The Problem With “Asian,” “Hispanic,” or “Black” as Single Risk Categories

Hepatocellular carcinoma risk is rarely distributed evenly across a population. Clinicians see this every day: patients with chronic hepatitis B, hepatitis C, metabolic dysfunction-associated fatty liver disease, alcohol-related liver disease, cirrhosis, immigration-related viral hepatitis risk, and uneven access to surveillance do not fit neatly into broad epidemiologic labels. Yet much of the research literature and many health-system dashboards still group patients into large racial or ethnic categories that may conceal clinically meaningful differences.

The article “Detailed racial/ethnic disparities in hepatocellular carcinoma risk with an electronic health record-based cohort” by DeRouen and colleagues was listed in Clinical Gastroenterology and Hepatology as an Article in Press in early August 2026. The official journal listing identifies the article title and first author, and a ScienceDirect listing describes the study question as assessing sex-specific disparities in hepatocellular carcinoma risk according to detailed race and ethnicity within a multi-institutional electronic health record cohort.

The available detailed abstract from the same EHR-based cohort describes a large linked cohort assembled from three health care systems, with race and ethnicity characterized into more granular groups, then linked to cancer registry data for incident HCC. This is observational epidemiology, not an intervention trial. It should therefore be interpreted as evidence of associations and disparity patterns, not as proof that race or ethnicity itself causes HCC.

A Large EHR-Registry Linkage, Not a Conventional Clinic Cohort

The cohort described in the accessible abstract included adults with at least one in-person encounter between 2000 and 2017 within three health care systems: San Francisco Health Network, Sutter Health Northern California, and Kaiser Permanente Hawai’i. EHR data were linked to population-based state cancer registry data to identify incident HCC. The cohort included 4,249,671 adults, with a median follow-up of 6.8 years; 55% were female, and 2,916 incident HCC cases were identified.

This design has clear strengths. First, the sample size is large enough to examine subgroups that would be invisible or underpowered in smaller cohorts. Second, linkage to cancer registry data strengthens ascertainment of incident HCC compared with relying only on diagnosis codes within a single health system. Third, use of EHR data allows investigators to incorporate clinical and sociodemographic variables not always available in cancer registries.

But the design also carries limitations. EHR cohorts are shaped by who enters the health system, how often they receive care, how diagnoses are coded, and how race and ethnicity are recorded. A person’s documented racial or ethnic category is not a biologic exposure; it is a social, demographic, administrative, and sometimes imperfectly captured variable. Therefore, the findings should be used to identify disparities and research needs, not to biologize race or to assign deterministic risk to individuals.

The Exposure Was Detailed Race and Ethnicity—But the Outcome Was Incident HCC

The main exposure of interest was detailed racial and ethnic classification, derived from EHR data. The investigators used EHR information to define 17 detailed racial and ethnic groups, alongside other sociodemographic and clinical factors. They used Cox proportional hazards regression with age as the time scale and sex-specific models adjusted for length of active follow-up, with stratification by site, birth cohort, baseline year, and number of encounters. They also calculated sex-specific hazards and population attributable fractions for HCC risk factors among racial and ethnic groups.

The outcome was incident hepatocellular carcinoma identified through linked cancer registry data. This distinction matters because the study was not evaluating treatment response, survival, surveillance adherence, transplant referral, or liver-related mortality. It was focused on HCC risk occurrence in a very large, diverse, real-world cohort.

For clinicians, this means the findings should primarily inform thinking about risk stratification, prevention research, surveillance equity, and public health targeting. They do not by themselves answer whether a particular subgroup should undergo surveillance outside established guideline indications.

The Most Striking Signal: Vietnamese American Patients Had the Highest Relative Risk Reported

The accessible abstract reports that, compared with non-Hispanic White males, Vietnamese American males had greater HCC risk, with a hazard ratio of 7.42 and 95% confidence interval 4.25–12.96. Increased risk was also reported among American Indian/Alaska Native, Black, Chinese American, Hispanic, Native Hawaiian, Pacific Islander males, and males of multiple races or ethnicities.

