Select Committee on Science and Technology Minutes of Evidence

Memorandum submitted by Professor Martin Bobrow CBE, Head of Department of Medical Genetics, Wellcome Trust Centre for Molecular Mechanisms in Disease, Addenbrooke's NHS Trust


  1.  Genetic tests can predict future ill health. The subject has been previously considered by the Science and Technology Committee (Third Report, 1995). The Human Genetics Advisory Commission reported on the implications of genetic testing for insurance in December 1997.

  2.  Because of fears that insurers would use genetic information to create a "genetic underclass" who would be disadvantaged in access to insurance products, several countries have banned insurance companies from using genetic information. The HGAC recommended that access to genetic test results should be permitted, but only when a strong evidence base for the specific insurance product had been established to the satisfaction of an impartial external controlling body. This led Government to establish GAIC (Genetics and Insurance Committee). HGAC also recommended a two year moratorium, to draw a clean line under past practise and to allow the collection and examination of evidence. Government did not agree this moratorium, and the ABI has continued to use tests for seven sentinel conditions, while awaiting the verdict of GAIC on whether or not this use is justified.

  3.  A few years ago there were many examples of poor decisions by insurance underwriters, on the basis of misunderstanding genetics. Since then, the ABI has established a code of practice, and an advisory committee. They have also agreed a "voluntary concession", relating to mortage related life policies up to an arbitary £100k. (No limit has been discussed with regard to disability or employment insurance). Although not enough, these are welcome improvements.

  4.  I would like to address the following questions:

  4.1.  Is this field something that can be left to market forces?

  4.2.  What is the potential for harm in current practice?

  4.3.  What are the defects in current and future genetic tests for insurance usage?

  4.4.  Why is there so much discussion about a small number of relatively uncommon disorders?

  4.5.  What are the arguments for and against a total ban on the use of genetic information by insurers?

  5.  Can this be left to market forces? If insurance were simply a fashion accessory, a competitive market and normal consumer protection may suffice. However, in today's society, insurance access is necessary for full participation as a home owner, and as a responsible family member. Increasingly, we will rely on insurance to provide pensions, and care in old age. Insurance is an instrument of social policy.

  6.  What harm is done by current practice? Currently, ABI insist on insurers right of access to genetic test results. This stance causes significant public unease. In genetics clinics, it is common for patients considering genetic testing, to raise the question of whether they would be disadvantaged in the insurance market. Medical care might well become distorted by fear of loss of insurance access. Secondly, making good use of knowledge derived from the Human Genome Project will require substantial clinical studies, correlating genetic combinations with health outcomes of research volunteers. Fears of insurance consequences may deter people from consenting to such studies, slowing the realisation of potential health gains from human genetics. Thirdly, this concern fuels a public unease about potential misapplication of genetic science, "giving genetics a bad name"—and may therefore add to a public climate which is not conducive to maximising health gain from the potentials of genetic science in the UK.

  7.  What are the weaknessess of current genetic tests? The tests under discussion (eg in relation to the ABI "7 conditions") are used for medical purposes, and their technical characteristics are not in doubt. The question asked in a medical context is however very different from that asked of the test in an insurance context. For medical purposes, the test is used in conjunction with expert interpretation, and intensive counselling, to decide whether a person carries an altered gene, whether this poses a risk to future health, and what can be done about it. For life insurance purposes, the issue is (put crudely) "will this person die within 20 years?"

  8.  Genetic tests are very good at distinguishing those who carry a particular gene from those who do not. They are somewhat less accurate at identifying who will and who will not eventually get the disease—even for these seven highly heritable conditions, a minority of people who have the defective gene do not develop a clinical problem. They are not at all good at predicting when the disease will start, how rapidly it will progress, and thus when death will occur. Specific research correlating gene changes with mortality experience is necessary before insurance implications can be safely assessed. The medical genetic literature is poor in this area, insurers appear to have conducted little research of their own, and actuarial research is also thinly supported.

