The literature search covered the period up to October 2014. The search was performed using electronic databases (PubMed, Web of Science, and Embase) and the Google search engine tool, Google Scholar. On Google Scholar, we investigated all the results obtained by the databases, but considered only the first 500 results because the number of relevant articles declined substantially after the first 300 results and because this search engine displays results by relevance using a link analysis system or algorithms [23].

We used “Direct-to-consumer genetic test” as the key term for each database and for Google Scholar. We scanned the reference lists for relevant articles up to the second level, and we considered the “related articles” of relevant ones in the PubMed database or Google Scholar when the paper was not present in PubMed.

We included all articles relevant to the subject of the research where the key term was anywhere in the text of the paper, written in English, with the abstract and full text available. We included only scientific articles, excluding popular articles published in daily newspapers or weekly and monthly magazines. Papers included articles associated with health-related genetic tests available online and offered direct-to-consumer. We selected only the articles reporting original data, excluding those with speculative discussion about the problem or citing data from other studies (ie, editorials, letters, comments, articles about regulation issues, and reviews).

Two investigators read the papers (LC and UG) and independently assessed the potential relevance of all publications, identified during the database search, based on the information provided in the titles and abstracts. Disagreement was resolved by consensus.

After screening by titles and abstracts, the first author critically reviewed the full texts of the remaining articles and extracted the information required to perform the review. The methodological quality of each study was assessed by 2 authors (LC and EC) using the Kmet tool for evaluating quantitative and qualitative research [24]. A score of between 0 and 1 was assigned to each paper based on a series of questions related to the type of study. Case studies and descriptive reports (a total of 38 papers) were excluded from the evaluation. Disagreements were resolved through discussion among the authors until consensus was reached. As shown in Figure 1, we identified 118 articles that fit the inclusion criteria.

A total of 20 articles explored health professionals’ opinions of DTC-GT as reported in Table 3 in Multimedia Appendix 2. Two of these are qualitative studies [96,97]. Out of 20 studies, 14 were mainly conducted in the United States (70%) [32,41,97-108]. Half of the surveys were implemented online [32,41,98-101,107-110].

Four studies recruited more than 1000 subjects [32,102,108,111], and five studies recruited more than 300 subjects [52,98,101,105,107]. Almost half of the studies (44%) had a response rate of around 40% [100-102,104,109,110,112] and other 6 studies (38%) had a response rate less than 20% [41,99, 103,105-107,111].

The awareness of DTC-GT among physicians was high (around 90%) in three studies [41,100,110]. In other surveys of physicians not specializing in genetics, fewer than 55% of respondents were aware of DTC-GT [102,105,111,112]. In general, the percentage of physicians who have discussed GT results with a patient or have ordered a GT for a patient was quite low. Sixteen percent of physicians recruited by Bernhardt et al [98] ordered one test a week or more, and only 7% reported having seen a DTC genetic risk assessment report. In five surveys, fewer than 19% of physicians, both specializing in genetics and not, reported having patients request genetic consultations [32,99,105,107,108]. Forty-four percent of clinical geneticists from different European countries had been contacted by at least one patient regarding DTC-GT services after purchasing the test [110], and 46% of genetic counselors from the United States had worked with patients who had initiated a discussion of DTC-GTs: however, only 15% of the latter had suggested them to their patients [101]. Forty-two percent of primary care physicians enrolled in an online survey in the United States had ordered a GT for a patient, and one third had ordered them for themselves [100]. Only 0.5% of general practitioners and 1% of clinical geneticists from Japan ordered DTC-GT [111]. A large percentage of neurologists (74%) and the 14% of psychiatrists recruited from the American Medical Association ordered a GT for a patient [107].

