Association Between Social Determinants of Health and Cancer Risk in Middle-Aged and Older Chinese Adults: Evidence From a Nationally Representative Cohort Study

Abstract

Introduction

This study examined the association between social determinants of health (SDHs) and cancer incidence in a nationally representative Chinese cohort.

Methods

We conducted a prospective cohort study using data from 12,335 participants in the China Health and Retirement Longitudinal Study (2011-2020). SDHs scores reflected favorable social conditions. Cox proportional hazards models estimated cancer risk across SDHs levels, with subgroup and sensitivity analyses.

Results

Higher SDHs scores were associated with a lower cancer risk (HR 0.69, 95% CI 0.51-0.95), particularly among men (HR 0.60, 95% CI 0.39-0.92), rural residents (HR 0.54, 95% CI 0.33-0.87), and individuals who smoked (HR 0.59, 95% CI 0.37-0.95) or drank alcohol (HR 0.47, 95% CI 0.27-0.82). No significant association was observed in participants with hypertension (HR 0.74, 95% CI 0.44-1.24) or hyperlipidemia (HR 0.85, 95% CI 0.33-2.18).

Conclusions

Favorable social conditions were linked to reduced cancer risk, emphasizing the need to improve social environments alongside lifestyle interventions.

Plain Language Summary

This study looked at how social and living conditions affect the chances of getting cancer in middle-aged and older adults in China. We used data from over 12,000 people who took part in a national health study between 2011 and 2020. People with better social conditions— such as higher income, more education, and stronger social support— were less likely to develop cancer. This link was especially strong among men, people living in rural areas, and those who smoked or drank alcohol. However, we did not see the same protective effect in people with high blood pressure or high cholesterol. Our findings suggest that improving social conditions may help lower cancer risk. This highlights the importance of not only encouraging healthy lifestyles, but also creating healthier social environments.

Introduction

Cancer is one of the leading causes of death and a major economic burden in China. ¹ In 2022, approximately 4.82 million new cancer cases and 2.57 million cancer deaths were reported nationwide. ² Between 2005 and 2020, the total number of cancer-related deaths increased by 21.6%, reaching 2,397,772 deaths, while years of life lost (YLLs) rose by 5.0% to 56,598,975 years. ³ Although age-standardized mortality and YLL rates have declined significantly for most cancers in urban areas, they have increased for approximately half of all cancers in rural areas ³ Compared to the United States and the United Kingdom, China has a lower overall cancer incidence but higher cancer mortality and disability-adjusted life year (DALY) rates, indicating suboptimal cancer prognosis and healthcare outcomes. ¹ Population aging is one of the most critical drivers of the increasing cancer burden, as the growth of the elderly adult population contributes substantially to the rising number of cancer deaths. ⁴

The main modifiable risk factors associated with cancer development include infectious agents, smoking, alcohol consumption, obesity, unhealthful dietary habits, and inadequate physical activity. ⁵ Social determinants of health (SDH)—non-medical factors such as socioeconomic status, education, neighborhood context, and social support—have been increasingly recognized as crucial contributors to both the incidence and prognosis of various cancers[6]. Within-country socioeconomic inequalities in cancer burden have increased consistently across all regions worldwide. ⁶ Systematic reviews and meta-analyses have shown that residential environment, educational attainment, socioeconomic position, and access to healthcare are significantly associated with the risk and outcomes of breast, pancreatic, prostate, lung, liver, and oral cancers.^7-16 Notably, lower socioeconomic status has been consistently linked to higher cancer incidence, reduced access to treatment, and poorer survival. Similarly, lower education levels have been associated with increased risks of oral and hepatocellular cancers. However, nationally representative longitudinal evidence from East Asian populations, particularly among middle-aged and older adults in China, remains limited, hindering a comprehensive understanding of cancer inequalities in this rapidly aging population.

