Research article
Validation of ICD-9-CM/ICD-10 coding algorithms for the identification of patients with acetaminophen overdose and hepatotoxicity using administrative data
-
* Corresponding author: Robert P Myers rpmyers@ucalgary.ca
1 Liver Unit, Division of Gastroenterology, Department of Medicine; University of Calgary, Calgary, Alberta, Canada
2 Department of Community Health Sciences; University of Calgary, Calgary, Alberta, Canada
BMC Health Services Research 2007, 7:159 doi:10.1186/1472-6963-7-159
Published: 2 October 2007Abstract
Background
Acetaminophen overdose is the most common cause of acute liver failure (ALF). Our objective was to develop coding algorithms using administrative data for identifying patients with acetaminophen overdose and hepatic complications.
Methods
Patients hospitalized for acetaminophen overdose were identified using population-based administrative data (1995–2004). Coding algorithms for acetaminophen overdose, hepatotoxicity (alanine aminotransferase >1,000 U/L) and ALF (encephalopathy and international normalized ratio >1.5) were derived using chart abstraction data as the reference and logistic regression analyses.
Results
Of 1,776 potential acetaminophen overdose cases, the charts of 181 patients were reviewed; 139 (77%) had confirmed acetaminophen overdose. An algorithm including codes 965.4 (ICD-9-CM) and T39.1 (ICD-10) was highly accurate (sensitivity 90% [95% confidence interval 84–94%], specificity 83% [69–93%], positive predictive value 95% [89–98%], negative predictive value 71% [57–83%], c-statistic 0.87 [0.80–0.93]). Algorithms for hepatotoxicity (including codes for hepatic necrosis, toxic hepatitis and encephalopathy) and ALF (hepatic necrosis and encephalopathy) were also highly predictive (c-statistics = 0.88). The accuracy of the algorithms was not affected by age, gender, or ICD coding system, but the acetaminophen overdose algorithm varied between hospitals (c-statistics 0.84–0.98; P = 0.003).
Conclusion
Administrative databases can be used to identify patients with acetaminophen overdose and hepatic complications. If externally validated, these algorithms will facilitate investigations of the epidemiology and outcomes of acetaminophen overdose.