Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review

Area of Science:

• Medical Informatics
• Public Health Surveillance
• Epidemiology

Background:

• Administrative data are crucial for health research but require accurate coding.
• The validity of International Classification of Diseases (ICD) codes for stroke identification in administrative databases needs systematic evaluation.

Purpose of the Study:

• To systematically review studies assessing the accuracy of ICD codes for identifying stroke in administrative health data.
• To determine the sensitivity, specificity, and predictive values of ICD codes for various stroke types.

Main Methods:

• A systematic literature search of MEDLINE and EMBASE (inception to February 2015) was conducted.
• Studies evaluating stroke code validity in administrative data, reporting key validation statistics, were included.
• The Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool was used for quality appraisal.

Main Results:

• Seventy-seven studies (1976-2015) were included, evaluating ICD-9 (430-438) and ICD-10 (I60-I69) codes.
• High sensitivity (≥82%) and specificity (≥95%) were observed for any cerebrovascular disease.
• Positive predictive values varied, with lower accuracy for acute stroke (≤68%) compared to specific subtypes like subarachnoid hemorrhage (≥93%).

Conclusions:

• ICD codes 430-438/I60-I69 collectively detect prevalent cerebrovascular disease well.
• Specific ICD codes are necessary for accurate identification of acute stroke.
• Linking administrative and vital statistics data may enhance the detection of fatal stroke cases.