Given the growing cancer burden and aging population in China, it is crucial to understand how social determinants of health influence cancer risk among middle-aged and older adults, who face both increased incidence and substantial disparities in access to prevention and care. To date, evidence on the association between social determinants of health and cancer risk has largely come from Western populations,^17-19 and most studies in China have been either regional or cross-sectional in design.^20,21 There is a lack of nationally representative longitudinal studies that have systematically examined the association between multiple social determinants of health and cancer risk in this population. Therefore, using data from a large, nationally representative cohort of Chinese adults, this study aimed to evaluate the prospective associations between SDHs and the risk of developing cancer.

Methods

Study Design

We conducted a prospective cohort study to examine the association between SDHs and the incidence of cancer. The analysis was based on data derived from the China Health and Retirement Longitudinal Study (CHARLS), covering survey waves from 2011 to 2020. CHARLS is a large-scale, nationally representative study targeting Chinese residents aged 45 years and older. ²² The participants were selected using a four-stage, stratified, cluster sampling procedure with probability proportional to size (PPS), following the design of the China Health and Retirement Longitudinal Study (CHARLS). In the first stage, county-level units were sampled from all provinces using PPS. In the second stage, villages and urban communities within each selected county were chosen. In the third stage, households were randomly selected within each community. In the final stage, individuals aged 45 years and older and their spouses were interviewed. The survey gathers comprehensive information on various domains, including health status, economic conditions, retirement, and social relationships. Data collection is carried out biennially since its initiation in 2011, employing household interviews, physical examinations, and additional assessment methods. Participants were recruited from 28 provinces and 150 counties across China, with detailed measures encompassing demographics, socioeconomic indicators, self-reported health, and objective physical and biomarker assessments. Further details regarding CHARLS methodology and protocols are available on the official website (https://charls.pku.edu.cn/) and in previous publications. ²³ The CHARLS team had fully de-identified all personal information prior to data release, and no individual could be identified in any way. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ²⁴

Inclusion and Exclusion Criteria

The baseline sample comprised 17,750 participants from the CHARLS cohort (2011 wave). Participants were excluded if they had missing data on social determinants of health (n = 3778), had a history of cancer at baseline (n = 1972), were younger than 45 years at baseline (n = 192), had impaired kidney function (n = 277), or developed cancer within the first 2 years of follow-up (n = 1103). Given that cancer diagnoses were self-reported in this study, the possibility of underreporting cannot be excluded. In addition, because previous evidence indicates a potential association between cancer and renal function impairment., ²⁵ participants with impaired kidney function were excluded to minimize reverse causality. After these exclusions, 12,335 participants were included in the final analysis. The detailed participant selection process is presented in Supplemental Figure 1.

Variable Definition

All exposures and covariates were derived solely from the baseline survey (2011 wave).

Exposure

SDHs ²⁶ in this study were defined based on multiple dimensions, including socioeconomic status, healthcare accessibility, and psychosocial factors. Specifically, we assessed housing tenure, per capita household income, employment status, medical insurance coverage, educational attainment, access to healthcare services, marital status, living arrangement, depressive symptoms, and social participation. Detailed definitions and scoring criteria for each SDHs variable are provided in Supplemental Table 1. The distribution and scoring of each component are presented in Supplemental Table 2. For each participant, we calculated a composite SDHs score by summing the individual scores of the ten SDHs variables, with higher scores indicating more favorable social determinants of health. The total score ranged from 0 to 10. Based on the median value of 7, participants were categorized into 2 groups: low SDHs score (≤7) and high SDH score (>7).

Main Outcome Measures

The primary outcome of this study was the incidence of cancer, excluding non-melanoma skin cancer. Cancer diagnoses were self-reported by participants during follow-up. Individuals were tracked from baseline until either the first reported cancer diagnosis or their last completed survey before 2020, whichever occurred earlier.

Covariates

Age was calculated as the difference between the baseline year (2011) and the participant’s birth year. Anthropometric data were collected following standardized procedures: weight and height were measured to the nearest 0.1 kg and 0.1 cm, respectively, using calibrated equipment, with participants in light clothing and barefoot. ²⁷ BMI was computed as weight divided by height squared (kg/m²) and categorized as overweight (BMI ≥23.0 kg/m²) or obese (BMI ≥27.5 kg/m²) according to WHO guidelines for Chinese adults. ²⁸

Smoking status was classified as never, former, or current, based on whether participants had smoked at least 100 cigarettes over their lifetime. Alcohol use was defined as current (drinking at least once per month), former, or never drinkers.

