Protecting Human Research Participants: Reading vs Understanding the Consent Form

In 1998, in this Journal, Davis and colleagues ( 1 ) published their findings on a comparative study of a standard vs a simplified consent form (CF) in 53 patients with cancer and 130 individuals who were apparently healthy. Interviews conducted with the participants after reading a standard Southwestern Oncology Group CF and a shortened version of the same CF in varying sequence indicated an overwhelming preference for the simplified form ( 1 ). The standard form was 7 pages, contained 3438 words with sentences that averaged 21 words, and was without graphics. Application of Flesch–Kincaid methodology indicated a 12th grade reading level. By contrast, the shortened version was prepared as a 7-page booklet that contained 524 words with an average sentence length of 12.5 words, culturally sensitive instructional graphics, and colored headers. Readability was determined at a 5th grade level. CF preference was independent of race but varied by reading and educational level. Surprisingly, participant responses to 10 comprehension questions, which included key elements of study design, drug allocation, and risk, showed that their understanding of the basic information was very low, irrespective of form preference. The authors, supported by editorial comment by Taylor et al. ( 2 ) in the same issue of the Journal, concluded that there are serious questions regarding the adequacy of the design of written informed consent documents for the substantial proportion of Americans with low-to-marginal literacy skills. Fifteen years later, it seems that little has changed to alter the conclusions of Davis et al. ( 1 ). In fact, CFs in cancer clinical trials are even longer (and now fattened by Health Insurance Portability and Accountability Act [HIPAA] legalese), more complex, and at a readability level that is still far beyond that of the average US adult, which remains no higher than the eighth grade [and for Medicaid enrollees, perhaps as low as the fifth grade ( 3 )]. More important, concerns about patients’ actual understanding of the process remain largely unaddressed. In this issue of the Journal, Koyfman et al. ( 4 ) analyzed 197 CFs from 56 cooperative group (CG) trials that were open in two or more participating institutions, comparing the characteristics of the standard CG template for informed consent with those of the CG templates modified by the institutional review board (IRB) at each of seven participating academic institutions. Together, these sites represent the major US centers for oncology research. They found that, compared with the recommended original CG template, the modified CFs were substantially longer (17 vs 13 pages), at a higher grade level (10.3 vs 9.4), and were more difficult to read by the Flesch Reading Ease Score (53.1 vs 57.1, where 100 is very easy and 0–30 is very difficult). Not only that, but paragraph-by-paragraph comparisons and even within-paragraph analysis indicated that the degree of interinstitutional variation was even more marked than whole document evaluation would indicate. So what does all this mean in the real world of cancer clinical trials? The methodology applied by Koyfman et al. has been validated in many differing environments, including the legal field. Embedded within writing software programs such as Microsoft Word, the Flesch–Kincaid algorithm calculates reading grade level by sentence length and the number of words with two or more syllables. It may be confounded by use of a semi colon instead of a period at the end of a sentence, or by the use of bullet points [in fact, in the Koyfman et al. article ( 4 ), the “Risk” sections of the CFs were omitted from specific software analysis because they were often presented in listed, bulleted format]. In addition, the Flesch–Kincaid scoring method in Microsoft Word artificially truncates readability at the 12th grade and therefore underestimates the reading level required for complex texts ( 5 ). Despite these limitations, the methodology in this paper should be considered sound and the conclusions valid: as CFs progress from National Cancer Institute–sponsored CG to local IRB, they become longer and less readable; interinstitutional heterogeneity in CF readability is substantial and widespread. These findings would support the recent proposals for change to the regulations governing research with human subjects, citing the lack of evidence that having multiple IRBs review a single study has led to improvement in the ethics review process ( 6 ). Human research protection hinges on the extent to which the patient is able to provide consent that is truly based on an understanding of what the research is all about. Readability of the CF is a part of this, but the process of informed consent goes far beyond a single conversation and a signature on a piece of paper. In their systematic review, Flory and Emmanuel undertook a MEDLINE search for pertinent publications from 1966 to 2004 and ultimately evaluated 30 studies in 42 trials that met their inclusion criteria ( 7 ). Interventions to improve comprehension were categorized into five groups: 1) multimedia; 2) enhanced CFs; 3) extended discussion; 4) test/feedback; and 5) miscellaneous. Of these, the authors concluded that the most effective way of improving research participants’ understanding was having a study team member or neutral educator spend more one-on-one time talking about the details of the trial. The use of multimedia and enhanced CFs had only limited success. Given the very recent explosion in the use of the smart phone in the daily lives of ordinary people, the replacement of the printed page with the e-reader and the worldwide impact of social networking on communication, perhaps the time is ripe to reconsider a new paradigm for formulating and sustaining consent as a dynamic and ongoing process, with inherent opportunities for “instant” feedback, communication, and education in modes that could be readily standardized. It is strange, in a way, that after all these years since the Belmont report, informed consent is still largely in the format of a single conversation and a (very long) piece of paper. In the conclusions to their article in support of an evidence-based approach to human research protection, McDonald and Cox state that “almost all of health research ultimately depends on the participation of human volunteers as research subjects. Search for evidence of what happens to these volunteers and potential volunteers is ethically mandatory. We cannot rest content with the absence of evidence. The onus is on the research community to collect and use that evidence to good effect” ( 8 ). It is ironic perhaps that, with few exceptions, the process of informed consent in randomized clinical trials has itself not been the subject of a similarly randomized approach with regard to the primary intent—the protection of research participants ( 9 ). Without this, where is the evidence that current practices of seeking consent are in fact achieving their moral obligation—providing information that allows the individual to make a genuinely informed choice about his or her research participation?

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.