Nguyen et al.1 recently described an increased risk of acute liver injury (ALI) due to acetaminophen overdose (AO) in patients with hepatitis C virus (HCV). ALI occurred in 16.7% and 7.1% of HCV-positive and HCV-negative patients, respectively [odds ratio (OR), 1.64; 95% confidence interval (CI), 1.18–2.26]. Alcoholic and nonalcoholic fatty liver diseases also conferred 5- to 8-fold risks. This commendable study adds to the literature supporting an association between liver disease and acetaminophen-related ALI.2–4 In an analysis of 1543 AO admissions identified with a Canadian database,4 we reported an increased risk of ALI (OR, 3.50; 95% CI, 1.57–7.77) and mortality (15.2% versus 0.5%; P < 0.0005) in patients with pre-existing liver disease. Although this study included a small number of HCV patients (n = 10), it provided insight into additional risk factors. Specifically, ALI was more common in patients with unintentional overdoses (OR, 5.18; 95% CI, 3.00–8.95) and alcohol abuse (OR, 2.21; 95% CI, 1.30–3.76).4 Because patients with liver diseases including HCV are more likely to overdose unintentionally,4 we wondered whether the increased risk reported by Nguyen et al. could be explained by differences in overdose circumstance. Therefore, we reanalyzed the Nationwide Inpatient Sample database5 with adjustment for this factor using codes for external causes of injury (ICD-9-CM E850.4, E935.4) that have been validated for defining suicidal intent due to poisoning.4, 6, 7 To reduce unmeasured confounding, we adjusted for race, hospital characteristics, and alcohol abuse8 in addition to previously reported factors.1 These characteristics differed significantly according to HCV status. We also adjusted for case mix with the Elixhauser algorithm rather than the Charlson/Deyo algorithm because it is more discriminative in patients with liver disease.8, 9 Finally, we performed a sensitivity analysis excluding patients with acute HCV (ICD-9-CM codes 070.41 and 070.51) because differentiating viral ALI versus drug-related ALI is difficult in this setting. All analyses employed SAS-callable SUDAAN (version 9.0.1, Research Triangle Institute, Research Triangle Park, NC) to account for the complexities of Nationwide Inpatient Sample data.5
Between 1998 and 2005, we identified 42,718 AO hospitalizations; 705 (1.7%) had HCV, and 36% of these infections were acute. HCV patients had a higher prevalence of unintentional ingestions (26% versus 13%), alcohol abuse (42% versus 22%), and comorbidities (≥3: 24% versus 13%) and were more frequently hospitalized in urban teaching centers (52% versus 41%; P < 0.0001). ALI and liver failure7 occurred in 8.2% and 4.5% of patients, respectively; 1.4% died. ALI (16.7% versus 8.0%; P < 0.0001), liver failure (11.4% versus 4.4%; P < 0.0001), and death (3.0% versus 1.4%; P = 0.01), but not transplantation (0.13% versus 0.08%; P = 0.68), were more common in HCV patients. The risk of ALI was higher in HCV patients with intentional (12.2% versus 6.4%) and unintentional (29.6% versus 18.4%; P < 0.005) overdoses. In multivariate analysis, HCV remained a significant predictor of ALI (OR, 1.42; 95% CI, 1.12–1.79; Fig. 1) and liver failure (OR, 1.60; 95% CI, 1.22–2.10) but not mortality. Sensitivity analyses excluding patients with acute HCV (OR for ALI, 1.52; 95% CI, 1.15–2.00) and cirrhosis revealed similar results. Unintentional AOs (OR, 2.94; 95% CI, 2.69–3.21), alcohol abuse (OR, 1.10; 95% CI, 1.00–1.20), and African American (versus whites: OR, 0.72; 95% CI, 0.62–0.0.84) and Hispanic race (OR, 0.53; 95% CI, 0.44–0.64) were also associated with hepatotoxicity.
In summary, our reanalysis supports Nguyen and colleagues' findings1 of an increased risk of acetaminophen-related hepatotoxicity in patients with HCV, although the magnitude of this risk was attenuated. We confirmed observations that AO-related ALI is more common in patients with unintentional overdoses and alcohol abuse.4, 10–12 These findings may have substantial public health importance because rates of unintentional AO are rising10, 13 It emphasizes the necessity of clear package labeling of acetaminophen content and perhaps reductions in the maximum daily dosage recommended in at-risk populations.2 If this is confirmed, studies exploring the mechanisms behind the racial differences that we observed may help explain heterogeneity in AO outcomes.