, log (− log) plots, and testing scaled Schoenfeld residuals [13]. After assessment, all variables satisfied the PH assumption. Age-stratified analyses were performed due to the known biological differences in tumor behavior across age groups and differential socioeconomic influences across the lifespan. Patients were categorized into age strata (20–39, 40–49, 50–59, 60–69, ≥ 70 years), and stratum-specific Cox models were used to examine whether the association between socioeconomic variables and survival differed significantly by age. All statistical tests were 2-sided, and only a P value less than 0.05 was considered statistically significant.

Results

Patient characteristics

From an initial pool of 540,240 patients with CRC aged 20 years or older in the SEER database (2007–2021), we identified a final analytic cohort of 50,671 patients after applying the inclusion and exclusion criteria. Primary exclusions involved non Tis or T1 or missing (N = 339,885), unknown N stage or ≥N1(N = 7,277), unknown M stage or ≥M1 (N = 134,980), unknown race (N = 5,291), unknown population of the residential area (N = 802), and missing follow-up information or a follow-up period of less than 1 month (N = 1,334). The 81,511 cases were excluded due to missing T-stage information. We compared key baseline characteristics (including age, gender, race, marital status, area-level median household income and population of the residential area) between the final analytic cohort (n = 50,671) and the cohort excluded due to missing T-stage (n = 81,511). The comparison showed no clinically meaningful differences in these baseline characteristics between the two groups. In the final cohort, 36,404 (71.84%) were right-censored for the OS analysis and 46,463 (91.70%) were right-censored in the CSS analysis.

The demographic, socioeconomic, and clinical characteristics of SCRC patients were presented in Table 1. Specifically, 45,155 patients (89.12%) were aged 50 years or older; 26,601 (52.50%) were male; 26,114 (51.54%) were married; and 39,817 (78.58%) identified as White. Notably, 6,462 SCRC patients (12.75%) had a tumor size greater than 3 cm. In terms of disease stage, 4,712 patients (9.30%) were at Tis stage, while 45,959 patients (90.70%) were at T1 stage. 80.56% of the patients had well-differentiated or moderately differentiated tumors. Additionally, 65.19% of the patients resided in counties where the median household income exceeded US$70,000. And 57.03% of the patients resided in counties within metropolitan areas with a population of 1 million or more. Regarding treatment patterns, 47,396 patients (93.54%) underwent primary surgical resection, while 2,671 (5.27%) received chemotherapy (any sequence), and 2,264 (4.47%) received radiotherapy (any sequence) during their treatment course.

Table 1.

Demographic, socioeconomic and clinical characteristics of SCRC patients

Variable	N=50,671	%
Age (years old)
20-39	2019	3.98
40-49	3497	6.90
50-59	11,912	23.51
60-69	13,681	27.00
70 or older	19,562	38.61
Gender
Male	26,601	52.50
Female	24,070	47.50
Race
White	39,817	78.58
Black	5768	11.38
Asian/Pacific Islander	4797	9.47
American Indian/Alaska Native	289	0.57
Marital status
Married	26,114	51.54
Unmarried	17,504	34.54
Unknown	7053	13.92
Tumor size (cm)
≤1	19,520	38.52
1-2	11,509	22.71
2-3	6153	12.14
3-4	2884	5.69
>4	3578	7.06
Unknown	7027	13.88
T score
Tis	4712	9.30
T1	45,959	90.70
Histologic grade
Well differentiated	13,921	27.47
Moderately differentiated	26,902	53.09
Poorly differentiated	2481	4.90
Undifferentiated	314	0.62
Unknown	7053	13.92
Area-level median household income(2022 US dollars)
< 40,000	913	1.80
40,000–49,999	3122	6.16
50,000–59,999	5190	10.24
60,000–69,999	8415	16.61
> 70,000	33,031	65.19
Population of the residential area
Urban > 1,000,000	28,899	57.03
Urban 250,000–1,000,000	11,087	21.88
Urban < 250,000	3970	7.83
Rural adjacent to urban area	3792	7.48
Rural not adjacent to urban	2923	5.78
Surgical treatment
Yes	47,396	93.54
No/Unknown	3275	6.46
Chemotherapy
Yes	2671	5.27
No/Unknown	48,000	94.73
Radiotherapy
Yes	2264	4.47
No/Unknown	48,407	95.53

Socioeconomic disparities in survival of SCRC patients

Female patients demonstrated higher 5-year CSS and OS rates compared to male patients, with absolute differences of 0.8% points and 2.5% points, respectively (Table 2; Fig. 2A and B).

