Table 2 – Summary of papers identified for systematic review
Reference / Purpose of study / Methodology includingsample information / Data collection method and
method of data analysis / Main findings / Quality
including Kmet score[1] /
Bloss et al (2010) / To assess consumer response to DTC personalised genomic risk assessment.
(Scripps Genomic Health Initiative - this paper reports on the baseline findings only) / Quantitative
Longitudinal cohort study
n=3640
(4884 enrolled – response rate of 74.5%).
Adults 18-85.
Eligibility criteria:
18+
Valid email address
Ability to provide a co-payment for test
Recruited from employees of large health & technology companies: highly educated, well off, adequate access to healthcare, in good health. / Online health assessment questionnaire.
Baseline health assessment; behavioural health measures; health care status; perception of DTC genetic testing.
Statistical analysis on SPSS, R and Dimension Research. Data screened for extreme cases.
Descriptive statistics and bivariate associations using chi-squared and Mann-Whitney U tests.
Logical regression for predictors. / Concerns:
Overall, almost 50%;
13% about learning of disease risk;
16% about unknown reaction to results;
16% about quality and reliability of data;
36% about privacy issues.
Concern highest in women, health-related occupations and individuals who perceived their health as ‘less than good’. Also, younger age, lower income and higher trait anxiety.
Concern decreases with age and increases with level of trait anxiety. Lower education – less likely to express concern.
Knowledge of risk:
82% would want to know their risk, no-one said definitely not.
Uncertainty highest among women, white, health-related occupation; also younger age, higher trait anxiety.
Non-white individuals less likely to purchase and undergo DTC tests.
It is suggested that if clinical validity and utility of DTC GWAS-based tests is demonstrated, consumers could benefit from tailored education and counselling services. / 0.68
This sample is not representative of the whole population and findings cannot therefore be generalised.
Bloss et al (2011) / To examine the psychological, behavioural and clinical effects of ‘risk scanning’ with a DTC genomics company (Navigenics)
(Genome-wide scan, uncertain clinical validity and utility).
(Scripps Genomic Health Initiative - this paper reports on the baseline and 3 month follow-up findings) / Quantitative
Longitudinal cohort study
n=2037
(3639 enrolled -
44% attrition).
Adults, as previous paper. / Online health assessment questionnaire - baseline, 3 month post-test and 12 months post-test (only reporting the 3 month follow up in this paper).
The test:
Analysis focused on 2 risk information formats: estimated lifetime risk (% age) and colour-coded risk for 22 conditions.
Primary outcome measures:
Changes in anxiety symptoms, dietary fat intake and exercise behaviour.
Secondary outcomes:
Test-related distress and subsequent use of screening tests.
Scores adjusted for age, sex, education, ancestry, income, health related occupation.
Various statistical tests for related samples and increase of use of screening tests (Wilcoxon signed rank);
to assess relationship between follow up scores (anxiety etc) and average estimated lifetime risk of all conditions, proportions of conditions color-coded orange (>20% above average risk, overall lifetime risk >25%) and estimated lifetime risk and color-coded risk for each of 23 individual conditions (linear regression);
for correlation between use of screening tests and the 2 composite risk estimates (Spearman’s rank correlation coefficients); relationship between people accessing genetic counsellor or physician and behavioural scores (linear regression).
Descriptive statistics on subjects who accessed genetic counsellor or spoke to their physician, and tested to see if this was associated with behavioural scores (using linear regression).
Used p<0.05 as significant. / No significant difference between levels of anxiety, dietary fat intake or exercise behaviour between baseline and follow up.
Actual and intended use of screening post-test:
About 50% intended to undergo additional screening and the number of screening tests was significantly increased from zero.
No significant associations between risk scores and behavioural outcomes.
90% showed no test related distress.
No significant association between risk scores and total number of screening tests actually completed after genetic testing.
BUT there was correlation between:
· risk scores and no. of screening tests subjects intended to complete post-test.
· risk scores and proportion of orange coded conditions.
· test-related distress with lifetime risk.
· test related distress with orange coded conditions.
10% discussed results with a counsellor,
26% shared results with physician.
Sharing of results with a physician was associated with lower fat intake and increased exercise.
In this sample, there was no evidence that DTC genome testing produced any measurable behavioural changes. / 0.77
See above – this is a biased sample.
