Validity of medico-administrative data related to amyotrophic lateral sclerosis in France: A population-based study

Amyotroph Lateral Scler Frontotemporal Degener. 2017 Feb;18(1-2):24-31. doi: 10.1080/21678421.2016.1241280. Epub 2016 Oct 31.

Abstract

The accuracy of French medico-administrative data concerning amyotrophic lateral sclerosis (ALS) is to date unknown. We aimed to assess the validity of hospital discharge data (HDD) and health insurance data (HID) related to ALS. A retrospective population-based study was performed. The French register of ALS in Limousin (FRALim) was used as gold standard (2000-2013 period). All patients discharged from the regional hospitals with a 'G12.2' code in their HDD (according to the International Classification of Disease-10th version) or having a G12 HID code were considered. In the study period, the register included a total of 322 incident ALS patients. Among 451 subjects identified through HDD, 290 were true incident ALS cases, corresponding to 90.1% (95% CI 86.3-93.1) sensitivity and 64.3% (95% CI 59.7-68.7) positive predictive value (PPV). A total of 184 subjects were identified through HID, 142 of which were true ALS cases. This corresponded to 44.1% (95% CI 38.6-49.7) sensitivity and 75.5% (95% CI 68.7-81.5) PPV. The combination of both HDD and HID led to 93.8% (95% CI 90.6-96.2) sensitivity and 60.8% (95% CI 56.3-65.1) PPV. This study shows that French HDD and HID, even if combined, are not per se suitable for accurate and exhaustive direct identification of ALS cases.

Keywords: Amyotrophic lateral sclerosis; administrative database research; epidemiology; hospital morbidity data; insurance data; positive predictive value; sensitivity.

MeSH terms

  • Adult
  • Aged
  • Aged, 80 and over
  • Amyotrophic Lateral Sclerosis / epidemiology*
  • Analysis of Variance
  • Community Health Planning
  • Female
  • France / epidemiology
  • Humans
  • Male
  • Middle Aged
  • National Health Programs / statistics & numerical data*
  • Odds Ratio
  • Patient Discharge / statistics & numerical data*
  • Registries / statistics & numerical data*
  • Reproducibility of Results
  • Retrospective Studies