Type 2 diabetes was determined by fasting plasma glucose ≥126 mg/dL or use of antidiabetic medication. ²⁹ Blood pressure was defined using the mean of three measurements, with hypertension identified as systolic BP ≥140 mmHg and/or diastolic BP ≥90 mmHg or current antihypertensive treatment. Dyslipidemia was defined as TC >6.2 mmol/L, TG ≥2.3 mmol/L, LDL-C ≥4.1 mmol/L, HDL-C <1.0 mmol/L (men) or <1.3 mmol/L (women), or self-reported dyslipidemia.^30,31Renal function was assessed using eGFR derived from the CKD-EPI equation, with an eGFR <80 mL/min/1.73 m² indicating kidney impairment ³²

We divided the participants into 2 groups (rural and Urban Community).

Data Analysis

Missing data were handled using Multiple Imputation by Chained Equations (MICE), with details on the proportion of missingness presented in Supplemental Table 3. Baseline characteristics were described across social determinants of health (SDHs) categories and compared using the Mann–Whitney U test for continuous variables and Chi-square test for categorical variables.

Participants who were lost to follow-up were censored at the date of their last completed interview. Cox proportional hazards regression models were employed to estimate the association between SDHs and cancer incidence, after confirming the proportional hazards assumption. Death was considered a competing risk event, with dates of death obtained through proxy reports from participants’ relatives or acquaintances. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were calculated. ³³

Three models were fitted: Model 1: unadjusted; Model 2: adjusted for sex and age; Model 3: additionally adjusted for hypertension, diabetes, dyslipidemia, BMI, smoking status, drinking status, and residential area (urban/rural).

SDHs were analyzed both as a continuous variable and as a dichotomous variable based on the median score (score <7 vs ≥7). Subgroup analyses were performed stratified by sex (male, female), age (≤60, >60 years), hypertension status, diabetes status, dyslipidemia status, and urban/rural residence. Interactions between SDH and stratification variables were tested with SDH treated as a continuous variable. Forest plots were generated to visually display subgroup-specific HRs and corresponding interaction effects. Linearity of the association between continuous SDH scores and cancer incidence was assessed using restricted cubic spline (RCS) regression with three knots, and P-values for overall and non-linear associations were reported.

As a sensitivity analysis, we repeated the primary Cox regression analyses using the dataset with missing values to assess the robustness of the findings. Results from the sensitivity analyses were compared with those from the imputed datasets to ensure consistency.

Population Attributable Fraction (PAF) was calculated as following: $PAF=\frac{\mathrm{p}\left(HR-1\right)}{\mathrm{p}\left(HR-1\right)+1}$ p: percentage of exposure group, HR: the relative risk of exposed vs unexposed groups

All descriptive and main Cox regression analyses were weighted using individual sampling weights to account for the multistage, stratified sampling design of CHARLS and ensure national representativeness. Subgroup and restricted cubic spline (RCS) analyses were conducted without weighting due to smaller sample sizes and model complexity, consistent with prior CHARLS-based analytical practice. All statistical tests were two-sided, with a P-value <0.05 considered statistically significant. Analyses were performed using R software version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

During 9 years’ follow-up (2011-2020), a total of 253 incident cancer cases were identified among 12,335 participants. Participants with higher SDH scores (>7) were more likely to be male, younger, and living in urban areas. They also had higher levels of education, greater BMI, and a higher prevalence of current smoking, current alcohol consumption, and hyperlipidemia, compared with those with lower SDH scores (≤7). Conversely, participants with lower SDH scores were more often female, older, rural residents, and had a higher prevalence of hypertension. Although the overall cancer incidence was low (2.09%), it was slightly higher among those with lower SDH scores (2.26% vs 1.66%, P = 0.044). Details are presented in Table 1. Participants with higher SDH scores (>7) were more likely to have higher education, stable income, health insurance coverage, social participation, and supportive living arrangements, whereas those with lower SDH scores (≤7) were more disadvantaged across multiple domains (Supplemental Table 4).