Another interesting topic is how confident physicians are in interpreting GT results. In the study by Bernhardt et al [98], 16% of primary care physicians declared themselves to be “very confident,” along with 15% of family physicians in the study by Powell et al [106] and only 7% of physicians specialized in genetics from the study by Brett et al [109]. In a study on nutrigenomics, health professionals reported a lack of competency to provide information on nutritional genomics. Inability to support a patient in managing genetic risk information also emerged from interviews with 18 clinicians providing genomic risk assessment services to their patients [97]. The study by Salm et al [107] reported the need to have more training in interpreting GT results; although in the context of predictive genomic testing, the United States has promising training programs for genetic counselors [103].

Health professionals’ opinions on the clinical utility of DTC-GT were contrasting, and the percentage of those who were in favor of these services was different among the studies. Giovanni et al [99] found that 52% of health care providers described the genetic test as clinically useful. The majority of respondents (86%) mentioned usefulness in the context of breast cancer susceptibility, in agreement with findings from Mainous et al [102]. The latter also found that 30% of participants perceived GT utility in detecting Alzheimer’s disease and 25% of participants for heart disease or diabetes. Half of genetic counselors recruited by Hock et al [101] said that GT should be limited to a clinical setting, 23% of the sample was neutral, and 27% disagreed. Furthermore, 56% of the sample considered a DTC-GT acceptable only with the provision of genetic counseling, 31% were neutral, and 13% disagreed. In the study by Bernhardt et al [98], physicians thought that genetic tests would be helpful in managing patients; in particular, 70% felt it would be useful with pharmacogenomics and 40% with disease risk assessment. However, only one third of physicians in both cases would order such testing for a patient. About 47% stated that genetic testing would be helpful for patients, motivating them to adopt healthy behaviors. Also, the clinicians interviewed in a recent study [97] were enthusiastic about the potential of GTs to enhance the personalized, preventive, and wellness orientations of their clinical practices.

In a study conducted in Australia [109], the majority of genetic health professionals did not consider DTC-GT useful for individuals who want anonymous testing (54%), are driven by curiosity (54%), or are geographically isolated (60%). Forty-three percent of physicians in the Powell et al [105] study considered DTC-GT clinically useful. In a Greek study by Mai et al [52], only 13% of medical practitioners were in favor of DTC-GT. Similarly, 86% of clinical geneticists recruited from 28 European countries [110] considered it unacceptable to provide a predictive test without face-to-face medical supervision, and all respondents expressed the unacceptability of offering DTC-GT for conditions neither treatable nor preventable.

Ohata et al [111] carried out a survey on 1145 general practitioners and 294 clinical geneticists in Japan. Convenience scored highest in both groups as the reason behind users’ ordering DTC-GT, and general practitioners rated the benefits of DTC-GT higher than clinical geneticists (score 2.54 vs 1.96 on a scale 1-4, 1=disagree). Among the risks, the concern for understanding results scored highest in both groups (score >3). Furthermore, reliability of results and provision of information/counseling were a source of concern greater in clinical geneticists than general practitioners (score 3.13 and 3.78 vs 2.77 and 3.48 respectively).

In a study conducted in New Zealand [112], general practitioners who had not received training in genetics agreed that convenience was a benefit, more than those with training (72% vs 38%, P<.005). At the same time, misinterpreting results and inadequate delivery of information were perceived to be the greatest risks associated with DTC-GT by the majority of respondents (around 90%). In general, only 19% agreed that DTC-GT provides a useful service in the delivery of health care, and 26% agreed that results encourage patients to take responsibility for their health. Clinical validity of the test (25%) and counseling (20%) were the most selected aspects regarding advertising regulation of DTC-GT.

In another study [108] dealing with GT in children, genetic counselors appeared less prone to GT compared with non-genetic physicians.

There was one study exploring the knowledge of and attitude toward personal genomics on a small group of medical students enrolled in a human genetic course [104]. The percentage of students who thought that genotyping information would be useful to physicians and consumers decreased after the course (32% post-course vs 63% pre-course and 52% vs 84% respectively). The majority of students, both before and after the course, expressed concerns about reliability and utility of results. They agreed that tests needed interpretation (around 70%) and DTC companies had to provide genetic counseling (71% pre- and 80% post-course).

