POS1429 IMPROVING ACCURACY OF SELF-REPORTED DIAGNOSES OF POLYMYALGIA RHEUMATICA AND GIANT CELL ARTERITIS IN THE FRENCH PROSPECTIVE E3N COHORT: A VALIDATION STUDY

Abstract

BackgroundGiant cell arteritis (GCA) and polymyalgia rheumatica (PMR) are two associated inflammatory diseases that probably share common pathophysiological mechanisms. Data on environmental risk factors are lacking. Population based cohort studies are the most adequate and less biased sources for identifying such factors. But case validation of disease diagnoses is the first necessary step for running such studies, even though it is not easy to perform.ObjectivesTo assess the accuracy of self-reported GCA/PMR diagnoses and to develop algorithms to ascertain GCA/PMR in a large French population-based cohort, using combined data of a dedicated questionnaire and medication reimbursement database.MethodsThe E3N cohort study (Etude Epidémiologique auprès des femmes de la Mutuelle générale de l’Education Nationale) includes 98,995 healthy French women born between 1925 and 1950, recruited in 1990 and was designed to investigate lifestyle and environmental factors associated with chronic conditions. Participants completed biennially mailed questionnaires to update their health-related information, lifestyle characteristics, and newly diagnosed diseases. Women who self-reported a diagnosis of GCA and/or PMR were sent a specific validation questionnaire designed to ascertain the diagnosis including clinical, biological, and therapeutic data, along with ACR 1990 classification criteria for GCA and ACR/EULAR 2012 classification criteria for PMR. We then devised algorithms based on self-reported answers and a medication reimbursement database, and evaluated their accuracy, comparing them with diagnoses obtained from medical chart review.ResultsAmong the 98,995 participants, 1,392 women self-reported GCA/PMR. The specific questionnaire was sent to 1,143 (82.1%) of the eligible women (249 women could not be contacted because of death or withdrawn consent) and response was obtained for 830 women (59.6%). Among them, 202 women provided sufficient medical data to ascertain a diagnosis and study accuracy of developed algorithms. 56 women were classified as ACG and 121 as PMR. Self-reported diagnoses alone had an accuracy of 87.6% with medical chart review. If women additionally self-reported a diagnosis confirmation by a physician and the use of glucocorticoids for ≥ 3 months, the accuracy was improved to 89.9%. For patients who did not respond to validation questionnaire, adding the use of glucocorticoids for ≥ 3 months in the reimbursement database also improved the diagnosis accuracy to 92.8%. These two designed algorithms also had the benefit of reducing the number of false positive cases by 10 and 16 respectively. Finally, 589 GCA and/or PMR cases were confirmed by our two devised algorithms: 401 cases with algorithm using the specific GCA/PMR questionnaire and 188 with medication reimbursement database. The mean age at diagnosis was 70.3 (± 8.0) years [73.4 (± 6.2) years for cases detected using the specific GCA/PMR questionnaire and 68.9 (± 8.3) years for cases detected with medication reimbursement database]. Demographic and clinical data were similar between our population of validated cases by medical chart review and the cases detected by our algorithms in the cohort.ConclusionThe accuracy of self-reported diagnosis of GCA/PMR was high in the E3N-cohort. Using additional data such as medication reimbursement and/or other self-reported data from a specific questionnaire, particularly the prolonged use of glucocorticoids led to a better accuracy with a very small number of false positive cases and seemed to be sufficient to correctly ascertained GCA and/or PMR diagnoses. With the validation of nearly 600 GCA and/or PMR cases in our cohort, we will be able to conduct epidemiological studies to identify risk factors of these diseases.AcknowledgementsThe authors are indebted to all participants for their continued participation. The authors would like to thank Pascale Gerbouin-Rerolle, Mariam Alyaniakian, Sofiane Harizi and Roselyn Rima Gomes for their help on data management. The present work was performed using data from the Inserm E3N cohort and support from the MGEN, Gustave Roussy, and the Ligue contre le Cancer for setting up and maintaining the cohort. The cohort was supported by a state grant ANR-10-COHO-0006 from the Agence Nationale de la Recherche within the Investissement d’Avenir program. The present work was conducted thanks to a research grant from the Agence Régionale de Santé – Île de France.Disclosure of InterestsNone declared