Social deprivation among socio-economic contrasted french areas: Using item response theory analysis to assess differential item functioning of the EPICES questionnaire in stroke patients.

Background: Multiple approaches have been proposed to measure low socio-economic status. In France the concept of precariousness, akin to social deprivation, was developed and is widely used. EPICES is a short questionnaire that was developed to measure this concept. This study aimed to evaluate Differential Item Functioning (DIF) in the EPICES questionnaire between contrasted areas: mainland France, French West Indies (FWI) and French Guiana (FG).
Methods: The population was taken from the INDIA study, which aimed to evaluate the impact of social inequalities on stroke characteristics and prognosis. Eligible people were patients referred to neurology or emergency departments for a suspicion of stroke. We assessed the DIF using hybrid ordinal logistic regression method, derived from item response theory.
Results: We analysed 1 553 stroke patients, including 768 from FWI (49.5%), 289 from FG (18.6%) and 496 from mainland (31.9%). We identified five items with a moderate to large DIF in area comparisons: "meeting with a social worker", "complementary health insurance", "home-owning", "financial difficulties" and "sport activities". Correlation between EPICES score and the latent variable was strong (r = 0.84).
Conclusion: This is the first attempt to assess the DIF of the EPICES score between different French populations. We found several items with DIF, which can be explained by individual interpretation or local context. However, the DIFs did not lead to a large difference between the latent variable and the EPICES score, which indicates that it can be used to assess precariousness and social deprivation between contrasted areas.
Competing Interests: We have read the journal’s policy, the authors of this manuscript have the following competing interests: Yannick Béjot received honoraria for participation to advisory boards or as a symposium speaker for AstraZeneca France, Daiichi-Sankyo, Pfizer-BMS, Bayer, Covidiem and MSD France. Other authors have declared that no competing interests exist. This does not alter our adherence to PLOS ONE policies on sharing data and materials.