Streamlined versus traditional consent for low-risk comparative effectiveness trials: a randomized experimental study to measure patients' and public attitudes

Participants were recruited from three populations: patients from Johns Hopkins Community Physicians (JHCP), patients from Geisinger (GHS) and an online, nationally representative panel. We created original animated videos portraying seven approaches to consent and authorization (Figure 1). Participants were randomly assigned to watch one of seven video modules and then asked to complete a survey. At JHCP and GHS, randomization of the seven video modules was executed through Qualtrics; for the national online panel, automatic randomization was programmed by Growth from Knowledge (GfK).

Including JHCP and GHS patient populations enabled comparisons based on patient demographics and health system design. JHCP is part of an urban, academic health system with a large population of both white and African–American patients, whereas GHS is an integrated health system with a demonstrated commitment to the learning health system model serving, primarily, a rural, predominantly white patient population [23]. The online sample enabled measurement of attitudes of a large, nationally representative sample. Analyses examining differences by subpopulation in willingness to participate in PCOR are reported elsewhere [24]. This paper focuses on differences between study arms in terms of respondents' views of consent where the study sample includes a broad spectrum of adults from a range of backgrounds and regions.

The online sample was recruited between 27 April and 10 May 2017 using GfK's KnowledgePanel, a standing, probability-based, nationally representative sample of USA adults [25]. Panel members were randomly selected using address-based sampling methods. GfK processed and weighted the data using Current Population Survey (CPS) benchmarks to adjust for known selection deviations during sampling, and noncoverage and nonresponse bias resulting from panel recruitment methods and attrition. Online participants received a US$5 incentive, consistent with GfK norms.

Self-administered surveys were completed electronically using touch-screen tablets at four JHCP clinics between June and December 2017 and at three GHS primary care clinics between July and December 2017. At JHCP, patients were recruited through posted signs and by clinic administrators telling patients about the survey and directing interested patients to study staff. At GHS, research assistants approached patients before or after scheduled medical appointments. At both JHCP and GHS, study staff described the survey to patients to determine interest in participating in the study. Interested patients-participants were provided with headphones and a tablet containing a link to the Qualtrics survey with videos embedded and asked to complete the survey in the clinic waiting room. Initial questions within the survey screened for eligibility (being 18 or over, being a patient at JHCP or GHS, and speaking English) and provided the survey disclosure statement. Once eligibility was appropriately met, patients were granted access to the remainder of the survey. JHCP and GHS participants received US$30 gift cards to reimburse for their time.

In this project, we aimed to detect at least a 10% difference in a key outcome of interest (e.g., understanding), assuming a baseline level of 0.8. Power calculations suggested that our target project sample of 2400 survey respondents was more than sufficient for 80% power in detecting differences among arms.

To model the hypothetical study comparing two blood pressure medications and how participants would be invited to participate, we created seven animated narrative whiteboard videos with Booster Shot Media, a production company specializing in multimedia patient education with previous experience in creating videos for another study on informed consent [21]. The videos were refined and revised based on feedback from patient advisory and focus groups. Once the seven videos were finalized, participants were randomized to one of seven study arms, and after watching their respective video, all participants then completed the same survey (see Table 1 for summary; full videos can be accessed here https://bioethics.jhu.edu/learning-health-systems/projects/patients-views-on-consent-and-respect/)). The core content of all videos was identical, with variations among arms linked to key research questions. In all videos, a hypothetical patient ‘James’ was told by his doctor, ‘Dr Williams’, that his blood pressure was still high and that it was time to begin antihypertensive medicine. Dr Williams explained to James that the health center was participating in a randomized blood pressure study. Randomization was explained using the analogy of a gumball machine.

The Arms 1–6 videos represented variations on streamlined, ‘opt-out’ consent approaches for the blood pressure study. For Arms 1–6, streamlining informed consent involved limiting study disclosure to the most important study information for potential participants by providing a shorter description of the study than traditional consent forms. The streamlined approach emphasized that the two approved drugs being compared had comparable benefits and similar side effects or risks, but did not specify what those risks were; used clear and simple language; presented information orally, rather than through a written form; and did not require the participant to sign a consent form. In these arms, Dr Williams explained that James would be in the study unless he objected, but that the decision was up to him (opt-out); she gave information orally, during the medical appointment, and she explained that he would not need to sign a consent form to participate. The videos presented in Arms 1–6 varied by whether they also: provided background explaining why comparing existing medicines through CER was important for improving patient care (opt-out CER); described processes within the health system to enhance ETA around research (opt-out ETA); emphasized to James that it was his choice whether or not to participate (opt-out choice) and depicted James as being provided with additional information through a waiting room video James watched before his appointment with Dr Williams, and not directly from Dr Williams (opt-out-previsit video). Videos explicated ETA as follows: engagement meant patients were involved in decisions about what types of research a health system should conduct and how patients should be informed about and invited into such studies; transparency meant health systems used multiple media outlets to regularly inform patients about all ongoing efforts to generate new knowledge and improve care; and accountability meant health systems translated findings of CER research into changes in practice to drive care improvements.

