1 INTRODUCTION

The sharing of patient health information or biospecimens across academic institutions, government, and industry can improve patient outcomes and accelerate breakthroughs in medical research.¹ The adoption of electronic medical records (EMRs) across most medical centers has steadily increased the number of hospitals electronically exchanging patient health information for clinical purposes, with over 75% of engaged physicians reporting they “experienced improvements in quality of care, practice efficiency, and patient safety.”²

Legally, entities are allowed to share patient data and biospecimens for research purposes without express consent under certain conditions, such as if they are de-identified or the research poses low risk to individuals.³ Because of the potential benefit of such research to advancing science and medicine and the legality of such an approach, many institutions share patient data and biospecimens broadly.⁴ Ethics scholars have argued that patients may have a moral obligation of reciprocity to participate in low-risk research if they have benefited from receiving care improved by the research themselves,^{5, 6} although this is hotly contested as it applies to biomedical research, generally.^{7, 8} While not legally required, some hospitals share basic information about research use of patient data and biospecimens in their clinical consent forms.⁹

That said, patients generally do not understand whether or how their clinical health information or biospecimens are used for research, and, if they do, many report discomfort with how the current system is structured.¹⁰ Patients report particular discomfort with data or biospecimen commercialization by private entities^{11, 12} and have expressed concerns about who profits from their data, even when de-identified, and how those profits are used.^13-15 The majority of patients want to be notified about health information and biospecimen commercialization, but very few are comfortable with such use.^{16, 17} However, some patients view the sale of health data as acceptable if profits are put back into public health services.¹² Tiered consent approaches have been explored for different types of research,^{18, 19} but implementation could be challenging for hospital systems. Hospitals have counter-argued that setting up a consent process for all secondary research use of data and biospecimens would be both costly and expensive and would likely result in less data being available for research.²⁰ Therefore, many hospitals continue such sharing, even with the knowledge that it makes some patients uncomfortable.

But, even if specific consent structures for secondary research use of health data and biospecimens are untenable, that does not mean that entities lack the moral obligation to at least notify patients about such use.²¹ Yet the best methods of going about such notification, and having that notification enable patients to better understand the system, have yet to be fully elucidated. This information is critical for establishing trust between patients and the medical system, a key priority identified by the National Academy of Medicine.²² Identified methods of increasing transparency and understanding for patients include detailed information on clinical informed consent forms, poster or visual campaigns, or further information on hospital websites.

The goal of this mixed methods project was to assess the impact of these communication and marketing materials on patient understanding of health data and biospecimen uses—as well as their attitudes and perceptions regarding sharing their health data and biospecimens for research—embedded within the Michigan Medicine (MM) system at the University of Michigan (U-M).²³ Developing our understanding of patient data-sharing practices and integrating patient preferences into health system policy, through this work and continued exploration, contributes to building infrastructure that can be used to support the development of a learning health system across hospital systems nationally.

2 METHODS

This study employed a convergent mixed methods approach,²⁴ merging qualitative data from Zoom focus groups and quantitative data from an online survey of MM patients. Survey respondents and focus group participants were recruited separately from the U-M Health Research (UMHR) website,²⁵ a platform that helps connect researchers with interested volunteers for U-M research studies. Participants for this study had to be 18 years of age or older, fluent in English, and must have seen a U-M provider sometime in the past. This study was deemed exempt by U-M IRBMED (HUM00236265 and HUM00235071).

3 RESULTS

3.2 Survey outcomes

3.2.1 Baseline

At baseline (before seeing either intervention), 60.6% (n = 291) of respondents agreed that they know how MM uses health information and biospecimens for research; 25.6% (n = 123) disagreed. In a “Select all that apply” question, half of respondents reported that they learned about it from the MM informed consent form (50%, n = 240) and one-third from a healthcare provider or researcher (34.4%, n = 165). A minority reported learning about it from an MM website or other advertising (17.1%, n = 82), or from the news (1.3%, n = 6). In addition, 75.2% (n = 361) reported they carefully read clinical forms, and those who read forms were also more likely to report knowledge (Spearman's rho = 0.193, p < 0.001). Those who reported that they were a fan of U-M sports were more likely to report trust in MM (Spearman's rho = 0.261, p < 0.0001).

The majority agreed that MM is transparent about how information and biospecimens are used for research (59.0%, n = 283). While a majority also agreed that patients should share their health info and biospecimens for health research that is low risk to them (72.3%, n = 347), almost all agreed that hospitals should tell patients how they are used (97.9%, n = 470). Respondents who reported that MM's research improves patient care were also more likely to report that patients should share their data and biospecimens (Spearman's rho = 0.197, p < 0.0001).