More than 80% of physicians recruited by Powell et al [105] expressed concerns about possible misinterpretation of test results and increased anxiety in patients. Almost half of physicians (neurobiologists and psychiatrists) surveyed by Salm et al [107] thought that GT could cause psychological harm to their patients and they could be exposed to possible insurance discrimination. This was further confirmed in the study by Bernhardt et al [98].

Uncertainty about clinical utility concerned the majority of primary care physicians (around 60%) in the study of Haga et al [100], with a recommendation for health care professionals to act as intermediaries also when discussing DTC nutrigenomic tests [96].

A total of 16 articles were identified regarding issues and marketing strategies related to the type of information provided by the DTC-GT websites (see Table 4 in Multimedia Appendix 2). The number of websites analyzed ranged from three [113] to 38 [114].

Goddard et al [115] found 27 health-related DTC-GT distributor websites and evaluated those that sold tests for thrombosis risk. Liu et al [116] analyzed 46 websites, but only 20 of them allowed consumers to order directly from the company. Sterling et al [117] identified 64 organizations hosting websites promoting nutrigenomic services, but only 29 offered or promoted at-home testing.

Borry et al [118] and Howard et al [119] investigated online companies focusing mainly on their policies in regard to GT for minors. The former analyzed 29 companies obtained from a list published by the Genetics and Policy Center, and the latter sent a questionnaire to 37 DTC-GT companies. Both studies emphasized a lack of exhaustive information on the privacy policy regarding minors, which is a deviation from the professional guidelines on this issue. Also, in a recent systematic Internet search for DTC genomic services, limited information on privacy policies was found [120]. This evidence contrasts with other studies that found the majority of sites selected provided this information [115-117,121,122].

Most studies assessed the quality of information provided by online GT companies through a content analysis of websites, with a focus on the provision of genetic counseling, suggestion for a physician’s consultation, and the description of risks, benefits, and limitations of GTs.

In relation to online genetic counseling, Geransar et al [123] showed that of the 24 online companies studied, 75% recommended and arranged for counseling services. However, only one-third of the companies directly provided counseling services and just one of them provided a face-to-face format. Half of the websites analyzed by Covolo et al [124] provided this service pre- and/or post-test, with 20% offering this service for an extra fee. In other cases, fewer than 39% of online companies provided genetic counseling [114,119,121,125,126]. Pre-test counseling was rarely offered in studies conducted by Hennen et al [114], Lachance et al [121], and Liu et al [116]. None of the 29 companies offering nutrigenomic services examined by Sterling et al [117] provided genetic counseling.

Additionally, except for the websites analyzed by some studies [115,123,124,127], very few companies suggested a physician’s consultation [114,117,119,121,125]. Sometimes the GT sale was accompanied by recommendations associated with disease prevention or health improvement (eg, nutritional supplements). This trend was found in the majority of websites (from 60-74%) investigated by Lewis et al [122] and Singleton et al [127]. Of 64 websites promoting nutrigenomic services identified by Sterling et al [117], 53% provided recommendations for dietary intake or supplementation.

Genetic discrimination, emotional consequences, risk of behavior changes, and confidentiality of test results are possible risks associated with GT. In general, all studies that searched for this information found that the risks were poorly cited, ranging from about 20% [115,116,122,128] to about 30-36% of the websites [117,124,125,127]. Of the company websites analyzed by Hennen et al [114], 47% provided information on consequences and actions to be taken in the case of a positive test result, and 37% in the case of a negative test result.

