Abstract
Background:
Liver cancer incidence and mortality are rising in the United States, with marked racial and socioeconomic disparities. Persistent poverty, defined as communities in which ≥20% of residents have lived below the federal poverty line for ≥30 years, is an important but understudied contextual determinant of cancer survival. This study examined the association between census tract–level persistent poverty and liver cancer survival.
Methods:
We analyzed 73,695 adult, population-based patients with liver cancer ages ≥20 years, diagnosed between 2006 and 2019 in the Surveillance, Epidemiology, and End Results 17 Registries. Persistent poverty status was assigned by residence at diagnosis. Cancer-specific and overall survival were estimated using Kaplan–Meier and Cox models, adjusting for demographic factors, stage at diagnosis, and initial treatment. We evaluated effect modification by age, sex, race/ethnicity, rurality, and stage at diagnosis.
Results:
Over 168,715 person-years, 56,373 deaths occurred, of which 81% were cancer-specific. Eleven percent of patients lived in persistent poverty tracts. The median cancer-specific survival was 16 months in persistent poverty tracts versus 22 months in nonpersistent tracts (P < 0.001). In fully adjusted models, persistent poverty was associated with 9% higher cancer-specific mortality (HR = 1.09; 95% CI, 1.06–1.12) and 11% higher all-cause mortality (HR = 1.11; 95% CI, 1.09–1.14). Associations were strongest in patients <65 years and with localized disease, with no significant differences by race/ethnicity, sex, or rurality.
Conclusions:
Living in a persistent poverty census tract is associated with worse liver cancer survival, particularly among younger patients and those with localized disease.
Impact:
Interventions and policies targeting economically disadvantaged communities may be important for reducing liver cancer survival disparities.
Introduction
Liver cancer is the sixth leading cause of cancer deaths in the United States, with an estimated 42,240 incident cases and 30,090 deaths in 2025 (1). Liver cancer is also one of the fastest rising causes of cancer mortality in the United States, and it is estimated to become the third leading cause of cancer related death by 2040 (2). There are marked racial and ethnic disparities in liver cancer incidence and survival in the United States (3). Specifically, liver cancer incidence rates are 1.5 to 2 times higher in American Indian/Alaska Native, non-Hispanic Black, Hispanic, and Asian and Pacific Islander populations compared with non-Hispanic Whites (4). Whereas Asian and Pacific Islander populations have the highest liver cancer survival rates among all racial and ethnic groups, non-Hispanic Black individuals experience significantly worse survival outcomes (3, 5). There is also mounting evidence that liver cancer survival rates vary based on contextual factors such as rurality and area-level poverty, which contribute to the noted racial and ethnic disparities (6–9).
Recently, there has been growing interest in understanding the specific impact of living in areas of persistent poverty, typically defined as communities in which 20% or more of the population has lived below the federal poverty line for at least 30 years, on cancer outcomes. Compared with areas experiencing only recent or short-term poverty, persistent poverty areas tend to have larger minority populations, lower educational attainment, higher unemployment, and greater exposure to cancer risk factors such as obesity, smoking, and alcohol use (10–12). These areas are also characterized by long-standing disinvestment in critical sectors, including education, healthcare, and economic development, which may further exacerbate cancer risk and limit access to prevention and treatment resources (13–15). One recent study found that living in a persistent poverty county was associated with higher overall cancer mortality and increased mortality for specific cancers, including lung and bronchus, colorectal, breast, stomach, and liver and intrahepatic bile duct cancers (10).
Yet, there is currently limited data on the relationship between liver cancer survival and persistent poverty at a more granular geographic level, such as the census tract. This is an important research gap given that county-level measures reflect relatively broad geographic areas and may obscure substantial variation in socioeconomic conditions within counties. In contrast, census tract–level data offer a more precise measure of neighborhood-level deprivation and may better capture the localized social and environmental factors that may influence individual cancer outcomes. For example, research has shown that twice as many non-Hispanic Black and Hispanic individuals reside in persistent poverty census tracts compared with persistent poverty counties (16), highlighting the importance of using finer geographic units to assess disparities. Furthermore, there is limited evidence on whether the relationship between persistent poverty and liver cancer mortality varies by race or ethnicity, which is important for understanding and addressing inequities in cancer outcomes across diverse populations.
To address these gaps, we analyzed case listing data from the Surveillance, Epidemiology, and End Results (SEER) program to examine the association between census tract–level persistent poverty and survival among individuals diagnosed with liver cancer. The study had two primary aims: (i) to assess whether cancer-specific survival and overall survival differ between patients with liver cancer living in persistent poverty census tracts and those in nonpersistent poverty tracts and (ii) to evaluate whether the associations between persistent poverty and liver cancer survival vary by key factors such as age, sex, race and ethnicity, rurality, and stage at diagnosis.
