Data were collected from clinicians at the time of consultation about the care that they provided in 12,813 encounters in a general practitioner (GP) staffed casualty department and 719 primary care encounters in two emergency departments (Bolton, 1999). Data were collected by the GPs themselves in general practice, and by a research officer located in the emergencydepartments. Patients seen in the emergency department were ambulatory patients whom the triage nurse assessed would not suffer an adverse outcome if they had to wait an hour or longer for care. Comparison of these two patient populationsestablishedthatthey were similar in terms ofage, gender, ethnicity, andreason for encounter. We observed that the rate of use of the terms "advice" and "counselling" as a description of the management provided to these patients varied between the two sites. At both sites the term "advice" was used both generally (that is, "patient advised") and for specific conditions (for example, "advice re smoking", "head injury advice"), while counselling was exclusively used in a general way. In contrast, the rate of pharmaceutical prescribing did not vary between the two sites (Bolton, 1999). "Advice" was given in 767 (6.0%) encounters in the GP staffed casualty and 13 (1.8 %) encounters in the emergency departments. "Counselling" was provided in 109 (0.9%) encounters in the GP staffed casualty and 1 (0.1 %) encounter in the emergency departments. Logistic regression analysis was used to assess the difference in the use of the term "advice" between the two sites, controlling for age and sex, and clustering at the level of the treating doctor. A difference was found (p=0.003). Fisher's exact test was used to assess the difference in the rate of the use of the term "counselling" between the two sites. Fisher's test was used because of the small expected