Among females, nearly every group except American Indian/Alaska Native, Asian Indian American, and Pacific Islander females had higher HCC risk than non-Hispanic White females. Reported hazard ratios ranged from 1.68 among Filipino females to 6.38 among Vietnamese American females.

The clinical interpretation should be cautious but attentive. These hazard ratios indicate associations within this cohort after the specified modeling approach. They do not mean that Vietnamese ethnicity itself causes HCC. Rather, they likely reflect a combination of underlying liver disease burden, viral hepatitis epidemiology, migration history, structural determinants, health care access, surveillance patterns, and other measured or unmeasured factors.

For GastroAGI readers, the practical message is that broad categories can be misleading. A clinician looking only at “Asian American” as a single category may miss the magnitude of risk observed in some subgroups while overgeneralizing risk to others.

Why Disaggregation Matters for Hepatology Practice

Hepatologists already practice a form of individualized risk assessment. We ask about HBV status, HCV history, cirrhosis, MASLD, alcohol use, family history, country of birth, prior antiviral therapy, fibrosis stage, and surveillance adherence. Yet health systems often operationalize equity work through broad demographic fields that are too coarse for targeted prevention.

This study strengthens the argument that disaggregated data are clinically and epidemiologically important. If health systems cannot distinguish between detailed Asian American, Native Hawaiian, Pacific Islander, Hispanic, American Indian/Alaska Native, Black, and multiracial groups, they may fail to identify where HCC prevention efforts are most needed.

That does not mean race or ethnicity should replace clinical risk factors. It means these variables may help identify where risk factors cluster, where screening gaps persist, and where outreach needs cultural, linguistic, or system-level adaptation.

Risk Factors Did Not Behave as a Single Uniform Story

The abstract reports that HCC risk according to clinical risk factors was broadly similar for males and females, but one notable difference was observed: the population attributable fraction of HCC for alcohol disorders was much higher for females than males in the available analysis. The abstract reports a PAF of 43% for females compared with 2% for males. It also notes variation in factor-specific HCC risk by racial and ethnic group in preliminary analyses. For example, metabolic-associated fatty liver disease and ever-smoking were associated with greater HCC risk only among non-Hispanic White and Asian American males in preliminary subgroup analyses.

These observations should not be overinterpreted. Population attributable fractions depend on both the strength of association and the prevalence of the exposure in the studied population. They are not individual-level risk predictions. They also depend on how alcohol disorders, metabolic syndrome, MAFLD, viral hepatitis, and smoking were captured in EHR data.

Still, the broader point is highly relevant: the drivers of HCC disparities may not be identical across groups. If viral hepatitis, metabolic risk, alcohol-related liver disease, and surveillance access contribute differently across populations, then a single prevention message is unlikely to be sufficient.

What Clinicians Should Conclude

Clinicians can reasonably conclude that this study supports more granular assessment of HCC disparities. It reinforces the need to avoid assuming that broad race and ethnicity categories are clinically homogeneous. It also supports the use of linked EHR and cancer registry data as a powerful method to identify population-level disparity patterns.

The study may influence practice indirectly by encouraging health systems to improve race and ethnicity data capture, stratify HCC prevention metrics more precisely, and examine whether surveillance eligibility and surveillance completion differ across detailed subgroups.

However, clinicians should not conclude that the study creates new surveillance indications based on race or ethnicity alone. Current HCC surveillance decisions should still be anchored in established risk conditions such as cirrhosis, chronic HBV infection in guideline-defined risk groups, and other validated clinical criteria. This study is best understood as a health-equity and epidemiologic signal that may guide future research and implementation strategies.

What the Study Cannot Prove

The study cannot prove causation. Race and ethnicity are not biologic mechanisms. They are markers that may correlate with ancestry, migration, geography, viral hepatitis exposure, socioeconomic context, structural racism, language access, insurance status, health-system engagement, neighborhood factors, and clinical risk profiles.

