Reporting of Informed Consent and Ethics Committee Approval in Clinical Trials

Context To determine whether journals have improved their disclosure of ethical protections in clinical trials.

Methods Comparison of clinical trials published before and after 1997 (July 1995 to December 1996 and January 1998 to June 1999) in Annals of Internal Medicine, BMJ, JAMA, The Lancet, and The New England Journal of Medicine. Sixty articles per journal per period were randomly selected and assessed for rate of reporting on informed consent and on ethics committee approval.

Results Informed consent was not described in 79 articles (26%) published before 1997 vs 53 (18%) published after 1997 (P = .01), and ethics committee approval was not mentioned in 93 (31%) before 1997 vs 54 (18%) after 1997 (P<.001). Neither protection was described in 48 articles (16%) published before 1997 vs 28 (9%) after 1997 (P = .01). In subgroup analyses, those journals with the worst initial rates generally improved the most. BMJ did not describe informed consent in 25 articles (42%) before 1997 vs 15 (25%) after 1997 (P = .05), and JAMA did not describe ethics committee approval in 25 (42%) before 1997 vs 13 (22%) after 1997 (P = .02). BMJ, JAMA, and Annals had the lowest initial rates of reporting on both protections in the same article, with 25 (42%), 32 (53%), and 34 (57%), respectively, but improved markedly to 38 (63%), 43 (72%), and 45 (75%) (P = .02, .04, and .03, respectively).

Conclusions Major medical journals have improved their reporting on informed consent and ethics committee approval; however, 9% of studies still report neither.

The World Medical Association issued the Declaration of Helsinki in 1964,¹ with subsequent updates,² to establish international regulations for human experimentation. It specifically identified 2 protections: that all participants in trials should understand the risks, benefits, and alternatives of the experiment and, following this, should enroll in the trial under their own free will, by giving informed consent, and that a disinterested party unconnected with the trial, termed the ethics committee or institutional review board, should have approved the experimental protocol after assuring its appropriateness of design.

When publishing reports of experimentation, journals have an obligation to define what constitutes medical research of the highest quality and to make the ethical conduct of trials a part of this definition. The Declaration of Helsinki charges journals with this important responsibility, stating: "Publishers have ethical obligations. . . . Reports of experimentation not in accordance with the principles laid down in this Declaration should not be accepted for publication."² It is essential that journals inform the public that, to the extent they can determine, the trials they publish are ethical. They can do so in part by requiring that all published trials describe informed consent and ethics committee approval. This duty is supported by the International Committee of Medical Journal Editors (ICMJE).³

Despite this obligation, in 1997, on the 50th anniversary of the Nuremberg Code,⁴ there was growing evidence that journals were not meeting their commitment to these guidelines.^5-8 Data at the time⁵ and further documented since⁹ suggested that articles reporting ethical protections were of higher methodologic quality than those that did not. The current study assesses whether journals have improved in their reporting of ethical protections from the time these deficiencies were highlighted.

Clinical trials were studied because their use of interventions makes the reporting of safeguards particularly important. We selected a random sample of trials published in the 18 months before and after 1997 (July 1995 to December 1996 and January 1998 to June 1999) in Annals of Internal Medicine, BMJ, JAMA, The Lancet, and The New England Journal of Medicine. Clinical trials were defined as studies with interventions performed at the level of the patient. Brief communications, letters, case studies, and case series were excluded. Based on data from a pilot study, we selected a sample size of 300 articles per period (60 per journal) to have a 2-tailed α of .05 and an 80% power to detect a 10% difference in the primary outcome measures across the pooled, not individual, journals. A random numbers table was used to select eligible issues of each journal. All articles that met the inclusion criteria within a given issue were selected until the sample size had been reached.

