It is known that the GRB equivalent hydrogen column density (N_HX_) changes with redshift and that, typically, N_HX_ is greater than the GRB host neutral hydrogen column density. We have compiled a large sample of data for GRB N_HX_ and metallicity [X/H]. The main aims of this paper are to generate improved N_HX_ for our sample by using actual metallicities, dust corrected where available for detections, and for the remaining GRB, a more realistic average intrinsic metallicity using a standard adjustment from solar. Then, by approximating the GRB host intrinsic hydrogen column density using the measured neutral column (N_HI,IC_) adjusted for the ionization fraction, we isolate a more accurate estimate for the intergalactic medium (IGM) contribution. The GRB sample mean metallicity is =-1.17+/-0.09 rms (or 0.07+/-0.05 Z/Z_sol_) from a sample of 36 GRB with a redshift 1.76<=z<=5.91, substantially lower than the assumption of solar metallicity used as standard for many fitted N_HX_. Lower GRB host mean metallicity results in increased estimated N_HX_ with the correction scaling with redshift as {Delta}log(N_HX_cm^-2^)=(0.59+/-0.04)log(1+z)+0.18+/-0.02. Of the 128 GRB with data for both N_HX_ and N_HI,IC_ in our sample, only six have N_HI,IC_>N_HX_ when revised for realistic metallicity, compared to 32 when solar metallicity is assumed. The lower envelope of the revised N_HX_-N_HI,IC_, plotted against redshift can be fit by log(N_HX_-N_HI,IC_cm^-2^)=20.3+2.4log(1+z). This is taken to be an estimate for the maximum IGM hydrogen column density as a function of redshift. Using this approach, we estimate an upper limit to the hydrogen density at redshift zero (n_0_) to be consistent with n_0_=0.17x10^-7^cm^-3^.