|Home | About | Journals | Submit | Contact Us | Français|
Although the informed consent process is supposed to help potential research participants make informed and voluntary decisions about participating in research, little is known about how participants react to language in the informed consent document and whether their reactions are related to their willingness to enroll in clinical trials. We examined the relationship between patients' reactions to standard informed consent language and their willingness to participate in a hypothetical clinical trial.
We simulated the consent process for a hypothetical cardiology clinical trial with 470 patients in an outpatient cardiovascular medicine clinic at a large academic medical center. We analyzed the spontaneous comments and questions that participants made during the interviews about each section of the informed consent document. Few participants made positive comments. Participants made the most negative comments about the sections on risks, study purpose or protocol, and payment for injury. Having a negative reaction to any section was associated with a lower likelihood of participating in the clinical trial. Using a multivariable model, we found that negative reactions in the patient rights, financial disclosure, and confidentiality sections predicted willingness to participate (P < .001).
Recognizing elements of informed consent that elicit questions and concerns from potential research participants may help investigators design clinical research trials and model language in a way that reduces concerns or increases participant understanding, and thereby enhancing informed consent for research.
The informed consent process is intended to help potential research participants make informed and voluntary decisions about whether to participate in research studies.1 What transpires during the consent process has implications for both the integrity of potential research participants' decisions and the ability of investigators to enroll sufficient numbers of participants in clinical trials. Enrollment has been a particular concern in cardiology clinical trials.2 Previous studies have found that the handling of the consent process (eg, reading the informed consent document vs having it explained) can influence decisions to enroll.3,4 However, these studies did not examine how potential research participants react to specific components of the consent process and how those reactions relate to decisions to enroll in cardiology trials.
US federal regulations require several elements of disclosure in the informed consent process, including statements about the study's purpose, procedures, and potential risks and benefits; alternatives to participation; the voluntary nature of participation; and whether treatments will be made available for research-related injuries.1,5,6 The disclosure of whether treatments will be made available for research-related injuries is a longstanding requirement, but institutions continue to struggle with questions about the costs of treating research-related injuries.7,8 Academic institutions increasingly disclose to research participants that they will not assume these costs or offer compensation for research-related injuries, sometimes in a section termed “patient rights,”7,8 an approach that has invited some criticism.9-13
Since the enactment in 2003 of the Privacy Rule of the Health Insurance Portability and Accountability Act of 1996 (HIPAA), the federal government has also required investigators to describe what protected health information will be collected, how the information will be used or disclosed, and who can use or disclose the information. The investigator must also explain that regulations may not protect the information after disclosure to other parties.14 Previous research has suggested that these policies and procedures may have slowed recruitment and increased the costs and logistical complexity of conducting research.2-4,14
A relatively new element of informed consent is the disclosure of financial conflicts of interest. Professional societies, governmental bodies, and others have suggested that conflicts of interest could be managed, in part, by requiring investigators to disclose financial relationships to potential research participants.15-17 Research coordinators have reported that financial disclosure during the consent process is commonplace,18 but there is limited evidence about how much importance potential research participants attach to this information in the context of the informed consent process.19,20
Using a simulated informed consent process for a hypothetical but realistic clinical trial, we examined how potential research participants spontaneously react to different parts of the informed consent document and how participants' comments about these elements of informed consent were related to their likelihood of participating in the hypothetical clinical trial.
