Promoting prevention in primary care

SIR,-With regard to the comments on our paper by Dr Ann-Louise Kinmonth and colleagues (13 June, p 1551) we accept the lack of absolute proof that the difference in the improvement of recording of cardiovascular disease risk factors in intervention compared with control practices was due entirely to our "facilitator package." We are confident, however, that it made a major contribution for three reasons. Firstly, although we agree that "case-control" matching ofpractices is difficult, our matching was done on the basis of considerable knowledge of the practices, and the similarity of baseline data from all intervention and control practices suggests that matching was close. Systematic differences between control and intervention practices were unlikely to have been greater than those between individual practices, whether intervention or control. Secondly, we agree that the design of our study does not allow distinction between the effects of the initial audit and the "facilitator package," but other evidence suggests that the effects of audit are small compared with that of our intervention.' We did not consider it to be justifiable to recruit and audit control practices at the outset without offering them some other intervention. Thirdly, subsequent application of our model in a large number and variety of practices has produced similar results. GODFREY FOWLER ELAINE FULLARD MUIR GRAY Department ofCommunity Medicine and General Practice, University of Oxford, Oxford OX2 6HE