I have found some scientific papers in which a spline regression is performed with a categorical dependent (death) and a continuous predictor. In the output graph, the dependent is measured as an Hazard Ratio, like this (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1626441/pdf/ehp0114-001538.pdf, figure 1). Does anyone know if this is possible to do with SPSS? I know how to do a spline regression with SPSS but I have always worked with continuous dependent variables, I would not know how to obtain single HRs for all the subjects.

They used a Cox-proportional hazards model. It's a regression model that assumes a constant relative hazard over time. Try looking it up. It should be in almost all statistical packages since its presently established as the preferred method for modeling mortality rates.

Thanks for your reply, but I was rather wondering how to obtain a similar graph and if it is possible at all to get it with SPSS. As far as I can see, the Hazard Ratio is the dependent variable.