As part of the Conflict of Interest Notification Study, we studied patients in the outpatient cardiovascular medicine clinic of a large academic medical center.18-23 The institutional review boards of the Duke University Health System, the Johns Hopkins Medical Institutions, and Wake Forest University approved the study. Details of the study design have been published previously.20
We developed an informed consent document for a hypothetical clinical trial using a template from a previously approved trial at the Duke University Medical Center. We adapted the simulated informed consent document to describe a clinical trial evaluating an oral medication for the treatment of coronary artery disease (Appendix A). The consent document consisted of an introduction; a description of the study's purpose and protocol; risks and benefits of participation; alternatives to participation; confidentiality; patient rights; and authorization. The chair of the institutional review board of the Duke University Health System reviewed the document to ensure that the language was consistent with other approved informed consent documents for clinical trials. We randomly assigned participants to receive the consent document with no financial disclosure statement or with 1 of 2 previously tested financial disclosures based on previous research and input from an expert panel.19-22
The study sample was composed of patients diagnosed with coronary artery disease at the outpatient cardiovascular medicine clinic of a large academic medical center. Potential research participants received letters from their cardiologists asking them to participate in a telephone survey. Interested patients were informed that they would be asked to read an informed consent document for a hypothetical clinical trial. Those who agreed to participate provided verbal consent, scheduled a telephone interview, and received a copy of the informed consent document for the hypothetical clinical trial. During the audio-recorded telephone interview, the interviewer verified that the participant received and read the consent document. The interviewer then simulated the informed consent process by summarizing each section of the consent document, consistent with typical practice in clinical trials. After reviewing each section, the participant could ask questions and comment on the section. After the simulated consent process, participants expressed their likelihood of participating in the hypothetical clinical trial on a scale from 1 (“I definitely would not participate”) to 5 (“I definitely would participate”),20 responded to the 4-item Trust in Medical Research measure using 5-point Likert scales,24 and rated the importance of key aspects of the study for their decision making (Appendix B).19 The first 40 interviews served as pilot data for survey and protocol refinement.
Participants' reactions to the informed consent document during the simulated consent process were transcribed and dual-coded using a coding scheme developed through an iterative process of code definition and refinement. The coding scheme included codes for questions and positive and negative comments. Comments in which participants explicitly stated that a section of the consent document made them hesitant to participate were also identified. The coding structure allowed us to associate comments and questions with the section of the consent document to which the participant was referring (ie, research purpose and study protocol, risks, benefits, financial disclosure, confidentiality, and patient rights). To assess interrater reliability, we calculated kappa statistics for the 24 possible reactions (ie, question, positive comment, negative comment, hesitation comment) in each of the 6 sections from a randomly selected 12% of the transcripts. We examined the proportion of items in the kappa categories “slight,” “fair,” “moderate,” “substantial,” and “almost perfect” agreement.25
We used a χ2 test with exact P values to determine whether the frequencies of comments differed by section of the informed consent document. To determine whether comments about a section were due to a participant's overall tendency to voice reactions, we first created a dichotomous indicator for each section coded “1” if the person asked a question or made a comment about that section. We then computed all pairwise correlations (phi coefficients) between these dichotomous indicators for each section. Using χ2 tests and t tests, we compared distributions of sex, age, race, education level, annual household income, previous participation in a research study, and a trust score among patients who did or did not make at least 1 negative comment and did or did not ask at least 1 question. We used a t test to examine the relationship between asking at least 1 question and willingness to participate. To determine whether asking questions about particular sections of the consent document was related to willingness to participate, we used a general linear model in which we regressed willingness to participate on dichotomous indicators of whether a question was asked in each section.
We used a t test and a general linear model for similar analyses examining the relationship between making a negative comment or asking a question and willingness to participate. (There were not enough positive comments to examine the relationship between positive comments and willingness to participate.) To facilitate interpretation, we also expressed group differences between continuous variables in standard deviation units as standardized effect sizes (d). Thus, a d = 0.30 means that 2 groups were approximately one third of a standard deviation apart from each other. For the general linear model, it was important to determine whether relationships between negative comments (or questions) and willingness to participate could be due to confounding effects of patient education level, ethnicity, race, sex, age, and previous participation in a research study. Thus, we used t tests and χ2 tests to examine bivariate relationships between each of these characteristics and making a 1 negative comment (or asking a question). Variables with P < 0.10 were included in the multivariable model. The final model for making a negative comment included education level (high school or less, college degree, or graduate or professional degree), race (white or nonwhite), age, and previous participation in a research study. No patient characteristics were selected into the final model for asking a question.
In a supplementary analysis, we dichotomized the likelihood of participation to reflect our interest in understanding people who would likely decline participation. That is, we created categories of “probably” and “definitely” would not participate vs “uncertain” and “probably” and “definitely” would participate. We then used a χ2 test to examine the relationship between negative comments about a section and refusal to participate in the clinical trial. To control for other factors that might explain willingness to participate, we used a multiple logistic regression model in which we modeled the dichotomized likelihood of participation by using dummy codes representing the occurrence or nonoccurrence of a negative comment separately for each section of the consent document.
