Thanks for your response, aerba.
ATOMREMO works but when you calculate the defect formation energies you end with values (for anion vacancies) that are higher than with using GHOSTS or the ATOMSUBS, with the latter values agreeing more with published defect formation energies.