While the majority of respondents reported they were not concerned about their health privacy (64.4%, n = 309), nor have done anything in response to such concerns (67.7%, n = 325), a minority did report having privacy concerns (17.7%, n = 85) and having taken some action. Reported actions (presented as “select all”) included: kept some health information from their clinician (16.7%, n = 80), looked for additional information about privacy (11%, n = 53), chose a different healthcare provider (11%, n = 52), avoided seeking healthcare (8.3%, n = 40), or asked their clinician not to record or share private information (6%, n = 29).

3.2.2 Poster or media intervention

Viewing either the informational poster or the news article resulted in statistically significant decreases in trust (mean difference ± sd: −0.077 ± 0.663, p = 0.011, Cohen's d = 0.12) and comfort (−0.065 ± 0.591,p = 0.017, d = 0.11), and an increase in concern for privacy (0.163 ± 1.07, p < 0.001, d = 0.15) and perception of transparency (0.096 ± 0.8, p = 0.009, d = 0.12), but all effect sizes were small. The type of intervention (poster vs news) made no statistically significant difference in its influence on any characteristic (privacy, transparency, comfort, respect, or trust). However, 94.6% (n = 452/478) of respondents said they would rather have MM tell them about patient data and biospecimen research sharing as opposed to hearing it from another source (like on the news). U-M sports fans were more likely to report continued trust in MM both after seeing the poster (Spearman's rho = 0.205, p < 0.0001) or the news (Spearman's rho = 0.225, p < 0.0001).

3.2.3 FAQ webpage

Half of respondents who viewed the poster reported that they would visit the FAQ webpage, advertised via QR code and a link on the poster itself (50.2%, n = 120/239); respondents who reported that they carefully read forms at baseline were more likely to do so (r = 0.231, p < 0.0001). Across both groups, a large majority of respondents wanted to know the answers to the following, select all, questions: “Who is given access?” (72.3%, n = 347), “How is my privacy protected?” (67.1%, n = 322), and “Does Michigan Medicine sell my health information?” (65.2%, n = 313). Fewer respondents wanted to know answers to contextual questions such as: “Why are health information and biospecimens important for research?” (27.7%, n = 133) and “Is Michigan Medicine the only hospital that uses patient health information and biospecimens for research?” (27.5%, n = 132) (Figure 1). After reviewing the webpage, respondents were significantly more likely to report knowledge about MM use of data and biospecimens for research (p < 0.0001).

4 DISCUSSION

In this mixed methods project, we examined patient understanding, attitudes, and perceptions regarding sharing their health data and biospecimens for research through exposure to an informational poster, a FAQ webpage, and a revised consent form excerpt (focus group only). Overall, both our survey and focus group participants reported a high level of trust and comfort with MM. That trust might not always be specific to the health system—as indicated by our previous qualitative research,²⁹ we found that being a fan of U-M sports was associated with a higher trust in the medical center, both before and after the interventions.

Consistent with previous ethical arguments for reciprocity in the learning health system,⁵ many survey respondents agreed that patients should contribute to low-risk research, especially those that believed that such research would translate into improved care. But almost everyone both in the survey and the focus groups reported that hospitals should tell patients this – and that they would certainly rather hear it from the hospital than from another source, like the news. Previous studies have found contrasting views on this point with some finding that a notification-only approach would be unacceptable to a majority,³⁰ while others finding that notification is acceptable where written consent is not feasible.³¹ These types of findings have also been associated with self-identified race and ethnicity.³² There was also a low level of reported baseline concern for privacy, and the majority reported that they would not change behavior if they were concerned. However, a substantial minority reported potential behaviors, like keeping relevant information from their clinicians, that could have a significant impact on clinical care.³³

Focusing on the clinical consent form, the majority of survey respondents subjectively reported that they knew how MM uses patient data and biospecimens for research, and that half learned it from the consent directly. But our focus group participants, who read the revised consent form excerpt as part of this research, emphasized the consent document's low readability and comprehensibility. It is hard to harmonize these two findings together. One possibility is that, because our focus group participants had lower educational attainment on average than our survey respondents, it was in fact true that our survey respondents were more likely to learn relevant information from the informed consent. Another possibility, because the survey respondents did not see the informed consent form and the focus group participants did, that the survey respondents suffered from recall bias. Or, perhaps the additional information that we added to the focus group's informed consent excerpt—intended to increase understanding and disclosure—was more likely to result in confusion. All these hypotheses warrant further review and research.

In addition, focus group participants highlighted the inherent coerciveness of requiring patients to agree to research with their data and biospecimens to receive clinical care. Our research findings continue to underscore that while patients might be conceptually supportive of contributing to minimal risk research with their clinical data and biospecimens, they remain adamant about wanting the opportunity to know and to opt-out.