Table 1.

Baseline Characteristics of Participants Stratified by Social Determinants of Health (SDH) Score

Characteristic	SDH
	≤7			>7			p-value a
	Unweighted n = 12,335	Weighted n (weighted %)	Unweighted n = 8900	Weighted n (weighted %)	Unweighted n = 3455	Weighted n (weighted %)
Cancer, n (%)							0.044
No	12,082.00	369,085,326.74 (97.91%)	8700.00	260,683,239.03 (97.74%)	3382.00	108,402,087.71 (98.34%)
Yes	253	7,859,704.15 (2.09%)	200	6,031,766.91 (2.26%)	53	1,827,937.24 (1.66%)
Social determinants of health, median (Q1, Q3)	7	7.00 (6.00, 8.00)	7	6.00 (6.00, 7.00)	8	8.00 (8.00, 9.00)	<0.001
Gender, n (%)							<0.001
Male	5979.00	181,302,331.78 (48.10%)	4048.00	119,874,917.37 (44.94%)	1931.00	61,427,414.41 (55.73%)
Female	6356.00	195,642,699.12 (51.90%)	4852.00	146,840,088.58 (55.06%)	1504.00	48,802,610.54 (44.27%)
Age, median (Q1, Q3)	58	58.00 (51.00, 65.00)	59	59.00 (53.00, 67.00)	54	54.00 (49.00, 60.00)	<0.001
Smoking, n (%)							<0.001
Never smoked	7464.00	231,390,762.38 (61.39%)	5512.00	168,539,036.39 (63.19%)	1952.00	62,851,725.99 (57.02%)
Currently smoking	3812.00	112,024,441.62 (29.72%)	2642.00	75,899,562.34 (28.46%)	1170.00	36,124,879.28 (32.77%)
Ever smoked	1059.00	33,529,826.89 (8.90%)	746	22,276,407.21 (8.35%)	313	11,253,419.68 (10.21%)
Alcohol intake, n (%)							<0.001
Never drink	8590.00	267,446,760.76 (70.95%)	6344.00	194,548,513.00 (72.94%)	2246.00	72,898,247.76 (66.13%)
Currently drinking	3181.00	94,276,932.09 (25.01%)	2134.00	60,863,623.64 (22.82%)	1047.00	33,413,308.45 (30.31%)
Ever drunk	564	15,221,338.04 (4.04%)	422	11,302,869.30 (4.24%)	142	3,918,468.74 (3.55%)
Education, n (%)							<0.001
Junior high school and below	8070.00	235,601,963.18 (62.50%)	7186.00	209,996,617.54 (78.73%)	884	25,605,345.64 (23.23%)
Senior high school	3943.00	128,156,697.89 (34.00%)	1648.00	53,658,196.29 (20.12%)	2295.00	74,498,501.61 (67.58%)
College and above	322	13,186,369.82 (3.50%)	66	3,060,192.12 (1.15%)	256	10,126,177.70 (9.19%)
Area, n (%)							<0.001
North	5692.00	162,230,352.86 (43.04%)	3929.00	109,575,081.70 (41.08%)	1763.00	52,655,271.16 (47.77%)
South	6643.00	214,714,678.03 (56.96%)	4971.00	157,139,924.25 (58.92%)	1672.00	57,574,753.79 (52.23%)
Hypertension, n (%)							0.002
No	7646.00	232,256,313.59 (61.62%)	5427.00	161,016,682.45 (60.37%)	2219.00	71,239,631.14 (64.63%)
Yes	4689.00	144,688,717.30 (38.38%)	3473.00	105,698,323.49 (39.63%)	1216.00	38,990,393.81 (35.37%)
Diabetes, n (%)							0.033
No	10,950.00	334,839,324.52 (88.83%)	7942.00	238,310,738.39 (89.35%)	3008.00	96,528,586.13 (87.57%)
Yes	1385.00	42,105,706.37 (11.17%)	958	28,404,267.56 (10.65%)	427	13,701,438.81 (12.43%)
Hyperlipidemia, n (%)							<0.001
No	11,179.00	339,410,093.39 (90.04%)	8192.00	244,769,178.71 (91.77%)	2987.00	94,640,914.68 (85.86%)
Yes	1156.00	37,534,937.50 (9.96%)	708	21,945,827.24 (8.23%)	448	15,589,110.26 (14.14%)
Body Mass index (BMI), median (Q1, Q3)	23.31	23.31 (20.97, 25.86)	23.06	23.06 (20.70, 25.59)	23.94	23.94 (21.60, 26.35)	<0.001
Urban/Rural, n (%)							<0.001
Rural	7411.00	194,165,792.51 (51.51%)	5744.00	151,240,432.39 (56.70%)	1667.00	42,925,360.11 (38.94%)
Urban	4924.00	182,779,238.38 (48.49%)	3156.00	115,474,573.55 (43.30%)	1768.00	67,304,664.83 (61.06%)
Heart disease, n (%)							0.046
No	10,977.00	336,045,034.91 (89.15%)	7949.00	239,198,065.28 (89.68%)	3028.00	96,846,969.63 (87.86%)
Yes	1358.00	40,899,995.98 (10.85%)	951	27,516,940.67 (10.32%)	407	13,383,055.31 (12.14%)

