Paperonium

Breast cancer · JAMA Netw Open · 2022

Cancers between scans: the summary says 3.73, range 0.41 to 2.73. Ranges must contain their estimates.

The summaryoutside its own rangeIts own chart (Figure 3)the same 3.733.730.41 to 2.73should be inside1.87.80 per 1,000 scans2468
The summaryoutside itsown rangeIts own chart(Figure 3)the same 3.733.730.41 to 2.73should be inside1.87.802468per 1,000 scans

Cancers found between scans after breast rebuilding, per 1,000 scans

The paper says

For patients whose breast was removed and rebuilt, the paper’s opening summary gives the rate of cancers found between scans (after a normal scan, before the next): 3.73 per 1,000 scans, with a range of plausible values of 0.41 to 2.73.

Its own data show

Measured on the paper’s own chart (its Figure 3), this rate is 3.73 with a range of 1.8 to 7.8: so 3.73 stands and its range does not. The printed 0.41 to 2.73 is, digit for digit, the range of a different rate in the same patients: cancers that could not be felt.

A range shows where the true rate plausibly lies, so it must contain the estimate. This rate is the second question the paper set out to answer. Anyone quoting the summary, or combining it with other studies, takes a range that belongs to another rate. The main conclusion, built on rates of cancers that could not be felt, is unaffected.

“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 →

Yield of Surveillance Imaging After Mastectomy With or Without Reconstruction for Patients With Prior Breast Cancer · JAMA Netw Open (2022) · doi:10.1001/jamanetworkopen.2022.44212