Neither as harmful as feared by critics nor as empowering as promised by providers: risk information offered direct to consumer by personal genomics companies

Abstract

In this paper, I investigate ethical and policy aspects of the genetic services and web-rhetoric of companies offering genetic information direct to consumer, and I do so with a special focus on genetic risk information. On their websites, the companies stress that genetic risk testing for multifactorial complex medical conditions such as cardiovascular disease and cancer may empower the consumer and provide valuable input to personal identity. Critics maintain, on the other hand, that testing can be psychologically harmful, is of limited clinical and preventive value, and vulnerable to misinterpretation. I stress the importance of empirical studies in assessing the pros and cons of direct-to-consumer testing and point out that recent empirical studies indicate that this testing is neither as harmful as feared by critics nor as empowering as promised by the companies. However, the testing is not entirely harmless. Remaining problems include testing of third parties without consent and ownership of genotypic and phenotypic information. Moreover, the testing, although not particularly empowering, may still provide input to self-understanding that some people find valuable. Regarding policy-making, I suggest that self-regulation in terms of best practice guidelines may play an important role, but I also stress that national and international regulation may be necessary.

Introduction

In recent years, a substantial number of companies have emerged offering DNA information direct to consumer (DTC) via the internet (Hogarth et al. 2008; Lachance et al. 2010; Genetics and Public Policy Center 2010). Some use genome-wide scanning technologies. Well-known examples of DTC companies (personal genomics companies or consumer genomics companies) are “the big four”: deCODEme, 23andMe, Navigenics, and Pathway Genomics. An aim of this paper is to investigate ethical and policy aspects of the genetic services and web-rhetoric of DTC companies with a special focus on genetic risk information. Another aim is to discuss the traditional criticism (“Criticism 1.0”) of this risk information in the light of new empirical studies and present a more updated and balanced criticism that takes these new empirical studies seriously (“Criticism 2.0”).

The new generation of genetic testing

The DNA information offered by the DTC companies may be health or non-health related. The health-related information may concern risk to develop multifactorial complex (“multiplex”) conditions like cancer and cardiovascular disease. It may also concern drug response. The non-health-related information may be about genes for non-disease traits like baldness and bitter taste perception or about people’s deep ancestry thousands of years back in time (“genetic ancestry tracing”; for an ethical analysis, see Nordgren 2010).

Regarding genetic disease risk information, which is the focus of this paper, the company Navigenics has made a useful distinction between “the new generation of genetic testing” that it (and one might add, other consumer genomics companies) offers and the traditional genetic testing in clinical genetics. This is how Navigenics describes the differences (for a similar distinction, see Grimaldi et al. 2011).

1. While the traditional testing concerns “rare, highly penetrant mutations”, the new testing concerns “common, low penetrant variations”.

2. While the disease outcome in traditional testing is “highly predictable”, it is merely “probabilistic” in the new generation testing.

3. While the impact on family inheritance is “specific” and “predictable” in traditional testing, the new testing “doesn’t yield predictable family inheritance patterns”.

4. While the “public health impact is low” of the diseases targeted in traditional testing, it is “high” of diseases targeted in the new generation testing.

5. While the “individual impact is high” in traditional testing, it is “low to high depending on genotypes” in the new testing.

6. Finally, in distinction to traditional testing, the “interaction with environmental risk factors” is of vital importance in the new generation testing (Navigenics 2012b)

It should be noted, however, that recently some DTC companies, including Navigenics, have started to provide information not only about multiplex conditions but also about variants linked to rarer, more highly penetrant diseases as well as carrier states. Examples are BRCA1, cystic fibrosis, PKU and Tay-Sachs (cf. Lerner-Ellis et al. 2010).

