Public health datasets for research
The right source depends on what kind of health question you are asking. For U.S. individual-level exam and lab data, use NHANES. For global disease burden and comparisons across countries, use the Global Burden of Disease study through IHME. For population and maternal-child health in low- and middle-income countries, use the DHS Program. For clinical, record-level critical care data, use MIMIC-IV — but budget time for its access approval process, since it is not an open download. Below we cover each source's coverage, format and access requirements.
The short answer
NHANES, WHO's Global Health Observatory and Our World in Data's health datasets are free, open downloads with no application process. GBD data requires only a free IHME account. DHS survey microdata requires registration and a short review of your stated research purpose, typically granted within one to two business days. MIMIC-IV and UK Biobank are the two sources here with real friction — MIMIC-IV requires completing a human-subjects training course before you get credentialed access, and UK Biobank charges an access fee and requires a signed data agreement. Plan your timeline accordingly, especially if a deadline depends on clinical-record access.
Population and survey-based sources
NHANES (National Health and Nutrition Examination Survey)
NHANES is run by the National Center for Health Statistics, part of the U.S. CDC, and is the only national U.S. survey that combines interviews with physical examinations and laboratory testing. It has run continuously since 1999, with new data released every two years across five domains: demographics, dietary intake, physical examination, laboratory results and questionnaire responses. Public-use files are free to download directly from the CDC website in SAS transport format, no registration required. More granular restricted-use files, which include identifiers that could support re-identification, require a research proposal submitted to the NCHS Research Data Center and are reviewed on a longer timeline. NHANES is well suited to U.S.-focused nutrition, chronic disease and biomarker research; it is not designed for tracking the same individuals over multiple waves, since each two-year cycle draws a new cross-sectional sample. Official site: cdc.gov/nchs/nhanes.
DHS Program (Demographic and Health Surveys)
The DHS Program, funded primarily by USAID, has run nationally representative household surveys on population, health and nutrition in more than 90 low- and middle-income countries since the mid-1980s, with a particular strength in maternal and child health, fertility, and HIV indicators. Data are free for legitimate academic research; you register on the DHS website, submit a dataset request naming your project and intended analysis, and access to unrestricted files for the chosen country is typically granted within one to two business days. A small number of geographically detailed or otherwise sensitive datasets are restricted and require a separate authorization step, with delivery by secure FTP once approved. Because DHS surveys are repeated periodically in many countries with broadly comparable questionnaires, it is also a practical source for before/after or cross-country comparisons in global health work. Official site: dhsprogram.com/data.
UK Biobank
UK Biobank is a very large, deeply phenotyped prospective cohort of about 500,000 UK adults recruited from 2006 to 2010, with genetic, imaging, biomarker and linked health-record data, and is a common choice for genetic epidemiology and biomarker discovery. It is not free: standard access costs £9,000 plus VAT for three years for researchers at eligible institutions, with a reduced fee of around £500 for researchers based in lower-income countries and for some student applications, plus additional compute and storage charges if you use UK Biobank's own Research Analysis Platform. Access requires a signed Material Transfer Agreement between UK Biobank and your institution, in addition to the standard application describing your proposed research. This is a source to plan for early in a grant or project budget, not something you can add at the last minute. Official site: ukbiobank.ac.uk/use-our-data/apply-for-access.
Global and comparative sources
Global Burden of Disease (GBD) via IHME
The Global Burden of Disease study, produced by the Institute for Health Metrics and Evaluation (IHME), estimates mortality and disability across 292 causes of death, 375 diseases and injuries, and 88 risk factors, for 204 countries and territories, with annual estimates back to 1990. It is the standard reference for cross-country disease burden comparisons and health-policy prioritization work. Non-commercial use is free after creating an IHME account; the GBD Results Tool lets you build custom CSV downloads, capped at 100,000 rows per request, so large multi-country, multi-cause pulls may need to be split. The companion Global Health Data Exchange (GHDx) catalogs the underlying input sources GBD draws from, which is useful when you need to trace an estimate back to its original survey or registry. Official site: healthdata.org/data-tools-practices/data-access.
WHO Global Health Observatory (GHO)
The Global Health Observatory is WHO's own data repository, covering more than 1,000 indicators across mortality, the Sustainable Development Goals, noncommunicable diseases, epidemic-prone diseases, health systems and health equity for its 194 member states. Data are open and free, downloadable in Excel, CSV, JSON and XML, with an OData API for programmatic access. GHO is a good complement to GBD when you need official country-reported indicators rather than modeled burden-of-disease estimates, or when you specifically need WHO's own health-system and SDG tracking series. Official site: who.int/data/gho.
