Phase separation of the ultrasoft restricted primitive model (URPM) with gaussian charges is re-investigated in the random phase approximation (RPA)--the "Level A" approximation discussed by Nikoubashman, Hansen, and Kahl [J. Chem. Phys. 137, 094905 (2012)]. We find that the RPA predicts a region of low temperature vapour-liquid coexistence, with a critical density much lower than that observed in either simulations or more refined approximations (we also remark that the RPA critical point for a related model with Bessel charges can be solved analytically). This observation suggests that the hierarchy of approximations introduced by Nikoubashman et al. should be analogous to those introduced by Fisher and Levin for the restricted primitive model [Phys. Rev. Lett. 71, 3826 (1993)], which makes the inability of these approximations to capture the observed URPM phase behaviour even more worthy of investigation.