It is known that the GRB equivalent hydrogen column density (NHX) changes with redshift and that, typically, NHX is greater than the GRB host neutral hydrogen column density. We have compiled a large sample of data for GRB NHX and metallicity [X/H]. The main aims of this paper are to generate improved NHX 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 (NHI,IC) adjusted for the ionisation 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/Zsol) 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 NHX . Lower GRB host mean metallicity results in increased estimated NHX with the correction scaling with redshift as log (NHX cm-2) = (0.59+/-0.04)log(1+z) + 0.18+/-0.02. Of the 128 GRB with data for both NHX and NHI,IC in our sample, only 6 have NHI,IC > NHX when revised for realistic metallicity, compared to 32 when solar metallicity is assumed. The lower envelope of the revised NHX - NHI,IC, plotted against redshift can be fit by log(NHX - NHI,IC cm-2) =20.3 + 2.4 log(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 (n0) to be consistent with n0 = 0.17 x 10-7cm-3.