Materials and Methods
Study population
We used patient case listing data from Research Plus Specialized Data with Census Tract Attributes released in November 2022 from the National Cancer Institute’s SEER program (RRID: SCR_006902). The case listing data include patients with cancer from 17 registries across the United States who were diagnosed between 2006 and 2020, which was the period that census tract attributes for persistent poverty was available. Our study population included patients with liver cancer ages 20 years or older diagnosed from 2006 to 2019 (n = 99,135). The latest date of diagnosis was December 31, 2019, to ensure at least one year of follow-up on the patients. The primary site of diagnosis was the liver according to the International Classification of Diseases for Oncology (ICD-O-3), coded as C220. We excluded individuals who were missing data on persistent poverty (n = 136), survival time (n = 16,687), or cause of death (n = 1). We further excluded individuals with missing data on stratifying factors, including race and ethnicity (n = 102), rurality (n = 11), and stage at diagnosis (n = 8,413). This resulted in a final analytic sample of 73,695 patients. We received approval to use deidentified cancer incidence data through a SEER Public Use Research Data Agreement. The study was exempt from ethical review and complies with Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines (17).
Persistent poverty
We used the persistent poverty variable, available in the specialized SEER Incidence Data with Census Tract Attributes Database (18), which identifies census tracts as being persistently poor if 20% or more of the population has lived below the poverty line for approximately 30 years, based on the 1990 and 2000 decennial censuses and the American Community Survey 5-year estimates (2007–2011 and 2015–2019; ref. 19). Participants in our study were assigned to a census tract based on their residential address at the time of cancer diagnosis, and each census tract was then categorized as either persistent poverty or nonpersistent poverty. This variable was developed by the National Cancer Institute in collaboration with the US Department of Agriculture’s Economic Research Service (20).
Survival outcomes
Our primary outcomes were cancer-specific survival and overall survival for patients with liver cancer. Cancer-specific survival was defined as the time from diagnosis to death from liver cancer, whereas overall survival was defined as the time from diagnosis to death from any cause. We calculated survival time in months from the date of diagnosis to the date of death or censoring (December 31, 2020).
Covariates
Covariates included age, race and ethnicity, sex, rurality, stage at diagnosis, and treatment type. Age was categorized as <65 or ≥65 years. Race and ethnicity were categorized as follows: American Indian/Alaska Native, Hispanic (all races), Native Hawaiian/Pacific Islander, non-Hispanic Asian American, non-Hispanic Black, and non-Hispanic White. The stage at diagnosis was categorized as localized, regional, and distant based on the 2000 version of the SEER Summary Stage variable (21). Three variables on the initial course of treatment were included as covariates, including receipt of surgery, radiotherapy, and chemotherapy. These variables were each categorized dichotomously (with treatment versus without treatment or unknown). Rurality was determined based on the Urban Rural Indicator Code based on Census Bureau’s data (22). Urban was defined as greater or equal to 50% of the population living in urban areas, and vice versa for rural areas.
Statistical analysis
We compared the baseline characteristics of patients with liver cancer in persistent poverty census tracts with those in nonpersistent poverty census tracts using the χ2 test. Kaplan–Meier survival curves and log-rank statistics were computed to compare cancer-specific and overall survival outcomes for patients in both groups and to obtain stratified median survival times. We then used multivariable Cox proportional hazards regression to estimate hazard ratios (HR) and 95% confidence intervals (CI) for the association between residence in a persistent poverty census tract and both cancer-specific and all-cause mortality. We fitted a series of nested models to assess how covariate adjustment influenced the association between persistent poverty and survival outcomes: Model 1 was unadjusted; model 2 adjusted for age and sex; model 3 further adjusted for race and ethnicity; model 4 further adjusted for rurality; model 5 further adjusted for stage at diagnosis; and model 6 further adjusted for treatment modality. To assess potential multiplicative effect modification, we included interaction terms between persistent poverty and age, sex, race and ethnicity, rurality, and stage at diagnosis; each tested in a separate model. Statistical significance was evaluated using a two-sided hypothesis test, with a P value threshold of <0.05.
We tested the proportional hazards assumption using Schoenfeld residuals. The proportional hazards assumption was met in models 1 to 5. However, after additional adjustment for initial course of treatment in model 6, there was evidence that the association between census tract persistent poverty and both cancer-specific and overall survival violated the proportionality assumption, indicating that the effect may vary over time. This violation was addressed by stratifying the analysis based on follow-up time (<36 months vs. ≥36 months) using the stsplit command in Stata. We selected 36 months as the cut point based on visual inspection of time-dependent HR estimates, which indicated that the proportional hazards assumption began to diverge around the third year of follow-up. As a sensitivity analysis, we examined associations between persistent poverty and cancer-specific mortality using Fine and Gray’s proportional subdistribution hazard models to account for competing risks. All analyses were performed using the Stata 15.1 statistical software package (Stata Corporation 2017).
Results
Descriptive statistics of the study population
There were 56,373 deaths, representing 76.5% of the sample, over 168,715 person-years of follow-up. Of these deaths, 81.0% (n = 45,646) were attributed to cancer. A total of 11.3% (n = 8,580) of patients with liver cancer lived in persistent poverty census tracts at the time of diagnosis. Among patients living in persistent poverty neighborhoods, 7,001 (81.6%) died during follow-up, with 80.2% (n = 5,616) of deaths attributed to cancer. Among patients living in nonpersistent poverty neighborhoods, 49,372 (75.8%) died during follow-up, with 81.1% (n = 40,030) of deaths attributed to cancer.
