We examine the effects of polymer layer anisotropy (bond correlations) on the numerical mean-field predictions of the interactions between two physisorbed polymer layers under restricted equilibrium conditions. Lattice mean-field calculations are presented for two models based on second-order Markov chains (i.e., without backfolding), one in an isotropic mean field and another in an anisotropic mean field. The results from lattice Monte Carlo simulations are used as a reference to examine those obtained from the numerical mean-field theories. It is found that the numerical mean-field theories predict the structural features of the polymer layer fairly well but are limited in the predictions of the interaction forces. The effects of anisotropy near and away from the adsorbing surface differ from each other but are still insufficient to capture the interactions accurately, Introduction of anisotropy results in higher segment densities of trains but lower segment densities of loops, tails, and bridges and, consequently, lower steric interactions. It is seen that the mean-field theories can at best offer qualitative information on the structure of the adsorbed layer and of the interaction forces at conditions where steric interactions become important (at high coverages). At conditions where bridging interactions dominate (at low coverages), the meanfield theories fail to capture the bridging attraction between the surfaces (as expected), in contrast to some of the earlier published reports.