Prediabetes is one of the commonest reasons people start looking at these drugs, and “high risk” is the phrase that prompts it. A post hoc analysis of a three-year trial put numbers on what that phrase predicts [1], and the answer is less than most people assume. Who is reaching these drugs at all is in who the price filtered out.
Three criteria were assessed against what actually happened in the placebo group. A one-hour glucose reading of 155 mg/dL or above caught 83% of those who developed diabetes — but flagged so many people that only 15% of those it identified went on to develop it. Fasting glucose was the opposite: it missed half the cases while being right about 37% of the people it flagged. HbA1c fell between, at 19%.
The more useful property is the one nobody quotes. All three tests had negative predictive values between 0.90 and 0.92, meaning a reassuring result is fairly reliable. These are instruments for ruling out rather than ruling in, and that asymmetry is worth understanding before treating a screening result as a reason to buy anything.
Tirzepatide did prevent diabetes in this trial, and substantially: new-onset cases fell by 90% to 99% in the high-risk groups and 68% to 91% in standard-risk groups, depending on the criterion used. Those reductions are relative, though, and in a group where only 15% would have developed diabetes anyway, a 90% relative reduction prevents far fewer cases than the number suggests — the arithmetic set out in the falls and fractures cohort.
One technical point separates two figures that look comparable. The predictive values were calculated in the placebo group, so they describe what happens without treatment. The prevention percentages come from the treated arms. They answer different questions, and a seller quoting both in one sentence would be blending natural history with drug effect. What prevention is worth against what it costs is in is a GLP-1 worth what it costs and what a GLP-1 actually costs.