Table 4.

Potential Data Sources for Liver Disease Epidemiologic Research

Data source	Country/Region	Strengths	Weaknesses
National Health and Nutrition Examination Survey (NHANES)	United States	Nationally representative sample of non-institutionalized individuals Cirrhosis definition based on interview, examination, laboratory data Provides estimates of undiagnosed cirrhosis	Cross-sectional design Small sample size (~5,000) Potential for selection bias Potential misclassification of mild liver disease
Veterans Affairs (VA)	United States	Large sample size Includes clinical notes, laboratory data, and imaging Well-validated ICD coding strategies for liver disease and its associated complications	Predominantly male population VA enrollees may differ from general population regarding access or delivery of care Limited information on care received outside VA system
Medicare	United States	Large sample size Nationally representative of population age ≥65 years Patients followed until death	No laboratory data available Relies on diagnosis and procedure codes alone and is subject to misclassification
Medicaid	United States	Large sample size Includes sample of patients with low socioeconomic status	No laboratory data available Relies on diagnosis and procedure codes alone and is subject to misclassification
Private-insurance claims data	United States	Large sample size Nationally representative of privately insured population	Unable to ascertain death Enrollment relies on ongoing insurance coverage No laboratory data available Relies on diagnosis and procedure codes alone and is subject to misclassification
National Inpatient Sample (NIS)	United States	Large sample size Nationally-representative sample of inpatient population Includes all payers	No laboratory data available Relies on diagnosis and procedure codes alone and is subject to misclassification Inability to link hospitalizations to individual patients limits longitudinal follow-up post-discharge
National Readmissions Database (NRD)	United States	Large sample size Accurate assessment of patient readmission	No laboratory data available Unable to account for events that may preclude readmission (e.g. death) Relies on diagnosis and procedure codes alone and is subject to misclassification
Medical Expenditure Panel Survey (MEPS)	United States	Nationally-representative of non-institutionalized individuals Includes health care expenditures from all payers	Potential for recall bias Subgroup analyses among certain groups (e.g. race/ethnicity minorities) may not be possible
Surveillance Epidemiology, and End Results (SEER) program	United States	Includes information on clinical information, tumor stage at diagnosis, first treatment and survival Allows linkage to Medicare	Unable to determine etiology or severity of liver disease May not be entirely representative of US given that it only covers selected subset of population Generalizability for Medicare-linked data limited by age of enrollees (≥65 years old)
US Cancer Statistics registry	United States	Nationally-representative data source on HCC incidence from all 50 states (~97% of population)	Unable to determine etiology of liver disease due to lack of laboratory data Unable to determine HCC stage at diagnosis
Organ Procurement and Transplant Network (OPTN)	United States	Granular data on waitlisted individuals, liver transplantation, and post-liver transplant outcomes Linked by UNOS to social security death index	Potential for selection bias by transplant centers
National patient registries	Denmark, Finland, Iceland, Norway, Sweden	Longitudinal, nationwide clinical data with individual-level linkage Includes detailed information on clinical characteristics, laboratory data, imaging, procedures and outcomes	Resource-intensive
Clinical Practice Research Datalink (CPRD)	United Kingdom	Nationally-representative data Granular data on diagnoses and prescriptions of patients seen by participating providers	Covers only a subset of the population Longitudinal follow-up depends on ongoing treatment by a participating practice Limited data on liver-specific information
European Liver Transplant Registry (ELTR)	Europe	Large sample size (155 centers from 28 countries) Standardized questionnaire used Detailed information on liver transplant indications, transplant types and complications	Potential for misclassification due to inaccurate completion of questionnaire No information on patient ethnicity or socioeconomic information
NORDCAN database	Denmark, Finland, Faroe Islands, Greenland, Iceland, Norway, Sweden	Population-based incidence, prevalence and mortality data for HCC Longitudinal data allows for examination of HCC trends	Inclusion of only Nordic countries limits generalizability, particularly to racial/ethnic minorities
Global Burden of Disease (GBD) project	Worldwide	Captures data from a wide range of sources to estimate incidence, prevalence and mortality from liver disease and HCC	Data quality highly variable, particularly in resource-limited areas In many areas, relies on verbal autopsy (post-mortem interview with relatives and/or witnesses of death)