Paperonium

Cancer trials · JAMA Network Open · 2022

Paper: trials had too few Asian or Pacific Islander patients. Its own count shows the opposite.

Trial patients, 4.5%The paper’s share of cases5.9%, trials fall shortIts own count of cases3.0%, trials have more0%1%2%3%4%5%6%
Trial patients, 4.5%The paper’sshare of cases5.9%,trials fall shortIts own countof cases3.0%,trials have more0%2%4%6%

Share Asian or Pacific Islander, 2000 to 2018: trial patients vs new cancer cases

The paper says

Of 8,309 patients in early US cancer-drug trials (2000–2018), 4.5% were Asian or Pacific Islander. The paper says 5.9% of new cancer cases in a North American registry were, so trials fell short by 1.5 percentage points, a gap chance alone would rarely produce.

Its own data show

In that same sentence the paper prints the registry’s count beside the 5.9%: 145,568 of 4,891,486 cases. That is 3.0%. So trials, at 4.5%, had a larger share, by 1.5 points: the same gap, the other way round. The count is exactly the paper’s 2000–2011 (2.7%) and 2012–2018 (3.4%) counts added together.

Pool two groups and the combined share always lands between the two: a 2.7% group and a 3.4% group cannot make 5.9%. So 5.9% cannot be right, yet the paper’s conclusions still list too few Asian or Pacific Islander patients overall. Its main finding, too few Black patients, does not depend on this error.

“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 →

Participation of Patients From Racial and Ethnic Minority Groups in Phase 1 Early Cancer Drug Development Trials in the US, 2000-2018 · JAMA Network Open (2022) · doi:10.1001/jamanetworkopen.2022.39884