Clearly, the benefits of testing are described more than risks [115-117,124,127,128]. In particular, empowerment over one’s health was highlighted by several authors [113,116,124,128]. Almost all of the sites identified by Lachance et al [121] and Singleton et al [127] listed at least one benefit to consumers by undergoing testing. Three-quarters (76%) of websites analyzed by Lachance et al [121] highlighted the fact that test results can help inform consumers in making a health decision. In the second study, prevention of the onset of a disease was the most common benefit presented (96%). Interestingly, 52% of websites stressed the consumer’s ability to use the results to make informed decisions. The concept of patient empowerment also appeared in the Sterling et al [117] study. In fact, 73% (47/64) of organizations analyzed mentioned that consumers could use test results in their own diet and lifestyle decision making.

Over three-quarters (78%) of websites analyzed by Singleton et al [127] and about half of the websites analyzed by Lachance et al [121] and Lewis et al [122] mentioned limitations of test. None of the websites selling DTC-GT for thrombosis reported limitations [115].

Very little information or scientific evidence was provided on the clinical validity of tests [114-117,121-123,128]. Some websites referred to a laboratory certification, such as Clinical Laboratory Improvement Amendments (CLIA) standards, to indicate legitimacy [114,115,117,120-122,124,128].

A total of 14 papers, including two reports from the US Government Accountability Office (GAO) [11,129], question the scientific quality, clinical validity, and utility of DTC-GT (see Table 1). This issue was addressed in different ways.

Seven studies [11,129,130,132,133,136,138] focused on the comparison of GT results from DTC companies for one or more individuals. The first study, executed by the GAO in 2006 [129], evaluated the results of nutrigenetic tests purchased from four DTC companies for 14 fictitious consumers with different characteristics obtained from two DNA samples. Interestingly, all 14 results predicted the risk of developing different medical conditions. These predictions were similar for all of the fictitious consumers, no matter which DNA or lifestyle description was used. Only one of the four companies gave contradictory results.

In a more recent report by GAO [11], 5 individuals purchased 10 tests manufactured by four different DTC companies. The tests were specific to 15 diseases. The analysis found a large variation in prediction risk from company to company. In agreement with the GAO report, Ng et al [138] found a modest concordance among the results (50% or less) from two DTC companies on 13 diseases for 5 individuals.

Similarly, Imai et al [133] compared the relative common disease risks obtained from three DTC-GT companies for the same individual and found comparable results from the single nucleotide polymorphism (SNP) analyses from different companies. However, in a similar recent study [132], they also pointed out a large variation in relative risks for some of the diseases investigated, possibly due to different SNPs used to calculate the same disease, the choice of the reference population, and the risk calculation methodology.

Bloss et al [131] compared the DTC genomic risk estimates with self-reported disease from individuals who purchased a GT. The risk estimates were significantly associated with self-reported family or personal health history in only five out of 15 conditions studied. Two studies [135,140] examined the risk assessment of common diseases in DTC-genomic services and found that the predicted risks differed among the companies due to different methodologies used, different loci, and SNPs selected for analysis.

In an evaluation of type 2 diabetes risk prediction from commercial companies offering genome-wide scanning [137], it was shown that the individual risk prediction changed depending on the number of polymorphisms used to calculate the risk and characteristics of people (eg, age and gender). In particular, 39% of 5297 individuals switched between risk categories once and 11% switched twice. A study by Palomaki et al [139] of type 2 diabetes, genomic risk profiles advertised by DTC-GT companies highlighted a lack of analytical validity and clinical utility in the tests through the Evaluation of Genomic Applications in Practice and Prevention Working Group. This approach was established to support the development of a systematic process for assessing the available evidence for GT in clinical practice.

Other studies focused on the scientific evidence of genetic polymorphisms used to estimate the disease risk by DTC companies. In particular, Janssens et al [7] looked for meta-analyses supporting 69 polymorphisms tested by seven companies and found inconsequential scientific evidence. Similarly, it was found that only 18 out of 88 SNPs identified as known SNPs in genomic databases associated with a disease were present in a commercially available test [139].

The quality scores of the evidence reviewed ranged from 0.55-0.95. The majority of the studies that could be evaluated (69/80, 86%) had a score >0.7 (data not shown). Overall, all the studies are adequate in terms of methodological quality. The bias that was mostly present was a selection bias due to the recruitment of convenience population or a small sample size that did not allow a generalization of the results.

