Introduction

Biobanks and biospecimens are critical components of much medical and translational research.1 Numerous biobanking efforts are under way around the world,2,3 but informed consent for the collection and storage of materials for future research use remains a challenge. This challenge will be magnified if proposed changes to federal regulations governing human subjects research move forward, as these would require consent for research use of biospecimens, regardless of whether they are initially collected for clinical or research purposes.4

The goal of informed consent is to enable competent individuals to decide for themselves whether to participate in research. However, obtaining consent is not simply a matter of disclosing information; promoting comprehension is integral to ensuring that the individual’s decision is actually “informed.” Valid consent can be critical for protecting the rights and welfare of individuals whose biospecimens are used in research—recent headlines provide a stark reminder of what can otherwise happen.5,6,7,8

For research involving biobanking, government regulations and best practice guidelines identify key categories of information that should be communicated to prospective participants during the consent process.9,10 But little attention has been paid to identifying which of this information prospective participants must grasp to demonstrate adequate understanding. Although participants would perhaps ideally understand every detail of the information deemed important for inclusion in a consent form, such comprehensive knowledge is not necessary to give valid consent.11,12

As long as the threshold for adequate comprehension remains unspecified, researchers may fail to identify participants whose consent is not valid or, conversely, deny participation to individuals whose understanding is, in fact, adequate. Significantly, uncertainty as to what constitutes adequate comprehension also undermines efforts to evaluate the sufficiency of consent forms and processes and the development and assessment of interventions to improve them. To address this gap, we conducted a Delphi process with a diverse panel of experts to define “adequate comprehension” in the context of informed consent for biobanking.

Materials and Methods

The Delphi method is an iterative process involving a series of surveys (or “rounds”) to gather opinions and seek consensus from a panel of experts.13,14,15,16,17,18 The results of one survey are used to construct the next, and panelists are asked to consider their subsequent responses in light of the results of previous rounds. The absence of an obligation to meet in person lessens constraints on group size and composition. Panelists can be allowed substantial time to consider evidence and reflect on their answers, and the anonymous nature of the process ensures that a single influential panelist does not have a disproportionate impact.

Selection of experts

Our main consideration in selecting the size and composition of our panel was to ensure the results would be regarded by others as reflecting the consensus of experts representing a range of relevant perspectives. To that end, we recruited panelists from existing committees and consortia that play a leading role in biobanking and human research protections in the United States. We identified and invited panelists based on recognized expertise in (i) biobanking operations/research; (ii) ethics/policy in human subjects research; and/or (iii) community/participant perspectives. The latter included members of community advisory boards for major US biobanks, as well as scholars with extensive experience studying community and participant attitudes and opinions. We confirmed prospective panelists’ expertise through an “about you” questionnaire. Those who rated themselves less than “somewhat familiar” with informed consent for biobanking were deemed ineligible. We compensated panelists $200 per round; those who did not complete a round were not included in subsequent rounds.

Procedures

We conducted three online surveys to define “adequate comprehension” for biobanking consent. The surveys were structured around model consent language (Supplementary Appendix S1 online) adapted from established sources19,20,21 and purposely contained ample detail to accommodate a range of opinions about comprehension.

Across all surveys, the panelists’ task was to review the model language and, for each of the 16 sections, indicate the minimally correct answer if a prospective participant were asked to explain that topic in his own words. Specifically, we asked panelists to identify the answer that reflects the minimum understanding necessary to provide valid consent. We asked them to imagine that, during the consent process, the study coordinator would compare the participant’s answer with the panelist’s answer. If the participant’s answer was missing key elements of information, the coordinator would review that topic with him in an effort to improve understanding. If the person was still unable to provide the minimally correct answer, he would not be allowed to participate.

A detailed description of the content and analysis of each survey is provided in Supplementary Appendix S2 online. In summary, our approach moved from more qualitative elicitations in early rounds to more quantitative elicitations in later rounds. Results from each were provided to panelists and used as the basis for subsequent surveys. The round 3 survey was explicitly designed to assess consensus quantitatively. Panelists were shown the model consent language, together with one statement per section of what prospective participants would need to say if asked to explain that topic. We constructed this statement to reflect the greatest degree of agreement developed through the prior rounds. Panelists were asked to rate on a four-point scale how much they agreed or disagreed that the statement generally represented adequate comprehension. There are no established criteria for defining consensus in a Delphi process; suggestions in the literature range from 51 to 80%.17 We selected 70% agreement as our predetermined threshold.

