Usually, in age-adjustment (whether it be direct or indirect), we hold the assumption that we have data on the age distribution of a population. However, if I am performing a prevalence analysis globally (i.e. I can only know the mean age of each country), how do I adjust for age in this case?
Currently, the only data I have are the prevalence of this particular disease and the mean ages of every country.
Thank you!
I would suggest that if the disease prevalence in question has a near linear (or at least monotonic) relationship with age , do a simple linear regression of prevalence v age and compare country prevalences as actual/ predicted from model at mean age of country.
if the relationship with age is more complex it will be much more difficult to do age adjustment.
for another predictor variable do a multiple linear regression of prevalence against mean age and mean value of this orher variable (not population adjusted.)
this is an ecological study.
Steven, I agree with James that it's better to start with the simplest model and build up from there, keeping in mind all the environmental (ecological) relationships. My own impression is that certain diseases are a function of age (younger being more vulnerable), but that this may be due to a lack of understanding proper sanitation among the parents. Simply boiling water kills the amebae, yet (at least somewhat recently) amebic dysentery was a primary cause of infant and older mortality.
I agree with James. It is unfortunate that you only have mean age of country, this will reduce the statistical power or information from the data. However you can build an adjusted regression model to see the independent association of the "mean age" with the prevalence and include the "country" as another variable in the model but factor it (for example in stata, regress prevalence mean_age i.country ....).
The data for each country population, including age-groups based on gender are widely available. I suggest using that data to have more robust raw data to work with. In this case, you can adjust the prevalence/incidence based on each age group population.
Dear Steven,
I do not know with disease prevalence you are looking at. But I like Seyed's suggestion very much. The mean age per country does not count particular for childhood mortality and age groups add more information. However, I believe, even though you might delete some countries as you did not find corresponding age groups. But all in all the results will be more trustful.
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology