Generating near-misses is a very expensive proposition computation-wise, especially when using large dictionaries, or when a word is encountered (say, in source code embedded in an English document) that is nowhere near any English words in its spelling. Yet much spell-checking done today does not need suggestions all of the time — say, when an interactive program is checking words just to underline them or otherwise flag a problem for the user. For example, if Emacs "flyspell" mode is used with a big hunspell dictionary, the editor can pause for a large fraction of a second each time the cursor encounters a mis-spelled word.
There ought, therefore, to be an option like "terse-mode" that turns off near-miss calculations in "-a" mode and just returns whether a word is a mis-spelling or not. It could just return zero for <count> and then list no mis-spellings, so that its output was compatible with the current format.
Or, if you wanted to avoid creating yet another mode, you could create a variant of the "^" command that never generates misses; I note that underscore "_" is available and is a semantic opposite of "^" in many programmer's minds. :-)