Written and maintained by CASRAI Editorial Board
Last updated
Short answer: The Framingham Heart Study is a long-running prospective cohort study of cardiovascular disease, sponsored by the National Heart, Lung, and Blood Institute (NHLBI). It began in 1948 with 5,209 adults from Framingham, Massachusetts, and later added their children, grandchildren and an Omni cohort. Researchers can request its data through BioLINCC, dbGaP or the study’s own research application.
This page is for researchers and students who want to understand the study’s design and work out which data route fits a project. It is not medical advice.
Definition and scope
The study follows participants through repeated examinations and event follow-up, which makes it a textbook example of a cohort study and of longitudinal study design. The NHLBI describes the original cohort as examined every 2 to 6 years. The study’s researcher page states that more than 15,000 people have enrolled over the decades.
The historical literature links the 1948 start to the death of President Roosevelt in 1945 (Mahmood, Levy, Vasan and Wang, Lancet 383:999-1008). That review also credits a 1961 publication by Dawber and Kannel with popularizing the term “risk factor”.
Cohorts
- Original cohort (1948): 5,209 adults from Framingham, Massachusetts.
- Offspring cohort (1971): 5,124 adult children of original participants and their spouses.
- Omni cohorts (1994 and 2003): the first Omni cohort enrolled 507 people of African-American, Hispanic, Asian, Indian, Pacific Islander and Native American descent.
- Third Generation (2002): grandchildren of the original participants.
Source: NHLBI Framingham Heart Study page, read 2026-11-07.
What did the Framingham Heart Study find?
The historical review in the Lancet summarizes these findings from the study’s data:
- Hypertension is linked to coronary heart disease and stroke, and systolic blood pressure is a stronger predictor than diastolic.
- Hypertension precedes most cases of heart failure.
- Non-rheumatic atrial fibrillation was associated with a more than five-fold higher stroke risk.
- HDL cholesterol is inversely associated with coronary heart disease.
- Obesity is an independent risk factor.
Framingham risk score and “normal” values
The Framingham Risk Score grew out of earlier risk functions (1967 and 1976). It was published in 1998 by Wilson and colleagues as a 10-year coronary heart disease risk estimate, presented as lookup tables with low, intermediate and high categories, and adopted by the National Cholesterol Education Program. It is a population-derived estimate, not a diagnosis, and what a particular score means for an individual is a question for a clinician. This page does not give threshold values.
How researchers get Framingham data
Three routes exist, and they hold different material. The table compares them using each custodian’s own description.
| Route | What it holds | Key conditions |
|---|---|---|
| BioLINCC (NHLBI repository) | Study datasets only: examination data from the first 32 clinical exams, selected ancillary data, and event follow-up through 2018; last updated July 21, 2025 | Free registration; requires full or expedited IRB or ethics committee review and approval, and waivers or exempt determinations do not suffice; consent for commercial use is tiered |
| dbGaP | Genetic data plus major risk-factor information (per NHLBI) | Listed by NHLBI as a Framingham data source |
| FHS Research Application | Data reviewed through the study’s list of internal and external repositories | Join the researchers listserv, review available data, submit the web-based application; sponsored projects need Executive Committee approval before funding submission, unsponsored projects need Research Committee approval; funding is reported to the study; costs go through the FHS Service Center |
Sources: BioLINCC Framingham Cohort page, NHLBI FHS page and FHS Researchers page, all read 2026-11-07. For a comparable access process, see UK Biobank data access.
Which exam count should you cite?
Official pages differ: the NHLBI page refers to cohort exam data from the first 30 exams, while BioLINCC currently lists the first 32. The NHLBI page gives the original cohort age range as 30 to 62 and BioLINCC gives 28 to 62. Cite the dataset version you actually received and attribute counts to the page you took them from.
Design strengths and limits
| Design feature | Stated strength or limit |
|---|---|
| Long prospective follow-up | Strength: repeated exams and events over decades |
| Women made up more than half of the original cohort | Strength |
| Family-based design (offspring, grandchildren) | Strength; critics also questioned the family-based approach |
| Consistent heart-failure adjudication since the 1960s | Strength |
| Omni cohorts and SHARe genotyping | Added diversity and genetic data |
| Original cohort predominantly white, of European descent | Limit: representativeness |
| Town chosen partly as representative of 1940s America | Limit: critics questioned representativeness |
Source: Mahmood, Levy, Vasan and Wang, “The Framingham Heart Study and the Epidemiology of Cardiovascular Diseases: A Historical Perspective”, Lancet 383:999-1008 (PMC4159698), read 2026-11-07. Related design concerns for cohorts are discussed in immortal time bias and what is epidemiology.
Common misunderstandings
- It is one dataset. It is several cohorts and several access routes, with different contents and approvals.
- A waiver is enough for BioLINCC. BioLINCC states that waivers and exempt determinations do not suffice.
- The risk score is a diagnosis. It is a 10-year risk estimate built for population use.
FAQ
Is the Framingham Heart Study still going on?
The NHLBI describes ongoing examinations. The BioLINCC dataset contains event follow-up through 2018.
What are the flaws of the Framingham Study?
The original cohort was predominantly white of European descent, and the town was chosen partly as representative of 1940s America, so critics questioned representativeness. The Omni cohorts were added to broaden the population.
Is the Framingham Heart Study an example of a cohort study?
Yes. It is a prospective, longitudinal cohort study; see cohort study for the design.
Where is the Framingham Heart Study dataset?
Through BioLINCC, dbGaP or the FHS Research Application, as in the table above.
Sources
- NHLBI, Framingham Heart Study (FHS): nhlbi.nih.gov/science/framingham-heart-study-fhs (read 2026-11-07)
- Framingham Heart Study, For Researchers: framinghamheartstudy.org/researchers/ (read 2026-11-07)
- BioLINCC, Framingham Cohort: biolincc.nhlbi.nih.gov/studies/framcohort/ (read 2026-11-07)
- Mahmood SS, Levy D, Vasan RS, Wang TJ. Lancet 383:999-1008, PMC4159698 (read 2026-11-07)
