A novel approach to the epidemiology of people living with spinal cord injuries in France based on an original algorithm from public health insurance data.

Abstract

Introduction: French Public Health Insurance gathers health, demographic and economic data based on codes from the 10th version of the international classification of diseases (ICD-10), specific nomenclature for each health-care (medical or surgical) procedures, medical expenses and justifications for full coverage of medical care. We aimed to build an algorithm that could identify the French population of people living with spinal cord injury (SCI) relying on public health insurance metadata.

Material and methods: The SNDS (in French, Système National des Données de Santé) was searched for the time-period 2012-2019, looking for: full-coverage motives, ICD-10 codes, and health-therapeutic procedures specific of our population of interest. We built a step-by-step algorithm that identified i)including codes, ii)excluding codes, iii)codes needing confirmation. A group of 3 physicians recognized as experts in this field contributed with data scientists to the selection of pertinent codes and their association. Including codes were ALD-20 (full-coverage 'paraplegia', in French, Affection de Longue Durée), G114 (spastic paraplegia), Q05.x (spina bifida), spinal cord trauma (S14.x; S24.x), vascular myelopathy (G951), degenerative myelopathies (M47.x). Autoimmune, other disabling neurological diseases with a specific ICD code, and oncologic patients were excluded. Neurological symptoms (G82.x) needed confirmation. We identified 6 categories of SCI regarding their etiology, based on ICD-10 code combinations (congenital, genetic, tumoral, traumatic, acquired and symptomatic) Finally antibiotics consumption and hospitalizations of persons identified as SCI were compared to a control sample from overall population (with a 1:5 ratio).

Results: Among almost 245 000 persons with putative SCI, we identified 133 849 living individuals with SCI aged>16 (55.8% men, age 57 yo [44;70]) by 2019. Confirmed traumatic SCI were 21 459 (67% were men, age 53 yo [39;67]), acquired non-traumatic were the most frequent (n=62 561, 46.7%). SCI consumed 1.5 to 3-times more antibiotics and were 4-fold more hospitalized than controls. Also, when hospitalized, they remained twice longer in rehabilitation facilities and 3-times longer in acute care.

Conclusion: Using multiple code entries, our algorithm allowed an exhaustive identification of the French adult SCI population, with an updated epidemiology. This innovative method opens the field for large-scale studies regarding medical history of persons living with SCI by the prism of medical expenses and habits.