3 study areas - changes in incidence/prev over time within AND between study areas?
Hi everyone, I just thought it's easiest if I go through my research plan and the aim to put things into context. I've been told one thing by one supervisor, and other things by outside people so I'm trying to get to the bottom of it
I have 3 purposively selected study areas, each containing multiple communities, and my time period is 1995-2011. I am getting hospital admission data and emergency department data for each of these communities (and data from each community will be lumped together the 3 respective study areas). For the selected conditions I will convert these into incidence and prevalence. Obviously the 'participants' in the study will be based upon how many people presented to the ED or were admitted to hospital and will not be the same study population over time. I am looking for: 1) changes in incidence/prevalence of selected conditions WITHIN each of the study areas over time, and; 2) changes in incidence/prevalence of selected conditions BETWEEN each of the study areas over time. I may also look at these same two things but by gender, age, SES, etc. So looking at all of these trends over time.
I have been told by someone to use a mixed methods ANOVA, so the variables would be study area, time, and selected conditions, with the hospital admission or ED rates as the dependent variable. This would allow looking within and between groups. BUT another person told me that we would not do this because of repeated measures - we do not have the same subjects over time.
So, my supervisor has said to do either: 1) logisitic regression for basic level with time as the predictor variable and presence/absence of disease as the outcome variable....giving us an OR to describe the association before and after the intervention we are looking at and the health condition(s). Then we would add in a third variable of the study area type - producing an OR for each study area....BUT this is quite crude (and we will have to make definite cut-offs for time periods. So she said about doing more sophisticated modelling (but not really an indication of what) and looking at changes over time (per year) within each study area and whether those changes are equivalent to changes over time (per year) compared to the other study areas.
I am truly at a loss as I feel like I keep getting suggestions of different things and then when I look up and discuss what one person suggested, the next person tells me that that's not what should be done! Any input/thoughts would be greatly appreciated. Thanks for your help!!