Health care organisations, pricing and reimbursement agencies, and the pharmaceutical industry all have interest in the economic burden of diseases, and how intervention can impact the cost trajectory. For the rheumatology field, this has attracted much interest after the introduction of the efficacious but expensive biologic therapies.
Historically, economic costs related to a disease have generally been collected using questionnaires in samples of limited size. Non-response as well as recall bias may cause problems using such designs. Study cost may also become an issue if the sample size is to be expanded and/or efforts made to minimise non-response. Furthermore, attribution of costs to the disease is also difficult and risks becoming arbitrary.
The aim of this presentation is to describe the potential and pitfalls of register-based estimation of costs associated with RA. This description will be exemplified by use of nationwide Swedish clinical quality registers for RA (the Swedish Rheumatology Register and the Swedish Biologics Register ARTIS), and how these can be linked to other national data sources on health care utilisation, drug dispensation, and productivity losses (sick leave and disability pension).
Examples will be given for the prevalent cohort of patients with RA in Sweden (n$≈ $50,000), as well as incident patient cohorts (incident in terms of their RA disease or certain treatment exposures). The distribution of total costs across dispensed drugs (biologic/non-biologic), inpatient and outpatient care utilisation, sick leave and disability pension will be described from a register-perspective, as will the challenges posed by non-normal cost distributions. The importance of subgroup analyses, allowed for by the large sample size, will also be illustrated.
Furthermore, an alternative to arbitrary attribution of costs to RA will be presented using individual-level matched comparators from the general population. This alternative is an empirical approach to the issue of which costs that are to be deemed as RA-related, and its strengths and weaknesses will be outlined.
Disclosure of Interest None Declared