Registration, reporting, and replication in clinical trials: The case of anorexia nervosa

Treatment outcomes for anorexia nervosa (AN) remain modest, and recent research suggests that clinical trials may be of limited methodological quality. With increasing evidence illustrating the irreproducibility of psychological research, no research to date has systematically examined the cumulative effect of bias in research relating to the treatment of AN.

We identified all AN trials listed on ClinicalTrials.gov between 2000 and 2018 and examined rates of (a) the noncompletion of clinical trials, the (b) nonpublication of trials once listed as completed, (c) the nonprospective registration of clinical trials, and (d) the nonreplication of findings.

We note that of 201 trials listed on ClinicalTrials.gov, only 101 have been completed, and of those, only 41 have been published. Moreover, of these 41 published trials, only eight demonstrated evidence of prospective trial registration, and only seven have had their primary findings replicated in other studies.

These results illustrate the profound cumulative effect of methodological bias in registered trials for AN, which may have a significant impact both on what appears in the current evidence base, and on the reproducibility of studies comprising this evidence base.