(Regression? Correlation?) Analysis for prevalence I've got this table of data and I'd like to run a regression analysis on it looking for trends in the data. Mainly if there are differences in the prevalence of lesions when divided into decade categories. 
I'm trying to figure out how to statistically analyze the prevalence and the categorical data of the decades. A combination of categorical (decade) and continuous data (prevalence). Which statistical test should I use?
Thank you!
Here is the data table:
                   <30  30-39   40-49   50-59   60-69   70-79   80-89 total
without lesions    58    52      56      65      49      28      7      315
with lesions        6    5       3       4       17       9      0       44
total              64    57      59      69      66      37      7      359
lesions prevalence 0.09 0.088   0.051   0.058   0.26    0.24     0     1.768

 A: It's not clear what question(s) you hope to answer with additional analyses.. The omnibus chi-square (with all the rows and columns in one table) will tell you if the proportions of lesions you observe are unrelated (or related) to decade category.
You could use logistic regression to predict the presence/absence of lesion as an outcome with the decade variable as a predictor. It answers a slightly different question: 'what are the log odds of being in the lesion group for each additional decade.'
Which question are you asking? 
A: It seems pretty clear that your objective is to characterize how prevalence of lesions varies with age in a sample of people, presumably with some form of disease, e.g. HPV.
This aggregate representation of prevalence is a common form of contingency table that is amenable to many forms of analysis. I must correct you, however, in calling the y table values "decade" and "categorical". Firstly, it is age and not year you are discussing, so you must be clear in describing exactly what is being measured. Secondly, this is not a categorical variable, per se but a form of "grouped linear" coding. The underlying variable is continuous and observable, but it's simply been cut into groupings to make presentation facile.
Given all that, you can easily study the relationship between age and prevalence of lesions by using a variety of models. If you use a relative risk model, you can estimate a prevalence ratio describing the incremental change between subsequent age categories in prevalence.
A logistic model for the same outcomes will produce an odds ratio, which has a more contrived interpretation, but will give you results equivalent to the Armitage test of trend. Treating the "decade age" variable as categorical will give you a global test of whether there is heterogeneity in the sample, or that at least one specific age group has prevalence that varies significantly from other age groups. This is a more powerful test than a trend test, but does not produce any useful summaries, except if using the less powerful Tukey Post-hoc test. 
A: I ended up using a chi-square test to analyze the distribution of lesions across the different subgroups.
