Citation and adherence to TRIPOD guidelines by published radiological prognostic models: Systematic review.

Abstract

Objectives: We aimed to establish the extent to which prognostic models published in highly indexed radiological journals cite and adhere to generally accepted reporting guidelines.

Methods: We identified articles reporting multivariable prognostic models, developed using regression and published in the top 3 indexed general radiological journals, December 2022 to May 2023 inclusive. We determined whether they cited the generally accepted reporting guideline, TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis). We scored adherence to individual TRIPOD domains to determine reporting quality both overall and for specific areas.

Results: We included 140 articles. Only 4% (n = 6) cited TRIPOD, with just one including the checklist. TRIPOD adherence was poor overall, with a median score of 57% (inter-quartile range, IQR 48% to 64%, range 30% to 87%). Individual domains particularly poorly reported were, title (2% adherence), abstract (3%), and statistical analysis (5%). Only 38% articles (n = 53) named a statistician author. Only one journal mentioned TRIPOD guidelines in their "Instructions For Authors" but did not mandate checklist submission.

Conclusions: The large majority of prognostic models published in highly indexed radiological journals did not cite TRIPOD, nor fulfil its recommendations.

Advances in knowledge: Authors should adhere to the TRIPOD statement so that their work is reported with sufficient clarity, and radiological journals should stipulate adherence for authors submitting prognostic models for publication.