This review summarized the scientific literature on DTC-GT with a comprehensive view meant to unravel the complexity of the DTC-GT market. Previous systematic reviews dealt with this topic by focusing on certain aspects, particularly position statements, policies and recommendations [15], user perspectives [18], and health professionals’ perspectives [19]. This systematic review aimed to give an overall view of the DTC-GT market to include studies that analyzed the content of the websites offering these products, as well as studies focused on the scientific evidence and clinical utility of such tests. The large number of reports retrieved on this issue indicates a strong interest in the topic.

Thanks to the prevalence of the Internet over the past decade, the availability of health-related products on a DTC basis has become increasingly common. However, the fact that the promotion of these products such as drugs [141] or nutritional supplements [142] is comparable to the sale of any commodity is a cause for concern.

In terms of marketing, we must discuss the results considering the product, the offer, and the potential customers, in addition to the opinion of health professionals as product experts.

Advances in genomic technology made GT available for both monogenic disorders and common complex diseases, in addition to nutrigenetic and pharmacogenetic tests. To date, the majority of these tests have provided a poor predictive value, and the assessment of the clinical validity and utility is still a work in progress. However, many commercial companies have begun to bring these tests to market. Their lack of scientific evidence was confirmed by some studies focused particularly on commercially available GTs [7,134]. Overall, all studies comparing the results of GT of the same people from different companies showed a modest concordance in risk predictions and sometimes reported contrasting results [11,129,130,132,133,136,138].

In general, the studies focused on the content of websites selling DTC-GT agree that these companies do not provide complete information to the consumer. This emphasizes the poor quality of information on the scientific evidence and clinical validity to support the tests. It also highlights the lack of attention to the risks related to the performance of the tests, genetic discrimination, emotional consequences, behavior changes, and confidentiality of test results. Furthermore, genetic counseling requirements were often missing. As with other sales-oriented companies, these websites contain marketing strategies that accentuate the benefits of the product. The benefits of testing were described more than the risks, and the theme of patient empowerment is highly emphasized as a good reason for testing [113,116,124,128]. The main emphasis on genes, without consideration for environment, might lead consumers to misinterpret test results, as was found by the majority of studies that addressed this issue.

Thirty-seven papers examined consumer knowledge of and attitude toward DTC-GT; likewise 37 papers discussed the impact of these tests on users. In general, it was confirmed that consumers have an interest in DTC-GT and that their main motivation is curiosity, as well as some interest in monitoring and improving health.

It should be noted that interest is highest among employees of biotechnology companies [30] or people at risk for cancer [35] and other diseases [25]—not the general public. Additionally, study participants were highly educated.

As for the impact of DTC-GT on users, nine papers and eight case studies were reviewed, in addition to those researched by Goldsmith et al [18]. The research remains limited because it evaluates the actual consequences of having a DTC test. Other studies reported situations where participants were exposed to mock reports about their genetic susceptibility or were asked to voluntarily submit to testing for research purposes. It seems that the feared negative consequences, such as psychological impact or increased anxiety for consumers, were not confirmed. Similarly, positive consequences, such as adoption of healthier lifestyle behaviors, were not observed, although a large proportion of people expressed the intention to change lifestyles [43,55,63,80].

Negative consequences may arise from misinterpretation of test results, which is another aspect addressed by several studies. The majority of studies showed that participants did not have particular difficulty understanding the test results. In fact, only a small number of people shared their results with a physician and reported worry after receiving them. Yet, it was also determined that the presence of a professional provided better interpretations of results compared to participants who received results online [87]. Interestingly, incorrect interpretation of results was confirmed in actual DTC-GT customers [43,91].

