According to Jim Crow”s 2006 article, base substitutions are mostly (overwhelmingly) from males and increase with paternal age, but small deletions are contributed about equally by males and females, with no noticeable age effect. Probably the deletions happen during meiosis.
So, with a huge gene like those involved in Duchenne’s muscular dystrophy or neurofibromatosis I, which have many exons (79 for dystrophin), many of the mutations are caused by deletions. The paternal age effect is weaker for those syndromes (since less than half of the causal mutations are base substitutions)
Looking again at the loss-of-function paper in Science by whatisname, the Yoruba have about 42% more stop mutations than CEU (all base substitutions), while the number of large deletions is not higher in the Yoruba (26.6 for the Yoruba, 28.3 in CEU). Smells like paternal age.
This pattern is obviously not brand new. At least a couple of hundred years old, because you see it in both the Yoruba and African-Americans. Obviously a good deal older than that.