A systematic literature review of researchers and healthcare professionals attitudes towards the secondary use and sharing of health administrative and clinical trial data | Systematic Reviews | Full Text
This search identified 4019 articles, of which 241 underwent full-text review; 73 articles met the inclusion criteria for the larger review. five systematic reviews of the literature and one article that presented duplicate results were excluded; this left a total of 67 articles eligible for review. see figure 1 for the prism diagram describing study selection.
This systematic review of the literature was originally developed to identify attitudes toward the secondary use and sharing of clinical and health administrative trial data in breast cancer. however, as a paucity of material relating specifically to this group was identified, we present the multidisciplinary results of this search, highlighting results specific to breast cancer and cancer in general where possible. We believe that the material identified in this search is relevant and reflects broader attitudes towards data sharing within the scientific and medical communities and can be used to inform data sharing strategies in breast cancer.
Reading: Literature review on health care
eighteen [18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35] of the 67 articles addressed the perspectives of clinical and scientific investigators, data custodians, and ethics committees and were analyzed for this article (Table 2). the majority (n = 16) of the articles focused on the opinions of researchers and health professionals, [18,19,20,21,22, 24,25,26, 28,29 ,30,31 ,32,33,34,35], only one article focused on data custodians [27] and ethics committees [23] respectively. four articles [18, 19, 21, 35] included a discussion of the attitudes of both researchers and health professionals and patients; Only results related to investigators/clinicians are included in this analysis (Fig. 1).
design, location and disciplines of the studio
various study methodologies were used, including surveys (n = 11) [24,25,26,27, 29,30,31,32,33,34,35], interviews and group approach (n = 6) [18,19,20,21,22,23], and mixed methods (n = 1) [28]. studies were conducted in various countries and regions; a breakdown by country and study is available in table 3.
In addition to articles focusing on general health and science [18, 21, 22, 24,25,26, 29,30,31,32,33,34], two articles included both scientific and not scientists. disciplines [27, 28]. multiple sclerosis (ms) [19], mental health [35] and human immunodeficiency virus (hiv)/acquired immunodeficiency syndrome (aids)/tuberculosis (tb) [20] were each the topics of an article
study quality
Quality assessment results are provided in Table 2. Qualsyst [15] scores ranged from 0.7 to 1.0 (possible range 0.0 to 1.0). although none were blinded, most provided clear information on respondent selection, methods of data analysis, and justifiable study design and methodology.
themes
four key issues were identified, barriers, facilitators, access and ownership; 14 subthemes were identified. a graphical representation of the topics of the article is presented in fig. 2. two articles reflect the perspective of research ethics committees [23] and data custodians [27]; the concerns pointed out by these groups are similar to those highlighted by researchers and health professionals.
barriers and facilitators
reasons not to share
eleven articles identified barriers to data sharing [20, 22, 24, 25, 27, 29,30,31,32,33,34]. concerns cited by respondents included having their results taken by other researchers [24, 25], data being misinterpreted or attributed [24, 27, 31, 32], missed opportunities to maximize intellectual property [24, 25, 27] and loss of publication opportunities [24, 25] or funding [25]. the results of a qualitative study showed that respondents emphasized the competitive value of research data and its ability to advance an individual’s career [20] and the potential for competitive disadvantage with data sharing [22]. Systematic problems related to increased data sharing have been noted in several articles suggesting that the barriers are “deeply embedded in the practices and culture of the research process, as well as in the researchers themselves” [33] ( p. 1), and that scientific competition and a lack of incentives in academia to share data remain barriers to further sharing [30].
Lack of time, lack of funding, limited storage infrastructure, and lack of procedural standards were also noted as barriers [33]. the quantitative results demonstrated that the investigators did not have the right to make the data public or that the study sponsor did not require sharing [33]. maintaining the balance between the interests of the investigator and the funder and the protection of the research subjects [31] were also mentioned as barriers. privacy concerns were noted in four articles [25, 27, 29, 30]; one study indicated that clinical researchers were significantly more concerned about privacy issues compared to scientific researchers [25]. Results from a qualitative study indicated that physicians were more cautious than patients about including personal information in a disease-specific registry; the authors suggest that this may be the result of potential legal challenges in the context of a lack of explicit consent and consistent guidelines [19]. researchers, particularly clinical staff, indicated that they did not consider sharing data in a repository relevant to their work [29]
Trust was also identified as a barrier to further data sharing [32]. rathi et al. identified that investigators were likely to withhold data if they mistrusted the intent of the investigator requesting the information [32]. ethical, moral, and legal issues were other potential barriers cited [19, 22]. in a quantitative study, 74% of respondents (n = 317) indicated that ensuring proper use of data was a concern; other concerns included data that was not fit for the purpose requested [32]. data quality concerns were also cited as a barrier to data reuse; some respondents suggested a perceived negative association with data reuse among health scientists [30].