We considered health professionals’ perspectives as expert opinions, considering the strong recommendation for involvement of a health professional in the order process and interpretation of test results [143]. Compared to previous systematic reviews [19], 12 additional articles were retrieved that focused on health professionals’ perspectives and the overall scenario described by these authors were confirmed by our findings. As stated by Goldsmith et al [19], the level of awareness of DTC-GT remains inconsistent, even with three studies [41,100,110] in which the majority of physicians are aware of DTC-GT (but the sample size was ˂50 participants). In addition, few respondents have had direct experience with DTC-GT. The overall opinions regarding the utility of the tests are contrasting. Some professionals are in favor of some GT [99,101,102], while others considered it unacceptable to provide a predictive genetic test without genetic counseling [110] and were concerned about possible psychological harm [107], misunderstanding of results, and insurance discrimination [98]. Understandably, clinical geneticists expressed more concerns than general practitioners [107,110,111]. It should be noted that few physicians considered themselves confident in interpreting GT results and reported the need for more training [97,106,107,109]. In fact, an increase in the incorporation of instruction about application and technique in predictive genomic testing was presented in a recent study [103].

As discussed by some authors [97,144], these concerns raise the question of whether a non-geneticist physician involved in the commercial distribution of GT is properly equipped to offer test information to patients. So the presence of a physician does not guarantee the provision of adequate information. This is a cause for concern considering the recent shift from selling tests directly to the consumer to a direct-to-provider marketing model [97,118,144].

We believe this review highlights the important aspects in considering the regulation of DTC-GT from a policy perspective. More specifically, there are at least three main issues to address to improve DTC-GT for a better service for the public:

Overall, research and practice must collaborate toward policy making in a field that is already open to the public despite its serious pitfalls.

Through a systematic approach we aimed to provide a comprehensive look at the DTC-GT market in order to better understand its actual impact on population. Although the number of articles retrieved is relatively large, some limitations related to studies design should be underlined. The majority of the studies used a cross-sectional design. It is known that response rate as well as sample size and sample selection are critical points in this kind of design [145]. Considering the total of the surveys investigating the awareness, use, and perceptions of health professionals and consumers, only 9 studies [30-33,40,52,56,59,81,92] out of 56 surveys (16%) (see Multimedia Appendix 2) have a response rate of at least 50% and a large sample size (more than 1000 subjects).

Most of the subjects recruited were highly educated and sometimes selection bias was present (eg, employees of health and technology companies [81]). All these aspects mean a poor representativeness of population [145]. Furthermore, few health professionals and consumers had direct experience with DTC-GT, so as previously argued [18,19], the responses of participants based on hypothetical scenarios make it difficult to draw conclusions about the actual impact of DTC-GT market.

Based on the evidence collected, it seems that DTC-GT is neither beneficial nor detrimental to potential users. It should also be noted that the development of online companies is rapidly changing, most likely due to pressure from government agencies such as the FDA. Some companies have also changed their delivery model to include the health profession in the order process [143].

However, regardless of the large amount of data available on this issue, the actual experiences of DTC-GT users are still limited and this market is still in the early stages of distribution to the general public. Furthermore, some limitations on previous studies must be addressed. For instance, the majority of studies are characterized by people who do not represent the general public (participants were often convenience samples), featuring low sample size or limited response rate. Additionally, the prospective studies typically employ relatively short-term follow-up in the majority of the cases, not sufficient to evaluate the impact of DTC-GT on behavioral changes.

On the other hand, it is unacceptable that online companies offer GT lacking scientific evidence, no proven clinical utility, and misleading marketing claims. As underscored by Janssens and van Duijn [146], the expected benefits of whole genome scanning may be larger when tests are targeted only to specific at-risk populations, and not to populations-at-large, because of the moderate predictive ability of these current tests.

According to global industry analysts, the global genetic testing market is expected to reach more than US $230 million by 2018 [147]. Combined with the rapid decrease in biotechnology costs, this revenue stream will eventually allocate testing accessibility to all socioeconomic classes. It is important, therefore, to remain cautious and vigilant about this growing, influential health care market.