Paperonium

Diabetes and heart · Circulation · 2018

The summary’s type 1 diabetes death figures are really type 2’s, and point the wrong way

Men · the summary’s “type 1”type 2’s figureMen · results for type 2, identicalMen · results for type 1nearly doubleWomen · the summary’s“type 1”0.98×: type 2’s figureWomen · results fortype 2, identicalWomen · results for type 11.31×: 31% higher0.96×0.951.91×1.682.180.971.051.650.8×1×1.2×1.5×2×2.5×same as no diabetes; over 1× = more deaths
Men · thesummary’s“type 1”type 2’s figureMen · results fortype 2, identicalMen · results fortype 1nearly doubleWomen · thesummary’s“type 1”0.98×:type 2’s figureWomen · results fortype 2, identicalWomen · resultsfor type 11.31×: 31% higher0.96×0.951.91×1.682.180.971.051.650.7×1×2×3×same as no diabetes;over 1× = more deaths

Odds of dying within 30 days of a heart-failure admission, compared with no diabetes

The paper says

The paper’s summary says that, of people admitted to hospital with heart failure for the first time, more with type 1 diabetes die within 30 days than without diabetes. Its figures: men 0.96 times the odds of dying, women 0.98 times.

Its own data show

In its results, those figures and the margins of error beside them are, digit for digit, its type 2 diabetes results; it calls type 2 similar to no diabetes. Its type 1 results: 1.91 times the odds for men, 1.31 for women. The model figures in its online appendix agree: about 1.91 and 1.31.

Each figure compares the odds of dying (the chance of dying divided by the chance of surviving) with those of people without diabetes; below 1 means fewer deaths. So the summary’s words are right but its numbers say the opposite, and anyone reading only the summary never sees the real type 1 figures, a finding the paper calls new.

“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 →

Incidence of Hospitalization for Heart Failure and Case-Fatality Among 3.25 Million People With and Without Diabetes Mellitus · Circulation (2018) · doi:10.1161/CIRCULATIONAHA.118.034986