Paperonium

Pregnancy · Sci Adv · 2023

Summary: a preeclampsia test’s “yes” is right 55% of the time. The paper’s own counts: 86%.

The 123 pregnancies testedThe summary55%86%Its own spreadsheet: 66 of 77impossibletoo low30%40506070809056worst case: a false “yes” for all 52
The 123 pregnancies testedThe summary55%86%Its ownspreadsheet66 of 77impossibletoo low30%4070809056worst case: a false “yes” for all 52

When the test said “yes”, how often it was right

The paper says

A test for preeclampsia, a blood-pressure illness of pregnancy, was tried on 123 pregnancies, 71 with it and 52 without. The paper’s opening summary says it caught 93% of the 71, and that 55% of its “yes” results were right: the rest were false alarms.

Its own data show

The paper ties the 55% to these 123 pregnancies, where it is impossible. Catching 93% means 66 of the 71; 55% right would need 120 yeses, 66 real and 54 false, but only 52 pregnancies didn’t have preeclampsia. The spreadsheet published with the paper lists 11 false alarms: 66 of 77 yeses were right, 86%.

A test can catch most cases and still raise many false alarms. The paper says another preeclampsia test’s main weakness is that only 30 to 65% of its yeses are right. At 55% this test would be no better; at 86% it looks better, though the share runs higher when most of those tested are ill, as here.

“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 →

Discovery and verification of extracellular microRNA biomarkers for diagnostic and prognostic assessment of preeclampsia at triage · Sci Adv (2023) · doi:10.1126/sciadv.adg7545