I 100% believe there is such an effect, and that it’s confounding the measurement of population level changes. But I want to point out that the existence of such an effect is not really evidence _against_ the changes, it’s just another confounder. Your wording is very accurate — you “consider it realistic that depression did not increase”, ie it is possible that depression didn’t increase. But whether depression actually did increase or actually didn’t, is not possible to determine from these numbers directly. Without controlling for confounders and doing some fun statistics, we largely just have opinions