We used SAS version 9.1 (SAS Institute Inc, Cary, North Carolina) for all analyses.
Of the 1545 patients who provided verbal consent and were contacted for the telephone interview, 510 completed the interview and 862 declined or did not return 4 telephone calls.20 Excluding the 40 pilot interviews left 470 interviews for the analysis. Table 1 shows the participant characteristics. Consistent with real-world cardiology studies conducted at the study location, most participants were white men with relatively high education levels and incomes.20 One third of the participants had previously participated in a research study, and half of those had participated in a cardiology trial.
In the qualitative analysis of spontaneous reactions, interrater reliability was high. All but 2 reactions (92%) were in perfect or almost perfect agreement (kappa > 0.88). The remaining 2 reactions had substantial and moderate agreement (kappa = 0.66 and 0.48).
A total of 1285 questions and comments were expressed by 347 (73.8%) of the 470 participants. The frequency of these reactions differed across the sections of the informed consent document (Table 2). The few participants with positive comments most often referred to the potential benefits of the research (3.0%) and the financial disclosure (2.1%). Participants frequently had negative comments about the patient rights section (15.8%), the financial disclosure (15.8%), and the confidentiality section (9.8%). Sections that participants most often reported would affect their likelihood to participate in the hypothetical clinical trial were the sections describing the potential risks of the study (4.0%), how the study works (3.4%), and patient rights (3.2%) (For additional information about the nature of these questions and comments, see Appendix C).
In the purpose and protocol section, participants most often asked questions about what the study would entail (eg, the number of doctor visits and the length of the study). Participants also asked for clarification that the study was hypothetical. Almost half of the participants who had negative comments in this section expressed concern about taking another medication in addition to the medications they were already taking. In the risks section, participants most often had questions or concerns about the side effects of the medication. The most salient reaction in the benefits section involved questions or negative comments about the lack of incentive for participation, because the hypothetical clinical trial involved no monetary compensation and no guarantee of clinical benefit.
In the financial disclosure section, one third of the participants who had questions about the section asked how the investigator might benefit, and two fifths wanted clarification about per capita payments. Most participants who had negative comments about the financial disclosure expressed an aversion to the investigator's financial relationships or thought this relationship could bias the investigator.
In the confidentiality section, patients most often asked about and expressed concern regarding who would have access to their health information. Participants also questioned why there was not an absolute guarantee of confidentiality.
Almost three quarters of the participants who had questions about the patient rights section asked whether there would be compensation or free medical care in the case of an adverse event. Thirty-nine participants had a spontaneous negative comment about the lack of compensation or provision of free medical care in the event of research-related injury.
Whether a participant made a comment or had a question about one section was generally unrelated to whether he or she made a comment or question about another section (Table 3). The correlations ranged from essentially zero to 0.26, supporting our decision to model the occurrence of comments in each section separately. Moreover, asking at least 1 question was not associated with independent participant variables. However, participants who made negative comments tended to be slightly older, have more formal education, have participated in clinical trials, and have less trust in research (Table 4).
Participants who asked at least 1 question at any time did not differ in their willingness to participate in the hypothetical clinical trial (mean, 3.31; SD, 1.3) from those who did not ask questions (mean, 3.32; SD, 1.32; P = .12; d = 0.15). Figure 1 displays the mean willingness-to-participate scores for participants who did and did not ask questions. There was no relationship between asking a question about a particular section and willingness to participate.
Patients who made at least 1 negative comment at any time had a lower willingness to participate (mean, 2.86; SD, 1.34) than those who did not make negative comments (mean, 3.52; SD, 1.21; P < .001; d = 0.63). Figure 2 displays the mean willingness-to-participate scores for participants who did and did not make a negative comment. Negative reactions in the patient rights, financial disclosure, and confidentiality sections were uniquely associated with the reported likelihood to not participate.
