Personally I prefer calling it "Durkin's Attack", although I am reliable told that Durkin was in the process of writing a second book based upon the reference "Sodium Attack" shortly before he was believed to have died.
If you go through Durkin's games he rarely played 1.Na3 e5 2.Nc4, his preference was 2.e3, why? your guess is as good as mine. I have said a few times before I think 2.Nc4?! is a poor move because of 2...e4! intending d5.
if there is anything to be found in playing 1.Na3, i think it lies in those who play 1...d5 in reply. Many times I have reached reverse Benoni like structures with 1.Na3 d5 2.Nf3 c5 3.c4 d4 4.e3 or 4.Ne5!?
HTH
If, for some reason, U are obliged to play the Sodium attack 1. Na3, don't be scary, it's not as terrible as usually thought.
King's Sodium 1...e5 is supposed to be the best black reply (Explorer and ChessOK).
King's Sodium (Explorer vs Explorer test opening, d=20):
1. Na3 e5 (-0.33) 2. e3 (-0.36) Nf6 (-0.42) 3. d4 (-0.25) exd4 (-0.27) 4. exd4 (-0.28) Bxa3 (-0.18) 5. bxa3 (-0.16) O-O (-0.26)
So, Explorer preferes the Reversed French Sodium 1. Na3 e5 2. e3. It doesn't seem to be really problematic for white except for doubled pawns in the a-file.
Queen's Sodium:
1. Na3 d5 2. d4 (-0.18) Bf5 (-0.22) 3. Bf4 (-0.25) a6 (-0.31) 4. e3 (-0.31) e6 (-0.22) 5. Bd3 (-0.25) Bxd3 (-0.22) 6. cxd3 (-0.21)
Well, here it transposed into the Double London system (or Double Mason attack).
Transposition to Sicilian: Kronberger variation:
1. Na3 c5 2. e4 e6 (+0.00) 3. Nf3 (-0.05) Nc6 (+0.07) 4. Bb5 (+0.02) d5 (+0.00) 5. O-O (-0.02) a6 (+0.03)
4. Bb5 connects it also with the Rossolimo attack in the Old Sicilian. Btw 5. O-O is a good gambit which may be called Durkin-Kronberger-Rossolimo gambit.