Paperonium

Oral cancer · Cell Rep Med · 2024

A mouth-cancer test’s table counts 9 people twice: it clears 78.9% of cancer-free people, not 82.6%.

82.6%The results table: cleared, of 46 cancer-free people38cleared8false alarms, counted twice78.9%Its own patient list: cleared, of 38 cancer-free people30cleared8false alarms
82.6%The results table: cleared, of46 cancer-free people38cleared8false alarms, counted twice78.9%Its own patient list:cleared, of 38cancer-free people30cleared8false alarms

Everyone without cancer, and how many the test rightly told “not cancer”

The paper says

The paper’s summary says a mouth-cancer test caught 98.2% of cancers and rightly told 82.6% of people without cancer “not cancer”. Its table, said to cover 92 people with a spot in the mouth: 54 caught, 1 missed, 38 told “not cancer”, 8 false alarms.

Its own data show

Those four counts add up to 101, but the paper says the table covers 92 people: 54 with cancer and 38 without. Its 54 caught and 38 told “not cancer” are the full groups, so the 1 miss and 8 false alarms are counted twice. Its patient-by-patient list: 53 caught, 30 told “not cancer”.

The 82.6% is a share: of everyone without cancer, how many the test rightly told “not cancer”. Its own counts give 30 of 38 (78.9%); the table’s doubled count gives 38 of 46 (82.6%). The 98.2% barely moves (98.1%). So the paper’s estimate of needless biopsies (tissue removed for testing) rises from 17.4% to 21.1% of people without cancer.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

Beta-defensin index: A functional biomarker for oral cancer detection · Cell Rep Med (2024) · doi:10.1016/j.xcrm.2024.101447