Australia’s genomics professionals were generally favourable towards patient data ownership
A total of 184 respondents completed the survey, though only 117 respondents across the four groups of professionals completed the questions relevant for this analysis. Among them were 50 genetic counsellors, 30 researchers, 20 bioinformaticians, and 17 clinical geneticists. The respondents were distributed across six states and two territories in Australia. Most respondents were from New South Wales (43), followed by Victoria (29), Queensland (20), Western Australia (11), South Australia (9), Australian Capital Territory (3), Northern Territory (1), and Tasmania (1).
Respondents rated their perceptions of patient data ownership on a scale from 1 to 8. We merged “neither agree or disagree” and “don’t know” to create a new scale item of 4 = “neutrals & don’t know” to maximise the sample size for the analysis and for calculating means. We investigated whether the perceptions on each item of patient data ownership varied among the four professional groups. The results showed most respondents (>50%) across all professional groups had a favourable perception towards patient data ownership (Fig. 1). Over 90% of bioinformaticians and researchers agreed on all the three items. However, the responses of clinical geneticists and genetic counsellors were more varied across the three items, especially on patients’ rights to own data. Nearly one-third of clinical geneticists responded that patients should not have rights to own their data, and more than a quarter of genetic counsellors (26%) remained neutral on this question. These results indicated that respondents’ professional roles, experience, and responsibilities may influence how they view the benefits and drawbacks of patient data ownership.
Percentages on the right and left represent the percentages of respondents agreeing and disagreeing to the three items of patient data ownership respectively.
Support for patient data ownership decreases among professionals with higher degrees of direct patient interaction
The means of perceptions towards patient data ownership across the three scale items were computed. The mean of the three items of patient data ownership created an internally inconsistent single measure of perceptions towards patient data ownership (alpha = 0.79). Then we identified any outliers present in the data that may influence the analysis [44]. After removing three outliers (one researcher, one bioinformatician, and one genetic counsellor) to meet normal distribution assumptions for hypothesis testing, a total of 114 observations were used for this analysis. The mean values across the four professionals are shown in Fig. 2 (Also in Supplementary Table 2). Since 4 was the scale midpoint (neutrals & don’t know), any mean values over 4 represented agreement and values below 4 disagreements, on average. All professional groups were supportive of patient data ownership, as the means were above 4. The mean values were the lowest among clinical geneticists followed by genetic counsellors and highest among bioinformaticians and researchers. This finding further indicated a potential influence of professional roles, experience, and responsibilities on their perceptions.
The bars show the means and their standard errors for perceived patient data ownership. The results of Tukey’s post hoc tests are shown above the mean values, with the differences in means and the p values in the parentheses. For example, the difference in means between clinical geneticists and genetic counsellors is 0.39 and a p value of 0.32. The red horizontal lines show statistically significant results, while the black lines were not significant.
To test whether the perceptions of professional groups towards patient data ownership varied according to their professional roles and responsibilities, we performed a hypothesis testing:
The results of the ANOVA test confirmed that the perceptions of patient data ownership were statistically different among the professional groups at the 95% confidence level (p = 0.000014) (Supplementary Table 3); hence, HA was accepted. Further Tukey’s significance tests were used for post hoc comparison of means. The difference between means and their p values (significance level) are shown in Fig. 2; the Tukey test results are provided in the Supplementary Information (Supplementary Table 4). The results show that the difference in perceptions towards patient data ownership between clinical geneticists and genetic counsellors was not statistically significant (p = 0.32). A similar result was found between bioinformaticians and researchers (p = 0.67). Whereas statistically significant differences in perceptions towards patient data ownership were found between (1) clinical geneticists and bioinformaticians (p = 0.0001) (2), clinical geneticists and researchers (p = 0.0001) (3), genetic counsellors and bioinformaticians (p = 0.002), and (4) genetic counsellors and researchers (p = 0.02). This finding suggests that professionals with no direct interface with patients (i.e., working “behind-the-scenes”) have more favourable perceptions towards patient data ownership than “patient-facing” professionals.
Thematic analysis
Exploring the nature of disagreement with patient data ownership
Fifteen respondents expressed disagreements towards patients having rights to either own or control or decide who uses their data, of which, 14 respondents provided written comments about the reasons for their disagreement. We identified one major theme that explains the reasons for their disagreement, which was “control but not own”.
The theme of “control but not own” captured the views that patients have rights to control who uses their data, but this does not automatically translate to the ownership of that data. This was expressed by one genetic counsellor as follows: “I don’t think they [patients] need to own the data, just own the right to make decisions on what is done with it” (G14). These respondents also identified three challenges they saw being related to patients “owning” their data: “copyright”; “collective ownership”; and “potential misuse”.