Risk information on the websites

How is the risk information in the new generation genetic testing presented on the websites of the DTC companies? The risk information may be of two types, lifetime risk (a type of absolute risk) and relative risk. This is how deCODEme explains these two types of risk:

Disease risk is a way to describe how likely it is that a person will develop a particular disease. The chance that a person will develop a disease at some point during their lifetime is referred to as lifetime risk. Because the development of a disease can occur at different times in different people, risk is often calculated as an average among groups of people. The likelihood that a particular group of people will develop a disease compared to the average likelihood of developing the disease is called the relative risk … It is important to remember that risk is a statistical term that best applies to large groups of people, and that your individual risk of developing a disease and the estimated risk based on a group of people may be quite different (deCODEme 2012b).

So, lifetime risk is the risk to develop a disease during lifetime, and relative risk is the likelihood that a group of people develops a disease compared with the likelihood that the average population develops it. The last statement in the quotation that the risk of an individual and the risk based on a group might diverge is extremely important. However, this aspect is mentioned only rarely on the websites of the DTC companies (I will return to this later).

The risk information on the websites is rather technical. It concerns risk in the strictly technical senses mentioned above. Additional information about advanced risk calculations is available on some websites. However, illustrations are provided to facilitate understanding of the technical risk calculations. Let me give a few examples from the demos of deCODEme and Navigenics.

deCODEme uses a thermometer-like scale to illustrate relative risk. This is how deCODEme puts it:

Your relative genetic risk shows where you lie on this simple thermometer like scale. See if you lie below or above average (1.0) by looking at the arrow and the number (deCODEme 2012c).

In this example, “your relative risk” is on the average 1.00, but as a result of testing it could also be 2.8 or 0.7.

In order to illustrate lifetime risk deCODEme uses two bars. deCODEme states:

Your lifetime risk is a way for you to compare your results to the population. The colored bar of the lifetime risk diagram shows your results and the lower bar shows the population average for your ethnicity and gender (deCODEme 2012c).

In the demo, “your lifetime risk” coincides with the average 17.0 %, but in real testing, it could be, for example, 25.0 or 10.5 %.

Navigenics uses other illustrations and focuses on lifetime risk. The company defines lifetime risk as “the percentage change a person of your gender has of developing this condition over an average lifespan” (Navigenics 2012c). The lifetime risk of an individual is illustrated by boxes in different colours. This is how Navigenics describes it:

For most conditions, we show you how your estimated lifetime risk compares with the U.S. population average. When you are about or below average, the box will be gray… Although your risk of a condition seems low, you could still be at significant risk of uncommon forms of that disorder not screened for by our tests. So, it’s important to click even on the gray boxes to learn more about these conditions.

If the box is orange, it’s a condition you might want to pay closer attention to, either because your estimated lifetime risk is more than 20 percent above the population average or because your estimated lifetime risk is higher than 25 percent (even if your’re at or below average) or for conditions where lifetime risk estimates are not available, because you have increased genetic risk (Navigenics 2012c).

Thus, low risk is displayed with percentages in a gray box and high risk with percentages in an orange box. The meaning of the last statement (“for conditions where lifetime risk estimates are not available, because you have increased genetic risk”) is not quite clear. It might just mean that very high risk is not displayed in percentages.

I will discuss some particular aspects of this risk information on the websites later. Let us first turn to the more general pros and cons of the genetic services of the DTC companies, or, more precisely, of their provision of genetic risk information DTC.

Pro arguments: the web-rhetoric of the companies

Pro arguments can be found on the websites of the companies. I will highlight two important types. The first is that risk information direct-to-consumer empowers the consumer. The consumer gets control over his or her life and health (for a deeper analysis, see Nordgren 2012). The second is that risk information direct-to-consumer provides knowledge pertinent to personal identity. The consumer learns more about himself or herself (for a more thorough analysis, see Nordgren and Juengst 2009). Let me give a few examples.

Empowerment

This is how Navigenics appeals to empowerment:

Our goal is to empower you with personal, confidential genetic insights to help motivate you to improve your health (Navigenics 2012a).

Similarly, deCODEme states:

Calculate genetic risk—empower prevention (deCODEme 2012a).

And:

Getting to know your personal genome will empower you and provide you with a road map to improve your health (deCODEme 2012d).