The EHR-based design also means that risk factor measurement may be incomplete. Alcohol disorders, smoking, viral hepatitis status, metabolic syndrome, and MAFLD can be undercoded or inconsistently documented. Patients with more encounters may have more complete ascertainment of risk factors. Conversely, patients with limited access or fragmented care may have under-recorded exposures.

Another limitation is generalizability. The cohort came from three health care systems in California and Hawai’i. These settings are highly relevant for studying diverse populations, but findings may not directly generalize to every US region, rural population, uninsured population, or international setting.

Finally, the article’s detailed full text should be reviewed before translating its findings into institutional policy. The accessible data strongly support the concept of disaggregation, but local implementation requires careful attention to data quality, community engagement, and current surveillance guidelines.

Implications for Future Research and Implementation

The most important next step is not simply to replicate hazard ratios. It is to understand what lies behind them. Future studies should clarify the relative contribution of HBV, HCV, MASLD, alcohol-related liver disease, cirrhosis, socioeconomic position, language access, immigration history, health-system utilization, and surveillance completion across detailed racial and ethnic groups.

Health systems should also examine whether their EHR race and ethnicity fields are adequate for disparity research. If categories are too broad, inconsistently collected, or not patient-reported, risk patterns may remain hidden. The related abstract on race and ethnicity characterization in this cohort emphasized that researchers must critically assess racial and ethnic categories typically available from healthcare system repositories and use schemas that avoid masking smaller populations.

For clinicians and researchers, the implementation challenge is to translate epidemiologic insight into equitable care without stereotyping. That means using disaggregated data to identify populations who may benefit from better viral hepatitis testing, linkage to care, fibrosis assessment, surveillance navigation, language-concordant education, and culturally informed outreach.

Clinical Takeaway

This EHR-linked cohort study highlights a central problem in HCC disparities research: broad racial and ethnic categories can conceal clinically meaningful heterogeneity. Vietnamese American patients showed particularly high relative HCC risk in the accessible cohort data, and several other detailed groups also had elevated risk compared with non-Hispanic White reference groups. These are associations, not proof of causation.

For gastroenterologists and hepatologists, the study should not be interpreted as creating race-based surveillance recommendations. Its value is different: it argues for better data granularity, better understanding of subgroup-specific risk drivers, and more precise prevention strategies. The future of equitable HCC prevention will likely depend not only on knowing who has cirrhosis or viral hepatitis, but also on whether our health systems can identify and reach the communities in whom preventable liver cancer risk remains concentrated.

Five Key Clinical Takeaways

1. The verified article is “Detailed racial/ethnic disparities in hepatocellular carcinoma risk with an electronic health record-based cohort,” listed in Clinical Gastroenterology and Hepatology as an Article in Press in early August 2026.

2. The study used a large EHR-based cohort linked to cancer registry data to assess incident HCC risk across detailed racial and ethnic groups.

3. The accessible cohort abstract included 4,249,671 adults, median follow-up 6.8 years, and 2,916 incident HCC cases.

4. Vietnamese American males and females had the highest relative HCC risk reported in the accessible abstract compared with non-Hispanic White reference groups.

5. The study supports more granular disparity research and targeted prevention planning, but it does not establish new race-based clinical surveillance guidance.

Source Reference and Link

DeRouen MC, Thompson CA, and colleagues. Detailed racial/ethnic disparities in hepatocellular carcinoma risk with an electronic health record-based cohort. Clinical Gastroenterology and Hepatology. Article in Press, early August 2026. Official article listing:

Supporting accessible cohort abstract from the same research program: DeRouen MC et al. Examining factors that contribute to racial and ethnic disparities in liver cancer within an EHR-based epidemiologic cohort linked to cancer registries. Cancer Epidemiology, Biomarkers & Prevention. 2023;32(12 Suppl) B138.