Failure to report ethical approval in child health research: review of published papers

Methods and results

We reviewed all research reports related to child health (involving patients <18 years, pregnant women, or clinicians providing care to children) published in 1999 in five American journals (JAMA, New England Journal of Medicine, Pediatrics, Journal of Pediatrics, Archives of Pediatrics and Adolescent Medicine). Each report was reviewed independently by two research assistants for any statement about informed consent or ethics committee approval; if either statement was present, the report was categorised as having ethical approval. An article that referred to another publication for more details was not reviewed. This approach has been used previously.2 All five journals require studies with human participants to report ethical approval. Reports not describing approval were reviewed again by an investigator to ensure that errors in categorisation of the studies had not been made. Studies were also categorised according to design. The ethics committee at Boston University School of Medicine judged that this project did not represent research involving human subjects and did not require review.

The journals contained 561 research reports related to child health. Overall, 340 (61%, range 53-66%) publications had ethical approval (table). The proportion of studies reporting ethical approval was dependent on the study type.

Comment

Despite increasing concern about the protection of human subjects, 40% of papers did not report ethical approval. These findings are particularly unexpected, since all five journals explicitly ask authors to document approval—24% of medical journals do not include such instructions.3 Although we reviewed only American publications, we are not aware of data showing that papers published in non-American journals report ethical approval more frequently.

There are three possible explanations for the low proportion of papers that do not report ethical approval. Approval may have been obtained, but not reported—this situation is likely to be rare and is inconsistent with instructions to authors. Random non-reporting cannot explain differences between study types.

Investigators may have failed to obtain and report ethical approval because of confusion about the requirements. Nearly all investigators obtained approval for randomised controlled trials, a study design widely recognised to require safeguards. However, researchers may not realise that ethics committee approval and informed consent may be required for studies that do not enrol participants or that use large datasets. Some of the prospective cohort studies without approval could have been considered to be quality improvement projects (audits), and there is no consensus about when such projects become research. The Institute of Medicine recently said that the project represents research and approval is necessary if the investigator intends to publish the findings.4 Some researchers may have deliberately disregarded ethical approval standards and laws. Such occurrences are likely to be rare, but they are not inconceivable.

Unless we enhance our system of safeguards, an unethical study could be published. While the primary responsibility for assuring ethical conduct of research rests with investigators, peer review journals should be more active in protecting human subjects.

We have four recommendations to prevent unethical research being published. Just as every article has a list of authors, every research study should include a statement regarding human subjects. Articles should not refer to other publications for information regarding ethical approval. If the investigators believe that the research does not need to be reviewed by the ethics committee, the reason for exemption should be provided (however, we believe that investigators should not make such a decision). Investigators should document approval from the ethics committee and whether informed consent was obtained—as discussed previously,5 these are not synonymous.

Protecting participants in clinical research is, and must remain, one of the highest priorities of medicine. As the final conduit of most biomedical research, peer review journals can play a greater role in protecting human participants by ensuring that every report of clinical research contains a statement about the participants' protection.