I checked these histograms of 2,4,8 and 5 uA data. It seems like the HAND histos are not using the same bin width. I can not tell if those histo for BB are in the same bin width or not.
To get the idea if it worth going with 5uA or not, one needs to compare the rate of good triple coincident events without the background. I would suggest to fit those histos using a function of a linear line (pol1) + gauss then one can see how many good events under the peak. After that normalize to the total entries of the histo to get Good-Event-Fraction (how many good events per DAQ event).