Some comments:

1) I would be interested in using data from our five nursing homes to help with this study but the window of exposure is to long. Our dataset is only 4 years total and so we would have almost no one who meets the inclusion criteria.

2) I have concerns about the definition of treatment pathway since many of the antidepressant drugs have secondary indications (e.g., bupropion and smoking cessation, trazodone for sleep). It might make sense to distinguish in the analysis apparent treatment switches (e.g., sertraline to escitalopram) from treatment suplementation (e.g., sertraline for 6 weeks and then sertraline plus venlafaxine).

3) The definition of the condition is interesting. Does the condition need to appear more than once in the patient records during the observation period? Not sure what level of granularity the analysis would examine for treatment course.