^aPearson’s Chi-squared test; Wilcoxon rank sum test.

Bold indicates significant of p-values.

Higher SDH scores were associated with a lower risk of incident cancer. In Model 1, each unit increase in SDH score was associated with an 11% reduction in cancer risk (HR = 0.89, 95% CI: 0.78-0.99, P = 0.044). When comparing SDH categories, participants with SDH scores >7 had a lower cancer risk compared to those with scores ≤7 (HR = 0.73, 95% CI: 0.52-1.03, P = 0.077). After adjusting for age and sex (Model 2), and further for major confounders (Model 3), this association remained statistically significant (HR = 0.69, 95% CI: 0.51-0.95, P = 0.023) (Table 2).

Table 2.

Association Between Social Determinants of Health (SDH) and Cancer Incidence

Characteristic	Model 1					Model 2					Model 3
	N	Event N	HR	95% CI	P-value	N	Event N	HR	95% CI	P-value	N	Event N	HR	95% CI	P-value
SDHs	12,335	253	0.89	0.78, 0.99	0.044	12,335	253	0.91	0.80, 1.04	0.174	12,335	253	0.91	0.81, 1.02	0.096
SDHs category	12,335	253				12,335	253				12,335	253
≤7	8900		—	—		8900		—	—		8900		—	—
>7	3435		0.73	0.52, 1.03	0.077	3435		0.72	0.53, 0.98	0.037	3435		0.69	0.51, 0.95	0.023

Abbreviations: CI = Confidence Interval, HR = Hazard Ratio

The number of participants (N) represents the unweighted sample size. Hazard ratios (HRs), 95% confidence intervals (CIs), and P-values were derived from weighted Cox proportional hazards models using individual sampling weights provided by CHARLS to account for the complex survey design. Bold indicates significant of p-values.

Based on the weighted Model 3, approximately 24.5% of cancer cases (PAF = 24.5%) could potentially be prevented if all participants achieved favorable SDH levels (>7).

RCS analysis further examined the association between continuous SDH scores and cancer incidence. As shown in Figure 1, the overall association was statistically significant (p for overall = 0.007), while no evidence of non-linearity was observed (p for non-linear = 0.128).

Figure 1.

Restricted Cubic Spline Analysis of the Association Between Social Determinants of Health (SDH) Score and Cancer Incidence. Notes: RCS Analyses Were Conducted Using Unweighted Models

We performed stratified analyses to examine the association between higher SDHs scores and cancer incidence across various subgroups (Figure 2). The inverse association between higher SDHs scores and cancer incidence was generally consistent across most subgroups, although the magnitude of association varied.

Figure 2.

