Abstract WP353: Assessing the Accuracy of ICD-10 Codes for Ruptured Intracranial Aneurysm Location

• Published in: Stroke (1970) 2020-02, Vol.51 (Suppl_1)
• Main Authors: Roark, Christopher, Kubes, Sheila, Mayer, David, Wiley, Laura K
• Format: Article
• Language: English
• ISSN: 0039-2499; 1524-4628
• DOI: 10.1161/str.51.suppl_1.WP353

Abstract only The adoption of ICD10 in October 2015 offers important opportunities for administrative database and EHR-based aneurysm research as it includes more than twenty distinct codes for subarachnoid hemorrhage (SAH), for the first time labeling the specific location for common sites of aneurysm rupture. Aneurysm location is a significant risk factor for rupture, but was not specified in any previous ICD classification. This has limited the value of “big-data” research for aneurysm risk modeling. Currently, it is not clear how frequently - or accurately - the codes are being used in practice. This study reviewed three years of records (10/2015-9/2018) from the UCHealth system in Colorado to assess the accuracy and utility of these codes for stroke research. Over this time, ICD10 codes for SAH (I60.x) were used a total of 5,090 times in 1,842 patients. After removing non-aneurysmal SAH (I60.8), the majority (92.7%) of the codes had no location specified (I60.7, I60.9). Of the codes with a location (I60.00-I60.6), upon review of the clinical record 9 (2.5%) did not have any evidence of subarachnoid hemorrhage, 3 (0.8%) did not have an aneurysm, and 5 (1.4%) had a pseudoaneurysm. Of the 338 verified instances of aneurysmal subarachnoid hemorrhage, 82 (24%) had multiple aneurysms and were excluded from further analysis. Within those records that had only a single aneurysm 197/256 (77%) had the location coded correctly. Of the 23% that were incorrectly coded, 1 was coded as the correct artery on the wrong side, all remaining (n=58) were the wrong artery. Although ICD10 codes offers the possibility of more advanced analysis due to the inclusion of aneurysm location, based on these data, we found that codes specifying location are frequently incorrect. Future studies are needed to validate these findings in other settings. However, researchers should be cautious about their ability to detect aneurysm location from administrative billing data using ICD10 location mapping.