Overall, 75.1% of liver cancer cases occurred in males, 53% were diagnosed before age 65 years, and 50.7% were non-Hispanic White. On average, patients with liver cancer residing in persistent poverty census tracts were more likely to be less than age 65 years at diagnosis, non-Hispanic Black, Hispanic, or American Indian/Alaska Native, and living in urban areas, compared with those in nonpersistent poverty tracts. They were also more likely to be diagnosed at a distant stage and less likely to receive chemotherapy, radiation, or surgery (Table 1).
Table 1.
Descriptive characteristics of patients with liver cancer diagnosed from 2006 to 2019 in the SEER 17 Registries, overall and by persistent poverty census tract status.
| Characteristic | Persistent poverty census tract | Nonpersistent poverty census tract | Total | P valuea |
|---|---|---|---|---|
| (n = 8,580) | (n = 65,115) | (n = 73,695) | ||
| n (%) | n (%) | n (%) | ||
| Age at diagnosis | | | | <0.001 |
| <65 years | 5,215 (60.8) | 33,854 (52) | 39,069 (53) | |
| ≥65 years | 3,365 (39.2) | 31,261 (48) | 34,626 (47) | |
| Sex | | | | 0.59 |
| Female | 2,119 (24.7) | 16,254 (25) | 18,373 (24.9) | |
| Male | 6,461 (75.3) | 48,861 (75) | 55,322 (75.1) | |
| Race and ethnicity | | | | <0.001 |
| American Indian/Alaska Native | 160 (1.9) | 633 (1) | 793 (1.1) | |
| Hispanic | 2,538 (29.6) | 12,533 (19.3) | 15,071 (20.5) | |
| Native Hawaiian/Pacific Islander | 50 (0.6) | 615 (0.9) | 665 (0.9) | |
| Non-Hispanic Asian American | 705 (8.2) | 10,191 (15.7) | 10,896 (14.8) | |
| Non-Hispanic Black | 2,634 (30.7) | 6,294 (9.7) | 8,928 (12.1) | |
| Non-Hispanic White | 2,493 (29.1) | 34,849 (53.5) | 37,342 (50.7) | |
| Rurality | | | | <0.001 |
| Rural | 793 (9.2) | 6,913 (10.6) | 7,706 (10.5) | |
| Urban | 7,787 (90.8) | 58,202 (89.4) | 65,989 (89.5) | |
| Stage at diagnosis | | | | <0.001 |
| Localized | 4,605 (53.7) | 35,908 (55.2) | 40,513 (55) | |
| Regional | 2,577 (30) | 19,903 (30.6) | 22,480 (30.5) | |
| Distant | 1,398 (16.3) | 9,304 (14.3) | 10,702 (14.5) | |
| Chemotherapy | | | | <0.001 |
| No/unknown | 5,039 (58.7) | 36,232 (55.6) | 41,271 (56) | |
| Yes | 3,541 (41.3) | 28,883 (44.4) | 32,424 (44) | |
| Radiation | | | | |
| No/unknown | 7,635 (89) | 57,248 (88) | 64,883 (88) | 0.004 |
| Yes | 945 (11) | 7,867 (12.1) | 8,812 (12) | |
| Surgery | | | | <0.001 |
| No/unknown | 6,731 (78.5) | 46,412 (71.3) | 53,143 (72.1) | |
| Yes | 1,849 (21.6) | 18,703 (28.7) | 20,552 (27.9) | |
P value estimated using the χ2 test.
Survival curves stratified by census tract persistent poverty status
Figure 1 displays the Kaplan–Meier survival curves for patients with liver cancer, stratified by persistent poverty status; A shows cancer-specific survival, and B shows overall survival. Patients with liver cancer residing in nonpersistent poverty census tracts had higher cancer-specific survival and overall survival compared with those living in persistent poverty tracts. The median cancer-specific survival time was 22 months for patients in nonpersistent poverty census tracts compared with 16 months for those in persistent poverty census tracts (log-rank test P value <0.001). The median overall survival time was 16 months for patients in nonpersistent poverty census tracts compared with 12 months for those in persistent poverty census tracts (log-rank test P value <0.001).
Figure 1.
Kaplan–Meier survival curves by census tract persistent poverty among patients with liver cancer diagnosed from 2006 to 2019 in the SEER 17 Registries, n = 73,695. A, Cancer-specific survival curves and rates. B, Overall survival curves and rates.
Overall associations between census tract persistent poverty and survival
Before adjusting for covariates, residence in a persistent poverty census tract was associated with a 19% higher risk of cancer-specific mortality (Table 2, model 1: HR, 1.19; 95% CI, 1.16–1.22). After adjusting for patient demographics and stage at diagnosis, the association was modestly attenuated to a 16% higher cancer-specific mortality risk (model 5: HR, 1.16; 95% CI, 1.12–1.19). Further adjustment for initial treatment reduced the association to a 9% higher cancer-specific mortality risk (model 6: HR, 1.09; 95% CI, 1.06–1.12). Similar patterns were observed in the sensitivity analysis using Fine and Gray subdistribution hazard models, which accounted for competing risks in the associations between persistent poverty and cancer-specific mortality. Similarly, the association between persistent poverty and all-cause mortality showed a 21% higher risk in the unadjusted model (model 1: HR, 1.21; 95% CI, 1.18–1.24), which decreased to an 11% higher risk in the fully adjusted model (model 6: HR, 1.11; 95% CI, 1.09–1.14).
