Article Text

Download PDFPDF

AB0224 Simulation for choice of bdmards and tsdmards in order to success for three-year survival in rheumatoid arthritis treatment
  1. I Yoshii
  1. Rheumatology, Yoshii Hospital, Shimanto City, Japan


Background In rheumatoid arthritis (RA) treatment, bDMARDs and tsDMARDs (BIO) perform tremendous disease activity control, however, its effectiveness is uncertain and unstable, because their survival ratio is not good enough to tolerate. If more right choice is done in guided by some simulation.

Objectives This study aims more than 80% of three-year survival ratio (SR@3Y) in simulating risk of BIO by using statistical clinical data and post marketing surveillance (PMS) data with Bayes estimation.

Methods Infection risk and survival risk were harvested from Japanese PMS data, and our clinical data. All our cases were calculated with last observation carried forward method (LOCF). If BIO was continued for more than three years or discontinued by attaining clinical remission, it was evaluated as success, while other cases were evaluated as failure. Patient's clinical data and general status were calculated for each case, and SR@3Y for success was statistically evaluated with Binary Logistic Analysis for success. Evaluation methods for parameters were divided according to general risk and drug specific possibility. If calculated general risk went above 0.2, selection of BIO was discarded. In other case which had gone below, choice of BIO is done in according to point that had been cumulated by drug specific possibility in choosing what took maximum calculated expectation value.

If chosen drug have matched used BIO, it was evaluated as true, if not, it was evaluated as false, while if true case was in success, it was evaluated as true success, and if in failure, it was evaluated as true failure, while false case was in success, it was evaluated as false success, and if in failure, it was evaluated as true failure. Sensitivity in success cases and specificity in failure cases was evaluated in patients in whom BIO was administered. Statistical evaluation was done with chi-square test.

Results 188 cases have had enough data for simulating. In these, 108 were success and 80 were failure. In success cases, simulated TNF inhibitor (TNF-i) counted 73, Tocilizumab (TCZ) counted 11, Abatacept (ABT) counted 12, and Tofacitinib (TOF) counted 2 while real chosen cases were TNF-i counted 65, TCZ counted 11, ABT counted 21, and TOF counted 11. Overall success ratio was 57.4%. In these cases, true choice had been done in 97 cases of 108. In failure cases, simulated TNF-i counted 44, TCZ counted 16, ABT counted 2, and TOF counted 1, while real chosen cases were TNF-i counted 49, TCZ counted 17, ABT counted 10, and TOF counted 4. True success counted 97, and false success counted 11, while false success counted 37 and true failure counted 54. Then, sensitivity was 89.8% and specificity was 67.5% (<0.01).

Conclusions Drug choice of BIO supported with simulation was superior to real choice. If risk management was adequately performed, SR@3Y is expected more than 85%.

Disclosure of Interest None declared

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.