The number of stall/spin crashes due to mishandling engine-out procedures in a twin would seem to argue that less practice equals more risk. One thing I’ve never seen is an attempt to measure the rigor of instruction. I went through a flight school with very high expectations, and in recurrent training have insisted that my instructors make me sweat; I believe those high expectations had a positive impact on my confidence and personal standards, and made me a safer pilot. My first cross-country after getting my instrument ticket included 2 1/2 hours in IMC (in a 172 with a standard six-pack and dual nav coms), which I only undertook because I felt well prepared. I’ve had way too many conversations with pilots, especially those with just a private rating, where it seems the tone of their instruction was to expect just enough to get by.
With the proliferation of flight simulators today, I think insurance companies could play a large role in improving safety by discounting their rates for frequent recurrent training, especially for instrument and twin engine-out practice. It would need to be frequent to do any good.