Perturbation theory is good for small perturbations to the original Hamiltonian. If there is no multipole expansion the part that is treated as perturbation is quite large which will result in a worse approximation. It would probably also be much harder to calculate the perturbation with the whole interaction term.