Table 2.

5-year survival of SCRC patients based on demographic, socioeconomic characteristics

	Overall survival	Cancer specific survival
	5-year survival rate (%), [95% CI]	5-year survival rate (%), [95% CI]
Age (years old)
20-39	96.70 [95.50-97.90]	97.90 [96.90-98.80]
40-49	93.50 [92.50-94.60]	96.90 [96.10-97.60]
50-59	91.90 [91.30-92.50]	96.80 [96.40-97.20]
60-69	85.70 [84.90-86.40]	95.30 [94.80-95.70]
70 or older	65.50[64.70-66.40]	88.30 [87.70-88.80]
Gender
Male	78.80 [78.20-79.40]	92.80 [92.50-93.20]
Female	81.30 [80.60-81.90]	93.60 [93.20-94.00]
Race
White	79.50 [79.00-80.00]	93.20 [92.90-93.50]
Black	77.80 [76.50-79.20]	92.10 [91.20-93.00]
Asian/Pacific Islander	86.40 [85.20-87.70]	94.80 [94.00-95.70]
American Indian/Alaska Native	84.48 [79.34-89.96]	92.74 [88.79-96.86]
Marital status
Married	83.90 [83.30-84.40]	94.80 [94.50-95.10]
Unmarried	74.00 [73.30-74.80]	90.70 [90.20-91.20]
Area-level median household income (2022 US dollars)
< 40,000	75.70 [72.38-79.17]	89.87 [87.43-92.37]
40,000–49,999	74.70 [72.80-76.60]	91.50 [90.30-92.80]
50,000–59,999	75.90 [74.50-77.40]	91.50 [90.50-92.40]
60,000–69,999	78.30 [77.20-79.30]	92.20 [91.50-92.90]
> 70,000	81.60 [81.10-82.10]	94.00 [93.60-94.30]
Population of the residential area
Urban > 1,000,000	80.70 [80.10-81.20]	93.40 [93.00-93.70]
Urban 250,000–1,000,000	80.30 [79.40-81.20]	93.90 [93.30-94.40]
Urban < 250,000	79.50 [77.90-81.00]	92.30 [91.30-93.30]
Rural adjacent to urban area	75.90 [74.20-77.60]	91.50 [90.40-92.60]
Rural not adjacent to urban	77.20 [75.40-79.20]	92.10 [90.90-93.40]

Fig. 2.

OS and CSS of patients with SCRC based on gender, age group, race, marital status, household income and area of residence (A OS by gender; B CSS by gender; C OS by age group; D CSS by age group; E OS by race; F CSS by race; G OS by marital status; H CSS by marital status; I OS by area-level median household income(2022 US dollars); J CSS by area-level median household income(2022 US dollars); K OS by area of residence; L CSS by area of residence)

Compared to patients aged 20–39, older age groups showed progressively higher hazards of CSS, with AHR of 1.21 (95%CI 0.85–1.72) for the 40–49 age group, 1.45 (95%CI 1.05-2.00) for the 50–59 age group, 1.97 (95%CI 1.43–2.70) for the 60–69 age group, and 4.25 (95%CI 3.11–5.82) for those aged 70 or older. Statistically significant increases in CSS hazard were observed in the 50–59, 60–69 and ≥ 70 age groups. For OS, AHRs were 1.54 (95% CI 1.18–2.01) for ages 40–49, 2.24 (95% CI 1.75–2.86) for ages 50–59, 4.07 (95% CI 3.20–5.18) for ages 60–69, and 10.98 (95% CI 8.65–13.94) for patients aged 70 years or older (Table 3; Fig. 2C and D).

Table 3.

