Surgical care · JAMA Network Open · 2023
A blood-sugar result “for patients with diabetes” counts 17,491 operations; at most 11,600 involved such patients.
Operations in the blood-sugar result, against its own patient table
The paper says
In a trial of remote advice to anesthesia staff during surgery, a main result counts only patients with diabetes (“patients without diabetes were excluded”): 85.8% of operations avoided high blood sugar afterwards, with or without advice. The paper calls this no effect, “with high confidence”.
Its own data show
That result counts 17,491 operations: exactly the number with any blood-sugar test in the paper’s chart of who was analyzed. Its own patient table, which counts operations the same way, allows at most 11,600 on people with diabetes, even counting the 5,867 whose diabetes status is unknown as diabetic. Only 5,733 are recorded as diabetic.
A percentage describes only the group it counts. At least a third of these operations (17,491 − 11,600 = 5,891) were on people without diabetes, probably two-thirds. So 85.8% is not the figure for diabetic patients; theirs isn’t in the paper. Rough calculations suggest “no effect” may hold, but with a margin of error 2 to 3.5 times as wide.