Hello, I am conducting a retrospective cohort study using the General Practice Research Database. I have 17,000 exposed patients matched by age and gender to 85,000 non-exposed subjects. The outcome is time to malignancy which I want to relate to exposure which is recorded as a simple binary value. (i) Can Cox regression adjust for the matching?Should I account for the matching by bringing in a variable representing the matched subjects and use it as a STRATA variable in SPSS? Or do I just perform the standard regression to generate the hazard ratios? Dr John Belcher Department of Primary Care and Public Health North Wales Clinical School Cardiff University Gwenfro Unit 6/7 Wrexham Technology Park Wrexham LL13 7YP Tel +44 (0)1978 726074