  9.  The precision with which genetic test results can predict expected time of death is also compromised by a number of systematic biases in the way medical research studies must be conducted, which need great care when extrapolating the test results outside an expert medical context:

  9.1.  The likelihood of being affected (the gene penetrance), and the average severity of the disorder, is prone to being exaggerated by early medical studies, which for obvious reasons focus on the most florid disease (less possibility of mistaken diagnosis), and the largest families which are easy to study, but which over represent gene changes that are particularly likely to cause disease. Understanding of the factors which interact with the primary gene change to actually cause disease, is still very rudimentary.

  9.2.  They take no account of possible beneficial effects of early diagnosis and treatment, which for some conditions may be considerable. Because these tests are relatively new, there is little evidence on the mortality prospects of someone who is diagnosed today and appropriately managed. Evidence of such benefit will inevitably accrue many years after the development of the relevant test and its associated treatment, as the treated cohort of patients is followed through future life experiences.

  10.  If tests are developed for common disease, early data are again likely to be seriously biased:

  10.1.  Gene penetrance and disease severity will tend to be over-estimated by early studies (see para 9.1).

  10.2.  The beneficial effects of treatments which have not yet been fully developed or deployed will be underestimated (para 9.2)

  10.3.  For common gene variants associated with common disease, beneficial effects which counter-balance the harmful effects of the gene combination may be quite common—the genes would not remain common were that not the case. Because medical studies are usually conducted on groups of patients with specific diseases, harmful effects will be identified long before any counter-balancing beneficial effects. ("Cohort" studies, which follow through a group of healthy people to correlate their genetic makeup with future ill-health in an unselected and therefore unbiased way, are a slow but important way of avoiding this problem).

  11.  These possible distortions in the early data, which are very likely to occur and will take years to unravel, all happen to favour the insurer rather than the insured.

  In summary, I believe that genetic tests in insurance are currently being overinterpreted and being given a predictive weight that the scientific evidence does not yet support. I would suggest there is need for a great deal more reseach in this area, and until such research has been undertaken a moratorium should be enforced.

  12.  Why is there so much discussion about a small number of relatively uncommon disorders?

  The seven diseases seem too uncommon to represent a real financial threat to insurers. Some protection against unusually large policies, and the common catch-all clause which obliges insurance applicants to declare any relevant known health factor, should suffice to protect the companies. Several major insurers take this stance, do not ask for genetic tests results, and have come to no obvious harm.

  13.  I have heard two explanations. First, the "right to underwrite", a hypothesised freedom of the unfettered market to operate as it wishes, seems to me a principle long abandoned in other areas of activity with serious social impact.

  14.  Secondly, the expectation that better discriminatory tests are on the way, and that to concede anything in these simple conditions would lock the industry into a potentially serious position if in future many individuals gained access to high quality predictive information for common diseases. I believe this fear is unfounded, because the nature of genetic contribution to common disease is such that it will have only weak predictive power, but others take a different view and this remains to be worked through in practice. Some flexibility in the long term is therefore reasonable.

  15.  What are the arguments for and against a total ban on the use of genetic information by insurers? Notwithstanding some frustration, I remain unconvinced about legislative prohibition on insurers access to genetic information:

  15.1.  Supression of information is a dangerous and uncertain path to tread in any situation, in part because it is ineffective and open to abuse.

  15.2.  We have long conceded, perhaps without deep thought or discussion, the industry's right to ask for our family history. Family history is genetic information. I still wait to be persuaded that insurers use this information reasonably, but legislation that did not also ban this long established practice, would still allow access to genetic information.

  15.3  Legislation is rigid. Genetics is developing rapidly, and even the medium term future is hard to predict with any confidence.

  16.  Despite these reservations, the ABI stance, of continuing to use information on tests for their "seven conditions" until banned by GAIC, rather than agreeing not to use them until they had been agreed as valid, instils no feeling of confidence. Rather, it leaves the taste of arrogance, and seems to point ever more to the possibility that legislative intervention may be undesirable in principle but unavoidable in practice.

18 January 2001

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2001
Prepared 3 April 2001