reasons to share
eleven articles [19,20,21,22, 24, 25, 29,30,31,32,33] discussed the reasons identified by researchers and health professionals for sharing health data; Generally speaking, the principle of data sharing was considered a desirable norm [25, 31]. Cited benefits included improvements in care delivery, communication and receipt of information, impacts on care and quality of life [19], contributing to the advancement of science [20, 24, 29], validating scientific results, reducing duplicating scientific effort and minimizing research costs [20], and promoting open science [31, 32]. Professional reasons for sharing data included academic benefit and recognition, networking and collaboration opportunities [20, 24, 29, 31] and contributing to the visibility of your research [24]. several articles noted the potential of data sharing to allow faster access to a larger group of patients [21] for research, better access to population data for longitudinal studies [22] and greater responsiveness to public health needs [20]. in one study, a small percentage of respondents indicated that there was no benefit to sharing their data [24].
See also: Garmin Dash Cam 57 review | PCWorld
analysis of the quantitative survey data indicated that the perceived usefulness of the data was most strongly associated with the intention to reuse [30]. lack of access to data generated by other researchers or institutions was considered a major impediment to the progress of science [33]. in a second study, quantitative data showed no significant differences in the reasons for sharing the academic productivity of clinical investigators, geographic location, funding source or size of the trial, or the journal in which the results were published [ 32]. attitudes toward sharing for academic benefits or recognition differed significantly by respondent’s geographic location; Western Europeans were more willing to share compared to respondents from the United States or Canada and the rest of the world [32].
feedback on sharing
Seven articles [19,20,21, 29, 31, 33, 34] discussed the views of researchers and healthcare professionals regarding data sharing, noting a wide range of opinions. two articles, both qualitative, discussed the role of national registries [21] and data repositories [31]. overall, there was clear support for and acceptance of national research registries [21], with some respondents believing that de-identified data sharing through data repositories should be required and, when requested, investigators should share data [31]. sharing de-identified data for reasons beyond academic and public health benefit was cited as a concern [20]. two quantitative studies reported a proportion of investigators who believed that data should not be available [33, 34]. the researchers also expressed differences in how shared data should be managed; some preferred the requirement that the data be ‘kept at the gate’, while others were happy to relinquish control of their data once selected or upon publication [20]. quantitative results indicated that scientists were significantly more likely to classify data reuse as highly relevant to their work than clinicians [29], but not all scientists shared data equally or held the same views on sharing or reusing of data [33]. some respondents argued that not all data was created equal and therefore should only be shared in certain circumstances. this was in direct contrast to other respondents who suggested that all data should be shared all the time [20].
differences by age, background, discipline, professional focus, and world region
Differences in attitudes toward shared data were observed by age, professional focus, and world region [25, 27, 33, 34]. younger researchers, aged 20-39 and 40-49 years, were less likely to share their data with others (39% and 38% respectively) compared to other age groups; respondents older than 50 years were more willing (46%) to share [33]. Interestingly, although less willing to share, younger researchers also believed that lack of access to data was a major impediment to science and their research [33]. where younger researchers were able to place conditions on access to their data, rates of willingness to share increased [33].
respondents from the disciplines of education, medicine/health sciences, and psychology were more likely than others to agree that their data should not be available for others to use in the first place [34]. however, the results of one study indicated that researchers in the fields of medicine and the social sciences were less likely to share compared to other disciplines [33]. for example, results from a quantitative study showed that, compared to biologists, who reported sharing 85% of their data, medical and social sciences reported sharing their data 65% and 58% of the time, respectively [33 ].
one of the main reasons for controlling access to data, identified in a study of data custodians, was due to the desire to prevent misuse of data; this was cited as a factor for all data repositories surveyed except those of an interdisciplinary nature [27]. limiting access to certain types of research and ensuring that attribution was not listed as a concern for sociology, humanities, or interdisciplinary data collections [27]. issues related to privacy and sensitive data were only mentioned as concerns for data collection related to the humanities, social sciences and biology, ecology and chemistry; Concerns regarding intellectual property were also noted [27]. the disciplines of biology, ecology, chemistry, and social sciences had the most policy restrictions on the use of data stored in their repositories [27].
