# Trend test for ratios

I have binomial data infected/not-infected on individuals of a sample from different years for two different infections. I have used prop.trend.test to test for trend for each infection.

I want to test if there is a trend in the ratio of infected/not-infected between the two infections. Since a ratio could vary between 0 and ∞, what test is appropriate?

For testing trends in proportins, some have suggested glm with family binomial. Can i use glm with a different family?

What test can I use to prove there is an upward trend in time series of ratios assesment in R? suggests using the Cochran-Armitage test (which I think is the same as prop.trend.test) or logistic regression. Not sure if it is suitable in the case of ratios, though.

> fit<-lm(m$prop ~ m$score + I(m\$score^2))