Table 2.
Association between residence in a persistent poverty census tract and mortality risk among patients with liver cancer diagnosed from 2006 to 2019 in the SEER 17 Registries, n = 73,695.
| Model | Cancer-specific mortality | All-cause mortality |
|---|---|---|
| HR (95% CI) | HR (95% CI) | |
| Model 1: unadjusted | 1.19 (1.16–1.22) | 1.21 (1.18–1.24) |
| Model 2: adjusted for age and sex | 1.21 (1.17–1.24) | 1.23 (1.20–1.27) |
| Model 3: model 2 + race and ethnicity | 1.17 (1.14–1.20) | 1.19 (1.16–1.22) |
| Model 4: model 3 + rurality | 1.17 (1.14–1.20) | 1.19 (1.16–1.22) |
| Model 5: model 4 + stage at diagnosis | 1.16 (1.12–1.19) | 1.18 (1.15–1.21) |
| Model 6: model 5 + initial treatmenta | 1.09 (1.06, 1.12) | 1.11 (1.09, 1.14) |
Includes variables for receipt of surgery, radiation, and chemotherapy (yes vs. no/unknown).
Stratified associations between census tract persistent poverty and survival
In the fully adjusted model stratified by follow-up time, living in a persistent poverty census tract was associated with an 8% higher risk of cancer-specific mortality within the first three years after diagnosis (Table 3, HR: 1.08; 95% CI, 1.05–1.11) and a 22% higher risk among those who survived three years or more after diagnosis (HR: 1.22; 95% CI, 1.12–1.33). For all-cause mortality, living in a persistent poverty census tract was associated with an 10% higher risk of cancer-specific mortality within the first three years after diagnosis (HR: 1.10; 95% CI, 1.07–1.13) and a 22% higher risk among those who survived three years or more after diagnosis (HR: 1.22; 95% CI, 1.14–1.31). As shown in Supplementary Table S1, the use of different follow-up time cut points further supports that the association between census tract persistent poverty and mortality risk is stronger after three or more years of follow-up time compared with within the first three years after a liver cancer diagnosis.
Table 3.
Association between residence in a persistent poverty census tract and mortality risk, stratified by follow-up time, among patients with liver cancer diagnosed from 2006 to 2019 in the SEER 17 Registries, n = 73,695.
| Follow-up time | n | Cancer-specific mortality | All-cause mortality | ||
|---|---|---|---|---|---|
| Events | HR (95% CI)a | Events | HR (95% CI)a | ||
| <36 months | 54, 855 | 40,151 | 1.08 (1.05–1.11) | 48,338 | 1.10 (1.07–1.13) |
| ≥36 months | 18,840 | 5,495 | 1.22 (1.12–1.33) | 8,035 | 1.22 (1.14–1.31) |
Estimates are adjusted for age at diagnosis, sex, race and ethnicity, rurality, stage at diagnosis, and initial course of treatment (chemotherapy, radiation, and surgery).
When stratified by age at diagnosis, the associations between living in a persistent poverty census tract and both cancer-specific and all-cause mortality were stronger among individuals less than 65 years compared with those 65 years and older at liver cancer diagnosis (Table 4, interaction term P values <0.001). For cancer-specific mortality, the HR was 1.14 (95% CI, 1.10–1.19) for individuals diagnosed before age 65 years, compared with 1.02 (95% CI, 0.98–1.07) for those aged 65 years and older. For all-cause mortality, the HR was 1.17 (95% CI, 1.13–1.21) for individuals less than 65 years and 1.04 (95% CI, 1–1.08) for those 65 years and older. The association between census tract persistent poverty and mortality also differed by stage at diagnosis (interaction term P value = 0.03 for cancer-specific mortality and P value = 0.04 for all-cause mortality). For individuals diagnosed with localized disease, persistent poverty was associated with a 14% higher risk of cancer-specific mortality (HR: 1.14; 95% CI, 1.09–1.18) and a 15% higher risk of all-cause mortality (HR: 1.15, 95% CI, 1.11–1.19). Among those diagnosed with regional stage disease, the corresponding risks were 5% higher for cancer-specific mortality (HR: 1.05; 95% CI, 1–1.10) and 8% higher for all-cause mortality (HR: 1.08; 95% CI, 1.04–1.13). For patients diagnosed with distant-stage disease, the risk was 7% higher for both cancer-specific (HR: 1.07, 95% CI, 1–1.13) and all-cause mortality (HR: 1.07; 95% CI, 1.01–1.13). The association between census tract persistent poverty and cancer-specific and all-cause mortality did not vary significantly by sex, race and ethnicity, or rurality (all interaction term P values >0.05).
Table 4.