How may genetic risk information direct-to-consumer empower the consumer? As indicated on the websites, this can be done in two different ways. The first is by providing direct access to genetic risk testing without going through clinical geneticists and genetic counsellors in regular health care. In this way the customer gets direct control over her own genetic information. The second is by stimulating the customer to undertake certain medical measures, for example further screening, or to undertake certain preventive measures, for example changing diet and start exercising.

Input to identity perception

The second type of argument—the appeal to personal identity—is apparent already in the very names of two of the DTC companies, 23andMe and deCODEme. Both names indicate that genetics may provide important clues to self-knowledge. In the former case, it is indicated that “my” 23 chromosomes are closely linked to who “I” am as an individual person. In the latter case—deCODEme—it is indicated that “my” genetic code can be decoded and thereby inform the understanding of “my” personal identity. Moreover, 23andMe states:

By tapping into advances in DNA analysis and offering education, tools, and expertise, we at 23andMe want to help others take a bold, informed step toward self-knowledge (23andMe 2012a)

Regarding the part of personal identity that concerns health status, deCODEme encourages the potential customer to “deCODE your health” (deCODEme 2012a).

How may genetic risk information direct-to-consumer provide positive input to the consumer’s understanding of his/her identity, i.e. who he or she is? The websites indicate that this can be done in different ways. One way is by supporting the customer’s self-image regarding risk status and health condition. On the basis of the risk information, the individual may come to the conclusion that “I’m a healthy person. I’m not at risk” or the conclusion that “I believed that I was at risk. This is now confirmed and I can take preventive action”. Another way by which the customer may get input from genetic risk information is that it may change the customer’s self-image in a manner that is experienced as positive. The customer may conclude that “I believed that I was at risk, but I wasn’t” (Nordgren and Juengst 2009; for a more general analysis of genetics and identity, see Nordgren 2008).

Criticism put forward in the debate (“Criticism 1.0”)

Direct-to-consumer genetic testing via the internet has been discussed for many years. In this debate, several critical arguments have been put forward. Let me give a brief overview of some of these arguments. I will also comment on the arguments, in some cases extensively, in some only briefly. In many of the comments I refer to recent empirical studies. These studies may in some cases change the picture rather radically.

1. Some customers may be psychologically harmed. They may be confused or become anxious by their results (Ransohoff and Khoury 2010; Lippi et al. 2011).

   Comment: Bloss et al. carried out an important empirical study designed to examine the effect on consumers of direct-to-consumer genome-wide profiling to assess disease risk. They investigated the psychological, behavioural, and clinical effects of risk testing with the Navigenics Health Compass test. 3,639 subjects were recruited from health and technology companies and purchased the Health Compass test at a discounted rate. Of these 2,037 subjects completed the follow-up. Bloss et al. found that the testing did not result in any measurable short-term changes in psychological health (anxiety), diet or exercise behaviour, or use of screening tests (Bloss et al. 2011a). They acknowledge the limitations of the study in terms of study population and stress that additional studies are needed, but they also point out that their results so far a similar to those of studies of risk-disclosure for single-gene conditions (Heshka et al. 2008). The results are also in line with those of the interview study by McGowan et al. of the views of early users of DTC testing (2010). So, preliminary empirical studies indicate that consumers seem to experience no or only minimal psychological harm by DTC testing. However, this does not exclude the possibility that individual consumers still might experience more than minimal psychological harm, but the empirical studies referred to here at least give us reason to believe that this is rare.

2. The tests are of limited clinical or preventive value. In most cases, the added disease risk contributed by the genetic variants is only 10–40 % compared with the general population (Feero et al. 2008). Most known genetic variants constitute only a small part of the whole genetic variance of a disease.

