A polygenic score can arrive as a single, precise-looking number, sometimes with a percentile beside it. The number is real, but what it measures is narrower than it looks. It says where a person’s inherited variants place them compared with other people. It does not say whether, or when, they will become ill.
What a polygenic score adds up
Some diseases trace back to a variant in a single gene. Many common ones do not: the National Human Genome Research Institute (NHGRI) describes complex diseases as occurring “as a result of many genomic variants, paired with environmental influences (such a diet, sleep, stress and smoking).” Coronary artery disease is one of them.
Researchers find the variants linked to such a disease by comparing the genomes of people with and without it, and, NHGRI says, there can be “hundreds or even thousands of variants per disease.” NHGRI notes that a person’s risk can be estimated from these variants without knowing which genes are involved. The PGS Catalog describes the usual calculation as a weighted sum: the number of copies of each variant a person carries is multiplied by that variant’s effect size, as estimated from a genome-wide association study, and the results are added into a single number that “predicts genetic predisposition for a phenotype.”
Relative, not absolute, risk
NHGRI is direct about what the number means: “A polygenic risk score can only explain the relative risk for a disease.” It tells you how a person’s risk compares with that of others, but “polygenic scores do not provide a baseline or timeframe for the progression of a disease.” Its example is two people with the same high score for coronary heart disease, one 22 years old and one 98, who have different lifetime risks. It adds that “[p]olygenic risk scores only show correlations, not causations.”
Absolute risk is different. NHGRI’s example is a BRCA1 mutation, which carries a 60–80% absolute risk of breast cancer for women who have it. Most people’s polygenic scores fall in the middle of a bell curve; people at the high end “may benefit from discussions about this risk with their physicians and genetic counselors.” In NHGRI’s words, polygenic risk scores “will always be probabilities, not certainties.”
The top of the range
Khera and colleagues (2018) developed and validated genome-wide polygenic scores for five common diseases. Their scores identified 8.0, 6.1, 3.5, 3.2 and 1.5% of the population at more than threefold increased risk for coronary artery disease, atrial fibrillation, type 2 diabetes, inflammatory bowel disease and breast cancer, respectively. For coronary artery disease, that share was 20 times higher than the carrier frequency of rare single-gene mutations conferring comparable risk. The authors proposed that “it is time to contemplate the inclusion of polygenic risk prediction in clinical care.”
Threefold is a relative figure. It compares people at the top of the score range with others; it is not, by itself, the chance that any one of them will develop the disease.
Who the scores were built from
A score’s accuracy depends on who was in the studies behind it. NHGRI notes that “[t]he majority of genomic studies to date have examined individuals of European ancestry,” so there may not be adequate data to calculate a polygenic risk score for other populations.
Martin and colleagues (2019) identify “the major ethical and scientific challenge surrounding clinical implementation” of these scores: those available then were “several times more accurate in individuals of European ancestry than other ancestries,” a disparity they call “an inescapable consequence of Eurocentric biases in genome-wide association studies.” They argue that greater diversity must be prioritized in genetic studies, and that summary statistics must be shared publicly, so that health disparities are not increased for those already most underserved.
What a reader should be told
In 2021 the Clinical Genome Resource (ClinGen) Complex Disease Working Group and the Polygenic Score (PGS) Catalog published the Polygenic Risk Score Reporting Standards (PRS-RS). The authors note “notable heterogeneity in the application and reporting of these risk scores,” and the standards define “the minimal information that is needed to interpret and evaluate PRSs.” The items cover the study populations, the statistical methods used to develop and validate a score, and the score’s potential limitations.
The PGS Catalog itself is an open database of published scores, with the variants, alleles and weights needed to reproduce them. Its release of 23 October 2026 listed 7,029 polygenic scores for 820 traits from 850 publications. A score in the catalogue can be recalculated and compared by others; a score whose variants and weights are not shared cannot.
INSTAR’s research interest
INSTAR’s interest is in the questions a careful reader can ask before relying on a polygenic score. Is the result relative or absolute, and compared with whom? Were the people the score was built and tested on similar in ancestry to the person being scored? Is the score published, with its variants and weights, so others can check it? Does the report state the score’s limitations?
These questions connect to the genetics program page, which discusses biobanks, variant interpretation and fairness in evaluation. The brief on psychology replication asks similar questions about a single study, and the page on how this site checks a fact describes the checks applied to the sources cited here.