Cox regression analysis of overall and cancer-specific survival in patients with SCRC

Variable	Overall survival		Cancer specific survival
	AHR (95% CI)	p-value	AHR (95% CI)	p-value
Age (years old)		< 0.001		< 0.001
20-39	Reference		Reference
40-49	1.54(1.18-2.01)	0.001	1.21(0.85-1.72)	0.298
50-59	2.24(1.75-2.86)	< 0.001	1.45(1.05-2.00)	0.024
60-69	4.07(3.20-5.18)	< 0.001	1.97(1.43-2.70)	< 0.001
70 or older	10.98(8.65-13.94)	< 0.001	4.25(3.11-5.82)	< 0.001
Gender		< 0.001		< 0.001
Male	Reference		Reference
Female	0.77(0.75-0.80)		0.83(0.78-0.88)
Race		< 0.001		0.097
White	Reference		Reference
Black	1.08(1.02-1.14)	0.005	1.07(0.97-1.18)	0.160
Asian/Pacific Islander	0.79(0.74-0.84)	< 0.001	0.89(0.79-1.00)	0.052
American Indian/Alaska Native	0.86(0.68-1.09)	0.217	1.01(0.68-1.49)	0.971
Marital status		< 0.001		< 0.001
Married	Reference		Reference
Unmarried	1.48(1.42-1.54)	< 0.001	1.44(1.34-1.55)	< 0.001
Unknown	0.47(0.19-1.13)	0.091	0.24(0.03-1.68)	0.148
Tumor size (cm)		< 0.001		< 0.001
≤1	Reference		Reference
1-2	1.07(1.02-1.13)	0.003	1.19(1.07-1.31)	0.001
2-3	1.19(1.12-1.25)	< 0.001	1.31(1.17-1.47)	< 0.001
3-4	1.32(1.23-1.41)	< 0.001	1.65(1.45-1.88)	< 0.001
>4	1.36(1.28-1.45)	< 0.001	2.04(1.83-2.28)	< 0.001
Unknown	3.46(1.44-8.35)	0.006	11.48(1.61-81.88)	0.015
T score		0.312		0.038
Tis	Reference		Reference
T1	1.03(0.97-1.09)		1.12(1.01-1.25)
Histologic grade		< 0.001		< 0.001
Well differentiated	Reference		Reference
Moderately differentiated	1.06(1.01-1.11)	0.015	1.24(1.13-1.37)	< 0.001
Poorly differentiated	1.21(1.12-1.31)	< 0.001	1.62(1.41-1.87)	< 0.001
Undifferentiated	1.28(1.06-1.53)	0.009	1.47(1.04-2.09)	0.031
Unknown	0.47(0.19-1.13)	0.091	0.24(0.03-1.68)	0.148
Area-level median household income(2022 US dollars)		< 0.001		< 0.001
> 70,000	Reference		Reference
< 40,000	1.37(1.21-1.56)	< 0.001	1.52(1.23-1.88)	< 0.001
40,000–49,999	1.20(1.11-1.30)	< 0.001	1.14(0.99-1.31)	0.078
50,000–59,999	1.26(1.18-1.33)	< 0.001	1.28(1.15-1.44)	< 0.001
60,000–69,999	1.15(1.09-1.20)	< 0.001	1.23(1.13-1.33)	< 0.001
Population of the residential area		0.006		0.514
Urban > 1,000,000	Reference		Reference
Urban 250,000–1,000,000	1.03(0.99-1.08)	0.181	0.94(0.87-1.02)	0.161
Urban < 250,000	0.91(0.85-0.97)	0.005	0.91(0.81-1.03)	0.152
Rural adjacent to urban area	0.95(0.88-1.02)	0.143	0.98(0.86-1.11)	0.703
Rural not adjacent to urban	0.99(0.92-1.08)	0.953	0.96(0.83-1.12)	0.609
Surgical treatment		< 0.001		< 0.001
Yes	Reference		Reference
No/Unknown	4.29(4.07-4.51)		7.85(7.27-8.47)
Chemotherapy		0.001		< 0.001
Yes	Reference		Reference
No/Unknown	0.84(0.76-0.93)		0.54(0.48-0.62)
Radiotherapy		< 0.001		< 0.001
Yes	Reference		Reference
No/Unknown	1.34(1.20-1.49)		1.52(1.32-1.76)

Black patients showed comparable CSS to White patients. However, their OS was significantly lower (AHR = 1.08, 95% CI 1.02–1.14), corresponding to a 1.70-percentage-point reduction in the 5-year OS rate (Table 2). Asian/Pacific Islanders showed comparable CSS to White patients but experienced improved OS (AHR = 0.79, 95% CI 0.74–0.84), as illustrated in Fig. 2E and F.