   Comment: This objection seems to hold true. The empirical study by Bloss et al. indicates that not only does direct-to-consumer testing not lead to any change in psychological health, but neither does it lead to changes in diet or exercise behaviour or to further screening (Bloss et al. 2011a). Moreover, these findings are in line with the more general results of the Cochrane review by Marteau et al. regarding the effects of communicating genetic disease risk estimates on risk-reducing behaviours (Marteau et al. 2010) So, it seems that testing doesn’t have much value in practice. However, health and risk information might not be what is most important to the customers. Customers describing their experiences on the companies’ blogs rather seem to have a wish to know more about “who they are” (Gurwitz and Bregman-Eschet 2009). This is why the companies on their websites—as found above—in addition to an appeal to empowerment also often include an appeal to identity. Getting information about genetic risk might also be a matter of intellectual curiosity (Su et al. 2011).

3. A risk for misinterpretation of risk. The risk and severity of a disease may be exaggerated as well as underestimated. The customer may worry unnecessarily or develop a false sense of security. Misinterpretation could lead to unnecessary and expensive medical interventions or to neglect in undertaking proper preventative action (Hudson et al. 2007; Kaye 2008).

   Comment: The risk for misinterpretation should be taken seriously. An analysis of 29 websites of companies offering DTC testing found that the websites generally were quite demanding intellectually, indicating that many potential and actual customers may have problems in interpreting the genetic risk information (Lachance et al. 2010). Other evidence of difficulties of understanding risk and risk reduction—although without reference to genetics—has been provided by Gigerenzer and collaborators. A face-to-face computer-assisted interview study was conducted with more than 10,000 informants in nine European countries, aiming at assessing perceptions among the general public of cancer-specific mortality reduction associated with mammography and prostate-specific antigen screening. It was found that the vast majority of citizens in these countries systematically overestimate the benefits of screening (Gigerenzer et al. 2009).

4. The results from different companies are sometimes inconsistent. They cannot be trusted (US GAO 2010; Imai et al. 2011).

   Comment: The inconsistencies can primarily be explained by the fact that different companies partly focus on different single nucleotide polymorphisms and also use different risk prediction algorithms. Moreover, the companies may change their risk predictions over time due to further developments of their tests (Bloss et al. 2011b). Further developments of testing might reduce the inconsistencies.

5. Genetic counselling is seldom part of the services. Given the risk for misinterpretation, this is serious (Hudson et al. 2007; Beaudet 2010).

   Comment: It is true that the approach to genetic counselling differs among the companies. Most companies do not offer genetic counselling (Genetics and Public Policy Center 2010). Some companies do, however. Navigenics and Pathway Genomics provide genetic counselling as part of the service, and the information about this is easy to find on their websites (Navigenics 2012d; Pathway Genomics 2012). deCODEme provides free genetic counselling through their network of certified genetic counsellors, but this information is difficult to find on the website (deCODEme 2012f). 23andMe refers its customers to an independent network of genetic experts at an additional, but discounted, cost. However, even in this case the information is not easy to find on the website (23andMe 2012b). Given the risk for misinterpretation, to offer genetic counselling as part of the service package seems highly preferable, and information about this should be easy to find on the websites.

6. Supervision by a physician is seldom part the services. Genetic risk testing should involve a health care professional (Hudson et al. 2007).

   Comment: Most companies, for example deCODEme and 23andMe, do not require that the test is ordered via a physician. A few companies, for example Navigenics and Pathway Genomics, do require this, although they may recommend certain physicians who are prepared to order the test (Genetics and Public Policy Center 2010). In discussing this latter model of service provision, Howard and Borry point out, however, that it has proven rather difficult in practice to achieve “adequate medical supervision” in this context (Howard and Borry 2012). On the other hand, it is not self-evident that consumers would be better off if genomic testing were only available through a physician. Studies show that most physicians lack training and knowledge in genomics (Baars et al. 2005; Nippert et al. 2011; cf. Beaudet 2010). In addition, most physicians have clear short-comings in risk understanding and risk communication. Studies by Gigerenzer and collaborators indicate that doctors have problems understanding health statistics, including statistics from their own area of expertise. They conclude that these problems appear to be due, not to limited cognitive ability, but to the neglect of medical schools to teach risk communication (Wegwart and Gigerenzer 2011).

