PHREG test for trend

[spatsassy]

Hi there, I am running a PHREG model with mortality as the outcome and quartiles of protein X as my predictor (where quartile 1= ref). I was able to get the HR for quartile 2, 3, and 4, and their respective p-values- which are not significant (0.8, 0.4, 0.2) and now wondering if there is a way to test for a trend in the HRs?? Any suggestions/tips for the analysis would be appreciated!

[StatDave · Accepted solution]

Assuming that you have the original continuous values of your predictor, use that variable itself in your MODEL statement instead of your quartile variables and do not include it in the CLASS statement. The test of its parameter is a test of the linear effect of the predictor.