Working out risk factors for a medical complication?

#1
Hi All - HELP!

I have a large dataset on women in labour, and I'm interested in finding out risk factors for a particular complication of labour. The complication is either there, or not there.

I also have lots of information on possible risk factors for the complication, e.g.: length of labour, birth weight of baby and number of previous babies, together with things like ethnicity.

Q: How can I find out if there are particular risk factors for the complication?

(?Possible A): I was planning to do logistic regression in JMP, to calculate Odds ratios. Is this the right approach?
Also: would I need to break data down before analysing? (eg split into ranges of birth weights, such as 3,000g-4,000g, >4,000g). Or can I just compare all the birth weights for those with the complication and those without, and get a sensible answer?

After staring at JMP for several hours I'm now broken - so any help would be much appreciated!

:confused:
 
#2
If you perform logistic regression you do not necessarily have to categorize your independent variables. If your independent variables are continuous you can include them in the model as continuous variables, and the odds ratio calculated will be the odds for a 1 unit increase in the variable (afterwords, you could calculate the odds ratio for a specific unit increase). However, sometimes independent variables are categorized if the categories have clinical relevance; for example, serum cholesterol levels, as this allows you to compare your results with previous studies or guidelines.