7. Increased workload and financial burden for regular health care. Patients may ask questions about test results regarding risk and many physicians may have problems answering them. Moreover, the testing may lead to increased screening and financial burdens (McGuire and Burke 2008 and 2011; Lippi et al. 2011).

   Comment: When asking social networkers about their attitudes towards direct-to-consumer testing, McGuire et al. found that 78 % of those who would consider to take such a test intended to ask their physician for assistance in interpreting the results (McGuire et al. 2009). Similarly, Gollust et al. found that 92 % of the participants in the Coriell Personalized Medicine Collaborative study intended to share their results with physicians, mainly in order to get medical recommendations (Gollust et al. 2012). However, in their study referred to above, Bloss et al. observed no increases in health screening after genetic testing. On this basis, they criticize the view that DTC testing constitutes a significant burden for regular health care (Bloss et al. 2011a). Similarly, McGowan et al. came to similar results in their interview study of early users’ views. The vast majority (19 of 23) reported that they did not take any sort of preventive health action on the basis of the test results, for example contacting a physician (McGowan et al. 2010). A study by Giovanni et al. also indicated that the amount of health-care referrals was limited. They made a survey among specialists in clinical genetics about individuals who consulted them after DTC testing. Responses were requested only from those who had seen a patient after DTC testing. They invited 4,047 specialists, resulting in 133 responses describing 22 cases of interaction after DTC testing. Of these 59.1 % were self-referred and 31.8 % physician referred. Giovanni et al. stress, however, that it is not possible to make sure how representative the results are (Giovanni et al. 2010). More empirical research is certainly needed to settle the issue of the impact of DTC testing on regular health care. Anyhow, we should not overstate the magnitude of the potential problem. It seems that the general demand for direct-to-consumer testing is small (Wright and Gregory-Jones 2010), and that the impact on regular health therefore could be expected to be small compared with the overall costs of regular health care.

8. Societal control of the commercial risk testing business is by and large lacking, leaving potential customers in a vulnerable position (Human Genetics Commission 2007; Kaye 2008; Hogarth et al. 2008).

   Comment: It is true that regulation for the most part is lacking (Grimaldi et al. 2011). In France, Germany, Portugal and Switzerland, genetic testing can only be carried out by a physician after informed consent. In Belgium and the UK, DTC testing is allowed (Borry et al. 2012). The Additional Protocol issued by the Council of Europe is a step towards a European regulation (Council of Europe 2008). The question remains, however, to what extent regulation is necessary and what kind of regulation is preferable. Self-regulation could play an important role (see further below).

9. People might have third parties tested without their consent. Gurwitz and Bregman-Eschet give a number of examples of cases when third parties might be tested without consent: a politician might want to disqualify a rival candidate, an attorney might want to defend a client by providing “genetic evidence” of promiscuous behavioural traits of the crime victim, a person might want to secretly test potential partners before deciding with whom to conceive a child, the police might want access to a company’s genetic database to find family members of a terrorist, which might lead to interrogation of people who just happen to share some genetic relatedness. Moreover, people might get harmed if a consumer genomics company sells its database to another company with less strict privacy protection (Gurwitz and Bregman-Eschet 2009).

   Comment: Testing of third parties without their consent could be a serious problem, although it is difficult to assess its magnitude. This might occur despite the fact that companies point out in the service agreement sections on the websites that by purchasing the genetic service consumers also confirm that the DNA sample is their own or comes from a minor in their custody (23andMe 2012c). However, one cannot say that this information is highlighted and stressed on the websites. It is rather hidden in the “small print”.

10. It is not clear who owns the genotypic and phenotypic information in the database. Most companies do not address this issue (Gurwitz and Bregman-Eschet 2009; Gruber 2011).

    Comment: Although some companies, like 23andMe (2012c), are rather clear about ownership, most companies are strangely silent on this issue. But, as Gurwitz and Bregman-Eschet point out, if the DNA sample and the information retrieved from it is the property of the company, then the company is free to sell or transfer it to third parties like any other commodity (Gurwitz and Bregman-Eschet 2009). This may have harmful consequences that the potential customer might not be aware of. This is certainly a serious problem that has to be solved one way or another (see further below).

