Is the increase in background associated with a drift in gain, or peak-broadening (increase in FWHM) that might be indicative of a yellowing of the crystal (I assume that you're using an NaI probe)?