I'm skeptical of the benefits claimed for non-diabetics: the 15% difference in median survival time between metformin-using diabetics and control "non-diabetics

I'm skeptical of the benefits claimed for non-diabetics: the 15% difference in median survival time between metformin-using diabetics and control "non-diabetics" may just indicate undiagnosed diabetes in the control population. The abstract doesn't say how the controls were selected, whether they were verified to not have diabetes or were simply people with no diabetes diagnosis.