Suppose you want to look at the risk of glaucoma in 7 groups (e.g. grouped by weight: 100-110 lbs, 111-120 lbs, etc..). Is risk measured by the proportion of people that have glaucoma? Or should you look at excess risk/relative risk? I guess it shouldn't matter what you look at because the risk pattern would be the same?
I don't know if this is the right site to ask, as some domain knowledge about medical situations may be needed, but let me give it a try:
Classically, risk involves both the probability of a outcome and the severity of a bad outcome. In your situation, this means that risk relates to both the prevalence of the disorder (the chance of a random patient in the group getting glaucoma over a year, say) and the severity of the disorder (e.g., how significant the loss of visual function is).
Are you trying to determine the likelihood of developing glaucoma as a function of weight? If so, one thing you may need to take into account is the possibility of confounding factors, such as age, gender, or diet. As a made-up hypothetical, suppose that men were more likely to develop glaucoma than women. Since men tend to weigh more than women, you might discover that the higher weight groups have a higher incidence of glaucoma. However this would not necessarily mean that weight is a risk factor for glaucoma, or that (all else being equal) losing weight will reduce the likelihood of developing glaucoma: in my hypothetical it is possible that this once you control for gender, the dependence on weight disappears. This is only intended to be an illustrative example -- I'd imagine that the risk of glaucoma is not actually dependent upon one's gender, but for all I know it might well be dependent upon other factors that are correlated to weight. Therefore, you will need to be very careful in your statistical analysis to account for these potential confounding factors.