take it with a grain of salt.

- It's a retrospective cohort analysis, so it's inherently subject to selection bias.

- They exclusively used diagnostic codes for data collection. Docs tack on psych diagnostic codes willy-nilly.

- They excluded patients with a diagnosis of any psychiatric diseases within one year before or one month after the start of glp1-i. But they didn't screen for patients who were new to the healthcare system (i.e. hadn't had a chance to be diagnosed)

- They didn't assess if the side effects of the drugs contribute to the occurrence of depressive episodes.

- There is no data on change in BMI during the follow-up period--if the drug isn't working for a patient, they're more likely to become depressed

- Lastly, the study lacks data for patient compliance with the treatments. People prescribed a drug that should help them but find themselves unable to take it are at higher risk for depression / lack of hope