spssforum.com

Hi,
I did a study looking at 60 people of which 40 were exposed to medicine A and 20 to medicine B. Of the 10 people who ultimately developed a complication, all the 10 were exposed to A. Thus on the cross tabs with complication (yes/no) in the row and drug (A/B) in the column, I had a significant value of 0.10. Keep in mind that one of the cell values in the 2X2 table is zero (patients developing the complication had zero exposure to drug B).
When I do the binomial logistic regression with complication (yes/no) as the dependent varialbe and drug (A/B) as one of the independent variables, I get a very absurd p value of 0.998. I dont think this is right and is because of the presence of zero in one of the 2X2 cells. How can I go about correcting for this? I just put one more independent variable in the model that was not significant again but had more decent p value of 0.75. Any inputs will be appreciated, VA.