11. The risk testing has unclear purposes. In their disclaimers, deCODEme and 23andMe state that their genetic risk tests are to be not for medical purposes but for informational, educational or research purposes. They encourage the customer to seek the advice of physicians or other health care providers if they have any questions regarding diagnosis, treatment and prevention (deCODEme 2012e; 23andMe 2012c). Despite these disclaimers the risk tests offered by the companies appear to be medical tests (even if some of them currently are of limited clinical or preventive value). If the very same tests were to be used in regular health care, they would be considered medical tests.

    Comment: This could be a way of avoiding responsibility and regulation. The companies could argue that the tests are not for medical purposes and that they therefore are not subject to, for example, the regulations of the Additional Protocol, mentioned above (Council of Europe 2008). The situation is unclear. Is it the purpose of use or the type of test that determines whether or not a test is medical?

12. Some companies use consumer data for research and some consumers might not be fully aware of this. The information about participation in research might be hidden in the “small print” of service agreement sections (Howard et al. 2010).

    Comment: This could certainly be a problem in some cases. Studies show that people who consented to have their tissue used in one type of research might not be willing to have it used in other types (see for example, Goodson and Vernon 2004). The problem is linked to the problem of who owns the data, mentioned above.

A more balanced assessment (“Criticism 2.0”)

Given my comments on the traditional criticism (“1.0”), I will now articulate a new criticism (“2.0”). Criticism 2.0 is based on recent empirical studies to an extent that criticism 1.0 is not; the latter is more speculative regarding the consequences of testing. Although the distinction is not strict, it has some implications for some arguments. Some criticisms still hold true, some do not and some new criticisms emerge.

Rather harmless

The harms are overstated in Criticism 1.0, but the testing is not entirely harmless.

1. As indicated by the quantitative study of Bloss et al. (2011a) and the qualitative study by McGowan et al. (2010), the risk for psychological harm seems minimal. The studies have their limitations and more research is certainly needed (although the review of empirical studies of risk-disclosure for single-gene conditions by Heshka et al. (2008) point in the same direction), but it seems that the risk for psychological harm in terms of anxiety and confusion is overstated in the traditional criticism (1.0).

2. It might be that the risk for psychological harm increases when the tests become more developed and include more genetic variants. And the risk might increase even more now when some companies have started to test also variants linked to rarer, more highly penetrant diseases as well as carrier states (see above; cf. Lerner-Ellis et al. 2010).

3. There is a risk for misinterpretation of risk. Misinterpretation may harm people, possibly leading to disempowerment and distortion of identity (cf. the pro arguments for testing analysed above). Gigerenzer’s studies of the general public and doctors indicate that risk can easily be misunderstood (Gigerenzer et al. 2009; Wegwart and Gigerenzer 2011). Above I briefly presented the risk information that can be found on the websites. We saw that the companies use various visual illustrations to help the customer to understand the results of the technical risk calculations. Despite these illustrations, however, there is a risk for misinterpretation. Some aspects need to be emphasized more strongly. It needs to be stressed that risk is a statistical term that applies to groups of people. The reference to “you” on the websites can be misunderstood. The specific risk of an individual (“you”, “your relative risk”, “your lifetime risk”) and the estimated risk based on a group to which he/she belongs may differ quite substantially. As pointed out above, deCODEme mentions this but does not stress it enough (deCODEme 2012b). Take the example of cancer. Just because the estimated risk of developing cancer of the group to which you belong is x percent doesn’t mean that your individual risk of developing cancer is x per cent. Your risk of developing cancer depend also on many other risk factors including family history, habits and environment. But even the combination of these risk factors might not apply directly to you; two people with the same genotype, sex, age, ethnicity, socio-economic status and lifestyle might still differ in disease experience. A good example of how cancer risk can be explained to laypersons can be found on the website of the Mayo clinic (Mayo Clinic 2012).