Subgroup Analyses of the Association Between Higher SDHs Scores (>7 vs ≤ 7) and Cancer Incidence. Notes: Subgroup Analyses Were Conducted Using Unweighted Models

A statistically significant association was observed among males (HR = 0.60, 95% CI: 0.39-0.92, P = 0.018), individuals without hypertension (HR = 0.66, 95% CI: 0.44-0.98, P = 0.041), individuals without hyperlipidemia (HR = 0.68, 95% CI: 0.49-0.96, P = 0.026), rural residents (HR = 0.54, 95% CI: 0.33-0.87, P = 0.012), those who were current or former smokers (HR = 0.59, 95% CI: 0.37-0.95, P = 0.028), and those who were current or former drinkers (HR = 0.47, 95% CI: 0.27-0.82, P = 0.007).

No significant interactions were observed between SDHs scores and the stratification variables (data not shown).

To assess the robustness of our findings, we conducted sensitivity analyses using datasets that included missing values (Supplemental Table 5). The inverse association between higher SDHs scores and cancer incidence remained largely consistent across all models, despite slight reductions in sample size due to missing data. It suggests that the association between higher SDHs scores and reduced cancer risk was robust and not materially affected by missing data.

Discussion

In this prospective cohort study, we examined the association between social determinants of health (SDHs) and cancer incidence in a large sample of 12,335 participants. Our findings demonstrate that higher SDHs scores, reflecting more favorable social conditions, were associated with a reduced risk of developing cancer. Specifically, individuals with SDHs scores greater than 7 had a significantly lower risk of cancer incidence compared to those with scores of 7 or less, even after adjusting for potential confounders. Collectively, these results highlight the potential importance of favorable social determinants in lowering cancer risk.

Our findings align with prior research demonstrating that individuals with lower socioeconomic status (SES) face higher cancer risks, even after accounting for established risk factors such as smoking. This aligns with evidence from European cohorts reporting higher risks of lung, oral, and pharyngeal cancers in socioeconomically disadvantaged groups. ³⁴ Similarly, a Nordic cohort study observed a clear SES gradient in the incidence of head and neck cancers, with markedly higher risks in socially disadvantaged groups, especially for cancers of the oral cavity, pharynx, and larynx. ³⁴ Notably, the study reported that managers and higher SES groups exhibited substantially reduced standardized incidence ratios for various HNC subtypes. ³⁴

The inverse association between SDHs and incidence of cancer appeared stronger among men and rural populations in our study. The more pronounced protective association observed in rural residents in our study may be partially explained by evidence indicating that rural populations in China are subject to higher environmental health risks due to substantial urban–rural disparities in exposure to pollutants such as PM₂.₅-bound toxic elements. Prior studies have identified coal combustion as a dominant contributor to increased cancer risk in rural areas owing to elevated exposure to toxic elements like arsenic and lead. ³⁵ These findings suggest that individuals in rural and lower SES settings may be simultaneously burdened with greater environmental exposures and reduced access to healthcare resources, compounding their cancer risk. The observed stronger associations in men may further reflect gender-specific occupational or behavioral exposures, although prior studies suggest that sex differences in non-cancer and cancer risks from environmental pollutants are relatively modest. ³⁵ A recent study using data from the CHARLS demonstrated that low SES was associated with a significantly increased risk of incident physical multimorbidity (OR 1.22, 95% CI 1.05-1.42), and that participants with both low SES and unhealthy lifestyles had the highest risk (OR 2.19, 95% CI 1.57-3.04). ³⁶ Importantly, lifestyle only partially mediated the SES–multimorbidity relationship, underscoring that broader social determinants of health, including access to healthcare and environmental exposures, may play a critical role. ³⁶ These patterns are consistent with our observation that SDH-related cancer risks were concentrated in subpopulations with more disadvantaged profiles including smoking and drinking. In our stratified analyses, the protective association of higher SDH scores appeared stronger among individuals with current or former smoking and drinking behaviors. These findings suggest that disadvantaged social conditions and unhealthy lifestyles may have synergistic effects, compounding cancer risk in these subgroups.

