The genetics and insurance moratorium after 2006
Guy Thomas
r.g.thomas@kent.ac.uk
www.guythomas.org.uk
These notes are written in response to a request from the GAIC secretariat for comments on “resolving the moratorium.” I understand that the secretariat may draw on these notes, together with contributions from others, in preparing a discussion paper for GAIC. (I am, however, happy for these notes to be shown to GAIC or HGC.)
SUMMARY
Genetic testing is at present financially inconsequential to the overall operation of the insurance industry. It seems unlikely that this position will change radically in the next few years.
There is some credible evidence for an emerging deterrent effect, that is that concerns about insurance are discouraging people from taking tests which could be relevant to decisions about treatment. At present this is relevant most notably for BRCA1/2 testing, but other examples are likely to develop. The most important concern in formulating policy in relation to genetics and insurance is to minimise this deterrent effect.
The simplest and most obvious way to address this concern would be to extend the moratorium beyond 2006 on similar terms to the present.
Some people suggest more complicated mechanisms, which would enable data to be collected by insurance companies for actuarial research. For example, people might be required to report test results to their insurers, but the insurers promise not to use the information against them. Or an industry-wide arrangement might be established, whereby insurance companies pass on genetic risks to a single reinsurer, the cost of which is shared between all companies in proportion to their premium income.
In theory these solutions could work; but they are more complex, and they would require some initiative and effort on the part of the insurance industry. The viability of such complex solutions depends partly on whether the industry thinks they are worth the extra trouble as compared with a moratorium. In relation to minimising the deterrent effect, more complex solutions may be less effective if they are more difficult to understand.
The continuing moratorium could be subject to further review if and when the insurance industry can demonstrate that particular insurance markets are being severely impaired because access to genetic test results is not permitted.
Provided that the financial ceiling of the moratorium is high enough, relatively ‘light-touch’ regulation may be justified for policies above the ceiling.
FURTHER NOTES
1. Advantages and disadvantages of continuing the moratorium
The main advantages of continuing the moratorium are that
- It prevents avoidable deaths which are likely to result from the deterrent effect (see below for some evidence on the deterrent effect).
- It allows clinical practice to develop as far as possible unhindered by insurance concerns.
- Compared to more complicated pooling mechanisms, the moratorium is simple, and its benefits are easy to communicate. Simplicity and understanding are important features, particularly for alleviating the deterrent effect.
- For insurers, it should prevent politically damaging criticism of the type experienced in 2001 from the House of Commons Science & Technology Committee, and from HGC. This criticism arose partly because insurers (and perhaps DoH) had consistently misjudged the extent of public opposition to the use of genetic tests in insurance. Any plans to dismantle the moratorium after 2006 may reflect a similar misjudgment.
- It provides a robust starting framework for any changes in social welfare and insurance policy under future governments. For example, it seems to me entirely plausible that a future government will seek to promote private health insurance and reduce spending on the NHS.
The main disadvantages which are often suggested to follow from the moratorium are that –
- In principle it should result in very slightly higher insurance claims, and correspondingly increased premiums. It is not obvious that this is a disadvantage, either to insurance buyers in aggregate or to insurance companies[1]. However, in practice this effect is very much too small to measure anyway. All estimates of its magnitude appear immaterial in comparison to the large observed variations in premiums between different companies and over time (see below).
- Actuarial statistics relating particular genetic tests to prognoses are not collected. However, under a comprehensive moratorium such statistics have little value anyway (except possibly for policies above the financial ceiling). It is difficult to see a motivation for collecting such statistics, unless it is hoped or intended that the moratorium will later be abolished (for example, perhaps after a change of government).
2. Evidence for the deterrent effect
“The deterrent effect” refers to the idea that worries about insurance implications may discourage people from taking clinically or socially relevant tests. This is in my view the most important issue in formulating policy in relation to genetics and insurance.
The deterrent effect is difficult to measure because it is mainly concerned with private motivation and perceptions, rather than observable directly from behaviour. I first realised that it might be an important issue from anecdotal discussion with doctors. More recently I have seen some published evidence, which I summarise below.