Association between residence in a persistent poverty census tract and mortality risk, stratified by demographic and prognostic characteristics, among patients with liver cancer diagnosed from 2006 to 2019 in the SEER 17 Registries, n = 73,695.
| Stratifying variable | Cancer-specific mortality | All-cause mortality | ||
|---|---|---|---|---|
| HR (95% CI)a | Interaction term P value | HR (95% CI)a | Interaction term P value | |
| Age at diagnosis | | <0.001 | | <0.001 |
| <65 years | 1.14 (1.10–1.19) | | 1.17 (1.13–1.21) | |
| ≥65 years | 1.02 (0.98–1.07) | | 1.04 (1–1.08) | |
| Sex | | 0.14 | | 0.06 |
| Female | 1.05 (0.99–1.11) | | 1.07 (1.01–1.12) | |
| Male | 1.10 (1.07–1.14) | | 1.13 (1.10–1.16) | |
| Race and ethnicity | | 0.43 | | 0.65 |
| American Indian/Alaska Native | 1.05 (0.85–1.30) | | 1.08 (0.89–1.30) | |
| Hispanic | 1.05 (1–1.11) | | 1.08 (1.03–1.13) | |
| Native Hawaiian/Pacific Islander | 0.95 (0.65–1.40) | | 1.14 (0.82–1.57) | |
| Non-Hispanic Asian American | 1.12 (1.01–1.23) | | 1.14 (1.05–1.25) | |
| Non-Hispanic Black | 1.09 (1.03–1.15) | | 1.11 (1.06–1.17) | |
| Non-Hispanic White | 1.13 (1.07–1.19) | | 1.14 (1.09–1.19) | |
| Rurality | | 0.69 | | 0.73 |
| Rural | 1.11 (1.02–1.21) | | 1.13 (1.04–1.22) | |
| Urban | 1.09 (1.06–1.12) | | 1.11 (1.08–1.14) | |
| Stage at diagnosis | | 0.03 | | 0.04 |
| Localized | 1.14 (1.09–1.18) | | 1.15 (1.11–1.19) | |
| Regional | 1.05 (1–1.10) | | 1.08 (1.04–1.13) | |
| Distant | 1.07 (1–1.13) | | 1.07 (1.01–1.13) | |
Estimates are adjusted for age at diagnosis, sex, race and ethnicity, rurality, stage at diagnosis, and initial course of treatment (chemotherapy, radiation, and surgery).
Discussion
This study provides some of the first data on the association between living in a persistent poverty census tract and survival outcomes among patients with liver cancer in the United States. Overall, we found that living in a persistent poverty census tract was associated with a higher risk of both cancer-specific mortality and all-cause mortality among patients with liver cancer. These associations remained largely unchanged after adjusting for demographic characteristics, including sex, age, race and ethnicity, rurality, as well as stage at diagnosis. Further adjustment for the initial course of treatment, including chemotherapy, radiation, and surgery, attenuated the associations, but they remained statistically significant. Notably, after accounting for treatment differences in the fully adjusted model, the association between census tract persistent poverty and mortality was stronger among individuals who survived three or more years compared with those who survived fewer than three years after a liver cancer diagnosis. This suggests that, after accounting for treatment-related differences, which may have a greater impact on short-term survival outcomes, the impact of persistent poverty on mortality risk becomes more pronounced over time, particularly among longer-term survivors. However, additional research is needed to identify the downstream factors contributing to poorer survival outcomes in patients with liver cancer exposed to persistent poverty neighborhoods.
Whereas this study provides some of the first data on the relationship between persistent poverty at the census tract level and liver cancer survival, our findings are consistent with previous studies that have evaluated the association at the county level. For example, one previous study reported that liver cancer mortality rates were 25.8% higher in persistent poverty counties compared with nonpersistent poverty counties in the United States (10). Another study conducted a geospatial analysis of county-level mortality rates in the United States and found that hotspots for higher liver cancer mortality rates were more likely to occur in persistent poverty counties than in nonpersistent poverty counties (23). Our findings suggest that the negative effects of living in counties with persistent poverty on liver cancer outcomes are also evident at the census tract level. This suggests that factors operating at both broader and more local levels may contribute to disparities in liver cancer survival. These factors may include macro-level influences such as state or county health policies, healthcare infrastructure, and regional economic disinvestment, as well as micro-level influences such as neighborhood access to quality healthcare, exposure to environmental toxins, availability of healthy foods, and social stressors related to concentrated poverty. Future research should examine how these multilevel contextual factors work together to shape liver cancer outcomes.
Our findings also provide new evidence that the impact of living in a persistent poverty census tract on increased mortality risk may be stronger in patients diagnosed before age 65 years compared with those diagnosed at age 65 years or older. Younger patients may be particularly vulnerable to the adverse effects of persistent poverty, potentially due to reduced access to health insurance before Medicare eligibility, as well as unique social, economic, and biological factors that shape cancer risk, treatment access, and survivorship in younger adults. The observed age-specific differences in this study might also reflect differences in underlying tumor biology or underlying risk factors and comorbidities that may contribute to more aggressive disease and poorer prognoses among younger patients with liver cancer, potentially compounded by the effects of living in persistent poverty neighborhoods. However, because this is the first study to report age-specific differences in the association between area-level persistent poverty and liver cancer outcomes and given that we did not have data on individual-level risk factors, further research is needed to confirm these findings and identify the mechanisms underlying age-related differences in liver cancer survival.