The primary outcome measures were the rates of reporting on informed consent and ethics committee approval. Our definitions were fulfilled by any of the following types of descriptions (with examples): (1) informed consent—strict definition ("the patients gave written informed consent to participate in the study"), implied ("all subjects gave written consent"), and waived ("the investigators received approval from their institutional review boards to use deferred consent"); and (2) ethics committee approval—strict definition ("the study protocol was approved by the hospital's ethics committee"), implied ("ethical approval of the study was obtained from all study sites"), adhered to the Declaration of Helsinki ("the study was performed according to the guidelines of the Declaration of Helsinki"), and approved by a government or professional body ("the protocol was approved by the Office of the Surgeon General").

One secondary outcome measure was the rate of reporting on the presence, in each article, of both protections or neither protection. Another outcome was the rate of reporting on further details of informed consent—the types identified in the pilot study as written (required by the Declaration of Helsinki, in the absence of mitigating factors²), oral, familial, nonfamilial (eg, granted by a legal proxy), witnessed, or any combination of these. The final secondary outcome was the rate of naming of ethics committees. Because there is strong evidence of local variation in the practice and effectiveness of committees,^10,11 readers or others may want to know their identity to hold them accountable for their decisions.

Additional information was collected for use in subgroup analyses on study participants who belonged to vulnerable populations. The categories of vulnerable populations were identified through a review of the literature and government documents^12,13 and included children, pregnant women, adults who lacked decision-making capacity (eg, those with severe dementia), prisoners, and patients with human immunodeficiency virus, an intensive care unit level of disability, psychiatric disease, or genetic risks or disease. Data were also gathered on whether articles contained references to earlier publications on the same study. Such articles may have a plausible reason for not describing ethical protections, if these were stated in an earlier article. However, the difficulty of accurately identifying prior versions of the same article without the benefit of a reference has been well documented by meta-analysts.^14-16 Therefore, articles that do not mention protections should reference a prior article that does. We also collected data on the numbers of ethics committees per article. We then identified the funding source of each study, defined as industry only (eg, pharmaceutical company), nonindustry only (eg, government agency, university, or foundation), combination of industry and other, none, not mentioned, or unclear. Last, we assessed whether each journal's instructions for authors, published in the first issue of the first month of both study periods, explicitly stated that articles were required to describe informed consent or ethics committee approval.

Because the study did not involve human participants or records, it did not require review by the committee on human research at the University of California at San Francisco.

Data were gathered by using an extraction instrument developed during the pilot study. After an author (V.Y.) trained a research assistant (P.G.) to use the instrument, both tested it on a sample of journal articles that met the study inclusion criteria but had not been randomly selected for the study. After discussion of the most common areas of difference in categorization, both then applied the instrument to 50% of the study sample, with the test statistic for interrater reliability found to be excellent for all primary and secondary outcome measures (κ>0.95 for all). Differences were resolved by discussion. One of the authors (V.Y.) then completed assessments of the remaining articles. χ² Analyses were used to test for differences in the rates of reporting. All analyses were performed by using Microsoft Excel 98 (Microsoft, Redmond, Wash).

Ethical protections for participants in clinical trials were not reported in more articles published before 1997 than after (Table 1). Subgroup analyses showed that those journals with the lowest initial rates of reporting generally improved the most; for example, BMJ with informed consent JAMA with ethics committee approval. Only 2 journals, The Lancet on informed consent and Annals on ethics committee approval, added a new explicit statement regarding ethical protections to their instructions for authors between study periods.^17,18 Both journals demonstrated statistically significant improvements in reporting on these respective protections. The remaining journals retained roughly the same explicit statements, or lack thereof, regarding ethical protections.

All journals made progress in describing both informed consent and ethics committee approval in each article. Those journals with the worst initial rates improved the most: BMJ, JAMA, and Annals began with rates of 42%, 53%, and 57%, respectively, but improved markedly to 63%, 72%, and 75% (P = .02, .04, and .03, respectively). Overall, the rate of reporting improved from 175 of 300 (58%) before 1997 to 221 of 300 (74%) after 1997 (P<.001).