Armstrong et al (2003)[2] provide some evidence from Pennsylvania, USA of the deterrent effect in relation to BRCA1/2 testing. The main messages are –
- Fear of life insurance discrimination was rated a moderately or very important factor in the decision as to whether to take the test by 55% of the women who voluntarily attended for clinical risk assessment
- The women who expressed concern about life insurance discrimination were one-third less likely to decide to proceed with test.
The group of women attending for clinical risk assessment in this study had an average breast cancer lifetime risk of over 20%. The authors conclude that
“For this sample, then, fear of insurance discrimination may lead to underuse of prevention and avoidable cancer deaths.”
I have some comments on this study –
- The study necessarily focuses on women who had already decided to participate in voluntary programme of clinical risk assessment, with the possible outcome of taking a genetic test. It is plausible that the deterrent effect also operates on the earlier decision, ie many women who are worried about insurance discrimination may decide not to participate in clinical risk assessment in the first place.
- It is also stated that no women reported actual life insurance discrimination, although this is a little difficult to interpret. The authors state that life insurance was not “denied or cancelled”; but in a UK context, an existing life insurance policy could never be “cancelled” by the insurer. It is not clear to me how an insurer in the US would find out that a customer has participated in a stand-alone risk assessment, if the customer chose not to tell the insurer. It is also not stated what legal restrictions (if any) may apply on insurers’ access to genetic test results in Pennsylvania.
More generally, the distinction between actual insurance discrimination and perceptions about the possibility of insurance discrimination may be a useful distinction. It is perceptions which are most relevant to the deterrent effect. Perceptions may be vague or ill-defined and yet still be a potent influence on behaviour. Perceptions may be driven partly by wider public mistrust of the insurance industry, and also by public observation of government’s apparently ambivalent attitudes towards genetic discrimination.
3. Policies above the financial ceiling of the moratorium
There is also a question as to how, if at all, insurers should use genetic test results in relation to policies above the financial ceiling of the moratorium (£500,000 life insurance, and £300,000 / £30,000pa for critical illness / income protection insurance).
My view on this depends partly on the level of the ceiling. The present ceiling is sufficient that around 99% of new policy applications currently fall below the ceiling. I believe that this is appropriate and necessary to ensure that the moratorium largely negates the deterrent effect. However, if the ceiling is set high enough (as I believe it is, at the moment), this does raise the possibility that regulation above the ceiling may make little further difference to the deterrent effect.
Furthermore, a valid financial requirement for a policy above the current (high) ceiling implies some degree of affluence. Anyone in this position who is disadvantaged in relation to insurance is very likely to be able to address that disadvantage by other means.
Therefore subject to the ceiling remaining sufficiently high, I can see a case for relatively ‘light-touch’ regulation above the ceiling.
On the other hand, if the ceiling were reduced then more people would perceive regulation above the ceiling as potentially relevant to their decisions about testing. In this scenario, regulation above the ceiling would be more important, as it would have a greater influence on the deterrent effect.
In any event for historical reasons GAIC exists, and considerable work has already been done by insurers in preparing applications to GAIC. There are some benefits to this process of approving specific tests:
- The process has undoubtedly focused minds in the insurance industry on the justification (or lack thereof) for use of certain tests (eg the shelving of several of the applications which the ABI originally planned to submit). This encouragement to the industry to focus on evidence appears to have been beneficial.
- The application of evidence-based criteria in relation to genetic tests may have a wider beneficial effect in raising the standards of evidence in underwriting more generally.
However, my enthusiasm for this process is limited by the discrepancy between GAIC’s low thresholds and actual market variations in insurance prices, which I discuss below.
4. GAIC’s actuarial criteria
I believe that GAIC’s numerical thresholds for approving genetic tests appear ludicrously low in comparison with actual variations in prices in the insurance market. For example –
- Term assurance premiums in 2003 have typically reduced by between 20% and 45% compared with rates in January 1998 (source: Swiss Re).
- Critical illness insurance premiums have increased by between 50% and 100% from December 2002 to December 2003 (source: see my previous letters to GAIC).
- The variation in the half dozen cheapest term assurance rates for identical lives is up to 40% (source: see my previous papers to HGC).