Supplementary analyses of the dichotomized likelihood to participate supported these findings. Of participants who voiced negative comments about the patient rights section, 56% (29/52) “probably” or “definitely” would not participate in the clinical trial, compared with 24% (101/418) who did not make negative comments (P < 0.001). Of participants who voiced negative comments about financial disclosures, 52% (27/52) “probably” or “definitely” would not participate in the clinical trial, compared with 24% (103/418) who did not make negative comments (P < 0.001). Of participants who voiced negative comments about confidentiality, 54% (25/46) “probably” or “definitely” would not participate in the clinical trial, compared with 25% (105/424) who did not make negative comments (P < 0.001).
Little is known about the questions and concerns patients have about specific sections of informed consent documents and how such concerns relate to the likelihood to participate in research. Using a simulated informed consent document that reflects the typical content of such documents for clinical trials, we found that three quarters of the participants had a question or comment about a specific aspect of the hypothetical clinical trial. These remarks provide insights into the issues that are most troubling or confusing for potential research participants. We also found that negative reactions were strongly related to the stated intention to participate in the hypothetical clinical trial.
Broadly accepted models of informed consent require a core disclosure and an opportunity for potential research participants to request more information or to seek clarification.26 We found that participants asked the most questions about the study purpose/protocol, confidentiality, and patient rights sections of the consent document. Asking questions was not associated with participant characteristics or with the intention to participate in the hypothetical trial, which suggests that the typical consent process is capable of eliciting pertinent questions from a range of potential research participants.
Less frequent than questions were negative comments about the consent document. Negative comments were made significantly more frequently by participants who were older, had more formal education, had previously participated in clinical research, and had expressed less trust in research and researchers. These negative remarks were strongly associated with the stated intention of whether to participate in the hypothetical clinical trial. The sections of the consent document that evoked the most negative reactions and were most related to willingness to participate were the financial disclosure, patient rights, and confidentiality sections. As reported previously,20 most negative comments about financial disclosures were made by participants who were randomly assigned to receive a disclosure of the investigator holding equity in the company producing the drug, whereas per capita payments were not viewed as problematic.
Some commentators have expressed concerns about the effects of the HIPAA Privacy Rule on clinical research.27-30 Our study is the first to systematically investigate potential research participants' reactions to HIPAA authorization language that has been integrated into an informed consent document for research. We found that 1 in 10 participants made spontaneous negative comments about this section. These findings, combined with the nature of the comments, suggest that standard HIPAA Privacy Rule language raises concerns for patients that investigators and study coordinators should be prepared to address. Furthermore, having a negative reaction to the confidentiality language was associated more than any other element with a lower likelihood of agreeing to participate in the hypothetical clinical trial.
About 1 in 10 participants were bothered by consent document language that stated that care for research-related injuries would not be paid for by the institution or sponsor. This issue raised greater concerns than treatment risks and was negatively related to intent to participate. Despite repeated calls by ethics commissions for institutions to pay for research-related injuries, legal and financial pressures have led academic institutions to disclaim or limit their responsibility.8,12,31,32 Our findings about the impact on willingness to participate might provide a more practical reason for institutional officials to rethink their policies.
We attempted to replicate actual clinical trial enrollment by recruiting participants from an institution that conducts a high volume of clinical research. As a result, the demographic characteristics of the sample resemble many clinical trial samples in cardiology, suggesting that the results are generalizable to trial populations in this therapeutic area. However, because the sample was composed predominantly of well-educated white men, we had limited ability to examine the effects of demographic variables, such as race/ethnicity, on making a negative comment and reported willingness to participate.20
Statistical power was calculated to detect differences between the simulated informed consent document with no financial disclosure statement and the informed consent documents with 1 of 2 financial disclosure statements.20 Thus, we were limited in our statistical comparisons of the spontaneous reactions. For example, we did not have enough statistical power to compare having a comment or question in a particular section of the consent document with the participant's reported willingness to participate. However, we did have sufficient power to use a general linear model to compare relationships between negative comments or questions in any section and reported willingness to participate. In addition to the statistical analyses, we believe that the study provides valuable qualitative information about the nature and incidence of potential research participants' spontaneous reactions to informed consent documents.
