Following on my initial post on Genetic tests and personalised medicine, I’d like to offer some loose thoughts on science communication issues associated with genetic tests.

Ibn Sina (Avicenna) whose Canon of Medicine (1025) is one of the earliest examples of communication of disease risk. Image source: wikipedia

Ibn Sina (Avicenna) whose Canon of Medicine (1025) is one of the earliest examples of communication of disease risk. Image source: wikipedia (Statue in Dushanbe, Tajikistan)

When I think of genomes, genetic tests and medicine three obvious communication problems occur to me:

  1. A preference for ’black and white’ answers by the public, compared to the probabilistic answers genetic tests give.
  2. Generally, the more common the disease, the harder it is to resolve the genetic elements of the disease. This is the opposite of everyday experience, where familiar things are easier. Without understanding this people are likely to wonder why common diseases are not being addressed in the tests.
  3. I have a concern that better controls on claims made for ’alternative remedies’ are needed, especially as direct-to-consumer (DTC) genetic testing potentially enables people to bypass sound medical advice entirely. While this perhaps is more a government issue, the media will need to play their part.

A fourth might be for media to treat any data they receive ethically, but that’s in another category entirely (!) and let’s leave that aside.

Let’s briefly look at each in turn.

Test results are probabilistic and complex, not absolute and simple

It’s commonly noted that the media has mixed success at reporting statistics. (I’m being polite.) The overall impression is that they’re taught to avoid statistics. I can sympathise with that. This leaves them trying to translate the statistics into an analogy or comparison and often the comparison used is wrong. Trying to make a good analogy or comparison requires that you understand what the statistics are saying very well.

I think a key thing is to understand what is being measured. Try writing down what the statistics measure. If you can’t, then you’re in trouble. Understanding what they measure should usually lead you to the meaning of the results and hence to explaining it well.

Bear in mind that analogies invariably only work up to a point, then fail. I personally prefer people to explain things directly, partly for this reason.

The trouble really starts when you consider that most things, disease risks included, can’t easily be explained in one number and in fact depend on quite a few things at once. They’re complex, in other words.

It’s harder to resolve the genetic elements of common diseases

Basically, rare things tend to have a small number of underlying causes and common things tend to have a lot of different underlying causes.


(Source: Gene Tests ©University of Washington Seattle, from the example of co-segregation using Becker Muscular Dystrophy. See full copyright notice)

Geneticists hunt for genes that when defective are associated with a disease, by collecting DNA samples from families and scanning the DNA for changes that are present when the person has the disease (or is a carrier of the disease) that are not present or very rare in people without the disease, as illustrated in the image above.

For diseases that are very rare, genetic defects that cause the disease will usually only be found in a small number of closely-related locations in the genome and those few genetic changes will usually have a clear association with the person having the disease or not.

If the disease is very common, usually the genetics team will be looking for many defects scattered all over the genome with most of the defects each only having a small part of the overall effect on the disease. It’ll be more common for several defects interact in causing the disease. (That is, each defect might only be present in a small percentage of the people with the disease or need to be present with other defects to be associated with the disease.)

It’s a lot harder to find a large number of things that don’t have much effect on their own, than to find a smaller number of things that have each on their own have a strong association the disease.

All this said, there has been some impressive work in the last few years using whole genome scans to locate defects associated with disease. They’re big science efforts, however, involving large research teams and large collections of patients.

Care over “alternative medicine” claims

You can’t say enough about this issue. Other blogs already have, take yourself over to Science-based Medicine or Respectful Insolence and you could read for days.

The issue here is that people receiving genetic tests directly from direct-to-consumer companies could in principle bypass sound medical advice entirely. Some might say that those that are going to would anyway, but with DTC tests, they don’t even have to start with the medical profession.

Media can help, but I have to say I have little confidence, given their existing track record.

With so much to say about this, I’m just going to make one over-riding point. The ‘natural’ or ‘alternative’ in ‘natural remedy’ or ‘alternative remedy’ is not of the essence. Those words are more marketing labels that anything else. The essence is if the product has been tested, and with that if claims about the remedial ability of the product can be backed by substantive evidence.

If someone makes a claim about a health product that cannot be substantiated by evidence, then they cannot say it is a medicine, only that they hope that it might be a medicine.


This list is not comphrensive or even representing all the aspects. I’ve chosen articles that are commentaries or review articles so that they might be more easily read by an informed non-specialist. Only articles from this year are included (with one exception). All are not open-access (sorry).

Common sense for our genomes

Brenner, Nature 449:783a (2007) (No DOI available.)

[Older, but concise and easy to read.]

The promise and reality of personal genomics

Yngvadottir et al, Genome Biology 10(9)237 (2009) doi:10.1186/gb-2009-10-9-237

[From the perspective of computational biology.]

An agenda for personalized medicine

Ng et al, Nature 461(7265)724a (2009) (No DOI available.)

[Presented in a DTC genomics context; think more broadly when reading.]

Genetics and the general physician: insights, applications and future challenges

Knight, Q J, Med on-line in advance of publication (Sept 7, 2009) doi:10.1093/qjmed/hcp115

[Has case examples.]

Beyond odds ratios – communicating disease risk based on genetic profiles

Kraft et al, Nature Reviews Genetics 10(4)264 (2009) doi:10.1038/nrg2516

[From a position of how risk interpretations should be delivered.]

Musings on genome medicine: the value of family history

Clarke, Genome Biology 1(8)76 (2009) doi:10.1186/gm75

[Makes the point that family data should still matter.]