In short
Precision medicine means matching the decision to the individual. In practice that is less about genomics than the word suggests.
- Direct measurement beats prediction. A coronary calcium score outperformed polygenic risk scores for predicting coronary events in two large population studies.
- The calculator has to fit you. No single risk model is appropriate for all patients, and the one used should be validated for your population.
- Genetic testing earns its place in specific situations, mainly familial hypercholesterolaemia, where it lets relatives be screened.
Treatment Overview
What does precision actually mean here?
Three concrete things: using a risk model that fits you, adding the factors your model leaves out, and measuring disease directly where a prediction is not good enough.
None of those requires a genome. All of them change what gets prescribed.
Why does the choice of calculator matter?
Because different models give different answers for the same person, and that changes treatment. Newer models were built on more contemporary data because older ones overestimated risk in current populations.
In one nationally representative comparison, the newer model gave a mean 10 year risk of 4.3 per cent where the older one gave 8.0 per cent. The gap was largest in older adults, 10 versus 23 per cent at ages 70 to 75.
Which model for which population?
| United States | The 2023 PREVENT calculator, which includes kidney function, BMI and optionally urine albumin and HbA1c |
| United Kingdom | QRISK3, ages 25 to 84, which includes autoimmune disease |
| Western Europe | SCORE2 for 40 to 69, SCORE2-OP for 70 and over |
| China | China-PAR |
| Other regions | WHO risk charts, published for 21 world regions |
The thresholds differ between models too, so the risk cut off used has to be the one belonging to the calculator, not borrowed from another.
Where do calculators fail?
- LDL of 190 mg/dL or above, where risk is very high regardless
- Familial hypercholesterolaemia, where calculators do not apply at all
- Type 1 diabetes, underestimated by most models even under 40
- Type 2 diabetes, underestimated by models that omit kidney function, albuminuria and glucose control
- HIV, where risk is underestimated
- Anyone with plaque already visible on a scan done for another reason
Are polygenic risk scores useful yet?
Not for routine decisions. For cholesterol they are not standardised and not ready for clinical application. For predicting coronary events, a calcium score improved risk discrimination and reclassification in two large cohorts while the polygenic score did not.
They may become useful. Today, the more precise tool is a scan that shows what is actually in your arteries.
Where genetic testing does earn its place
Mainly in familial hypercholesterolaemia, and even there it is not routine, because the diagnosis is usually made on clinical criteria and the result rarely changes management.
It helps when the clinical picture is unclear, and when identifying the family variant allows targeted screening of relatives. A negative result does not exclude the condition, since up to half of people meeting clinical criteria have no identifiable variant.
Who Is a Good Candidate?
- Your risk has been calculated but you are not sure the right model was used for you
- You have diabetes, kidney disease or an autoimmune condition that standard models handle badly
- You have been given a genetic or polygenic risk report and want to know what it changes
- You have a strong family history and want to know whether genetic testing is warranted
- Two different clinics have given you different risk figures
- LDL cholesterol of 190 mg/dL or above, where treatment is indicated regardless of any risk model or scan
- Established cardiovascular disease, where the question is no longer risk prediction but secondary prevention
- An active symptom, which needs diagnosis rather than risk modelling
What Happens During the Consultation
-
1
Choosing the right model for you
Based on your population, age, and which risk factors you have. Models that omit a factor you actually carry will understate your risk, and that is corrected explicitly.
-
2
Adding the factors the model leaves out
Risk enhancers such as family history of early disease, and in diabetes, duration, albuminuria, kidney function, retinopathy and neuropathy. These adjust how the number is read.
-
3
Direct measurement where prediction is not enough
A calcium score at borderline or intermediate risk, where it reclassifies people in both directions and often settles the treatment decision.
-
4
Genetic testing only where it changes something
Considered when a clinical diagnosis is unclear, or when finding a family variant would let relatives be screened. Interpretation involves a specialist, because it is not straightforward.