4. Having third parties tested without their consent may lead to harm (see above). This must be prevented.

5. The unclear ownership of genetic and phenotypic information may harm people (see above). Ownership must be regulated.

Not as empowering as promised

The benefits are overstated by the DTC companies.

1. As indicated by the quantitative study of Bloss et al. (2011a) and the qualitative study by McGowan et al. (2010), the testing is not as empowering as promised.

   1. Most customers receive the result that they do not have a more than minimally increased risk. They do not need to undertake any medical or preventive measures. So, while the risk information might give them peace of mind, it is not empowering in terms of promoting action. This makes one wonder whether it perhaps is unnecessary for most people to take the test in the first place. Are they encouraged to purchase a service they actually do not need?
   2. In some cases, there is a slightly more than minimally increased risk but there is nothing to be done. No medical or preventive measures can be undertaken. The risk information is not empowering in terms of promoting action. As we saw above, Navigenics illustrates increased risk by percentages in an orange box. In a comment, the company states that “If the box is orange, it is a condition you might want to pay closer attention to.” This statement is very thin and cautious. I interpret this cautiousness as a symptom of the limited empowering capacity of the testing. But it is also a sign of the unwillingness of the company to recommend any preventive action since this would be contrary to the stated purpose of not testing for medical purposes but merely for informational ones (see comment above).
   3. In a few cases action might be suggested, but still most people do not do anything. They do not change their diet or start exercising, and they do not take part in further screening. As pointed about above, these findings of Bloss et al. (2011a) are consistent with the Cochrane review by Marteau et al. regarding the effects of risk communication on behaviour (Marteau et al. 2010). One might suspect that most people simply do not find themselves motivated enough. The risk information is not sufficiently empowering.

2. However, the risk information may give at least some new self-knowledge. The customers get some input to their self-perceived identity and their intellectual curiosity nourished (Su et al. 2011). As pointed out above, this might be what they are ultimately seeking for.

Implications for regulation

Since the demand for the genetic services of the DTC companies is rather low (Wright and Gregory-Jones 2010), the magnitude of the ethical problems should not be overstated, at least not at present. DTC risk testing seems rather harmless psychologically, although there are some potential harms in terms of testing third parties without consent and, at least in some cases, unclear ownership of genetic and phenotypic information. In addition, the testing does not seem to be as empowering as promised by the companies. However, it should be noted that these considerations are only preliminary and more research is needed. But given this Criticism 2.0, what regulations are reasonable?

Do not overregulate

First, at this early stage of development of DTC testing it seems important not to overregulate, i.e. steer politically more than necessary. The main reason is that some companies contribute to the further development of genetic risk testing and too strict regulation may hamper innovation (cf. Lerner-Ellis et al. 2010; Caulfield 2011; Javitt 2010).

Self-regulation: best practice

Second, some regulation seems necessary, but this might to some extent be solved by self-regulation. Grimaldi et al. point out that there is a broad, but not universal, support for such self-regulation among the providers (Grimaldi et al. 2011). Here are some reasons in favour of self-regulation.

1. Some companies have taken criticism seriously, and have, for example, started to offer genetic counselling as part of their service packages. This indicates that self-regulation might, to some extent, be possible.

2. Self-regulation might be easier to implement rapidly than national and international regulation (Gurwitz and Bregman-Eschet 2009). So, at least for the time being, self-regulation could be a good approach.

3. The global character of DTC testing indicates the value of self-regulation. Consumers may submit their DNA samples from almost anywhere in the world, while it might be virtually impossible to enact regulation and oversight in all countries. However, if the DTC companies regulate themselves by forming a common organisation and adopting “best practice” guidelines the problems may be solved much easier and more effectively. Moreover, serious companies could in this way get a marketing advantage for the benefit of consumers at the same time as their innovative ambitions are not hampered.

Gurwitz and Bregman-Eschet have suggested that best practice guidelines should include the following components.

1. An adoption of the trustee model with respect to ownership of genetic and phenotypic information and thereby placing limitations on its transferability. A trustee has a legal title to property hold in trust for another person and owes a fiduciary duty to that person.