Aside from the observed age-related differences, the association between persistent poverty and mortality risk did not significantly vary by other demographic factors, including race and ethnicity. This suggests that the detrimental impact of persistent poverty on survival may be widespread across diverse racial and ethnic groups within this population. Yet, it is important to note that racial and ethnic minority groups are disproportionately exposed to persistent poverty neighborhoods in the United States. In our study sample, 30% of non-Hispanic Black patients with liver cancer lived in persistent poverty census tracts, compared with only 7% of non-Hispanic White patients. Among American Indian/Alaska Native and Hispanic patients, 20% and 17% resided in persistent poverty census tracts, respectively. These data are consistent with previous research that has shown racial and ethnic minority groups are disproportionately exposed to area-level poverty and deprivation in the United States (10), in part due to racial residential segregation shaped by discriminatory housing policies (24, 25). As a result, although the relative impact of area-level persistent poverty on liver cancer survival may be similar across racial and ethnic groups, the absolute burden of exposure and its contribution to population-level disparities are disproportionately higher among minority communities.
When we examined the results by stage at diagnosis, living in a persistent poverty census tract was associated with higher mortality risk at every stage. However, the strongest association was observed among patients diagnosed with localized disease. This suggests that the impact of living in a persistent poverty neighborhood may be greatest among patients diagnosed with early-stage disease, who are typically the most likely to benefit from curative treatments. We found that patients with liver cancer residing in persistent poverty census tracts were less likely to receive chemotherapy, radiation, or surgery as part of their initial treatment compared with those living in nonpersistent poverty census tracts, supporting the hypothesis that treatment-related differences contribute to survival disparities. We also found that adjustment for the initial course of treatment attenuated the association between census tract persistent poverty and mortality risk. This supports that treatment-related differences may be contributing to liver cancer survival disparities. Furthermore, although we did not have data on the timing of treatment in this study, prior research has shown that therapeutic delays are more common in patients with low socioeconomic status and are associated with reduced overall survival in patients with liver cancer (7, 14, 26). Therefore, treatment delays may represent another key pathway through which living in persistent poverty neighborhoods contributes to increased mortality risk, potentially driven by underinvestment in healthcare infrastructure that hinders timely access to care (16).
Another potential pathway linking residence in a persistent poverty neighborhood to higher liver cancer mortality risk is increased allostatic load, which refers to the cumulative physiologic consequences on the body from chronic exposure to stress (27). Individuals living in communities with persistent poverty often face ongoing socioeconomic hardship and structural discrimination, which can lead to repeated activation of stress-response systems over time. Chronic inflammation, a core component of allostatic load, may promote cancer progression by altering the tumor microenvironment and increasing the risk of metastasis (28). A recent meta-analysis found that higher allostatic load was associated with increased risk of cancer-specific mortality (29). Chronic stress has also been linked to worse survival outcomes in patients with breast cancer and non–small cell lung cancer (30–32), though no studies to date have specifically examined this relationship in liver cancer.
Research studies on other cancer sites found similar associations between socially disadvantaged residence and cancer survival. Living in persistently poor census tracts was associated with a higher risk of breast cancer–specific and overall mortality in women diagnosed with stages I to III breast cancer (33). Cancer survival was also associated with persistent poverty in populations with oral and pharynx cancer (34) and hepatopancreatic biliary cancer (35), suggesting that exposure to area-level persistent poverty may influence cancer survival through shared pathways across multiple cancer types.
This study has several strengths, including being one of the first to utilize SEER case listing data to examine how living in persistent poverty census tract affects liver cancer survival outcomes. However, this study also has limitations. Due to constraints in the SEER database, we were unable to account for important individual-level variables such as insurance status, occupation, or education level, which may independently affect survival outcomes beyond the influence of neighborhood-level factors. Furthermore, the lack of data on healthcare utilization and detailed treatment beyond initial receipt limited our ability to directly assess this pathway. We also note that SEER provides only a dichotomous measure of persistent poverty, which limits our ability to examine potential gradients or dose–response relationships in exposure. Nevertheless, SEER uses the standard federal definition of persistent poverty, enabling meaningful comparisons with other studies using the same classification. Because the SEER 17 Registries cover selected geographic regions, covering approximately 26.5% of the US population (36), the findings may not be fully generalizable to the broader US population. We also could not disaggregate broad racial and ethnic categories (e.g., non-Hispanic Asian American), which combine multiple subgroups with distinct risk profiles, and thus future research is needed to better understand how poverty and race/ethnicity intersect to influence liver cancer outcomes. Another limitation is that SEER only captures the address at diagnosis, and thus neighborhood characteristics may be misclassified if individuals moved during follow-up. This also prevented us from evaluating exposure to persistent poverty earlier in the life course, prior to diagnosis. Neighborhood conditions at the time of diagnosis, however, may be particularly important for shaping access to healthcare and the treatment patients receive. Finally, we acknowledge that conducting multiple statistical analyses increases the likelihood of false-positive results; therefore, statistical significance should be interpreted with caution. Despite these limitations, this study provides important new evidence on the contribution of neighborhood-level persistent poverty to liver cancer mortality in the United States.
