[Effects of 2 reimbursement models on house calls by general practitioners].