BJR|Artificial Intelligence

Vollständiger Abstract

Worum geht es in dieser Arbeit?

<h4>Purpose</h4>In radiotherapy for bleeding gastric cancer, no standard response definition exists; studies have used transfusion-only or transfusion-plus-hemoglobin criteria. After inter-facility transfer or during home care-sometimes following palliative radiotherapy-laboratory results from other facilities are often unavailable, leaving patients inevaluable. We therefore tested whether omitting hemoglobin compromises the precision of response assessment.<h4>Methods and materials</h4>In criteria A, referring to our original study, patients were diagnosed as responders when all the following three conditions were met: (1) hemoglobin levels ≥ 8.0 g/dL; (2) 14 consecutive days without blood transfusion before blood sampling; and (3) no salvage treatment. In criteria B, patients were diagnosed as responders when only (2) and (3) of the above conditions were met. The strength of agreement between criteria A and B was estimated by calculating Gwet's first-order agreement coefficient (AC1) at 4- and 8-week follow-up for patients completing planned radiotherapy with assessable response.<h4>Results</h4>In 36 evaluable patients at 4-week follow-up, response rates were 69% (25/36) by criteria A vs. 78% (28/36) by criteria B (AC1, 0.86; almost perfect agreement). In 30 evaluable patients at 8-week follow-up, response rates were 83% (25/30) by criteria A vs. 90% (27/30) by criteria B (AC1, 0.91; almost perfect agreement).<h4>Conclusions</h4>Criteria A and B showed almost perfect agreement at 4- and 8-week assessments among evaluable patients with available hemoglobin data.<h4>Advances in knowledge</h4>Among patients with available hemoglobin data, criteria A and B showed almost perfect agreement. The observed agreement may not be fully generalizable to patients without hemoglobin assessment.

Abstract: PubMed · Datensatz