CLINICAL PREDICTORS OF MALIGNANT ETIOLOGY IN PATIENTS PRESENTING WITH OBSTRUCTIVE JAUNDICE: A CROSS-SECTIONAL STUDY

Abstract

Background: Malignant and benign biliary obstruction can present with similar cholestatic symptoms. Clinical features available at first contact may help prioritize patients for expedited tissue diagnosis and biliary decompression, but their discriminatory value varies across referral settings. Methods: This post hoc secondary analysis included 100 adults with obstructive jaundice enrolled in a tertiary-care cross-sectional study. Final etiology was established by tissue acquisition and/or operative findings. The primary outcome was malignant obstruction. Age, sex, pruritus, right upper quadrant mass, abdominal pain, anorexia, and fever were evaluated using Welch’s t test, Fisher’s exact test, odds ratios, phi coefficients, and Firth bias-reduced logistic regression. Discrimination was assessed using receiver operating characteristic analysis and bootstrap optimism correction. Results: Malignant obstruction was present in 76 patients and benign obstruction in 24. Patients with malignancy were older than those with benign disease (61.70 ± 9.33 vs 38.83 ± 11.24 years; t=9.03; p<0.001; Cohen d=2.33). Pruritus and a right upper quadrant mass were positively associated with malignancy, whereas abdominal pain and fever were negatively associated. The univariable Firth odds ratio for malignancy was 9.90 per 10-year increase in age, 16.21 for pruritus, and 67.09 for a right upper quadrant mass. An age cutoff of ≥48 years had 94.7% sensitivity and 83.3% specificity. An exploratory five-variable Firth model had an apparent area under the receiver operating characteristic curve of 0.958 and an optimism-corrected value of 0.944, although individual adjusted estimates were imprecise. Conclusions: Older age, pruritus, and a palpable right upper quadrant mass were strong univariable markers of malignant biliary obstruction, while abdominal pain and fever favored benign disease. The apparent performance of an exploratory clinical model was high, but sparse cells, referral bias, and absence of external validation preclude use as a stand-alone decision rule.