Percentage of small platelets on peripheral blood smear and Child-Turcott-Pugh class can predict the presence of oesophageal varices in newly diagnosed patients with cirrhosis: development of a prediction model for resource limited settings

Abstract

BACKGROUND:In cirrhosis upper-gastrointestinal-endoscopy (UGIE) identifies oesophageal varices (OV). UGIE is unavailable in most resource-limited settings. Therefore, we assessed prediction of presence of OV using hematological parameters (HP) and Child-Turcott-Pugh (CTP) class.METHODS:A prospective study was carried out on consecutive, consenting, newly-diagnosed patients with cirrhosis, in the University Medical Unit, Colombo North Teaching Hospital, Ragama, Sri Lanka from April 2014-April 2016. All patients had UGIE to evaluate presence and degree of OV, prior to appropriate therapy. HP (full blood count with indices using automated analyzer and peripheral blood smear using Leishmann stain) and CTP class were assessed on admission. Linear logistic regression model was developed to predict OV using HP and CTP class.RESULTS:54-patients with cirrhosis were included [14(26%), 24(44%) and 16(30%) belonged to CTP class A, B and C respectively]. 37 had varices [CTP-A 4/14(26.6%), CTP-B 19/24(79.2%), CTP-C 14/16(87.5%)] on UGIE. Generalized linear model fitting showed decreasing percentage of small platelets (%SP) (P = 0.002), CTP-B (P = 0.003) and CTP-C (P = 0.003) compared to CTP-A had higher probability of having OV. The model predicts the log odds for having OV = - 0.189 - (0.046*%SP) + 2.9 [if CTP-B] + 3.7 [if CTP-C]. Based on receiver operating characteristic (ROC) analysis, a model value > - 0.19 was selected as the cutoff point to predict OV with 89%-sensitivity, 76%-specificity, 89%-positive predictive value and 76%-negative predictive value.CONCLUSIONS:We constructed a model using %SP on peripheral blood smear and CTP class. This model may be used to predict the presence of OV, in newly diagnosed patients with cirrhosis, with acceptable sensitivity and specificity, to prioritize the patients who deserve early UGIE in limited resource settings