That's very interesting. Would you mind posting some sample data that people could test on? With made-up data (Which may not match your data structure), it looks like it is working for me.
library(tidyverse)
df <- tibble(a = 1:3, eosinophilsAverageValue = 3:5)
df %>%
select(contains("AverageValue"))
#> # A tibble: 3 x 1
#> eosinophilsAverageValue
#> <int>
#> 1 3
#> 2 4
#> 3 5
Created on 2021-10-05 by the reprex package (v2.0.0)