P: 7 Identifying Overt Hepatic Encephalopathy Using Administrative Data in the ICD-10 Era: A Study to Validate Diagnostic Algorithms

Abstract

BACKGROUND: Administrative datasets are necessary in order to study the outcomes of patients with hepatic encephalopathy (HE) at the population-level. Validated algorithms using ICD-9 diagnostic codes have facilitated prior studies but are now invalid given the switch to ICD-10 in 2015. To date, no study has validated any diagnostic coding algorithm for HE using ICD-10 codes. METHODS: From 2016–2017 at the University of Michigan outpatient liver clinic, we prospectively enrolled 300 persons with Child A-B cirrhosis and portal hypertension with no current or prior history of HE. All patients were followed for up to 3 years. Each patient was assessed every 3 months for clinical developments (including new overt HE, falls, hospitalizations, and liver cancer) and recorded all medications taken. Overt HE was defined as disorientation that was clinically defined as HE by the patient's hepatologist. We surveyed all ICD-10 billing codes generated for each patient during follow up. We sought to evaluate the sensitivity and specificity of codes and medications for the presence of HE: K72.90 (hepatic failure), K72.91 (hepatic failure with coma), G93.40 (encephalopathy, NOS), G93.49 (Other encephalopathy), and prescription of lactulose or rifaximin. RESULTS: Overall our cohort was aged 60 (52–66) years, 56.3% male, and 70% Child class A. All patients had portal hypertension, 76% had varices, and 41% had a history of ascites (predominantly well controlled). The median MELD-Na score was 9 (IQR, 7–13). Overall 68 out of 301 patients (22%) developed overt HE during follow-up. Only 1 patient was assigned the code K72.91, whereas codes G930.40 and G93.49 were not coded in our claims data. These 3 codes offered 0% sensitivity and specificity for HE. Conversely, the ICD-code K72.90 had moderate sensitivity (41.2%) and high specificity (97.9%), positive predictive value (PPV) 84.8%, and negative predictive value (NPV) 85.1%. Recorded lactulose and or rifaximin use was highly sensitive (94.1%) and specific (98.3%), PPV 87.7%, NPV 98.3%. CONCLUSION: In this prospective study, we define the performance of diagnostic codes for the identification of HE in the electronic health record study. Medications specific for HE therapy outperformed diagnostic codes.