It all has do to with...temperature. The hotter the drive, the harder you can throw fuel out the rear end.
Fusion and AM will both generate any temp. up to millions of degrees, so the basic issue without fantastical technology comes down to fuel availability and storage. Fusion wins here on both counts: Deuterium and plain-old hydrogen can be stored essentially inert (as water, ammonia, or methane) until needed, whereas AM has 'special needs'. If the ship takes a hit and breaks the AM containment, you are VERY VERY toasty.
Only way out of this is if you have some magic AM generator that converts some tiny fraction of your fuel into AM at burn time; AM just causes too many problems. Producing AM in mass beforehand is also problematic, whereas deuterium is anywhere you can find lots of hydrogen; all you have to do is mine it and refine it.
And what about maintenance? Any fancy-dancy AM containment/production gadget is going to need regular check-ups from your local neighborhood high-energy physicist, and I didn't notice many of those walking around CJC's books. I had the idea that the insystem drives were the reliable, easily operated part of the system. Any idiot could run a mining ship if he knew the right buttons to push...
Fissionables will also work in a NERVA-type engine: basically a runaway fission reactor using the reaction mass as coolant. But NERVA would only really work as a local in-system drive; fusion, on the other hand, produces enough heat to push a ship up to a good fraction of light speed.
1) v.rms = sqrt(3 * R * T / mu)
mu = molecular weight ~ 6 for disassociated water (H and O ions)
T = temperature of gas
R = Rydberg gas constant = 8315 (MKS)
v.rms = 'average' particle velocity at T
A rocket engine has to accelerate its exhaust at LEAST as fast as its top speed; and in order to get around in the times typical in CJC's novels, this has to mean basically acceleration at 1 gravity (~10 m/sec-sec) for, at most, 1 day.
2) v = a * t = 10 m/sec-sec * 86400 seconds = 864 km/sec = 0.288 % light speed
...sounds reasonable so far.... that means the exhaust velocity needs to be at least 1000 km/sec for sensible performance. Plug the numbers into equation 1) and you get:
(1000 km/sec) ^2 = 3 * 8315 * T / 6 =>
T = 6 * 10^12 / (3 * 8315) = 240 MILLION degrees !!!!!
Huh! Let's go slower, then...
100 km/sec => T = 2.4 million degrees
144 km/sec => T = 5 million degrees
200 km/sec => T = 10 million etc. etc.
That is at least possible..and about right compared to the containment technology talked about in fusion reasearch (check out Wikipedia articles on Robert W. Bussard, NERVA, and the Bussard Ramjet).
Any reasonable insystem drive has to burn at 2-5 million degrees to work efficiently at the right speeds. Fusion reactions are ONLY possible at these temperatures anyway, so... the NASA/DoE NERVA project collapsed in the 1970's because the engineers had a hard time containing at 50,000 K reaction, so....