ABSTRACT
We need to distinguish between the unit of analysis and the unit of reporting as they have different but overlapping sets of issues. If we want to compare clinics in terms of their staffing levels or hospital wards in terms of their bed numbers, for instance, the unit of analysis and the unit of reporting are typically the same. Analysis, reporting and interpretation are relatively straightforward. For other measures, the units of analysis and reporting are always necessarily different. For example, many figures reported by hospital, such as risk-adjusted mortality rates for acute myocardial infarction (AMI), involve analysis at the patient level and then reporting by hospital. In these examples, there is no choice in the unit of analysis, but in others there can be. We first consider the issues regarding the analysis because if that isn’t sound, then there’s little point in the reporting. Some issues affect both. After describing the issues, we give some potential solutions. Lastly, we discuss the special problem of attributing performance to a given unit when multiple units have been involved in the patient’s care. This is, of course, extremely common as patients are treated by multiple members of healthcare and other staff and can be transferred from one provider to another.