Similarly, fewer articles published after 1997 described neither protection (Table 2). Of articles on vulnerable populations, significantly fewer reported neither protection after 1997. The findings were similar for articles that contained no references to prior articles on the same study. In both cases, however, each type of article comprised a large proportion of all articles that mentioned neither ethical protection, even after 1997.

Of articles that reported informed consent, the majority provided further description of the type that was obtained (Table 3). Before and after 1997, informed consent was stated to have been given by someone other than the patient in 27 (9%) and 44 (15%) of all trials, respectively (P = .03).

In most articles, the number of ethics committees closely corresponded to the number of study sites. From before to after 1997, the mean number of ethics committees per article increased from 6.7 (SD, 13.6; range, 1-135) to 7.2 (16.4; 1-105), and the 75th percentile from 3 to 4. In contrast, for articles that met the strict definition of ethical approval, the rate of naming of ethics committees significantly decreased, from 155 of 198 articles (78%) to 156 of 234 (67%) (P = .007), possibly because articles with more than 1 ethics committee had roughly an inverse relationship between the number of ethics committees and the percentage that was named. For example, in articles for which the number of ethics committees could be determined, those that listed 3 or fewer were significantly more likely to name them than were articles with more than 3, with respective rates of 126 of 142 (89%) vs 27 of 41 (66%) before 1997 (P<.001) and 122 of 147 (83%) vs 31 of 63 (49%) after 1997 (P<.001).

Regarding study sponsorship, journals consistently improved their reporting of protections regardless of the type of funding (Table 4). Results were very similar for subgroup analyses performed by journal.

Our findings are consistent with the results of prior studies. We found that roughly 30% of articles published before 1997 did not report informed consent, and another 30% did not report ethics committee approval. Olson and Jobe⁵ reported that 70% of resuscitation articles published from 1966 to 1994 did not mention informed consent, and 50% did not mention ethical approval. Rikkert et al⁶ found that 40% of gerontology trials published from 1993 to 1994 failed to describe informed consent and 50% ethics committee approval. Matot et al¹⁹ reported that 40% of critical care articles from 1994 lacked a statement on informed consent, and 40% had no statement on ethical review. In articles from 1993 to 1994 in general medical journals, Ruiz-Canela and Gomez-Gracia²⁰ found no mention of informed consent in 20% and ethical approval in 30%. Finally, Bauchner and Sharfstein²¹ reported that only 3% of randomized controlled trials published on children's health in 1999 reported ethical protection compared with the 5% rate we found in studies of vulnerable populations published after 1997.

Our study had limited power to detect differences in reporting for individual journals or other subsets of articles. For example, in Table 1, BMJ experienced an absolute difference in its rate of reporting of ethical approval of 13% (relative change of 33%) but the finding was not significant.

We cannot comment on whether similar changes in reporting have occurred in other types of journals. We also do not know why the improvements have occurred. Finally, we do not know whether patients have benefited from having informed consent or ethical approval as safeguards, whether the ethical protections that are reported are actually being performed (or vice versa), or whether describing them in articles helps at all. However, as we have stated, transparency in the reporting of science has inherent benefits.⁸ The reporting of informed consent and ethics comittee approval attests publicly—in a forum where the methods can be openly challenged or discussed—to a minimum level of ethical consideration. Certainly, there is a moral imperative to assure participants in trials that meticulous attention is being paid to their safety.

Patients have died in studies that failed to adequately provide protection.^22-24 News stories, not surprisingly, have questioned the ethical conduct of medical research.^22-24 These events "have shaken the public's confidence in our ability to govern ourselves."²⁵ In describing the federal government's response, Shalala²⁶ stated "Clinical researchers and the institutions that support them must, without exception, maintain the public's confidence in our work, our competence, and most important, our ethics."

To this end, we recommend that journals publish in their instructions for authors the explicit requirement that all articles describe informed consent and ethics committee approval or why these were waived.⁸ Journals should also assess in-house practices for ways that might further improve reporting. For their part, authors should provide as many details as are necessary to accurately describe the implementation of these safeguards.