Prognostic score in liver cirrhosis developed using the Cox's proportional hazard regression model

In order to assess the prognostic value of clinical and laboratory variables in liver cirrhosis, 36 of these variables were statistically analyzed in 151 patients followed up for 8 years. The 'survival time' was taken as the reference variable. In a first step we analyzed by log-rank test and by Cox's proportional hazard regression model the data of 98 patients (study group), obtaining 7 prognostically significant variables (age, leukocytes, calcium, potassium, globulins, cholesterol and previous diagnosis). From the regression coefficients of these variables, a risk score was obtained for each patient. To validate the prognostic value of this score, we computed it, using the same coefficients obtained in the study group, in 53 subsequently examined patients (control group) showing that the prognostic score allows the classification of these patients in 3 risk classes with different observed survival times.