I'm not saying you're wrong, per se, but suggesting that you're right due to the absence of evidence is premature.
That said, while I don't know how official these numbers are, the NHTSA attributes 100,000 accidents to driver fatigue per year, and while the victim count is currently lower than your assertion for drunk driving, I think that driving while fatigued is underreported for the reasons I mentioned above.
Here's part of the problem. The actual statistic is that 10k deaths occur where one driver had a BAC of 0.08 or higher. It's purely an assumption that these deaths would not have occurred if the drivers were sober. Some, certainly; some, certainly not. We need better information.
The discussion lacks "scientific integrity", as Feynman would put it.