According to the Copyright Act 1968 of Australia, the copyright of genomic data rests with the compiler that has invested resources (human and financial) in collecting, validating, and storing the data. In a similar vein, one genetic counsellor stressed, “data is generated by the lab and therefore is theirs to own/manage, with responsible, pre-agreed terms that patients consent to at [the] time of test” (G1).
Respondents also reflected the perception that a patient’s genomic data is subject to the collective ownership of the patient’s family. One researcher described why it would be wrong for one person to hold ownership of their genomic data, “I believe my genome is not mine but my parents’ and their parents’ so looking at my genome as my own is not correct. I am just a custodian of the DNA passed onto me” (R12).
Additionally, respondents expressed concerns about the potential misuse of genomic data arising from patient ownership that would not occur if patients were simply able to control and decide who uses their data through existing consent processes. One clinical geneticist stated, “This is an issue of protecting patients from agents who might misuse their data. I am concerned that if individuals are given their genome data, they will hand it on to organisations that will misuse it without their [patients’] knowledge” (C5). This implies that respondents who disagreed that patient should own their genomic data may also believe that storing the data in a centrally located server is safer for patients than providing that data to patients directly.
Enablers and barriers in implementing patient data ownership
Four themes were identified in how respondents perceived of enablers and barriers to implementing patient data ownership in Australia. They were “patient readiness”, “genomics workforce”, “state of health system infrastructure”, and “coordinated policy environment”.
The theme of “patient readiness” refers to the state of patients’ knowledge and awareness about their own genomic data and the practice of clinical genomics, in general. Respondents perceived that patients’ current understanding of their genomic data could be a barrier to rollout mechanisms for them to responsibly and safely owning their data. Specific concerns about the utility of genomic data and its interpretation by patients were raised. A genetic counsellor and a bioinformatician feared that patients currently do not have “the ability to interact with the data in a useful way” (G15), which could lead to “a high degree of misunderstanding and misinformation unless handled carefully” (B6). As a result, respondents highlighted the need for education pathways to enable patient data ownership for both patients and professionals (as explored in the next theme). One researcher wrote, “there needs to be a pathway for educating people on safe ways to manage their genomic data” (R35). Additionally, clinical geneticists and genetic counsellors underscored the implementation of patient data ownership should be coupled with professional support.
In the “genomics workforce” theme, respondents explained that because clinical genomics is a relatively new field, there were not sufficient health professionals with genomics skills and knowledge to support the implementation of patient data ownership. As this should be enabled through appropriate professional education and support, the required genomic workforce should be expanded, particularly around “patient interface and more genetic counselling” (G40). Respondents cautioned of potential barriers in creating such a genomic workforce because this “will take several years of education to improve the general level of genetic and genomic literacy across non-genetic health professionals” (G23), which will require a long-term strategic investment.
The theme of “state of health system infrastructure” includes enablers and barriers relating to the technology and infrastructure required to support the implementation of patient data ownership. Bioinformaticians and researchers emphasised that the required technology to support patients owning their data is currently available, as one bioinformatician stated, “all of the technology and data management tools exist for this to happen” (B11). Researchers agreed that systems were in place to support this in research settings and identified the need to pilot these systems in clinical settings before rolling them out for mass uptake. However, for wider implementation of patient data ownership, respondents pointed to a key barrier of insufficient data storage and infrastructure. One genetic counsellor expressed, “we don’t have enough storage capacity to store such large data at the moment for people to ‘own’ their data” (G32). Similarly, a researcher commented that “Australia needs the right infrastructure to help people store and manage their data” (R9). Another barrier raised related to the challenges associated with embedding a user interface in the data storage system. This was expressed by a genetic counsellor, “the major barrier is online health system infrastructure is nowhere near ready to allow dynamic, interactive engagement to people’s genomic data” (G22).
In the theme of “coordinated policy environment”, respondents identified the lack of relevant legislation to support patient data ownership. One genetic counsellor commented that a barrier is “the time needed for developing appropriate protocols and legislation [and] implementing those at the federal and state levels” (G43). According to one researcher, there is “no legislation that guides data sovereignty” (R21). Most importantly, the ambiguity around patient ownership discussed earlier needs to be tackled, and as one researcher stated, “the legal and regulatory implications of data ownership as intellectual property of the individual vs. the lab generating the data needs to be resolved” (R3). The laboratory ownership model for genomic data was also discussed by respondents who expressed disagreement towards patients owning their data (under “copyright” concerns in the previous section).