Married patients exhibited 4.10% points and 9.90% points higher 5-year CSS and OS rates, respectively, compared to unmarried patients (Table 2). Unmarried status was associated with significantly increased hazards of both CSS (AHR = 1.44, 95% CI 1.34–1.55) and OS (AHR = 1.48, 95% CI 1.42–1.54) (Fig. 2G and H).

Patients in areas with a median household income below US$40,000 had 5-year CSS and OS rates that were 4.13% points and 5.90% points lower, respectively, than those in areas with median income above US$70,000 (Table 2). Multivariable regression showed an inverse association between area-level median household income and survival for both CSS and OS. Residents of most lower-income areas had a significantly higher hazard of death compared to those in the highest-income areas (> US$70,000), although this association was not statistically significant for CSS in the $40,000–$49,999 income bracket (AHR = 1.14, 95% CI 0.99–1.31) (Table 3; Fig. 2I and J).

Regarding residential population size, 5-year CSS ranged from 91.50% to 93.90%, and 5-year OS from 75.90% to 80.70% across categories (Table 2). In multivariable analysis, patients from counties with populations under 250,000 had a lower hazard of OS (AHR = 0.91, 95% CI 0.85–0.97) compared to those from metropolitan counties with 1 million or more residents (Table 3; Fig. 2K and L).

Interaction effects between age and socioeconomic factors

Significant interaction effects were observed between age group and gender (P for interaction < 0.001), age group and marital status (P for interaction < 0.001), and age group and area-level median household income (P for interaction < 0.001). In contrast, no significant interactions were found between age group and race (P for interaction = 0.228) or between age group and area of residence (P for interaction = 0.731).

Detailed analysis of these interactions revealed distinct age-varying patterns. For the interaction between age and gender, compared to male patients aged 20–39 years, the AHR for female patients was significantly elevated only in the oldest group (AHR = 1.65, 95% CI 1.01–2.71 for ages ≥ 70 years; P = 0.046), with no significant differences observed in younger age groups.

Regarding marital status, using married patients aged 20–39 years as the reference, the AHR for unmarried patients was highest in the middle-aged groups and attenuated with older age: 1.96 (95% CI 1.50–2.56) for ages 40–49, 1.98 (95% CI 1.76–2.24) for ages 50–59, 1.49 (95% CI 1.37–1.62) for ages 60–69, and 1.38 (95% CI 1.31–1.44) for patients aged 70 years or older (all P < 0.001).

For the interaction between age and area-level median household income (reference: age 20–39 years with income ≥$70,000), the hazard differed significantly across most combined strata. Only three specific age-income strata showed no statistically significant difference compared to the reference group: patients aged 40–49 years with an income <$40,000 (AHR = 0.81, P = 0.673); those aged 40–49 years with an income of $60,000-$69,999 (AHR = 1.14, P = 0.439); and those aged ≥ 70 years with an income of $40,000-$49,999 (AHR = 1.09, P = 0.061). (Supplementary Table S1).

The influence of socioeconomic factors on survival in SCRC patients stratified by age at diagnosis

The influence of gender on survival did not vary with age at diagnosis. Across all age groups, females exhibited a lower hazard of overall survival compared to males, with hazard ratios ranging from 0.60 (95% CI 0.35–0.95) to 0.84 (95% CI 0.80–0.88) (Fig. 3A). Similarly, marital status disparities in survival remained consistent across age groups. Unmarried patients faced a higher hazard of death compared to their married counterparts, with the highest AHR of 1.90 (95% CI 1.68–2.15) observed in the 50–59 age group and the lowest AHR of 1.39 (95% CI 1.32–1.45) in patients aged 70 or older (Fig. 3A).

Fig. 3.

The influence of socioeconomic factors on survival in SCRC patients stratified by age at diagnosis (A gender and marital status; B race; C household income; D population of the residential area)

Racial disparities in survival varied by age at diagnosis when compared to White patients. Black patients had a higher hazard of death in the 50–59 and 60–69 age groups, with AHRs of 1.23 (95% CI 1.07–1.42) and 1.17 (95% CI 1.05–1.30), respectively. No significant survival differences were observed between Black and White patients in other age groups. Asian/Pacific Islander patients showed a lower hazard of death compared to White patients, but this was only significant in those over 60 years old, with AHRs ranging from 0.76 (95% CI 0.65–0.88) to 0.80 (95% CI 0.74–0.87). American Indian/Alaska Native patients had a comparable hazard of death to White patients across all age groups (Fig. 3B).

