Network geometry shows evidence sequestration for medical vs. surgical practices: treatments for basal cell carcinoma

Basal cell carcinoma (BCC) is the most common cancer with 2 million treatments per year with little evidence-based guidelines for treatment. There are three classes of interventions (surgical, destructive, and topical) for BCC, and this study aimed to determine whether there are preferences or avoidances in comparisons of different types of treatments for BCC in randomized controlled trials (RCTs).

PubMed, Cochrane Central Registry of Clinical Trials, and ClinicalTrials.Gov were used to identify eligible published and registered ongoing RCTs.

Fifty-five trials (42 published and 13 registered trials) were identified. Only one unpublished registered trial compared a topical vs. a surgical intervention, and only one trial compared a topical vs. a destructive intervention. Conversely, 44 of the 55 trials compared interventions within the same treatment class and 9 of 55 trials compared surgical vs. destructive interventions. In most trials, selection of same-class comparators was not necessitated by the type of BCC lesions (nonaggressive superficial or nodular vs. aggressive, infiltrative, morpheic BCCs, P = 0.155) or their location (face vs. nonfacial, P = 0.137).

This is the first time that an evaluation of network geometry is applied to address issues of comparisons between different families of interventions that belong to different specialties and practices (medical vs. surgical). Previous evaluations of homophily have addressed different families of interventions, in which all interventions are medical (drugs) and performed in the same health-care settings. The noncommunicating bodies of evidence between medical and surgical interventions that we document highlight a problem of unnecessary sequestration of the evidence and the corresponding health-care practices.