The Duke University Health System Institutional Review Board deemed this study exempt (45 CFR 46.101(b)(2)). The surveys were implemented online between February and October 2013 using Qualtrics software (Provo, UT).

Results

Panelist characteristics

We invited 61 individuals to participate on the panel. Of these, three were determined ineligible, five did not respond, one was unable to participate, and one declined. Those enrolled (n = 51) represented a wide range of expertise and considerable familiarity with biobanking consent ( Table 1 ). We maintained a high participation rate throughout the study: 47 (92% of enrolled) completed all three rounds.

Table 1 Characteristics of Delphi panelists
Full size table

Early rounds

In round 1, we identified a total of 238 substantive elements of information in panelists’ qualitative responses regarding which specific details prospective participants must grasp. Of these elements, 84 were mentioned by at least 20% of panelists, our threshold for inclusion in the next survey. In round 2, of the 84 elements of information we incorporated, 51 (61%) were endorsed by fewer panelists and 33 (39%) were endorsed by more panelists compared with round 1. In general, panelists’ round 2 responses moved in the direction suggested by our round 1 results, that is, elements less often endorsed in round 1 tended to be endorsed even less often in round 2 and vice versa (Supplementary Appendix S3 online). Table 2 provides additional details regarding both section- and element-level results from these early rounds.

Table 2 Summary of early round results
Full size table

Final round

In round 3, consensus was achieved on 15 of the 16 consent topics ( Table 3 ). There was near unanimity on six topics, for which more than 90% of panelists agreed or strongly agreed that the statement given represented adequate comprehension. For five topics, more than 80% agreed or strongly agreed with the statement provided.

Table 3 Final round results (n = 47)
Full size table

Three topics exceeded our consensus threshold of 70% agreement but with slightly more variation of opinion (Certificates of Confidentiality, Blood Draw, Access to Biospecimens/Data). For each of these, panelists who disagreed that the statement represented adequate comprehension most often indicated that it reflected less than prospective participants need to know.

One other topic—Risk—met our consensus threshold, achieving 70% agreement. Most panelists who disagreed indicated that the statement reflected less than prospective participants need to know, although a nontrivial fraction said it reflected more than required for valid consent.

The topic for which consensus was not achieved was the Genetic Information Nondiscrimination Act (GINA). Results for this topic were disparate in the early rounds, and thus it was the only one for which we presented two separate statements in round 3. Over half (53%) of panelists agreed with the first option, “There is nothing in this section prospective participants must understand.” A slightly larger proportion (60%) agreed that the second option, “There is a law against discrimination based on my information,” represented adequate comprehension, although 26% felt it reflected less than participants need to know and 15% felt it reflected more. Panelists’ comments at the end of the survey suggested that these differences may be because of skepticism about GINA itself, as well as the opinion that participants should understand the limitations of the law22 ( Table 4 ).

Table 4 Examples of panelists’ comments on biobanking consent and Delphi process
Full size table

Panelists’ comments about the Delphi process were substantially positive ( Table 4 ). Many characterized the experience as enjoyable and thought provoking and the process as well organized and easy to follow. A few provided critiques, for example, about the length of the process, a preference for more qualitative input in round 3, and of Delphi techniques in general as “academic.” One wondered whether focusing on one consent topic at a time may have led to an overestimate of what participants need to know. However, most panelists looked forward to the final results with anticipation that they would be useful to the field.

Discussion

Valid informed consent requires that individuals understand and voluntarily agree to participate in research.23 Yet numerous studies have shown that participants in a variety of settings—including biobanking24,25,26—fail to grasp important aspects of the research to which they consented. Many commentators identify one source of the problem as equating informed consent solely with the requirement to disclose information.27,28,29 Disclosure is indeed a key aspect of informed consent, and federal regulations and best practice guidelines provide lists of topics that should be covered in consent forms. Although these are useful starting points, they are not identical to the elements that individuals must understand to give valid consent.23 As stated by Appelbaum12:

Consensus is needed on the question of what information must be understood and to what degree before subjects can be said to have adequate comprehension. . . . Research regulations generally elide this problem with lists of information to be disclosed as if all bits of data were equivalent. It is completely unrealistic, however, to expect research subjects to understand everything they are told, given the scope of most consent forms and the inherent complexity of projects, and some of what is communicated is clearly quite tangential to comprehension of the study itself. (p. 2)