Patients residing in areas with a median income below US70,000 had a higher mortality hazard than those in higher-income areas (> US$70,000), particularly in the 50–59 (Fig. 3C). Among patients aged 50–59 years, the hazard of all-cause mortality was 2.27 times higher for those in the lowest-income areas (< US$40,000) compared to their counterparts in the highest-income areas (> US$70,000) (AHR = 2.27, 95% CI: 1.62–3.18). This effect was strongest among unmarried individuals in the lowest-income areas, who faced a 3.63-fold higher mortality hazard (AHR = 3.63, 95% CI 2.17–6.10) compared to their married, high-income-area peers of the same age.

Additionally, patients residing in rural areas had a lower hazard of overall survival compared to those in large metropolitan areas with populations of 1 million or more. This association was most evident among patients aged 40–49, with AHRs of 0.51 (95% CI 0.33–0.80) for rural areas adjacent to urban regions and 0.51 (95% CI 0.30–0.86) for rural areas not adjacent to urban regions (Fig. 3D).

Sensitivity analysis

To assess the robustness of our findings, we performed a sensitivity analysis excluding patients with no or unknown surgical treatment and those who received chemotherapy or radiotherapy. The associations between socioeconomic factors and survival outcomes remained consistent with the main analysis (Supplementary Tables S2–S4). The significant interaction patterns between age and gender, marital status, and income were also robust. Notably, interactions between age and race, and age and area of residence became significant in the sensitivity analysis (Supplementary Table S5).

Discussion

This study provides a comprehensive analysis of socioeconomic disparities in survival outcomes among patients with SCRC. Our findings identified significant variations in survival outcomes based on demographic, socioeconomic, and clinical characteristics, indicating complex correlations between these variables and clinical outcomes of SCRC. Building on this foundation, our age-stratified approach was crucially justified by the finding of significant interaction effects between age and key socioeconomic variables (e.g., gender, marital status, area-level median household income), suggesting that the impact of these disparities on survival may be influenced by a patient’s life stage.

Consistent with previous research, older age was strongly associated with poorer survival outcomes [14, 15]. Our age-stratified analysis revealed unique patterns. Specifically, the survival disadvantages escalated sharply after age 60, with patients aged 70 or older having a 4.25-fold higher hazard of cancer-specific mortality than patients aged 20–39 years. Such observed age-based survival differences may be related to concurrent chronic diseases or varied treatment tolerance among elderly patients, while relevant underlying mechanisms remain to be further verified in subsequent research [16]. Beyond age, gender also emerged as a key predictor of SCRC survival. We found that female patients had higher 5-year OS (by 2.50% points) and CSS (by 0.80% points) rates compared to male patients. Although the absolute difference in CSS is 0.80% points, when projected to the population level, this modest improvement represents a substantial number of patients who could potentially avoid cancer-related death. This points toward notable public health implications of such gender-based disparities, which may warrant further exploration of gender-differentiated management approaches in future clinical practice. The survival advantage may be explained by established biological factors, including the protective effects of estrogen-mediated tumor suppression and the potential tumor-promoting role of androgen receptor signaling pathways in males [17]. Additionally, differences in tumor biology, such as variations in the tumor microenvironment or genetic and molecular profiles, may contribute to the observed outcomes [18]. Females also tend to exhibit stronger immune responses, which could enhance their ability to combat cancer [19]. Behavioral factors, including earlier healthcare utilization and lifestyle choices, as well as potential gender-specific differences in treatment efficacy and tolerance, may further explain this advantage. These findings support the rationale for incorporating gender factors into relevant research and clinical discussions, which may facilitate the development of more individualized care approaches worthy of further verification.

This study additionally explored racial survival disparities stratified by age subgroups. The results indicated that Black patients tended to have higher all-cause mortality hazards relative to White patients, and such correlative differences were most obvious in the 50–69 age bracket. Relevant observed disparities may provide preliminary clues for subsequent studies focusing on health service access and early disease detection status among different racial groups [20]. In contrast, Asian/Pacific Islander patients exhibited improved overall survival compared to White patients [21], suggesting potential protective factors, genetic differences, or differences in healthcare utilization within this population.