Cherkas et al (2010) / To explore the reasons why people would consider taking a commercial, internet based personal genome test (PGT). / Quantitative
n= 4050
(twins aged 17-91)
(62% response rate).
Sampling frame:
Database from TwinsUK Adult Twin Registry (age 16 and over).
Age, gender, family structure and socio-economic status were taken, and used to create sub-groups for the analysis of data.
Mean age 56, 89% female and lived all over UK.
Non-respondents younger on average - 50 (17-91), higher proportion of males.
79% of respondents had children. / Questionnaire
Analyses via STATA 10 software.
Respondents divided into under and over 50 for comparison purposes.
Spearman rank correlations to assess relationship between responses and actual age as well as between SES groups.
Chi-square to compare differences in responses between M and F, with and without children. / Awareness: Only 13% had heard of PGT, but younger people significantly more likely to be aware.
Level of interest in taking test clearly dependent on cost. If free, 48% expressed interest, 22% undecided, 30% unlikely.
Younger people and males significantly more interest than older and female (p<0.01).
Respondents in higher SES group significantly less likely to order test if it was free than those in lower SES group.
Reasons for testing:
(analysed from those who had expressed at least some interest, n=2814).
Most frequent reason – (93%) to adopt a healthier lifestyle if high risk result; younger people significantly more likely to endorse this reason (p<0.01).
Females more likely than males (p<0.01).
86% - to learn more about myself, again with younger significantly more likely to endorse this reason (p<0.01).
80% -Conveying risk to children,
79% for doctor to monitor health. Older more likely to endorse than younger (p<0.01), females more likely than males (p<0.01) and those with children more likely than those without (p<0.01).
For financial planning, 50:50 with only 20% expressing strong agreement (more likely if older). Also more likely if female or have children.
Of note: no significant trends with SES for any of the above reasons. / 0.90
The participants in this sample (TwinsUK database) are likely to be familiar with genetics as they have volunteered for a wide range of research.
Findings from ‘reasons for testing’ may not be robust, as participants had to choose between specific options.
Qualitative data may have been useful to identify the reasons in more detail.
Gollust et al (2011) / To assess the motivations, perceptions and intentions of participants at an enrolment event for the Coriell Personalized Medicine Collaborative (CPMC). / Quantitative
n=369 (response rate of 55.5%) from a sampling frame of people who registered for a CPMC enrolment event over the one year study period. / Internet based survey using Likert scales for:
Awareness and prior use of personalized medicine and DTC testing, perceptions of personal genomics and the risks and benefits of the CPMC study. There were also questions on respondents understanding of the CPMC study and whether or not they planned to share their results with an HCP.
Descriptive statistics; bivariate logistic regression and multivariate logistic regression models to see how respondent characteristics related to their perception of personal genomics. / Motivations: Curiosity, finding out about their disease risk and improving their health.
More than 50% took part because they hoped to find out their risk for a particular condition.
(Most common: heart disease, n=58; diabetes, n=24, general cancer, n=22; specific cancer, n=25. 12 people wanted to know their risk of Alzheimer disease.
Most people accepted that risk of common disease was multifactorial.
Most believed that the study would provide them with health-related benefits (behaviour change and personalised health plans).
Some had unrealistic expectations eg gene therapy (13%).
Concern about risks was moderate (31% believed no risk at all).
Most common:
Worry (30%) and unwanted results (29%).
Overall, 32% had misperceptions of personal genomics.
Predictors: less likely over 55, more likely, those not working in a health profession.
91.7% would share results with physician.
Only 25% would share to receive an explanation, but 65% wanted health advice based on the results, 79% wanted prescription of medicines based on their genes and 71% believe the result should form part of their medical record.
Of those who would not (25), 9 doubted their doctor’s ability to interpret DTC results and 8 were concerned about privacy. / 0.93
This sample is not representative of the whole population – recruited from people enrolling on a research study for a large genetic health research organisation.
Gray et al (2009) / To evaluate whether exposure to information on potential risks of DTC BRCA testing in different formats would alter women’s beliefs about online BRCA testing and intentions to get BRCA tested. / Quantitative
Randomised controlled trial: 3 conditions – no risk information (CC), unattributed risk information (URI) and expert risk information (ES).
17 did not meet inclusion criteria.
304 allocated to 3 experimental groups.
284 analysed.
(6% attrition)
Women, mean age 39 years.
