Administrative data in health research

What are administrative data?

Administrative health data, also referred to as “health claims data” are records of health care encounters that contain information on the patient, the provider, and the services that were provided. Think about the last time you went to the doctor. Information such as your personal health number, diagnosis, service rendered, the physician you saw, the date of your visit and more were recorded. These are examples of administrative health data.

Administrative data primarily exist for the purposes of registration, record-keeping, and billing. For instance, when a patient goes to the pharmacy and has a drug dispensed, the pharmacy generates a claim that bills a given payer for the cost of the drug. A payer could be the provincial pharmacare program, a private insurance company, or the recipient (out-of-pocket or uninsured). Likewise, when a patient goes to a hospital or physician, the care provider bills the health insurer for the cost of the medical service.^[4] Administrative data, when used in health research, is secondary data – information not originally collected for the purposes of research.

[Section 1_Image 1]

Why are administrative data used in health research?

Administrative data can provide a wide range of information on large samples of a population.^[4] Researchers can examine groups of individuals that have similar conditions, healthcare experiences and outcomes, with sample sizes larger than what might be readily attained using other data sources. Furthermore, since these data are by-products of existing administrative systems, they are relatively efficient and cost-effective to use for research when we consider the greater time and costs that are associated with primary data collection.^[4] Data are collected in real time, providing up-to-date health information. Lastly, administrative data are accumulated over time and are linkable to other types of data through unique encoded patient identifiers. By linking patients’ information from various healthcare encounters, we can create population-based data sets that give us a comprehensive picture of patients’ health status and health services utilization over long periods of time.^[1]