*Dossier de santé électronique basé sur l'algorithme pour identifier les patients avec trouble du spectre de l'autisme

Traduction: G.M.

PLoS One. 2016 Jul 29;11(7):e0159621. doi: 10.1371/journal.pone.0159621.

Electronic Health Record Based Algorithm to Identify Patients with Autism Spectrum Disorder

Lingren T¹, Chen P², Bochenek J³, Doshi-Velez F⁴, Manning-Courtney P^5,6, Bickel J^7,8, Wildenger Welchons L^7,8,9, Reinhold J^5,6, Bing N^5,6, Ni Y¹, Barbaresi W¹⁰, Mentch F¹¹, Basford M¹², Denny J³, Vazquez L¹¹, Perry C¹³, Namjou B^14,15, Qiu H¹¹, Connolly J¹¹, Abrams D¹¹, Holm IA^10,13,16,17, Cobb BA¹⁴, Lingren N¹⁸, Solti I^1,15, Hakonarson H^11,19, Kohane IS^16,20, Harley J^5,14,21, Savova G^16,20.

Author information

• ¹Cincinnati Children's Hospital Medical Center, Division of Biomedical Informatics, Cincinnati, Ohio, United States of America.
• ²Boston Children's Hospital, Center for Systems Biology, Boston, Massachusetts, United States of America.
• ³Vanderbilt University School of Medicine, Biomedical Informatics, Nashville, Tennessee, United States of America.
• ⁴Harvard Medical School, Center for Biomedical Informatics, Boston, Massachusetts, United States of America.
• ⁵University of Cincinnati, Department of Pediatrics, Cincinnati, Ohio, United States of America.
• ⁶Cincinnati Children's Hospital Medical Center, Division of Developmental and Behavioral Pediatrics, Cincinnati, Ohio, United States of America.
• ⁷Boston Children's Hospital, Pediatrics, Boston, Massachusetts, United States of America.
• ⁸Boston Children's Hospital, Developmental Medicine, Boston, Massachusetts, United States of America.
• ⁹Boston Children's Hospital, Neurology and Center for Communication Enhancement, Boston, Massachusetts, United States of America.
• ¹⁰Children's Hospital Boston, Division of Medicine, Boston, Massachusetts, United States of America.
• ¹¹Children's Hospital of Philadelphia, Center for Applied Genomics, Philadelphia, Pennsylvania, United States of America.
• ¹²Vanderbilt University Medical Center, Vanderbilt Institute for Clinical and Translational Research, Nashville, Tennessee, United States of America.
• ¹³Boston Children's Hospital, Division of Genetics and Genomics, Boston, Massachusetts, United States of America.
• ¹⁴Cincinnati Children's Hospital Medical Center, Center for Autoimmune Genomics and Etiology, Cincinnati, Ohio, United States of America.
• ¹⁵University of Cincinnati, College of Medicine, Cincinnati, Ohio, United States of America.
• ¹⁶Harvard Medical School, Pediatrics, Boston, Massachusetts, United States of America.
• ¹⁷Boston Children's Hospital, Manton Center for Orphan Disease Research, Boston, Massachusetts, United States of America.
• ¹⁸Cincinnati Children's Hospital Medical Center, Emergency Medicine, Cincinnati, Ohio, United States of America.
• ¹⁹Perelman School of Medicine, Pediatrics, Philadelphia, Pennsylvania, United States of America.
• ²⁰Boston Children's Hospital, Children's Hospital Informatics Program, Boston, Massachusetts, United States of America.
• ²¹United States Department of Veterans Affairs Medical Center, Medicine, Cincinnati, Ohio, United States of America.

Abstract

OBJECTIVE:

Cohort selection is challenging for large-scale electronic health record (EHR) analyses, as International Classification of Diseases 9th edition (ICD-9) diagnostic codes are notoriously unreliable disease predictors. Our objective was to develop, evaluate, and validate an automated algorithm for determining an Autism Spectrum Disorder (ASD) patient cohort from EHR. We demonstrate its utility via the largest investigation to date of the co-occurrence patterns of medical comorbidities in ASD.
La sélection d'une cohorte est difficile pour les analyse de dossier de santé électronique à grande échelle (DSE) , comme la Classification internationale des maladies, 9e édition (CIM-9) les codes de diagnostic sont des facteurs prédictifs de la maladie notoirement peu fiables. Notre objectif était de développer, d'évaluer et de valider un algorithme automatisé pour la détermination d'un trouble du spectre de l'autisme (TSA) dans une cohorte de patients à partir du DSE. Nous démontrons son utilité via la plus grande enquête à ce jour des modèles de co-occurrence de comorbidités médicales dans le TSA.

METHODS:

We extracted ICD-9 codes and concepts derived from the clinical notes. A gold standard patient set was labeled by clinicians at Boston Children's Hospital (BCH) (N = 150) and Cincinnati Children's Hospital and Medical Center (CCHMC) (N = 152). Two algorithms were created: (1) rule-based implementing the ASD criteria from Diagnostic and Statistical Manual of Mental Diseases 4th edition, (2) predictive classifier. The positive predictive values (PPV) achieved by these algorithms were compared to an ICD-9 code baseline. We clustered the patients based on grouped ICD-9 codes and evaluated subgroups.
Nous avons extrait les codes CIM-9 et les concepts issus des notes cliniques. Un ensemble de patients standard idéal a été caractérisé par des cliniciens de l'hôpital pour enfants de Boston (BCH) (N = 150) et de l'Hôpital et Centre médical pour enfants de Cincinnati (CCHMC) (N = 152). Deux algorithmes ont été créés: (1) à base de règles mettant en œuvre les critères TSA du Manuel diagnostique et statistique des maladies mentales 4e édition, (2) un classificateur prédictif. Les valeurs prédictives positives (PPV) obtenues par ces algorithmes ont été comparées à un code de référence de la CIM-9. Nous regroupé les patients sur la base des codes CIM-9 et des sous-groupes évalués.

RESULTS:

The rule-based algorithm produced the best PPV: (a) BCH: 0.885 vs. 0.273 (baseline); (b) CCHMC: 0.840 vs. 0.645 (baseline); (c) combined: 0.864 vs. 0.460 (baseline). A validation at Children's Hospital of Philadelphia yielded 0.848 (PPV). Clustering analyses of comorbidities on the three-site large cohort (N = 20,658 ASD patients) identified psychiatric, developmental, and seizure disorder clusters.
L'algorithme fondé sur des règles a produit le meilleur PPV: (a) BCH: 0,885 vs 0,273 (ligne de base); (B) CCHMC: 0,840 vs 0,645 (ligne de base); (C) combiné: 0,864 vs 0,460 (ligne de base). Une validation à l'Hôpital pour enfants de Philadelphie a donné 0,848 (PPV). Les analyses de regroupement de comorbidités sur vaste cohorte à  trois sites (patients N = 20658 TSA) ont identifiés des groupement de troubles psychiatriques, de développement, et de convulsion.

CONCLUSIONS:

In a large cross-institutional cohort, co-occurrence patterns of comorbidities in ASDs provide further hypothetical evidence for distinct courses in ASD. The proposed automated algorithms for cohort selection open avenues for other large-scale EHR studies and individualized treatment of ASD.
Dans une large cohorte interinstitutionnelle, les modèles de co-occurrence de comorbidités dans les TSA apportent une preuve supplémentaire hypothétique pour des trajectoires distinctes dans le TSA. Les algorithmes automatisés proposés pour la sélection de cohorte ouvrent des voies pour d'autres études de DSE à grande échelle et un traitement individualisé des TSA.

PMID: 27472449