Do you genuinely think a 50-fold difference in dementia incidence sounds remotely plausible?

That would probably be the strongest incidence gradient signal in medical history. Dementia research would be laser-focussed on the Albania-North Macedonia-Bulgaria axis and the Singapore-Malaysia axis as they'd be near-ideal comparative populations.

If the data isn't compatible, the table doesn't mean much at all, since it has merged reporting bias with the actual incidence rates.