Differences in data sharing practices were also observed by region of the world. respondents from neither North American nor European countries were more willing to place their data in a central repository; however, they were also more likely to place conditions on the reuse of their data [33, 34].
data sharing experience
the experience of sharing data between researchers was analyzed in nine articles [20, 24,25,26, 28,29,30,31,32,33]. Data sharing arrangements were highly individual and ranged from ad hoc and informal processes to formal procedures imposed by institutional policies in the form of contractual agreements, with respondents indicating data sharing behavior ranging from no data sharing to sharing all data. data [20, 26, 31] . quantitative data from one study showed that researchers were more inclined to share data before publication with people they knew compared to those they did not know; after publication, these figures were similar between groups [24]. while many researchers were prepared to share data, the results of one survey identified a preference among researchers to collect data themselves, followed by their team or close colleagues [26].
differences were observed in the reported rate of data sharing compared to the actual rate of sharing [25]. in a large quantitative study (n = 1329), almost a third of respondents chose not to respond if they made their data available to others; of those who responded to the question, 46% reported that they do not make their data electronically available to others [33]. By discipline, differences in the rate of refusal to share were greater in chemistry compared to non-scientific disciplines such as sociology [25]. Respondents who were most academically productive (>25 articles in the last 3 years) reported that they have or would withhold data to protect research subjects less frequently than those who were less academically productive or received industry funding [32]
Attitudes towards sharing de-identified data through data repositories were analyzed in two articles [29, 31]. the majority of respondents in one study indicated that de-identified data should be shared through a repository and that it should be shared on request. Lack of experience in uploading data to repositories was noted as a barrier [29]. when data was shared, most investigators included additional materials to support their data, including materials such as metadata or a description of the protocol [29].
Two articles [28, 30] focused on the processes and variables associated with the exchange. factors such as norms, data infrastructure/organizational support, and research communities were identified as important factors in a researcher’s attitude toward data sharing [28, 30]. a moderate correlation between data reuse and data sharing suggests that these two variables are not linked. furthermore, data sharing compared with self-reported data reuse was also only moderately associated (Pearson correlation 0.25 (p ≤ 0.001)) [26].
data exchange predictors and norms
Two articles [26, 30] discussed the role of social norms and an individual’s willingness to share health data. the perceived efficacy and efficiency of data reuse were strong predictors of data sharing [26] and the development of a ‘positive social norm towards data sharing support(s) [ed] researcher data reuse intention ‘ [30] (p. 400).
policy framework
establishing clear policies and procedures to support data sharing was highlighted in two articles [22, 28]. the presence of ambiguous data sharing policies was noted as a major limitation, particularly in primary care and the increased adoption of health informatics systems [22]. Policies that support an efficient sharing system that allows for the maximum amount of data sharing are preferred and can include incentives such as formal recognition and financial reimbursement; a framework for this is proposed in fecher et al. [28].
research funding
See also: Waterpik Complete Care 9.0 review | Live Science
The requirement to share publicly funded data was discussed in an article [25]. some cases of researchers refusing to share taxpayer-funded data were reported; reasons for refusal included a possible reduction in future funding or publication opportunities [25].
access and ownership
articles related to access and ownership were grouped together and seven sub-themes were identified.
access, information systems and metadata
ten articles [19,20,21,22, 26, 27, 29, 33,34,35] discussed the issues of access, information systems and the use of metadata. it was considered important to prospectively ensure privacy protection for data held in registries [19]. In the mental health setting, researchers suggested that patients should have more options to control access to shared record data [35]. some respondents considered important the use of guardianship committees [19] or guardians [20] to guarantee the security and access to the data kept in the registries; however, many have suggested that a researcher should relinquish control of data collection once selected or published, unless embargoed [20]. Reasons for maintaining control over registry data included ensuring attribution, restricting commercial research, protecting confidential (non-personal) information, and limiting certain types of research [27]. It was noted that security and confidentiality concerns were important and needed to be ensured in this regard; it is also necessary to include accountability and transparency mechanisms [21]. many respondents believed that access to registry data by pharmaceutical companies and marketing agencies was not considered adequate [19].
respondents to a medical and social sciences survey were less likely to agree to including all data in a central repository without restrictions [33]; in particular, this was also reflected in qualitative research results indicating that healthcare professionals were more cautious than patients about including personal data in a disease-specific registry [19].
