Authors' reply

EDITOR,—We note Karin Friedli and Michael King's concern about baseline values. The large numbers and the randomisation process in our study make important differences in baseline morbidity unlikely. The nurse gave the initial self rating scales to the patients; thereafter it was often a member of the clinical team who did this. We agree that the person providing the support should not undertake a subjective assessment of the intervention. We felt, however, that self rating scales would circumvent this problem. We would have expected that, if the nurse had influenced the scores, the scores in the group …