This is the situation. A classification model with 5 groups with some kind of order has recently been published in adult patients under mechanical ventilation (http://bit.ly/33E3T2V) and I want to test if this classification also suits for pediatric patients in terms of similarity of proportions. I wonder if Chi2 (in a 2 x 5 table) vs. Cochran-Armitage test would be the best way to test this situation (I think the latter, because of the ordered classification) and also which would be the minimum sample size for each test. I found here (http://bit.ly/2OWB7Gz) that sample size in this scenario gets smaller when number of groups increases, and that with a five-group classification a n=400 would satisfy the requirements for a power near 80%.