Proportion test for multiple groups

[Thushani Guruge]

Hi,

I want to compare how does use of drug "A" differs across age categories.

Total drug A users were 5100. The below table shows the number and percentage of users for each age category. I want to compare whether there is any difference in the use of drug A with regards to age? Do people younger than 49 years old have a significant drug use than people 65 years or older?

<49	1794/5100 (35.2%)
50-64	1102/5100 (21.6%)
≥65	1242/5100 (24.4%)
Missing	30/5100 (0.1%)

I'm not sure about the stata command to compare more than two categories. I would highly appreciate your thoughts!

Thank you in advance!

Thushani

[Carlo Lazzaro]

Thushani:
welcome to this forum.
Assuming that the regressand (that is, the dependent variable) is a yes/no two-level categorical one:

logit drug_use i.agecat

[Thushani Guruge]

Hi Carlo,

Thank you for your reply.

Unfortunately, this is a descriptive analysis and does not have a binary outcome. I have a drug A user population only and percentages of drug A users for each age group. I want to know whether people younger than 49 years old have a significant drug use than people 65 years or older?

Do you have any idea of such analysis?

Thank you in advance!

[Maarten Buis]

If you only have drug A users, then you have nothing to compare it with. There are different number of people in the different age categories in the general population. If you don't have that in your data, i.e. you have drawn a sample from only drug A users, then you could try to get something by treating this as a case control study, but now we are moving way beyond a descriptive analysis.