I think gem1939 has it right about what to look at in studies with large numbers of participants studied prospectively - and I would add in family history, both socioeconomic/educational and medical - and lifestyle including diet and smoking history.