Viewing the informational poster or the news article resulted in statistically significant decreases in trust and comfort, an increase in concern for privacy, and an increase in perception of transparency. That said, all findings had a small effect size. For example, while the mean of responses for “I trust Michigan Medicine” (on a 5-point Likert with “strongly agree” being 1) rose by 0.07, which was statistically significant, the average response went from a mean of 1.66 (between “strongly agree” (1) to “agree” (2)) to 1.74 (remaining between “strongly agree” and “agree”). None of the other characteristics (comfort, privacy, or transparency) with significant differences made a categorical difference either—suggesting little to no clinical impact of the results. Thus, while an argument against increased transparency about research use of data and biospecimens has been that it would impact patient care or the reputation of the hospital among patients, those concerns were not reflected in our data in a policy-relevant manner.

And, while a significant majority of survey respondents reported that they would rather hear about research use of their data and biospecimens from MM as opposed to another source, we did not find any significant difference in reported characteristics across interventions (e.g., respondents were no more likely to report decreased comfort if they learned about research use from the news versus the institution itself). This is likely because of the homogenous nature of our survey respondents, a high baseline rate of comfort and trust, and the neutral framing of the hypothetical news article. Often, when hospital data or biospecimen sharing policies are covered by media it is framed as a scandal, like the expose on Google's “Project Nightengale,” in which it collected millions of fully identified medical records from hospitals associated with Ascension.³⁴ While we intentionally chose to keep our hypothetical news article neutral—to focus our analysis on the comparison of modes of communication as opposed to content—it is highly likely that a negative news article would have had a greater impact.

Participants in the focus groups were also confused by the difference between transparency regarding how MM is already using their data and biospecimens for research and optional research recruitment (which also is often advertised by poster throughout the MM hospital). Their responses also underscored the inherent tension between designing an informational poster that will attract patient attention (i.e. with large photos of patient faces and minimal text) and wanting to immediately understand more detailed information (which would have required the inverse). Given this feedback, an image-centric poster that can immediately link to additional information on a website, might remain the right balance.

Promisingly, we found that reviewing the FAQ webpage was associated with a significant increase in subjective knowledge regarding MM research use of patient data and biospecimens in our survey respondents. Respondents were also most likely to report interest in questions regarding access to data and biospecimens, whether access is sold, and how patient privacy is protected. They reported less interest in questions that might provide helpful context into why the institution makes certain choices, such as why health data and biospecimens benefit research and whether MM is the only hospital that uses patient data in this way. These considerations seem to emphasize the importance of including contextual information on a permanent platform, such as the website's landing page, to ensure that it is continually available to users and does not require an additional “click” to reveal. Additional information regarding advances research has achieved would also likely add to such context and reinforce respondents' already high levels of reported altruism.

Focus group participants also reported both appreciating succinct answers to questions but also wanting to know all relevant information beneath whichever “accordion”-style question (i.e., the answer only appeared if the user selected the question) they selected—which might be improved by additional linking between answers. Participants also described reassurance regarding IRB oversight and the potential benefit of additional information and reassurances about review processes.

This study had limitations. First, the survey cohort was largely well-educated White women. In addition, when studying patient perspectives on enrolling in research, one inherently must rely on a cohort that has consented to research. These two factors combined suggest our research sample consisted of people who already have a high baseline trust of MM and are supportive of research overall. That said, we were able to purposively sample our focus group participants for more diverse input over gender, race, ethnicity, and education—and the findings from the focus group supported many of those from the survey. Second, while past studies have found differences in perspectives on enrolling in secondary research across race and ethnicity,^{32, 33, 35} we did not observe the same pattern of results, likely due to the homogenous nature of our cohort. Third, the poster and news article differed in imagery and language, which might have impacted results.

5 CONCLUSION

In this mixed methods project with survey respondents and focus group participants, we assessed the impact of a new campaign at a major academic medical center to promote transparency regarding how patient data and biospecimens are used for research. We found that while most survey respondents agreed patients should share data and biospecimens for minimal-risk research, almost all reported that hospitals should tell patients how they are used. After reviewing an informational poster or hypothetical news article disclosing how the hospital uses patient data and biospecimens, survey respondents reported no policy-relevant change in concerns for privacy, transparency, comfort, respect, or trust. But the vast majority reported they would rather learn this information from the hospital than the news. Our focus group participants also expressed largely favorable or ambivalent views after reviewing the same material and discussed related values—such as altruism and reciprocity—in depth. These findings can help inform hospitals acting as learning health systems to use validated methods to increase transparency and predict patient response.

CONFLICT OF INTEREST STATEMENT

The authors declare no relevant conflict of interest.