Marital status emerged as a significant predictor of survival, which was consistent with other studies [22, 23]. Our study further found that the marital disparity persisted across all age groups and that the protective effect of marriage was strongest in the 50–59 age group. This finding suggests that interventions targeting this subgroup may be beneficial. The protective effect of marriage may be attributed to enhanced social support, improved access to healthcare, and better adherence to treatment regimens. Collectively, these findings underscore the importance of addressing social determinants of health in SCRC management and highlight the significant impact of social support on cancer detection, treatment, and survival. Furthermore, lower area-level median household income was associated with significantly lower survival rates. This disparity was particularly pronounced in the 50–59 age groups, and our analysis uncovered a critical interactive effect. We found that marital status and income exhibited a synergistic effect. Notably, unmarried, low-income patients aged 50–59 years had a 3.63-fold higher OS risk compared to married, high-income peers of their age. This intersectional pattern, in which marital status and income jointly correlate with poorer clinical outcomes, has not been previously documented in SCRC. These correlational findings may imply that age-stratified income-targeted strategies combined with social support could serve as potential directions for future intervention exploration. Geographic location also influenced survival, with patients residing in rural areas demonstrating improved overall survival compared to those in large metropolitan areas. This counterintuitive advantage aligned with findings from other studies and may be attributable to factors such as potentially lower comorbidity rates in rural populations in this specific cohort or reduced exposure to environmental carcinogens [24, 25].

Our findings align with studies from other data sources and countries. Analyses using the National Cancer Database (NCDB) have similarly demonstrated that lower socioeconomic status is associated with worse survival in colorectal cancer patients [10]. International studies from European countries with universal healthcare coverage (e.g., UK, Denmark) also report persistent disparities, while studies from Asian populations show varying patterns potentially influenced by different social safety nets and screening programs [26–29]. These cross-country comparisons suggest that while healthcare access mediates some disparities, additional social and behavioral factors are also associated with survival differences.

Beyond socioeconomic factors, emerging evidence highlights the importance of mental health disorders in colorectal cancer outcomes [30]. Postoperative complications have been associated with increased risk of psychological distress, anxiety, and depression, which may in turn affect treatment adherence, follow-up attendance, and ultimately survival [31]. Patients from lower socioeconomic backgrounds may be particularly vulnerable to these mental health sequelae due to limited access to psychological support services. While our dataset did not include mental health variables, future research is warranted to integrate psychiatric assessment into CRC survivorship care, especially for socioeconomically disadvantaged groups. Additionally, a prospective study demonstrated that low postoperative butyrylcholinesterase levels on the first and third days after colorectal surgery were independently associated with a significantly higher risk of surgical site infections [32]. While our study focused on socioeconomic and demographic determinants, the potential role of such accessible biomarkers in refining risk stratification for SCRC patients warrants further investigation, particularly in relation to the disparities identified.

The persistent socioeconomic disparities in SCRC survival highlight a critical public health challenge that may warrant comprehensive attention and further investigation. These disparities are associated with inequities across healthcare access, early detection, and treatment, and disproportionately burden vulnerable populations such as unmarried individuals, low-income patients, and Black communities. To mitigate such health inequities, adopting multi-dimensional strategies may be worthy of consideration. Expanding population screening coverage in resource-limited areas could be a promising direction for future public health practice [33, 34]. Considering that SCRC has a high treatability rate when identified in its early stages, adherence to international screening guidelines may represent an important component of population-level strategies. These guidelines advocate for regular colonoscopies for individuals at average risk, commencing at age 45–49 [35, 36]. However, access to screening remains uneven, with socioeconomic barriers often preventing at-risk populations from receiving timely care.

In addition to the existing socioeconomic factors, two key factors may contribute to this screening access gap. First, differences in healthcare systems, such as the distinction between the public and private systems, may also play a role in the observed survival gaps. For instance, public healthcare systems might face resource constraints that limit the availability and frequency of screening services, while private systems may offer more comprehensive but potentially less accessible care due to cost barriers. These variations in the quality and accessibility of healthcare services between different systems could influence treatment outcomes. Second, limited patient education further widens the gap. Lower levels of patient education may lead to a lack of awareness about the importance of screening, reduced understanding of treatment options, and poorer adherence to treatment plans. Patients with limited education may also be less likely to participate in screening programs, either due to a lack of knowledge or misconceptions about the procedures.