Program Structure & Follow-Up
How results are used over time
The risk figure is not a permanent label. It is recalculated when something material changes, and at set intervals otherwise.
| Low risk | Reassess in four to six years, sooner if a risk factor changes |
| Borderline and intermediate | Yearly reassessment is reasonable |
| On medication | Sooner, for monitoring and adjustment |
| Calcium score of zero | Repeat at three to seven years only if it would change a decision |
Why lifetime risk gets calculated too
Between 30 and 59, when the 10 year figure is low or borderline, a 30 year estimate is added. Some people with a low 10 year risk carry a high lifetime one, and they show more subclinical atherosclerosis and more events.
The caveat is stated openly: there is limited evidence on the long term benefits and harms of starting medication on the basis of a 30 year figure alone.
What about a family history the calculator ignores?
It still informs the discussion. A well documented family history of early cardiovascular disease is used qualitatively even where the model does not include it and where it does not improve the model’s performance statistically.
Benefits, Limits & Safety
What precision actually buys you
Fewer people treated who do not need it, and fewer missed who do. Newer risk models reduce the number of people who qualify for statins and aspirin at a given risk level, which is a real change in both directions.
Direct measurement adds the most. A calcium score reclassified 52 per cent of intermediate risk older adults in one study and moved 20 per cent of low risk middle aged women into a more accurate category.
The honest limits
- Models are population tools. They estimate the average risk of people like you, not your risk
- Validation varies by location, and external validation of newer models has been mixed, underestimating incidence in three of four health systems in one study
- Breast cancer risk tools are inaccurate for women with several affected first degree relatives
- Genetic databases underrepresent people of African ancestry, so variants are more often classified as uncertain
What is not offered
Consumer genomic panels marketed as longevity testing, biological age scores, and any test whose result would not change a decision.
The test for whether a genomic result is worth having is simple: name what you would do differently. If nothing, it is not ordered.
What changes decisions
The right calculator for your population, the risk enhancers it omits, and a calcium score where prediction alone is not enough.
When risk is recalculated
Every four to six years if low, yearly at borderline or above, and immediately when a risk factor changes.
Not used for decisions
Polygenic risk scores, consumer genomic longevity panels and biological age tests. None of them is standardised or shown to improve outcomes.
Cost & Program Investment
The consultation and any tests are quoted before they are ordered. Genetic testing is not routine and is only proposed where a named decision depends on it, in which case the cost, the limits and the possibility of an uncertain result are explained first. Consumer genomic panels are not sold here, and a report you already have can be reviewed as part of the consultation.
Frequently Asked Questions
There is no evidence that it improves outcomes for a healthy adult. The useful question is what you would do differently with the result. If the answer is nothing, it is not worth doing.
Not ready for routine use. In cholesterol they are not standardised. For coronary events, a calcium score outperformed them in two large population studies and improved reclassification where the genetic score did not.
Almost certainly a different calculator. The same person can get 4.3 per cent from one model and 8.0 per cent from another, and the gap is largest in older adults. The thresholds belong to the calculator and cannot be swapped.
Possibly not. Most models underestimate risk in type 1 diabetes, and models that omit kidney function, albuminuria and glucose control underestimate it in type 2. Those factors are added explicitly.
Mainly in suspected familial hypercholesterolaemia where the clinical picture is unclear, or to identify a family variant so relatives can be screened for exactly that. Even then it is not routine.
No. Up to half of people who meet the clinical criteria for familial hypercholesterolaemia have no identifiable variant, and the absence of a definite variant does not mean none is present.
The word often is. The underlying idea, matching the decision to the individual rather than to an average, is sound and is what the calculators, risk enhancers and calcium scoring exist to do.
Evidence
Where this information comes from
Every figure on this page is taken from the sources below. All of them are free to read, so you can check them yourself.
This page is general medical information and does not replace a personal consultation. Trial results are averages across large groups, not a prediction for any one person.