The Arm 7 video depicted a traditional, written consent interaction in which Dr Williams briefly described the study and its purpose. James was then referred to a research nurse, ‘Jonathan’, who was portrayed as reviewing multiple pages of a longer consent document with James that required James's signature before enrollment in the blood pressure study (traditional opt-in). Visual cues were used to communicate time passing, rather than having the participant watch a review of all of the specific information that would have been on every page of the consent form. All videos ended before James was asked his decision about participation in the study.

The assigned video was embedded as a YouTube link into participants' surveys. A participant could watch their assigned video as often as they liked by clicking the ‘play’ button. After watching the assigned video, respondents completed self-administered survey questions assessing their understanding of the study described to James, their own willingness to join a study like the one offered to James, whether they thought James would join, their satisfaction with the amount of information provided to James, and the respectfulness of the interaction between Dr Williams and James (the full survey instrument can be found here https://bioethics.jhu.edu/wp-content/uploads/2018/12/SurveyDevelopment-20.June_.18_Clean.pdf). The videos and survey instrument were developed with input from patient advisory groups at both JHCP and GHS, and from a focus group at each institution. Group members provided input on video clarity and meaning, character depictions, how to make materials more engaging, and their understanding of survey questions. The study team conducted 40 cognitive interviews (24 at JHCP and 16 at GHS) in which participants viewed a version of the video and provided feedback on videos and survey questions including providing feedback on clarity of questions and meaning of key concepts being measured.

The study was designed to address two research questions. First, we were interested in whether small changes to the information provided in the streamlined video disclosure made a difference to participants, specifically: providing additional information about the importance of conducting a comparative effectiveness trial because little is still known about the relative merits of different blood pressure medicines (Arms 2, 5, 6); describing that additional respect-promoting procedures – ETA, or patient engagement around consent practices, transparency about ongoing studies, and accountability to implement what was learned – were in place (Arms 3, 5, 6); and further emphasizing that it was the patient's choice whether or not to join the study (Arms 4, 5, 6) (See Table 1). Our second research question was whether or how streamlined consent approaches compared with a traditional consent approach on outcomes that matter morally, such as understanding and perceived respectfulness.

Sociodemographic, health status characteristics and survey responses were summarized using standard descriptive statistics. We compared responses (binary) to questions about understanding, adequacy of amount of information provided about the study, perceived voluntariness and willingness to join the study across experimental arms using Pearson's chi-squared tests or Fisher's exact test if any cell contained fewer than ten observations. We used Kruskal–Wallis tests and, where applicable, post-hoc pairwise Dunn's tests to compare responses (four- or five-point Likert scale) to questions measuring attitudes toward the consent process and attitudes toward research and the doctor portrayed in the videos. To answer our first research question – whether there were differences among the six streamlined, ‘opt-out’ arms – we performed the above tests on responses from Arms 1–6. For our second question – whether the streamlined or opt out arms differed from the traditional opt-in consent arm – we performed the same tests on the combined responses from Arms 1–6 compared with Arm 7. All statistical analyses were conducted using STATA IC version 15.

Research reported in this publication was funded through a Patient-Centered Outcomes Research Institute (PCORI) Award (ME-1310-07763). The views in this publication are solely the responsibility of the authors and do not necessarily represent the views of PCORI, its Board of Governors or Methodology Committee.

The institutional review board at the Johns Hopkins Bloomberg School of Public Health reviewed and granted exempt status to this study prior to data collection. The institutional review board at Geisinger (IRB# 2017-0149) reviewed the study under an expedited review and granted approval prior to data collection. These IRBs did not believe it was necessary either to inform participants who were asked to complete this self-administered, anonymous survey that they were being randomized to different versions of the study video or to seek written consent for survey completion.

Over 90% of respondents across all seven arms answered correctly that: the study involved a comparison between two medicines and did not involve a placebo; Dr Williams believed James would do well on either medicine; the cost to James would be the same regardless of the medicine assigned; and if James were to experience any problems with his assigned medicine, Dr Williams would switch him to a different drug (Tables 3 & 4). Over 83% correctly understood that medicines would be selected by randomization, rather than assigned by a patient's doctor.

There were no statistically significant differences in understanding across the six streamlined consent opt-out approaches (Table 3). There were, however, significant differences in two of the understanding items between the streamlined, opt-out (combined 1–6) arms and traditional, opt-in consent Arm 7 (Table 4). First, while the vast majority of participants in all arms understood correctly that James could switch medicines if he experienced any problems, this was more likely to be true for participants in the traditional, opt-in Arm 7 (98%) than for participants in the streamlined, opt-out Arms 1–6 (95%) (p = 0.007). Also, approximately one quarter of participants across all six of the streamlined, opt-out arms reported an incorrect belief that James would need to sign a consent form to join, despite statements by Dr Williams during streamlined, opt-out videos that no signature was required. By contrast, in the traditional consent arm, 97% of participants correctly understood that a signature was required in the animation they watched (p = 0.000).