These huge variations demonstrate how imperfect and imprecise insurance pricing is in practice. In contrast to these figures, estimates of the aggregate cost of a moratorium are typically much less than 5% of premiums, even under extremely unfavourable assumptions[3].
Even the thresholds set by GAIC for individual cost increases (+50%/+25% pure risk premium, equivalent to perhaps +25%/+12% in the gross premium including expenses) are smaller than the market variation in premiums which typically occur for reasons which have nothing to with genetics, and indeed nothing to do with individual differences in risk.
I believe that when the facts about insurance pricing become more widely known, the very existence of GAIC will be called into question. The State’s promotion and endorsement of genetic discrimination, which many people find profoundly offensive in principle, will not be tolerated once people understand how fatuously inconsequential the matter is relative to other matters affecting insurance prices.
GAIC, however, is “not of the view that commercial variations are related to ”fairness” in determining insurance premiums.”[4] I cannot argue with such denial of the facts. But I think that others may be able to agree with me that the information has some relevance.
5. Costs to the “insurance industry”?
The enquiry from GAIC which led me to write these notes invited me to consider inter alia the “costs and benefits to the insurance industry” of any suggested policy. “The insurance industry” will no doubt speak with its own voice; but I do sometimes wonder who this voice represents.
My own economic interest in the insurance industry is solely as an owner, that is a shareholder. From the perspective of an insurance company shareholder, the regulation of genetics is an extremely low priority, in truth quite inconsequential. One illustration of this is that to the best of my knowledge, senior executives – chief executives and finance directors – have not concerned themselves with lobbying GAIC or HGC. They have more financially material matters to deal with.
The insurance people who lobby GAIC and HGC about underwriting are instead people whose work and career is in some way related to underwriting. Clearly these people are likely to emphasise the importance of the subject; but that does not mean that it is financially material to the owners of insurance companies.
As a shareholder, I do not find it obvious that “less regulation” of pricing is preferable to “more regulation.” I may prefer “more regulation” if it increases public acceptance of my product, and hence the size of the market. One illustration of this point is the growth of private medical insurance in Ireland as compared with the UK. In Ireland, the take-up of voluntary private medical insurance is about 50% of the population, compared with 11% in the UK. One of the reasons for this is that the pricing of private medical insurance is very strictly regulated in Ireland: almost no underwriting is permitted.
6. Work needed to clarify the costs and benefits of continuing the moratorium
Some areas in which further work could usefully be done are as follows.
(1) Comparisons of the aggregate increase in premiums from a moratorium with the variation in premiums which are observed between companies and over time. I have already made clear my view that the costs of a moratorium are inconsequential relative to other variations which affect the prices which consumers pay, but you may wish to study this independently.
(2) Discuss with clinical geneticists and patient groups the effectiveness of the moratorium in minimizing the deterrent effect, and how this might be affected by different policy options going forward.
- Does the moratorium currently address geneticists’ and patients’ concerns? What would be their reaction to different policy options?
- Is the current uncertainty about what happens after 2006 an issue in geneticist / patient discussions? (Someone involved in counselling people about genetic tests did raise this last point at the 22 September meeting.)
Guy Thomas
December 2003
[1] It is, however, obvious that there is in theory a small disadvantage to insurance customers who are not afflicted by genetic conditions. Most people lobbying on behalf of the insurance industry fall into this category.
[2] Armstrong, K. (2003). Life insurance and breast cancer risk assessment: adverse selection, genetic testing decisions and discrimination. American Journal of Medical Genetics, 120A, 359-364.
[3] Typically the aggregate cost of a moratorium has been estimated to be “less than 10% of premiums” for most markets (eg Macdonald, Institute of Actuaries). However this is in terms of premiums representing the pure risk cost, without expenses. The variations which I observe in actual market premiums are “gross” premiums charged to the customer, ie the cost of the pure risk plus insurer’s expenses, commissions, etc . Typically about half of the gross premium represents expenses, and half the pure risk cost (source: LangjÆr-ØhlenschlÆger, M. & Mcgaughey, see my previous letter to GAIC). Hence a 10% variation in the risk cost corresponds to a 5% variation in the gross premium charged to the customer.
[4] I wonder if the Consumers’ Association would agree with this statement?