2. Explicit warnings on the companies’ websites against submitting another person’s DNA.

3. To require a signed statement in which customers confirm that the DNA sample is either their own or from a minor in their custody.

4. To require customers to sign a statement in which customers confirm that they have consulted a healthcare provider before ordering a health-related test and that a healthcare professional with knowledge in clinical genetics will be available to consult them about the forthcoming test results (Gurwitz and Bregman-Eschet 2009).

The first three suggestions seem warranted but not the fourth one. To require pre-test consultation with a healthcare provider seems overly paternalistic and also flawed given most physicians’ limited understanding of genetics and statistics, as indicated by studies referred to above (Baars et al. 2005; Nippert et al. 2011; cf. Beaudet 2010; cf Bloss et al. 2011b; Wegwart and Gigerenzer 2011). Moreover, it seems better if the companies include genetic counselling in their service packages than that the customer is to confirm that a professional with knowledge in clinical genetics will be available post-testing. The responsibility to see to it that genetic counselling is available should lie with the company, not with the customer. With this in mind, I suggest the following revised version of (4):

(4: revised) Genetic counselling should be offered as part of the service package, but it is up to the customer whether or not to make use of this offer.

National and international regulation

Third, national and international regulation is also needed. There are at least two main reasons.

1. Not all DTC companies might be willing to take part in self-regulation, as indicated by a study of Lewis et al. (2011). Lewis et al. investigated how DTC companies met the recommendations by the American Society of Human Genetics for what they should disclose to potential customers. They found that only six of the 25 companies studied met even 70 % of the standards and that, generally, the companies complied with the recommendations only 44 % of the time (Lewis et al. 2011). These results give us reason not to be overly optimistic regarding self-regulation. It might work, but this is by no means certain. But, as indicated above, self-regulation could be a first step to be followed by national and international legally binding regulation.

2. Some issues need to be legally regulated because they may involve not only the customer and the company but also third parties. Examples of such issues are prevention of testing of third parties without consent and ownership of genetic and phenotypic information.

As mentioned above, the Additional Protocol issued by the Council of Europe is a step towards a European regulation (2008). It states that clinical utility, clinical validity, medical supervision and genetic counselling are necessary. However, the Protocol will affect only those countries that have signed the European Convention on Human rights and Biomedicine (Council of Europe 1997). And even for those European countries that have signed it, it is still optional (Kaye 2008). Moreover, it might not be applicable to the testing of the companies, because, as pointed out above, this testing is stated by the companies not to be for medical purposes. As mentioned above, while DTC testing is allowed in Belgium and the UK, the legislations in France, Germany, Portugal and Switzerland require that genetic testing is carried out by a physician (Borry et al. 2012). A common feature of the Additional Protocol and the legislation in these latter countries is the requirement that a physician should supervise the risk testing. However, as I have argued above I do not support this requirement, although I strongly support that genetic counselling should be part of the service packages offered by the companies.

Conclusions

I have discussed the pros and cons of the genetic risk testing offered DTC by personal genomics companies. I have argued that recent empirical studies indicate that the DTC testing is neither as harmful as feared by critics nor as empowering as promised by the companies. However, I have also argued that there are problems that need to be regulated such as testing of third parties without consent and ownership of genotypic and phenotypic information. In handling these problems, self-regulation in terms of best practice guidelines may play an important role, however national and international regulation will probably also be necessary.

We have seen that commentators have very different views on direct-to-consumer genetic testing. My key presumption in discussing these views has been that ethical assessment needs to be informed by empirical knowledge. We need to discuss the actual effects of direct-to-consumer testing rather than speculations. I have referred to some relevant recent empirical studies but more studies are certainly needed. We still know too little about the short-term and long-term effects of direct-to-consumer testing, psychologically, medically and at a societal level. The ethical assessments and policy proposals put forward in this paper may need to be revised in the light of these further studies and in light of new developments in this still rapidly emerging field.