Although many researchers stated that they commonly shared data directly with other researchers, most had no experience with uploading data to repositories [29]. the results of a survey indicated that younger respondents have more restricted access to data and thought that their data is much easier to access than older respondents [34]. In the primary care setting, concerns were noted that clinicians could block patients from participating in a registry by denying access to a patient’s personal data or by not giving permission for the data to be extracted from its clinical system [21]. there was also resistance in primary care towards health data amalgamation done for an unspecified purpose [22]; Respondents were not in favor of systems that included unwanted functionality (don’t want/need), inadequate attributes (capacity and responsiveness) of practice, or unwanted impact on the GP role (autonomy, status, control and workflow) [22].
access to ‘full metadata (needed) to support correct interpretation of data’ [26] (p. 4) at a later stage. when additional materials were shared, most investigators shared contextualizing information or a description of the experimental protocol [29]. the use of metadata standards was not universal and some respondents used their own [33].
healing
several articles highlighted the impact of data curation on researchers’ time [20,21,22, 29, 33] or finances [24, 28, 29, 33, 34]; these were seen as potential barriers to further adoption of registries [21]. Required tasks for curation included preparing data for dissemination in a usable format and uploading data to repositories. The importance of ensuring that data is accurately preserved for future reuse was highlighted; it must be presented in a reviewable and auditable manner [20]. the amount of time required to cure the data ranged from “no additional time” to “more than ten hours” [29]. in one study, no clinical respondent had their data in a shareable format [29]. In the primary care setting, health information systems that promote sharing were not considered beneficial if they required standardization of processes and/or sharing of clinical notes [22]. in addition, spending time on non-medical issues in a time-constrained environment [22] was identified as a barrier. six articles described the provision of funds or technical support to ensure data storage and maintenance and the ability to provide access to data upon request. all pointed to lack of funds and time as a barrier to sharing more data [20, 24, 28, 29, 33, 34].
consent
Qualitative research results indicated a variety of views regarding consent mechanisms for future data use [18,19,20, 23, 35]. consenting to future research can be complex as the exact nature of the study will be unknown and therefore some respondents suggested including a broad statement about future uses of the data [19, 20] during the consent process . conversely, other participants indicated that current consent processes were too broad and did not sufficiently reflect patient preferences [35]. the importance of respecting original consent in all future research was noted [20]. it was suggested that seeking additional consent for future use of the data may discourage participation in the original study [20]. differences in views were noted regarding the provision of detailed information on data sharing at the individual level, suggesting that researchers wanted to exercise some control over the data they had collected [20]. an opt-out consent process was considered appropriate in some situations [18] but not all; some respondents suggested that consent was not required to use a patient’s medical records [18]. there was support from some investigators for giving patients the option to “enroll” at different levels of participation in a registry setting [19]. it was considered important to give patients more granular options when controlling access to their medical data [35].
The attitudes of ethics and review boards (n = 30) toward the use of medical records for research were discussed in an article [23]. while 38% indicated that no further consent would be required, 47% required consent from the participants, and 10% said that the consent requirement would depend on how potentially identifying variables would be managed [23]. access by external investigators to medical record data was associated with a consent requirement [23].
acknowledgment
The importance of establishing mechanisms that recognize the use of shared data was discussed in four articles [27, 29, 33, 34]. a significant proportion of respondents believed that it was fair to use other investigators’ data if they acknowledged the author and funding agency in all disseminated work or as a formal citation in published work [33]. Other mechanisms to recognize the creator of the data included opportunities to collaborate on the project, reciprocal data sharing agreements, which allow the creator to review or comment on the results, but not approve derivative works, or the provision of a list of products that do data use and co-authorship [33, 34]. in the context of controlled data collections, survey results indicated that ensuring attribution was a motivator for controlled access [27]. more than half of the respondents in one survey believed that it was fair to release the results in whole or in part without the approval of the data provider [33]. no significant differences in recognition mechanisms were observed between clinical and scientific participants; mechanisms included co-authorship, acknowledgment in the acknowledgment section of publications, and citation in the bibliography [29]. no consent method to recognize shared data reuse was identified [29].
property
Data ownership was identified as a potential barrier to further data sharing in academic research [28]. within the framework of controlling data collections, respondents indicated that they wanted to maintain some control over the data set, suggesting that researchers have a perception of ownership of their research data [28]. examples of researchers extending ownership over their data include the right to publish first and access control to data sets [28]. dater et al. noted that the idea of researcher ownership of data is not always a legally supported position; it includes “ownership and usage rights, privacy, contractual consent, and copyright” [28] (p. 15). rather, data sharing is restricted by privacy law, which applies to data sets that contain personal data. legal uncertainty about data ownership and complexity of the law may deter data sharing [28].
promotion/professional criteria
The role of data sharing and its relationship to promotion and professional criteria were discussed in two articles [24, 28]. the requirement to share data is rarely a professional or promotional criteria, instead the systems are based on grants and publication history [24, 28]. one study noted that while the traditional link between publication history and promotion remains, it is ‘likely that funders will continue to see suboptimal returns on their investments, and that data will continue to be used and disseminated inefficiently’ [ 24] (p. 49).