82% white, 58% married, mean education of 3 years of college.
n=321 / Telephone interview for baseline data on potential co-variates.
Online survey following viewing of ‘stimulus material’ (“mock” website).
Descriptive statistics (frequencies) for participant characteristics.
Pearson’s chi-squared and ANOVA for differences between groups.
Multiple logistic regression to adjust associations. / Participants exposed to risk information had lower intentions to get BRCA tested and had less positive beliefs about online BRCA testing.
Participants in URI group had lower intentions than CC.
Women in ES group had higher preference for clinic testing and more negative beliefs about internet testing than women in CC.
This study shows that women’s beliefs about DTC genetic testing, intentions to get BRCA tested, and preference for where they get tested are altered by exposure to risk information. / 0.80
Kaphingst et al (2010) / To inform the ongoing debate re whether individuals offered DTC susceptibility testing can make informed decisions using online decision aids. / Quantitative
‘Observational’
6348 sampled individuals; 1930 completed baseline assessment.
612 visited website, 527 completed all 4 website-based assessments.
White participants more likely to complete all assessments than black (p=0.02).
Mean age 34.6 yrs, 263/526 white.
56.5% female, 63.9% in relationships. / Telephone assessments.
Web-based questionnaires.
Subsequent clinic attendance for blood test.
Multivariate analyses.
Primary outcome variable: ease of decision making.
Second outcome: attendance for blood test.
Primary predictor variable:
Number of websites viewed for each of 4 modules.
Mediating variables:
Perception of trustworthiness, satisfactoriness, helpfulness, clarity of information.
Demographic covariates:
Gender, age, education, race, marital status and family history of ‘multiplex’ health conditions.
Genetic self-efficacy, health information seeking, importance of genetic information. / Nearly half participants visiting website decided not to be tested.
Participants generally had positive perceptions of quality and usefulness of website information.
Viewing more of the information (no. of pages) was associated with easier decision making regarding having the test.
Engaged most with info re test, test procedures and what could be learned from results, less with health condition and genetic information. / 0.75
This sample is not representative.
Leighton et al (2011) / To investigate consumers’ perceptions and understanding of DTC test results. / Quantitative
n=145 (general public),
n=171 (genetic counsellors) (comparison group).
Members of the public recruited from Facebook (snowball sampling).
Genetic counsellors recruited via National Society of Genetic Counselors (US). / Online survey.
4 hypothetical results scenarios, based on actual wording taken from DTC websites.
Likert scales used to measure responses to these scenarios.
Differences between groups assessed using Mann-Whitney U test.
Chi square analyses to investigate association between public self-assessment of ability to interpret results and the accuracy of their interpretation. / Significant difference in the way the two groups interpreted results in 3 out of 4 scenarios.
The public more likely to consider that the test results would be helpful in managing future health care.
Although the majority of general public respondents interpreted the results correctly,
in 3 out of 4 scenarios, individuals did not have a higher probability of correctly interpreting the results if they thought the results were easy to understand. / 0.78
This sample is partly recruited from members of the public, so is more representative, but the sample number is low.
McBride et al (2009) / To evaluate what psychological and behavioural factors predict who is likely to seek SNP-based genetic tests for multiple common health conditions where feedback can be used to motivate primary prevention. / Quantitative
The multiplex genetic susceptibility test (MGST):
15 polymorphisms associated with increased risk for 8 common health conditions (type 2 DM, lung, colon and skin cancers, CHD, hypercholesterolaemia, hypertension and osteoporosis).
n=1959 (baseline survey); 612 subsequently visited website to consider testing.
Sampling frame: 350,000 commercially insured members of a health maintenance organisation.
Inclusion criteria: aged 25-40 yrs, enrolled for at least 2 years, assigned to primary care physician, and self-identified as being white or black. / Baseline survey, then if agreed to participate, sent brochure about web information site to consider genetic testing. Visit website to review modules (financial incentives); request testing, schedules for blood test; test feedback direct to subject by mail and phone FU.
FU phone survey 3 months after receiving results.
Dependent variables:
Accessing the website (Y/N),
Getting tested (Y/N).
Independent variables:
Gender, race, education;
Plus
1) beliefs about genetics as cause of disease
2) importance of learning about genetics
3) objective and subjective personal risk
4) self-rated competency in using the health system, including genetic competency
5) general health information seeking behaviour.