In conclusion, our study supports that neighborhood-level persistent poverty may be a significant determinant of liver cancer survival, particularly among patients diagnosed with localized disease and before age 65 years. These findings highlight how socioeconomic inequities shape cancer prognosis and reinforce the urgent need to address structural barriers to healthcare access and the affordability of curative treatment. Interventions and policies that target economically disadvantaged communities may be essential for reducing survival disparities and improving outcomes for patients with liver cancer.
Supplementary Material
Supplementary Table S1. Association between residence in a persistent poverty census tract and mortality risk, stratified by different follow-up times, among liver cancer patients diagnosed from 2006 to 2019 in the SEER17 Registries, n = 73,695
Acknowledgments
R.D. Kehm received grant R00CA263024 from the National Cancer Institute at the National Institutes of Health.
Footnotes
Note: Supplementary data for this article are available at Cancer Epidemiology, Biomarkers & Prevention Online (http://cebp.aacrjournals.org/).
Data Availability
The data generated in this study are available upon request from the corresponding author.
Authors’ Disclosures
R.D. Kehm reports grants from the National Cancer Institute during the conduct of the study. No disclosures were reported by the other author.
Authors’ Contributions
R. Wu: Conceptualization, data curation, formal analysis, writing–original draft, writing–review and editing. R.D. Kehm: Conceptualization, funding acquisition, writing–original draft, writing–review and editing.
References:
- 1. Siegel RL, Kratzer TB, Giaquinto AN, Sung H, Jemal A. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Rahib L, Wehner MR, Matrisian LM, Nead KT. Estimated projection of US cancer incidence and death to 2040. JAMA Netw Open 2021;4:e214708. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Rich NE, Carr C, Yopp AC, Marrero JA, Singal AG. Racial and ethnic disparities in survival among patients with hepatocellular carcinoma in the United States: a systematic review and meta-analysis. Clin Gastroenterol Hepatol 2022;20:e267–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. GBD US Health Disparities Collaborators . Burden of liver cancer mortality by county, race, and ethnicity in the USA, 2000–19: a systematic analysis of health disparities. Lancet Public Health 2024;9:e186–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ha J, Yan M, Aguilar M, Bhuket T, Tana MM, Liu B, et al. Race/ethnicity-specific disparities in cancer incidence, burden of disease, and overall survival among patients with hepatocellular carcinoma in the United States. Cancer 2016;122:2512–23. [DOI] [PubMed] [Google Scholar]
- 6. Pan C-W, Danpanichkul P, Cheung R, Pinheiro P, Jones PD, Wong RJ. Trends in urban-rural disparities in hepatocellular carcinoma mortality in the United States from 1999 to 2020. J Clin Gastroenterol 2025;10:1097. [DOI] [PubMed] [Google Scholar]
- 7. Wagle NS, Park S, Washburn D, Ohsfeldt RL, Rich NE, Singal AG, et al. Racial, ethnic, and socioeconomic disparities in treatment delay among patients with hepatocellular carcinoma in the United States. Clin Gastroenterol Hepatol 2023;21:1281–92.e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Wagle NS, Park S, Washburn D, Ohsfeldt RL, Rich NE, Singal AG, et al. Racial, ethnic, and socioeconomic disparities in curative treatment receipt and survival in hepatocellular carcinoma. Hepatol Commun 2022;6:1186–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Flores YN, Datta GD, Yang L, Corona E, Devineni D, Glenn BA, et al. Disparities in hepatocellular carcinoma incidence, stage, and survival: a large population-based study. Cancer Epidemiol Biomarkers Prev 2021;30:1193–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Moss JL, Pinto CN, Srinivasan S, Cronin KA, Croyle RT. Persistent poverty and cancer mortality rates: an analysis of county-level poverty designations. Cancer Epidemiol Biomarkers Prev 2020;29:1949–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Miller KK, Crandall MS, Weber BA. Persistent poverty and place: how do persistent poverty and poverty demographics vary across the rural-urban continuum. Measuring Rural Divers 2002:1–41. [Google Scholar]
- 12. Bennett KJ, Probst JC, Pumkam C. Obesity among working age adults: the role of county-level persistent poverty in rural disparities. Health Place 2011;17:1174–81. [DOI] [PubMed] [Google Scholar]
- 13. Moss JL, Geyer NR, Lengerich EJ. Patterns of cancer-related healthcare access across Pennsylvania: analysis of novel census tract-level indicators of persistent poverty. Cancer Epidemiol Biomarkers Prev 2024;33:616–23. [DOI] [PubMed] [Google Scholar]
- 14. Mohamed KA, Ghabril M, Desai A, Orman E, Patidar R, Holden J, et al. Neighborhood poverty is associated with failure to be waitlisted and death during liver transplantation evaluation. Liver Transpl 2022;28:1441–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Rawicz-Pruszyński K, Woldesenbet S, Endo Y, Munir MM, Katayama E, Waqar U, et al. Persistent poverty impacts access to minimally invasive surgery among patients with hepatopancreatobiliary cancer. J Surg Oncol 2023;128:823–30. [DOI] [PubMed] [Google Scholar]
- 16. Benzow A, Fikri K, Kim J, Newman D. Advancing economic development in persistent-poverty communities CASE. [cited 2026 Jan 9]. Available from:https://eig.org/persistent-poverty-in-communities/.
