Abstract
<title>Abstract</title> <p> Context Every medical student in India receives the Attitude, Ethics and Communication (AETCOM) curriculum, a professionalism programme mandated nationally by the National Medical Commission. Its evaluation literature is almost uniformly favourable. Setting aside whether AETCOM works, we asked a prior question: is this literature capable of answering it? Methods We synthesised empirical AETCOM evaluations of Indian undergraduates and interns published from 2015, searching seven databases, the Directory of Open Access Journals, a reproducible ISSN-level sweep of fourteen journals, and citation chains. Three analyses were specified in advance: coding the design and measurement features of every study contributing an effect, then estimating effects within design strata; adapting a spin taxonomy from the trials literature, coding what each report claimed before its methods were read; and mapping every report onto the inferences of an argument-based validity framework, overlaid with transfer-of-training theory. Quality was appraised using MERSQI, the JBI checklist, ROBINS-I and COSMIN. Results Effect magnitude tracked evaluation design rather than curriculum content. Standardised mean change fell from <italic>g</italic> = 3.49 (k = 4) where teaching staff rated performance without a comparator, through <italic>g</italic> = 1.90 (k = 7) for uncontrolled knowledge tests, to <italic>g</italic> = 0.33 and <italic>g</italic> = 0.12 in the only two studies comparing students who received AETCOM with students who did not. Among 53 reports with adequate full text, 40 (75%) contained at least one form of spin, 31 (58%) attributed change to the curriculum without any comparator, and 20 (38%) claimed a higher outcome level than they measured, even under a deliberately conservative rule. Evidence and claims were mirror images: 42% of reports supplied evidence about scoring, 5% about generalisation, 4% about extrapolation to practice and none about consequences, while 51%, 16% and 21% made claims at the latter three. In all, 49 of 55 reports (89%) claimed more than they supported. Transfer climate, supervisory support and opportunity to perform were measured in 5%, 5% and 12%. MERSQI averaged 11.3 of 18, seven of twelve studies reported no validity evidence for their outcome instrument, and both curriculum-exposure comparisons carried a critical risk of bias. Conclusions This evidence base is not merely thin but inverted: confident where it has least to say, and largest in effect where its designs are most permissive. Missing evidence about practice does not show that AETCOM fails, because no evaluation has been built that could detect transfer or measure the conditions on which transfer depends. We set out a phased evaluation architecture in which each component repairs a specific break in the argument. Clinical trial registration Not applicable. This is a synthesis of published research and not a clinical trial or any other form of prospective interventional study, so no trial registration was required. The review protocol itself was registered retrospectively, as described in Methods. </p>