Without consensus on which information is most crucial for participants to understand, it is difficult to operationalize researchers’ ethical obligation to ascertain comprehension30 or to make progress on improving forms and processes to ensure valid consent.31

We attempted to meet this need in biobanking consent by engaging a diverse group of experts in a systematic process to define “adequate understanding.” Our results reflect a high degree of consensus on a relatively small number of key points that prospective participants must grasp to provide valid consent. This outcome was achieved despite significant initial variability of opinion on the panel, which included genome scientists, biobank managers, ethics and policy experts, and community/participant representatives. Through a series of interrelated surveys in which panelists were able to consider results from prior rounds and perhaps think about informed consent in new ways ( Table 4 ), there was clear movement toward consensus and ultimately agreement. It is not uncommon for empirical studies of ethical issues to reveal discordant views among stakeholders—including our own research on the content of biobank consent forms.32 Our present findings suggest there is reason for optimism that common ground can be reached upon further work and reflection.

GINA was the only consent topic for which consensus was not achieved. Opinions were especially varied in early rounds, leading to this being the only topic for which two statements were presented for consideration in the final survey. It is possible that the lack of consensus for GINA in the final round was in part because of having two options to evaluate rather than one. Pending further research, whether knowledge of GINA is essential for valid consent will be a matter of investigator judgment.

Interpretation of our results is subject to several limitations. First, our panel lacked cultural diversity. Subject matter expertise was the primary inclusion criterion, and we identified panelists based on membership in existing committees and consortia; thus the cultural diversity of the panel reflects that of these underlying entities. Second, we asked panelists about existing model consent language for a general, non-disease-specific biobank such as might be established at any academic medical institution. Biobanks that focus on a particular disease or that are established within special populations5 may require different levels or types of understanding for valid consent. In general, investigators looking to assess comprehension may choose to tailor our results to fit their study context; interim findings from our earlier rounds ( Table 2 ) suggest candidates for additional elements of information that some may deem essential. Third, our study was conducted in the United States, with panelists and consent language based on the US context. Although further work is needed, our results may still be useful in other settings because approaches to biobanking are not only similar around the world but are being harmonized to enable global data sharing.33

The goal of our study was not to identify which information should be disclosed to prospective biobank participants, but rather which details among those disclosed are essential for participants to understand. We believe our results may be useful in three ways:

(i) Improve consent forms: Consent forms must meet federal regulatory requirements; in addition, they should be responsive to best practice guidelines and focus on information most important to prospective participants.32,34 Within these parameters, our results may be helpful in organizing and formatting forms to highlight the most essential information; less important details may be candidates for removal when not needed for another reason or moved to supplemental material. Items listed in Table 3 could also serve as an introductory “executive summary” ( Table 5 ) to a consent form or provide the basis for a brief, standard form, as has been advocated elsewhere.35,36

Table 5 Example of consent form “executive summary”
Full size table

(ii) Improve consent processes: Similarly, consent processes—which will increasingly be undertaken in a variety of scalable ways—could be directed to focus on effectively communicating the most crucial information. Items in Table 3 could serve as the basis for developing tools and assessing comprehension in practice (e.g., through a teach-back tool or online quiz).37,38

(iii) Facilitate research on biobanking consent: Lack of common definitions and standardized tools for assessing understanding have led to “conceptual and methodologic chaos”12 in studies of research participant understanding31 and interventions to improve informed consent.38 Our results help remedy this for biobanking; items in Table 3 provide an absolute metric of adequate understanding—an improvement over relative assessments of which intervention produces better scores than another without knowing whether either is adequate.

Next steps in our own program of research include using these results to develop practical ways to assess comprehension and to use these tools to determine the effect of a simplified biobanking consent form on participant comprehension. Further areas for future research include potentially using similar steps and methodologies to improve informed consent in other contexts (e.g., phase III clinical trials). Delphi techniques explicitly make use of expert opinion, that is, they involve individuals with substantial existing knowledge of the topic area. Additional research could make use of deliberative democracy methods39,40,41 to elicit informed and well-considered opinions about consent comprehension among the general public.

Disclosure

The authors declare no conflict of interest.