Testing a drug's effect on survival across multiple distinct studies

[user529]

I have survival and other data for between ~1000 patients from different studies with indications varying from organ transplantation to cancer. Patients within this cohort will have been drawn from many different studies in which the objective - long-term graft survival, overall survival rate, metatstasis free survival, organ acute rejection rate, graft loss, etc. - was different. I want to essentially control for all the variables of disease (cancer, organ disease and severity), sex, age, etc. to find out the effect of one explanatory variable - use of drug X as opposed to drugs Y, Z, etc. - on survival/time to death of patients. In the case of organ transplantation, drug X is used as an immunosuppressant: it's primary objective is not to treat disease as such. In the case of the cancer studies however, drug X has a curative objective.

How can I test drug X's effect on only survival within this cohort?

 

[Valkarie]

One approach could be to use a Cox proportional hazard regression model to measure the effect of drug X on survival. This type of model can account for various confounding factors such as age, sex, underlying disease, and other variables. Additionally, this type of model can control for the different objectives of the studies within the cohort by using stratified analyses. Stratified analyses involve the creation of separate models for each type of study objective, allowing for a more accurate assessment of the effect of drug X on survival.