Stomach cancer · Clin Cancer Res · 2025
11 of 16 patients responded, but the paper’s range of likely rates is calculated for 13.
Percent of patients who responded (dot) and the range of likely rates (bar)
The paper says
A 16-patient stomach-cancer trial’s main result: tumors shrank enough to count as a response in 11 patients (68.8%). With so few patients, the paper adds a range of likely response rates for patients like these, calculated from that count: 54.3% to 95.9%.
Its own data show
That range matches, to the last digit, a standard calculation for 13 of 16. The same calculation for 11 gives 41.3% to 89.0%, and other standard ones start at 46% or lower. Eleven is right: the results table counts 11 responses, 3 steady and 2 growing tumors; only the line beside the range says 13.
With few patients, luck can push a result up or down; the range shows how far. The trial’s plan said a rate under 40% would suggest the treatment isn’t working. The correct range starts at 41.3%, just above that line, not at 54.3%, well clear of it. Allowing for luck, the treatment still beats that line, but only just.