Session Type: Poster Session (Monday)
Session Time: 9:00AM-11:00AM
Background/Purpose: Sjogren’s syndrome (SS) is a chronic autoimmune disease, consisting
of primary SS (pSS) when it presents alone and secondary or associated SS when
accompanied by other connective tissue diseases. Prevalence of primary SS ranges
between 10 and 90 cases per 100,000 people in Europe. However, there is a lack of efficient
algorithm that can be used in claim-based database to identify SS patients. Our objective
was to establish an algorithm that can be used to identify SS patients by using a sub-sample
of the French healthcare database claim.
Methods: Potential SS cases were identified based on ICD-10 codes (M35.0) when used for
SS-related hospitalization or chronic disease status allowing full expenditure reimbursement
codage, from 2005 to 2016. Reimbursements of at least one drug of interest, number of
specialist prescriptions, reimbursement of Schirmer’s test, procedures on salivary gland and
research of antinuclear antibody were identified as variables of interest to build an algorithm.
Patients having at least exclusion criteria of pSS ACR-EULAR classification criteria were
excluded. Patients with associated code of rheumatoid arthritis, juvenile arthritis, ankylosing
spondylitis, systemic lupus erythematosus, systemic sclerosis and other overlap syndromes,
based on ICD-10 codes or biological therapy reimbursements, were classified as secondary
SS. The onset of the disease was identified as the first occurrence of ICD-10 M35.0. A
cross-validation was performed by using a logistic regression to estimate the accuracy of 15
different “diagnosis” algorithms.
Results: Among the 447 potential SS patients identified, 44 were excluded, 267 patients
were classified as pSS and as 136 secondary SS. The most efficient algorithm to identify
pSS (with an accuracy of 0.94) was: ICD code of SS + at least 2 prescriptions of at least one
of the drugs of interest in 4 years before, or the 4 years after the first occurrence of the ICD
code of SS. For secondary SS, the best algorithm (with an accuracy of 0.89) was: ICD code
of SS + at least 2 prescriptions by a rheumatologist or an internal medicine physician both
before and after the first occurrence of the ICD code of SS. With these algorithms, estimated
prevalences were 29.5 per 100,000 for pSS and 8.30 per 100,000 for secondary SS in 2016.
Conclusion: Using a sub-sample of the French healthcare claims database, we developed,
two algorithms to efficiently identify primary and secondary SS patients. Further analysis is
planned to estimate at national-level prevalence and incidence of SS in France and analyze
healthcare consumption. In addition, these algorithms can potentially be adapted for claims
data in other countries.
To cite this abstract in AMA style:Devauchelle Pensec V, Chiche L, Zhuo J, Lavrard I, Desjeux G, Seror R. Development of an Algorithm to Identify Sjögren’s Syndrome Patients in the French National Healthcare Claims Database [abstract]. Arthritis Rheumatol. 2019; 71 (suppl 10). https://acrabstracts.org/abstract/development-of-an-algorithm-to-identify-sjogrens-syndrome-patients-in-the-french-national-healthcare-claims-database/. Accessed October 23, 2020.
« Back to 2019 ACR/ARP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/development-of-an-algorithm-to-identify-sjogrens-syndrome-patients-in-the-french-national-healthcare-claims-database/