- 17. Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Int J Surg 2014;12:1495–9.25046131 [Google Scholar]
- 18. National Cancer Institute Surveillance, Epidemiology, and End Results Program . Incidence data with census tract attributes database. [cited 2026 Jan 9]. Available from:https://seer.cancer.gov/data/specialized/available-databases/census-tract-request/.
- 19. Dalaker . The 10-20-30 provision: defining persistent poverty counties. [cited 2026 Jan 9]. Available from:https://www.congress.gov/crs-product/R45100.
- 20. USDA Economic Research Service . Rural poverty & well-being. [cited 2026 Jan 9]. Available from:https://www.ers.usda.gov/topics/rural-economy-population/rural-poverty-well-being.
- 21. National Cancer Institute Surveillance . Epidemiology, and end results program SEER Summary staging manual. 2000. [cited 2026 Jan 9]. Available from:https://seer.cancer.gov/tools/ssm/ssm2000/SSSM2000-122012.pdf.
- 22. Moss JL, Stinchcomb DG, Yu M. Providing higher resolution indicators of rurality in the Surveillance, Epidemiology, and End Results (SEER) database: implications for patient privacy and research. Cancer Epidemiol Biomarkers Prev 2019;28:1409–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Ledenko M, Patel T. Association of county level poverty with mortality from primary liver cancers. Cancer Med 2024;13:e7463. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Faber JW. We built this: consequences of new deal era intervention in America’s racial geography. Am Sociological Rev 2020;85:739–75. [Google Scholar]
- 25. Swope CB, Hernández D, Cushing LJ. The relationship of historical redlining with present-day neighborhood environmental and health outcomes: a scoping review and conceptual model. J Urban Health 2022;99:959–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Govalan R, Luu M, Lauzon M, Kosari K, Ahn JC, Rich NE, et al. Therapeutic underuse and delay in hepatocellular carcinoma: prevalence, associated factors, and clinical impact. Hepatol Commun 2022;6:223–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Guidi J, Lucente M, Sonino N, Fava GA. Allostatic load and its impact on health: a systematic review. Psychotherapy and psychosomatics 2021;90:11–27. [DOI] [PubMed] [Google Scholar]
- 28. He X-Y, Gao Y, Ng D, Michalopoulou E, George S, Adrover JM, et al. Chronic stress increases metastasis via neutrophil-mediated changes to the microenvironment. Cancer Cell 2024;42:474–86.e12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Mathew A, Doorenbos AZ, Li H, Jang MK, Park CG, Bronas UG. Allostatic load in cancer: a systematic review and mini meta-analysis. Biol Res Nurs 2021;23:341–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Chen J, Elsaid MI, Handley D, Plascak JJ, Andersen BL, Carson WE, et al. Association between neighborhood opportunity, allostatic load, and all-cause mortality in patients with breast cancer. J Clin Oncol 2024;42:1788–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Chen J, Handley D, Elsaid MI, Plascak JJ, Andersen BL, Carson WE, et al. The implications of racialized economic segregation and allostatic load on mortality in patients with breast cancer. Ann Surg Oncol 2024;31:365–75. [DOI] [PubMed] [Google Scholar]
- 32. Obeng-Gyasi S, Li Y, Carson WE, Reisenger S, Presley CJ, Shields PG, et al. Association of allostatic load with overall mortality among patients with metastatic non–small cell lung cancer. JAMA Netw Open 2022;5:e2221626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Chen JC, Handley D, Elsaid MI, Fisher JL, Plascak JJ, Anderson L, et al. Persistent neighborhood poverty and breast cancer outcomes. JAMA Netw Open 2024 1;7:e2427755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Karanth S, Mistry S, Wheeler M, Akinyemiju T, Divaker J, Yang JJ, et al. Persistent poverty disparities in incidence and outcomes among oral and pharynx cancer patients. Cancer Causes Control 2024;35:1063–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Lima HA, Woldesenbet S, Hamad A, Alaimo L, Moazzam Z, Endo Y, et al. Hepatopancreaticobiliary cancer outcomes are associated with county-level duration of poverty. Surgery 2023;173:1411–8. [DOI] [PubMed] [Google Scholar]
- 36. National Cancer Institute Surveillance, Epidemiology, and End Results Program . Number of persons by race and hispanic ethnicity for SEER participants (2020 census data). [cited 2026 Jan 9]. Available from:https://seer.cancer.gov/registries/data.html.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Table S1. Association between residence in a persistent poverty census tract and mortality risk, stratified by different follow-up times, among liver cancer patients diagnosed from 2006 to 2019 in the SEER17 Registries, n = 73,695
Data Availability Statement
The data generated in this study are available upon request from the corresponding author.