Our World in Data — Health
Our World in Data aggregates and visualizes health indicators from WHO, the UN, GBD, World Bank and other primary sources, and republishes them as clean, versioned CSV downloads alongside interactive charts. Content and data produced directly by Our World in Data are under a CC BY license; data sourced from third parties carries that provider's own license, so check the citation panel under each chart before reuse. It will not replace a primary source for detailed methodology questions, but it is a fast way to get a clean, already-merged panel across many countries and years without writing your own harmonization pipeline. Official site: ourworldindata.org.
Clinical and record-level data
MIMIC-IV via PhysioNet
MIMIC-IV is a large, de-identified critical care database from a major U.S. academic medical center, distributed through PhysioNet (hosted by MIT), covering patient demographics, vital signs, laboratory results, medications, procedures and clinical notes for tens of thousands of ICU and emergency department stays. It is a leading benchmark for clinical machine learning research. Access is not immediate: you need a credentialed PhysioNet account, which requires completing the CITI Program's "Data or Specimens Only Research" training course, having a recognized reference or institutional affiliation, and signing a data use agreement that prohibits re-identification attempts and commercial redistribution. Once credentialed, the data itself is free. Budget at least several days, and often longer, for the training and credentialing steps before your first download. Official site: physionet.org/about/citi-course.
Public health sources side by side
| Source | Unit / scope | Access | Best for |
|---|---|---|---|
| NHANES | Individual, U.S., cross-sectional waves | Free, open download | U.S. nutrition, biomarkers, chronic disease |
| DHS Program | Household/individual, 90+ countries | Free, register + 1–2 day review | Maternal-child health, fertility, HIV in LMICs |
| UK Biobank | Individual, UK cohort, ~500,000 | Paid (from £500), MTA required | Genetic epidemiology, biomarker discovery |
| GBD / IHME | Country-level modeled estimates | Free account, 100k-row request cap | Cross-country disease burden comparisons |
| WHO GHO | Country-level official indicators | Free, open download + API | Official WHO health-system and SDG tracking |
| Our World in Data | Country-level, aggregated | Free, open download | Fast cross-source panels without custom merging |
| MIMIC-IV | Patient-record, single U.S. center | Free after CITI training + credentialing | Clinical machine learning, critical care research |
How to choose
Start from your unit of analysis. If you need individual clinical records, MIMIC-IV is close to the only realistic open option, but factor in the credentialing lead time before you commit to a deadline. If you need individual survey respondents in the United States, NHANES is free and immediate; outside the U.S. and especially in lower-income countries, DHS is the closest equivalent. If your question is a country-level comparison rather than an individual-level one, GBD and WHO GHO cover most needs for free, with GBD better for modeled burden estimates and GHO better for officially reported indicators; Our World in Data is worth checking first if you want an already-merged panel rather than raw source files. Reserve UK Biobank for genetic or deeply phenotyped cohort questions that specifically require its combination of genotype, imaging and linked health-record data, since the fee and agreement overhead are not worth it for questions the free sources can already answer.
FAQ
Do I need approval to use NHANES data?
No. Standard NHANES public-use files are free to download directly from the CDC website without registration. Only restricted-use files with finer-grained identifiers require a research proposal to the NCHS Research Data Center.
What does it take to access MIMIC-IV critical care data?
You need a credentialed PhysioNet account, which requires completing the CITI "Data or Specimens Only Research" training course, a recognized institutional affiliation or reference, and a signed data use agreement. It is free but not instant — plan for a review period before access is granted.
Is Global Burden of Disease data free to use?
Yes for non-commercial research and policy use. A free IHME account gets you access to the GBD Results Tool, with a 100,000-row cap per request, so very large custom pulls across many causes and countries may need to be split into multiple downloads.
Not sure which health dataset fits your study?
If you are weighing access cost and lead time against your research timeline, or you need to confirm a specific country or condition is actually covered before committing, you can give us your research question and required conditions, and we run a data availability assessment (3 free credits on sign-up) — searching these sources for real and judging matches and gaps against each requirement. Even when no perfectly fitting dataset is found, we present the search directions, approximate sources and item-by-item findings honestly for your reference.
New accounts get 3 free searches; every result comes with verifiable source links.