We first examined whether there were differences in attitudes about likeability, respect and dignity of the video's doctor–patient interaction among the six streamlined study arms. While no consistent patterns emerged that favored some arms over others, certain differences were found. For example, there was a statistically significant difference across the streamlined consent approach arms on the item assessing views of how Dr Williams talked to James about the study (p = 0.006). Responses here ranged from 76% of respondents in the opt-out basic Arm 1 ‘really liked’ how Dr Williams talked to James compared with 86% in the pre-visit video, opt-out enhanced Arm 5 (Table 3). Using pairwise analysis, Arm 1 respondents were significantly less likely to ‘really like’ the way the doctor spoke to James about the study than respondents in Arms 2, 3, 5 and 6 (Table 5).

In terms of respect, nearly 90% of respondents across all six streamlined arms reported that Dr Williams was ‘very respectful’ when talking with James about the study, with no statistical difference among streamlined arms (Table 3). In terms of dignity, while at least 54% of respondents in each streamlined arm ‘strongly’ agreed that Dr Williams treated her patients with dignity, there were significant differences across arms (p = 0.036) with pairwise comparisons highlighting that these were driven by variation between pre-visit video Arm 5 and Arms 1, 3, 4 and 6 (Table 5).

We detected no significant differences between respondent attitudes across all three of these measures when comparing the combined streamlined, opt-out consent Arms 1–6 and traditional consent Arm 7 (Table 4).

Most respondents in all arms said the amount of information Dr Williams provided to James was ‘just right’. In comparing the streamlined arms, participants calling the amount ‘just right’ ranged from 87% in Arms 1 (opt out basic) and 4 (opt out choice) to 93% in Arms 3 (opt out ETA) and 5 (opt-out pre-visit video) (p = 0.004, Table 3). There were no significant differences between the combined streamlined Arms 1–6 and the traditional consent Arm 7 (Table 4).

At least 97% of participants in every arm believed that it was up to James whether to join the study, and at least 93% in each arm said that if James did not want to join it would be easy for him to say no (Tables 3 & 4). Although streamlined Arms 4, 5, and 6 included an extra sentence emphasizing that it was James's ‘choice’ whether to participate, there were no statistically significant differences among the six streamlined arms on either of the perceived voluntariness measures. There were also no differences on these measures between the opt-out and traditional opt-in consent arms.

98% of all respondents thought that James would probably join the study, with no differences among the streamlined arms or between the streamlined arms and the traditional arm (Tables 3 & 4). When asked whether they would join the study described to James in the video, at least 85% of participants in all arms said they would probably join. There was a statistically significant difference among streamlined, opt-out Arms 1–6 (Table 3; p = 0.006), with more than 92% of participants in Arms 2 and Arms 5 reporting they would join. There were no significant differences between the streamlined arms and the traditional consent arm on this outcome (Table 4).

Among participants in the six streamlined arms, 70% reported it was ‘very important’ for patients to join CER studies (Table 3), ranging from 67% in Arms 1 and 4 to 77% in pre-visit video Arm 5 (Table 3, p = 0.034). Pairwise comparisons suggest that Arm 5, which contained all the enhancements in the pre-visit video, was driving this difference (Table 5). There were also differences among the streamlined Arms 1–6 in whether respondents felt Dr Williams was committed to doing what was best for her patients (p = 0.030, Table 3), with 66% of respondents in Arm 5 ‘strongly agreeing’ compared with 56% in Arm 4 and 56% in Arm 1. Pairwise tests suggest that Arm 4 was significantly lower than Arms 2, 3, and 5, while Arm 5 was significantly higher than 1, 4, and 6 (Table 5).

Nonetheless, across the six streamlined arms, and for all six of these items, there were no significant differences regarding whether respondents believed Dr Williams made good decisions for her patients (42% strongly agree), was a trustworthy doctor (45% strongly agree), communicated clearly (62% strongly agree), or was rushed with patients (44% strongly disagree) (Table 3). There was a significant difference across the streamlined arms in respondents ‘strongly agree[ing]’ that Dr Williams was ‘thorough and careful’ (Table 3; p = 0.003), ranging from 50% in Arm 1 to 61% in Arm 5.

There were no differences across the seven outcomes measuring attitudes toward CER or Dr Williams when comparing the combined Arms 1–6 to Arm 7 (Table 4).

Across all seven arms, and for all six of these items, almost all respondents answered either ‘strongly agree’ or ‘agree’ for the positively framed items, or answered ‘strongly disagree’ or ‘disagree’ for the negatively worded item. Fewer than 7% of participants in any of the arms expressed views that could be construed as critical of Dr Williams's conduct.