Interestingly, in our stratified analyses, the inverse association between higher SDH scores and cancer incidence was not statistically significant among individuals with hypertension or hyperlipidemia. Several plausible explanations may account for these findings. First, both hypertension and hyperlipidemia could act as intermediate variables in the pathway linking SDHs to cancer risk. Previous studies have shown that higher SDH scores are inversely associated with the prevalence of hypertension and hyperlipidemia^37-40; therefore, restricting analyses to subgroups with established hypertension or hyperlipidemia may block part of the indirect protective pathway of SDHs, attenuating the overall association. Second, stratification by these comorbidities may introduce collider stratification bias, as both hypertension and hyperlipidemia are influenced by SDHs and share common risk factors for cancer, which could distort the observed relationship in these subgroups. Third, although we excluded cancer cases diagnosed within the first 2 years of follow-up to minimize reverse causality, it remains possible that the follow-up duration was insufficient to fully account for individuals with undiagnosed, preclinical cancers or precancerous lesions at baseline. Such individuals—potentially already harboring early-stage malignancies or cancer-prone conditions—might have higher rates of comorbidities such as hypertension or hyperlipidemia,^41-43 which could further complicate the interpretation of subgroup analyses. These considerations emphasize the complex interplay between social determinants of health, chronic diseases, and cancer incidence, and suggest that future research should explore potential mediation, interaction, and reverse causation effects in greater depth. Moreover, cardiometabolic conditions such as hypertension and coronary vascular disease may act as potential mediators linking unfavorable social determinants of health to increased cancer risk, through shared mechanisms involving chronic inflammation, oxidative stress, and metabolic dysregulation. Elucidating these pathways in future longitudinal studies could provide new insights into the prevention of both cardiometabolic and oncologic diseases.

Strengths and Limitations

The major strengths of our study include the use of a large, nationally representative cohort from China, with nearly 10 years of follow-up (2011-2020), enabling robust estimation of long-term associations between social determinants of health and cancer incidence. However, several limitations should be acknowledged. First, cancer incidence was based on self-reported diagnoses without verification by medical records or cancer registries, which may have introduced misclassification bias. Second, detailed information on cancer subtypes was unavailable, as diagnoses were not coded using ICD classifications, limiting our ability to assess site-specific associations. Third, although we excluded cancer cases diagnosed within the first 2 years of follow-up to reduce reverse causation, this exclusion period might still be insufficient to completely eliminate preclinical cases. Extending the exclusion window to 4 years substantially reduced the sample size and number of cancer cases, limiting statistical power for robust analysis. Finally, we were unable to account for family history of cancer, an established risk factor with potential genetic and environmental influences, as this information was not available in the dataset. The lack of family history data may have led to residual confounding in our analyses. Although the construction of the SDH index followed established definitions in the literature, combining heterogeneous components into a single unweighted score may still limit interpretability, as individuals with the same total score could have different underlying SDH profiles.

Conclusion

In this large, nationally representative cohort with nearly 10 years of follow-up, we found that favorable social determinants of health were associated with a lower risk of cancer incidence. This protective association appeared stronger among men, rural residents, and individuals with smoking or drinking histories. Our findings suggest the potential importance of addressing broader social determinants—beyond individual lifestyle behaviors—to reduce cancer risk, especially in socioeconomically disadvantaged populations.

Appendix.

Abbreviations

SDH/SDHs

Social Determinants of Health

CHARLS

China Health and Retirement Longitudinal Study

BMI

Body Mass Index

HR

Hazard Ratio

CI

Confidence Interval

SES

Socioeconomic Status

DALY

Disability-Adjusted Life Year

YLL

Years of Life Lost

eGFR

Estimated Glomerular Filtration Rate

ICD

International Classification of Diseases

MICE

Multiple Imputation by Chained Equations

ORCID iD

Feng Gao https://orcid.org/0009-0004-1635-4775

Ethical Considerations

Ethical approval for all waves of the China Health and Retirement Longitudinal Study (CHARLS) was granted by the Institutional Review Board of Peking University (approval numbers: IRB00001052-11015 for the household survey and IRB00001052-11014 for biomarker collection). All participants provided written informed consent prior to participation. The present study analyzed data from the publicly available CHARLS database; therefore, no additional ethics approval was required.

Data Availability Statement

The data analyzed in this study were obtained from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative longitudinal survey conducted by the National School of Development at Peking University. The CHARLS data are publicly available at https://charls.pku.edu.cn/.