To address these barriers and potentially boost screening rates among vulnerable populations, targeted interventions would need to address both systemic and knowledge gaps if such approaches are to be effective. On the systemic side, healthcare systems could consider implementing targeted outreach initiatives and remove financial and logistical barriers. For example, it could be considered whether health checkup packages might include a digestive endoscopy as a standard component, which could help ensure that individuals have access to essential screening tools regardless of their socioeconomic status. On the knowledge side, educational campaigns could be designed to address the specific needs of different patient populations, taking into account their educational levels and cultural backgrounds. Beyond patient-facing initiatives, workplace-based screening programs for healthcare staff may also offer a model for promoting early detection. Despite their medical knowledge, healthcare workers face occupational barriers such as shift work and time constraints, and systematic screening protocols within healthcare institutions could help address these gaps [37].

Beyond resolving basic screening barriers, improving access to advanced endoscopic techniques may represent a potentially valuable direction to support early diagnosis and treatment. These techniques are often unavailable in resource-limited settings. Achieving equitable access to these technologies would likely require investment in healthcare infrastructure, training for medical professionals, and policies that prioritize resource allocation to underserved regions. Additionally, providing financial assistance or insurance coverage for advanced procedures can help bridge the gap for low-income patients. Advances in artificial intelligence, particularly deep learning algorithms, offer promising tools to improve diagnostic accuracy and efficacy in colorectal cancer. Recent studies have demonstrated that deep learning models can achieve high performance in classifying and diagnosing CRC from histopathology images, potentially reducing inter-observer variability and expediting diagnosis [38, 39]. In the context of socioeconomic disparities, such technologies could help bridge gaps by providing standardized, high-quality diagnostic interpretation regardless of the resources available at individual institutions, provided that the deployment of these technologies is accompanied by policies ensuring equitable access across diverse healthcare settings.

Further, targeted support for vulnerable populations may also be an important consideration. Unmarried individuals may lack the social support systems that facilitate timely healthcare access and treatment adherence. Community-based programs that offer navigation services, transportation assistance, and emotional support represent one potential approach that could be explored to address these challenges. Similarly, culturally sensitive interventions that build trust within Black communities and address systemic biases in healthcare delivery may help address racial disparities, but their effectiveness requires future investigation. To address cultural and racial barriers to screening participation (e.g., discomfort with body exposure during colonoscopies), educational campaigns that emphasize the clinical benefits of early detection may be considered as one potential strategy. More broadly, integrating social determinants of health (SDH) into routine clinical assessment—including income, education, housing, and transportation—could help identify patient-specific barriers to care [40]. However, whether such SDH-informed individualized strategies could ultimately optimize prognosis or reduce disparities requires prospective evaluation.

Strengths and limitations

The key strengths of this study lie in its distinct contributions to addressing socioeconomic disparities in SCRC survival, filling critical gaps in existing literature. First, it exclusively focuses on SCRC—a highly curable subset that has been understudied, unlike most prior research centered on advanced CRC. By targeting SCRC specifically, our study is among the first to conduct a thorough socioeconomic analysis in this population, revealing that modifiable socioeconomic factors are associated with survival gaps. This highlights actionable systemic inequities obscured in broader, stage-pooled analyses. Second, it conducts a systematic analysis of key socioeconomic factors and their interactions, uncovering novel intersectional patterns to precisely target high-risk subgroups. We uncover intersecting inequalities such as marital status benefits, racial disparities masked by non-cancer factors, and rural-urban OS gaps that are often overlooked in conventional analyses. Third, its age-stratified design clearly reveals life-stage disparity patterns that are obscured in broader analyses. By employing fine age strata, our study not only confirms the association between age and survival but precisely quantifies its profound dynamics in SCRC. We demonstrate a progressive and then sharply escalating survival disadvantage with increasing age, which is particularly severe in patients aged 70 or older. These findings provide a strong foundation for future research into the biological and healthcare-related mechanisms underlying these age-based disparities and for developing life-stage-specific interventions to ensure equitable outcomes for all SCRC patients.

