Validation of an algorithm that determines stroke diagnostic code accuracy in a Japanese hospital-based cancer registry using electronic medical records

BMC Medical Informatics and Decision Making

Research Article

Keichi Yamamoto¹  Daijiro Kabata²  Ayumi Shintani²  Hideki Mochizuki³  Yasufumi Gon³  Manabu Sakaguchi³  Kenichi Todo³ 

[1]Department of Drug and Food Clinical Evaluation, Osaka City University Graduate School of Medicine, Osaka, Japan

[2]Department of Medical Statistics, Osaka City University Graduate School of Medicine, Osaka, Japan

[3]Department of Neurology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, 565-0871, Osaka, Japan

关键词: Electronic medical record;    Diagnostic code;    Validation;    Clinical research;   

DOI  :  10.1186/s12911-017-0554-x

 received in 2017-06-30, accepted in 2017-11-19,  发布年份 2017

来源: Springer

【 摘 要 】

BackgroundThis study aimed to validate an algorithm that determines stroke diagnostic code accuracy, in a hospital-based cancer registry, using electronic medical records (EMRs) in Japan.MethodsThe subjects were 27,932 patients enrolled in the hospital-based cancer registry of Osaka University Hospital, between January 1, 2007 and December 31, 2015. The ICD-10 (international classification of diseases, 10th revision) diagnostic codes for stroke were extracted from the EMR database. Specifically, subarachnoid hemorrhage (I60); intracerebral hemorrhage (I61); cerebral infarction (I63); and other transient cerebral ischemic attacks and related syndromes and transient cerebral ischemic attack (unspecified) (G458 and G459), respectively. Diagnostic codes, both “definite” and “suspected,” and brain imaging information were extracted from the database. We set the algorithm with the combination of the diagnostic code and/or the brain imaging information, and manually reviewed the presence or absence of the acute cerebrovascular disease with medical charts.ResultsA total of 2654 diagnostic codes, 1991 “definite” and 663 “suspected,” were identified. After excluding duplicates, the numbers of “definite” and “suspected” diagnostic codes were 912 and 228, respectively. The proportion of the presence of the disease in the “definite” diagnostic code was 22%; this raised 51% with the combination of the diagnostic code and the use of brain imaging information. When adding the interval of when brain imaging was performed (within 30 days and within 1 day) to the diagnostic code, the proportion increased to 84% and 90%, respectively. In the algorithm of “definite” diagnostic code, history of stroke was the most common in the diagnostic code, but in the algorithm of “definite” diagnostic code and the use of brain imaging within 1 day, stroke mimics was the most frequent.ConclusionsCombining the diagnostic code and clinical examination improved the proportion of the presence of disease in the diagnostic code and achieved appropriate accuracy for research. Clinical research using EMRs require outcome validation prior to conducting a study.

【 授权许可】

CC BY   
© The Author(s). 2017