Table 1.

Distribution of RA patients selected through the four algorithms: the main analysis.

Algorithms*	Actual RA patients°, n (%)	Assumed RA patients§, n (%)	True positive patients#, n (%)
First	103 (37.2)	74 (71.8)	55 (53.4)
Second	103 (37.2)	96 (93.2)	79 (76.7)
Third	103 (37.2)	47 (45.6)	38 (36.9)
Fourth	103 (37.2)	123 (119.4)	96 (93.2)

*Out of patients with the first supply of bDMARD from 2014 to 2016 and at least one record of visit at the Rheumatology Unit of Pisa University Hospital from 2013 to the index date, we tested the performance of four index tests (algorithms):

First) RA according to hospital discharge records or emergency department admissions (ICD-9 code, 714*);

Second) RA according to exemption code from co-payment (006);

Third) RA according to hospital discharge records or emergency department admissions (ICD-9 code, 714*) AND RA according to exemption code from co-payment (006);

Fourth) RA according to hospital discharge records or emergency department admissions (ICD-9 code, 714*) OR RA according to exemption code from co-payment (006).

°The actual diagnosis of RA was that recorded in the medical chart (reference).

^§the assumed diagnosis was the RA diagnosis recorded in the hospital discharges (regardless of primary or secondary) and emergency department admissions or the co-payment exemption code related to RA in HAD.

^#True positive patients: These were assumed RA patients in the HAD with actual RA diagnosis in the reference.

bDMARD biologic disease modifying antirheumatic drug, HAD Healthcare Administrative Database, ICD-9 international classification of diseases 9th revision, n number; RA rheumatoid arthritis.