This study has several limitations. First, the SEER database lacks detailed information on comorbidities, treatment adherence, and lifestyle factors, which may influence survival outcomes. Second, our analysis was subject to limitations arising from missing data. A total of 81,511 cases (approximately 23.98%) were excluded due to missing T-stage information. Although our comparative analysis of baseline characteristics showed similar baseline characteristics, we cannot fully rule out residual confounding from this non-random missing data. The connection between different types of missing data (T-stage, comorbidity) as a collective limitation affecting the generalizability of the findings. Subsequently, the analysis of socioeconomic and racial disparities was conducted on the remaining cohort. Critically, if the missingness of socioeconomic variables like race within this analytical subset is not random but systematically linked to vulnerable groups that also experience poorer survival outcomes, it could introduce further selection bias. This would result in conservative estimates of racial disparities that may not fully represent the true effect in the entire population. Future studies using methods such as multiple imputation or linkage with supplementary data sources would be valuable to validate and extend our findings. Third, despite adjusting for available demographic, socioeconomic, and clinical variables, there remains the possibility of unmeasured confounding factors affecting the observed disparities. Fourth, approximately 6.5% of included patients had no or unknown surgical treatment. In these cases, N0 and M0 status were derived from clinical staging rather than pathological evaluation. This may have resulted in understaging, as clinically node-negative or metastasis-negative disease cannot exclude microscopic nodal involvement or occult metastases. Consequently, some patients with truly advanced disease may have been misclassified as SCRC, potentially biasing survival estimates. In addition, we could not exclude patients who received neoadjuvant therapy. This may have led to inclusion of downstaged tumors that were initially more advanced, potentially affecting observed survival disparities if downstaged patients had different socioeconomic profiles. While a sensitivity analysis excluding all patients receiving any chemotherapy or radiotherapy yielded consistent results, we cannot completely rule out residual bias from the inclusion of downstaged tumors. Future studies using datasets with explicit treatment timing variables are warranted.

Conclusion

In conclusion, this study found that demographic and socioeconomic factors, including age, gender, race, marital status, income, and geographic location, were associated with survival outcomes in SCRC patients, even within a patient population characterized by early-stage disease and a highly favorable prognosis. The persistence of these disparities after adjustment for tumor characteristics suggests that systemic and socially determined barriers may play a substantial role in these inequities, although residual confounding and unmeasured variables cannot be excluded. The observed associations suggest potential directions for future research to develop and assess targeted interventions optimizing care accessibility, strengthening social support and reducing economic treatment barriers, with the ultimate aim of promoting more equitable survival outcomes among all SCRC patients.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

SCRC

Superficial colorectal cancer

SEER

Surveillance, epidemiology, and end results

OS

Overall survival

CSS

Cancer-specific survival

HRs

Hazard ratios

CI

Confidence intervals

CRC

Colorectal cancer

SES

Socioeconomic status

IRB

Institutional review board

LNM

Lymph node metastasis

USPSTF

U.S. Preventive Services Task Force

ESGE

European society of gastrointestinal endoscopy

SDH

Social determinants of health

NCDB

National cancer database

Author contributions

C.S. Statistical analysis, drafting of the manuscript, and interpretation of data. Z.S. responsibility for the integrity of the data, and accuracy of the data analysis. L.J.T. and D.S.Y. Study concept and design. All authors reviewed the manuscript.

Funding

No funding.

Data availability

The data analyzed in this study were obtained from the Surveillance, Epidemiology, and End Results (SEER) 17-registry database (submission date: November 2023). The dataset used for this study was derived from the SEER research data files. Access to these data is not publicly available due to restrictions imposed by the SEER program’s data use agreement. However, qualified researchers can request access to the same underlying data by submitting the application through the SEER Data Access System. The direct link to the public description of the SEER 17-registry database (November 2023 submission) used in this study is: https://seer.cancer.gov/data-software/documentation/seerstat/nov2023/index.html. Access to the full raw data is restricted. Qualified researchers can submit an access request through the SEER Data Access System at: https://seerdataaccess.cancer.gov/seer-data-access. The specific data used in this study were from the SEER 17-registry database (November 2023 submission) [SEER*Stat 8.4.5]. Further information on data access is available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

This study utilized de-identified data from the Surveillance, Epidemiology, and End Results (SEER) database. The need for ethical approval and informed consent was waived by the Institutional Review Board (IRB) of the Ethics Committee of China-Japan Friendship Hospital because the data are publicly available and contain no identifiable patient information. All procedures performed in this study were in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments, as well as the guidelines of the aforementioned IRB. Not applicable. This study analyzed fully de-identified public data without any direct interaction with human participants, so individual informed consent to participate was not required.

Consent for publication

Not applicable. No identifiable clinical details, images or personal identifiers of patients were included in this manuscript, so separate consent for publication was unnecessary.

Competing interests

The authors declare no competing interests.