I believe people in the earlier cohort were more likely to die as children as child mortality was higher in the past. If so, that could apply a selection bias to the age 66 and up subset of this cohort, since those who survive their childhood could tend to have a natural proclivity to be healthier into old-age than those who don't.

Moreover, people became more educated over that 20 year timespan, so comparing people of equal absolute educational attainment across the two periods could be a case of comparing a higher socioeconomic group to a lower one.