Yeah, lots of risk for confounders and selection bias in a study like this. If you take the huge number of patients with a history of GI bleeding (for which PP

Yeah, lots of risk for confounders and selection bias in a study like this. If you take the huge number of patients with a history of GI bleeding (for which PPI therapy is well-proven), those are patients who already have an elevated all-cause mortality risk. They also selected patients who had at least one creatinine measure in their EHR; so probably people who are more likely to have background renal disease. And using US veterans makes it hard to generalize to the greater population. However, they also saw elevated risk in patients who didn't have a de-facto indication for PPIs (GERD, GI bleeding, etc), so maybe there is something to this. Honestly, it's kind of an amateur study design. I think the conclusion to draw here is we need a large randomized control study to look at this issue. Regardless, as a physician, I agree with the study authors that if a patient doesn't have a specific indication for PPI, they shouldn't be on one. Then again, I don't need a study to tell me that; it's my feeling about ALL medications.