This article is more than 5 years old. Information may no longer be current.

Hepatic encephalopathy risk score accurately predicts incidence

Elliot B. Tapper

A simple, four-component risk score stratified patients with cirrhosis by probability for hepatic encephalopathy, according to results from a recently published study.

“Hepatic encephalopathy is devastating and efforts to inform risk assessment are lacking,” Elliot B. Tapper, MD, from the University of Michigan, told Healio Gastroenterology and Liver Disease. “This score helps identify high and low risk patients. This score can be updated after a change in status. Patients can go from high to low risk if they quit alcohol or are cured of hepatitis C — and vice-versa.”

Between January 2005 and December 2010, 863 of 1,979 patients with cirrhosis from the Veterans Administration in Michigan, Indiana, and Ohio developed hepatic encephalopathy (HE).

“Because we used the VA data, we could follow people over time and update their risk continuously,” Tapper said.

Of the medications examined, patients who developed HE were more significantly less likely to use statins (13.3% vs. 19.2%; P = .0005) compared with those who did not develop HE. While it did not meet significance, more patients who developed HE used nonselective betablockers (13% vs. 10.4%) and most of those patients received propranolol.

“Medication use is a good way to improve risk assessment in administrative data and generate hypotheses for prospective trials,” Tapper said.

Multivariate analysis showed that baseline bilirubin (HR = 1.068; 95% CI, 1.048-1.088), albumin (HR = 0.543; 95% CI, 0.493-0.597), statin use (HR = 0.74; 95% CI, 0.608-0.901), and nonselective betablocker use (HR = 1.268; 95% CI, 1.036-1.551) correlated with HE development. The area under the receiver operating characteristics curve for these four predictors was 0.68. In a longitudinal model for annual risk, the AUROC was 0.73.

In the baseline risk score — including bilirubin, albumin, betablocker use, and statin use — a score of –10 or less correlated with a 5-year risk for HE of 27%, while a higher score correlated with a risk of 49% or higher.

In the longitudinal model, which factored in lab or medication changes, a score of 0 or less correlated with a 6% risk for HE in the following year, while a score of 1 correlated with a 25% risk for HE over the following year.

“If a patient is high risk, they should be monitored closely, neurocognitive testing should be considered, nutrition should be improved, and risky medications (such as sedating meds or opioids) should be discontinued if possible,” Tapper said. He advised that the results would need to be validated in other cohorts. – by Talitha Bennett

Disclosure: Tapper reports no relevant financial disclosures. Please see the full study for the other authors’ relevant financial disclosures.