I applaud this question. I have not really tried to find any solution, but I definitely recognize the trouble with water... nonsense/worthless reading. As far as why... easy; water absorbs like, all of the IR spectrum. Below is the IR spectrum of liquid water (from NIST's Chemistry Webbook).
I actually have an ATR IR Spec, which is (supposedly) meant for reading aqueous solutions...
...but it doesn't really work... it can't be used for figuring out dilutions (well maybe, give or take 20%). In other words, 50% water + sample looks a lot like 95% water + sample... which looks a lot like any sample + water. Water basically drowns the reading- so it all looks alike.
Don't get me wrong... it's super nice that I don't need to oven dry my sample or deal with KBr. Basically, I just put the sample on the top and I get a decent reading, but if the sample has a lot of water in it (like greater than 10%), it loses a lot of definition.