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

Algorithm accurately measures CTP score, predicts survival for cirrhosis

A new algorithm quantified Child-Turcotte-Pugh scores from administrative databases, appearing to be a useful predictor for survival among patients with cirrhosis.

Researchers tested multiple algorithms for determining each Child-Turcotte-Pugh (CTP) subscore by gathering International Classification of Diseases version 9 code for cirrhosis (ICD9-CM), Common Procedural Terminology (CPT), pharmacy and laboratory data from two patient cohorts; patients with cirrhosis and hepatocellular carcinoma and patients with cirrhosis without HCC. For each cohort, six sites were included from the following areas: Boston, MA, Bronx, NY, Brooklyn, NY, Philadelphia, PA, Minneapolis, MN, and West Haven VA Medical Center in CT, and provided cases from which they used to determine validity of diagnosis, laboratory data and clinical assessment of ascites and hepatic encephalopathy (HE), according to the research.

“To validate the algorithms, six site investigator teams were each provided 25 randomly selected cirrhotic cases with HCC to independently extract CTP subscores blind to the algorithm results as a first validation cohort,” the researchers wrote.

After the first validation, each of the site investigators were provided random cases of patients with cirrhosis without HCC to use for second validation. An optimal algorithm known as the electronic CTP (eCTP) was then applied to 30,840 patients with cirrhosis who had various survival data available for examination.

Overall, researchers evaluated four ascites and two HE algorithms that used all the patient data, including one of each in the final analysis. 

Spearman correlations for all data in the cirrhosis and HCC and cirrhosis-only cohorts were 0.85 and 0.92 for bilirubin; 0.92 and 0.87 for albumin; and 0.84 and 0.86 for international normalized ratio.

Among patients in the HCC and cirrhosis cohort, the overall eCTP score matched 96% of patients to within one point of the chart-validated CTP score and a Spearman correlation of 0.81. Among the cirrhosis-only cohort, 98% were matched to within one point of their actual CTP score with a Spearman correlation of 0.85.

When the algorithm was applied to the 30,840 patients with cirrhosis, unit changes were observed and each change in eCTP was associated with a 39% increase in the risk for transplantation or mortality.

“The Harrell C statistic for the eCTP was numerically higher than those for other disease severity indices for predicting 5-year transplant-free survival,” the researchers wrote. “Adding other predictive models to the eCTP resulted in minimal differences in its predictive performance.”

The researchers concluded: “When applied to an administrative database, this algorithm may be a highly useful predictor of survival for patients with advanced liver disease.” – by Melinda Stevens

Disclosures: The researchers report no relevant financial disclosures.