Pharmacy and health insurance databases

What is it?

Administrative claims databases are database systems built on data extracted from claims that are submitted by healthcare providers to payers when a patient uses health services. They include pharmacy databases and health insurance records, and are used to monitor health and disease management. These systems are primarily maintained for billing and administration, but can also be used by researchers, insurers, health authorities and other stakeholders to provide long-term data on the impact of health interventions on the healthcare system in ‘real-world’ (observational) studies.

Claims databases generally include information on the use of inpatient, outpatient, emergency room and pharmacy services. They contain information on, for example, the services performed during a clinician’s visit, surgical interventions, diagnostics, laboratory tests, hospitalisation and length of stay, and pharmacy filing.

Examples of claims databases

There are a large number of these types of databases. Commercial sources include data from private insurers whereas Medicare and Medicaid are examples of sources of non-commercial claims data. The US Food and Drug Administration’s (FDA’s) Sentinel Initiative is an example of claims data being used for safety surveillance.

Why is it useful?

  • Estimating clinical and economic factors: claims data can be used to estimate a number of clinical and economic aspects of healthcare, such as exposure to treatment, treatment patterns and medication adherence, treatment switches, medicines use, cost of care and the economic burden of diseases, and proxies for effectiveness and relative effectiveness of medical technologies.
  • Analysis and prediction: analysis of this data can also be used to identify chronic conditions, predict disease prevalence and incidence, assess gaps on quality of care and enable comparison of health outcomes with guidelines and other standard performance measures.

What are its limitations?

Health insurance and pharmacy records are largely maintained for administrative purposes and not for research. In addition to the shared limitations with electronic health records (see also the summary on healthcare databases here), additional limitations of claims data include:

  • Limited interpretability and generalisability: results may have limited generalisability outside of the specific context of their collection and there may be difficulties in linking data from different sources.
  • Actual use might not be reflected in the data: pharmacy data generally reflect what has been prescribed and picked up from the pharmacy but not necessarily what is actually used by the patient. Actual exposure to treatment might be misrepresented by such data. Furthermore, use of over-the-counter medicines might not be captured.
  • Limited clinical information and history: data may be lacking clinical information and the patient’s medical history, for example, pre-existing conditions and complications may not be recorded.
  • Coding issues: these may occur because of coding errors or because some coding systems are less detailed than needed to accurately identify a medical condition.

What do stakeholders say?

Claims databases are valuable sources for understanding healthcare and disease management, and guide healthcare decision-making. However, methods used to generate insights and identify outcomes of interest and medical exposure must be robust and appropriate.

When conducting clinical and epidemiological research with such data, it is important to understand how information from these different settings has been coded in order to allow for appropriate interpretation, and all limitations must be clearly stated.

The FDA and European Medicines Agency consider claims databases to be a useful source of data for benefit‑risk assessment and safety surveillance (for example, the Sentinel Initiative). Claims data have also been used in a number of pay-for-performance schemes (for example, in the US) and for monitoring adherence to prescription guidelines.

Key contributor

Denise Umuhire, Janssen EMEA