If health has declined and testosterone levels have declined then absent any other information I am going to assume that the decline in testosterone is at best uncorrelated with health. Now, maybe there is a reason to assume that lower testosterone levels are actually healthier but I'm going to trust my eyes for the time being.

By almost any measure, global health has improved and global testosterone levels have declined.

Absent any other information, you should therefore assume...

At the same time you are hiding information by taking the global average as the proper baseline for comparison. This is disingenuous here because the original study picked a US cohort. The factors that are responsible for global health indicators improving might be, at least in principle, cancelled out in the US.