Background Biologics registers have been established in many countries. However, the usefulness of these registers and the interpretation of results from analyses based on such data both hinge upon an understanding of the coverage and representativity of eligible patients, and on the accuracy and validity of the data entered. The Swedish setting with virtually complete national registers that can be linked together offers a possibility to compare clinical register data against de facto drug prescriptions filled.
Objectives To assess the coverage of ARTIS and the accuracy of the registered information on biologics exposure.
Methods Through cross-reference of ARTIS to virtually complete national registers on prescribed subcutaneous and oral drugs, nonprimary outpatient care visits, and death/residency, we assessed: (i) the coverage of ARTIS in terms of first time initiators of subcutaneous anti-TNF therapies 2006-2009, the validity of (ii) the registered start- and stop dates, (iii) “false positive” treatment registrations, and (iv) the accuracy of the registered information on concomitant DMARDs at start of the biological treatment. For each assessment we defined “coverage” and “accurate” using algorithms and pre-specified time-windows, and performed separate assessments for patients with register-identified RA and AS/SpA/PsA (requiring ≥1 visit in nonprimary outpatient care in 2006-2009 listing RA and AS/SpA/PsA, respectively). The biologics used in these analyses were the then available subcutaneous biologics (adalimumab and etanercept).
Results 8048 patients with RA and 3743 patients with AS/SpA/PsA had at least one prescription filled of a subcutaneous anti-TNF therapy 2006-2009. Of these, 94% of the patients with RA, and 84% of those with spondyloarthropaties were also found in ARTIS with a corresponding treatment. There was no age/sex difference between those captured by ARTIS and those not. Allowing a 90-day window, 93% of all the anti-TNF therapy prescriptions filled occurred between a start- and a stop date in ARTIS. Conversely, 2% of the registered treatments in ARTIS lacked a corresponding drug prescription. With respect to the first prescription filled of etanercept/adalimumab in relation to the registered start date in ARTIS, 5% of the patients had the first prescription >60 days prior, and 5% >60 days after the registered start date. 10% of patients had at least one prescription filled of the anti-TNF therapy >90 days after the registered stop date in ARTIS. 75% of all patients who started anti-TNF therapy in combination with a DMARD also had a DMARD prescription within 90 days after treatment start.
Conclusions Taking the routine clinical care setting and inevitable uncertainties such as non-compliance into account, we observed a high coverage and validity of ARTIS’ data on biologics exposure, both for RA and for spondyloarthropaties. Combining data from ARTIS, including also clinical data that cannot be derived from other registers, with prescription data from national registers offers a high quality measurement of de facto treatment in studies of subcutaneous biologics.
Acknowledgements ARTIS has agreements with Abbott, BMS, Merck, Sobi, AstraZeneca, Pfizer, and Roche
Disclosure of Interest None Declared