We selected the hypothetical clinical trial intervention (a cardiovascular medication) to resonate with the participants. Nevertheless, as in any hypothetical scenario, it is difficult to know how well participants understood and considered the hypothetical risks and benefits. A previous study found evidence that people make equivalent decisions in hypothetical and real settings33; however, no previous study has compared hypothetical and real research participation decisions.34 The hypothetical nature of the consent process might also have made participants more reflective and talkative. Only reactions that were strong enough to be voiced without prompting were included in the data set. As a result, we likely underestimated the extent of both positive and negative reactions. Thus, it is not certain that the participants' spontaneous comments or reported willingness to participate would be the same if they were being recruited to participate in a real clinical trial.
Despite these limitations, the simulated consent process seemed real to many participants, and research staff frequently had to remind participants that they were not consenting to enroll in an actual clinical trial.
Although the consent document we used for this study contained standard language on financial disclosures, HIPAA, and payment for injury, some components of the consent document were unique to the hypothetical clinical trial. Also, the standard consent language used by Duke University is not identical to that used by other institutions. Thus, the reactions to the consent document may not generalize across all institutions and clinical trials.
Previous research has documented research participants' difficulties understanding the informed consent document.35-39 We found that some participants' questions and negative reactions indicated a lack of information or a misunderstanding of elements of the clinical trial described in the informed consent document. For example, several participants assumed incorrectly that no one could access the participant's medical records without obtaining the participant's explicit permission. The high frequency of questions or comments about particular sections may indicate that the public has greater difficulty understanding certain elements of the consent document. This information may help study staff anticipate common questions or reactions to standardized parts of the consent document.
Given that the purpose of informed consent is to respect and promote participants' autonomy, it is important not to misinterpret negative comments about the consent document as meaning the consent language must be changed. Negative comments point to issues that are salient to potential research participants in enrollment decisions and thus deserve some attention. However, giving disproportionate attention to matters that should be of lesser concern might be seen as detracting from informed choice. The ethical balance involves the need to enroll and not inappropriately scare away participants versus the need for honest disclosure.
Regardless of how this balance is struck, identifying elements of the consent document that elicit questions and concerns from potential research participants may help investigators use model language in consent documents in a way that reduces concerns while increasing participant understanding, thereby enhancing informed consent for research.
We thank Larry W. Diener, MLS, MBA, of Duke University for assistance with database programming; and Damon M. Seils, MA, of Duke University for assistance with manuscript preparation. Messrs Diener and Seils did not receive compensation for their assistance apart from their employment at the institution where the study was conducted.
Funding/Support: Supported by grant R01HL075538 from the National Heart, Lung, and Blood Institute.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute or the National Institutes of Health.
Financial Disclosures: None reported.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Alice K. Fortune-Greeley, Center for Clinical and Genetic Economics, Duke University School of Medicine, Durham, North Carolina.
N. Chantelle Hardy, Center for Clinical and Genetic Economics, Duke University School of Medicine, Durham, North Carolina.
Li Lin, Center for Clinical and Genetic Economics, Duke University School of Medicine, Durham, North Carolina.
Joëlle Y. Friedman, Center for Clinical and Genetic Economics, Duke University School of Medicine, Durham, North Carolina.
Janice S. Lawlor, Departmentof Social Sciences and Health Policy, Wake Forest University School of Medicine, Winston-Salem, North Carolina.
Lawrence H. Muhlbaier, Duke Clinical Research Institute, Departments of Biostatistics and Bioinformatics, Surgery, Duke University School of Medicine, Durham, North Carolina.
Mark A. Hall, Wake Forest University School of Law, Winston-Salem, North Carolina.
Kevin A. Schulman, Center for Clinical and Genetic Economics, Medicine, Duke University School of Medicine, Durham, North Carolina.
Jeremy Sugarman, Berman Institute of Bioethics and Department of Medicine, The Johns Hopkins University, Baltimore, Maryland.
Kevin P. Weinfurt, Center for Clinical and Genetic Economics, Psychiatry and Behavioral Sciences, Duke University School of Medicine, Durham, North Carolina.