Reporting Randomized, Controlled Trials Of Herbal Interventions: An Elaborated CONSORT Statement

Randomized, controlled trials (RCTs) of herbal interventions often inadequately describe important aspects of their methods (1-4). Although the quality of reporting of these trials may be improving with time, many still lack important information, particularly about the composition of the herbal intervention (4, 5). Crude herbal drugs are natural products and their chemical composition varies depending on several factors, such as geographic source of the plant material, climate in which it was grown, and time of harvest. Commercially available herbal medicinal products also vary in their content and concentration of chemical constituents from batch to batch and when products containing the same herbal ingredient are compared among manufacturers (6-14). Even when herbal products are standardized for content of known active or marker compounds to achieve more consistent pharmaceutical quality, there is variation in the concentrations of other constituents. These variations can result in differences in pharmacologic activity in vitro (15) and in bioavailability in humans (16). Mindful of these issues, we elaborated on the 22-item checklist of the Consolidated Standards of Reporting Trials (CONSORT) statement (17) to help authors and editors improve reporting of RCTs of herbal interventions. Methods We developed these reporting recommendations in 3 phases that included premeeting item generation, a consensus meeting, and postmeeting feedback. The individuals who participated are listed in the Appendix. To generate items, 1 investigator conducted telephone interviews of 16 participants with expertise in the method and reporting of RCTs (5 participants), pharmacognosy (4 participants), herbal medicinal products (5 participants), medical statistics (1 participant), and herbal product manufacturing (1 participant). The investigator asked participants to suggest revisions to existing CONSORT checklist items and also to additional items required for reporting trials of herbal interventions. He asked participants to nominate revisions or new items on the basis of empirical evidence that not reporting the item would bias estimates of treatment effect. When no empirical evidence was available, commonsense reasoning was acceptable. After completing all telephone calls, the investigator thematically grouped items and circulated them by e-mail to each participant for review. Fourteen participants attended the consensus meeting. The meeting began with a review of the premeeting checklist item suggestions. We emphasized minimizing item elaborations and additions and basing elaborations on evidence whenever possible. Each item suggestion was presented and followed by debate for its inclusion, deletion, or modification. This process was repeated until all items were reviewed and a consensus emerged. After the consensus meeting, we circulated a draft summary report to all participants to ensure that it accurately represented decisions made during the consensus meeting. We then circulated the report to the wider CONSORT Group for input and revised it on the basis of their suggestions. Ethical approval was obtained from The University of Toronto Health Sciences Ethics Review Committee on 23 January 2004. Financial support for the consensus meeting was provided by the Canadian Institutes of Health Research. The funding body had no role in the design, conduct, or analysis of this study and did not influence the decision to submit the manuscript for publication. All researchers are independent of the funders. Results The group did not recommend any new CONSORT checklist items or modifications in the CONSORT flow diagram. We did, however, elaborate on 9 of the 22 CONSORT checklist items to enhance their relevance to trials of herbal interventions (Table, Figure; Appendix Table), including minor recommendations for 8 items (item 1 [title and abstract], item 2 [background], item 3 [participants], item 6 [outcomes], item 15 [baseline data], item 20 [interpretation], item 21 [generalizability], and item 22 [overall evidence]) and detailed recommendations for 1 item (item 4 [interventions]). Table. Proposed Elaboration of CONSORT Checklist Item 4 for Reporting Randomized, Controlled Trials of Herbal Medicine Interventions Figure. The high-pressure liquid chromatography chemical fingerprint for the extract of Ginkgo biloba L Appendix Table. Proposed Elaborations of CONSORT Items for Randomized, Controlled Trials of Herbal Medicine Interventions The Table shows the detailed recommendations for item 4 and an example of good reporting related to each recommendation. These recommendations begin with the words where applicable to indicate that all information suggested may not be applicable to every type of herbal medicine intervention. For example, an herbal medicinal product comprising crude herbal material (for example, leaves and stems) simply prepared as a tea or decoction does not require description of the type and concentration of solvent used and the ratio of herbal drug to extract (item 4B.3). Also, not every herbal medicine intervention will have a finished product or extract name or manufacturer (item 4A.2), but instead may be made by the investigators specifically for the study. In such circumstances, all methods used in preparing and formulating the product must be reported. Similarly, item 4F is not required for herbal interventions when the practitioner is not a part of the intervention. With these exceptions, we recommend that all information shown in the Table be reported for all herbal interventions. Discussion We developed recommendations to be used in conjunction with the existing CONSORT checklist when reporting RCTs of herbal interventions. In particular, we thought it imperative that reports of RCTs provide clear and complete descriptions of the herbal intervention. We think that our recommendations might also be relevant for reporting herbal interventions in other research designs, whether preclinical (for example, in vivo or in vitro) or clinical (for example, N of 1 trials), and refer interested readers to a detailed explanatory document that further describes each of our recommendations and provides additional examples of good reporting (22). We hope that authors find our recommendations instructive and that journals will endorse their use and modify their instructions to authors accordingly.