A model based on routine liver tests can reliably exclude intrahepatic cholestasis of pregnancy

•A retrospective cohort study among 640 women evaluated for ICP.•In all women serum bile acids levels were tested along with routine liver tests.•A combined laboratory score incorporating AST, GGT, ALK and TB was shown to reliably exclude the diagnosis of ICP.•The current study findings may be particularly important as serum bile acids levels testing is not readily available. Serum bile acid (BA) levels testing is used for the diagnosis of intrahepatic cholestasis of pregnancy (ICP). We aimed to determine the performance of routine liver tests in the evaluation of ICP. A retrospective cohort study conducted at a university hospital, including all pregnant women who underwent serum BA levels testing due to suspected ICP during 2007–2019. Liver tests were performed in all women including: aspartate aminotransferase (AST), alanine transaminase (ALT), alkaline phosphatase (ALK), gamma-glutamyl transferase (GGT), and total bilirubin (TB). The optimal combination of laboratory values was determined by an algorithm developed in the Python programming language. Of 640 women who met the inclusion criteria, 22% (n = 142) were diagnosed with ICP (serum BA>10 μmol/L). A combined laboratory score of: (TB>11 μmol/L) or (ALK>255 U/L) or (GGT>32 U/L) or (AST>31 U/L), had a sensitivity of 94%, negative predictive value (NPV) of 97%, specificity of 50%, positive predictive value of 35%, and a negative likelihood ratio of 0.11 for the diagnosis of ICP. The AUC of the laboratory model alone was 0.72 (95% CI: 0.69–0.75). The addition of history of ICP to the suggested laboratory score resulted in a sensitivity of 97%, NPV of 98% and a negative likelihood ratio of 0.06. The AUC of the final model was 0.76 (95% CI: 0.72–0.79). A combined laboratory score incorporating AST, GGT, ALK and TB was shown to reliably exclude the diagnosis of ICP. This may be particularly useful in settings with limited access to BA levels testing. Of 640 with suspected ICP, 22.2% (n = 142) were diagnosed with ICP (serum BA>10 μmol/L). A combined laboratory score of: (TB>11 μmol/L) or (ALK>255 U/L) or (GGT>32 U/L) or (AST>31 U/L), had a sensitivity of 94.4%, negative predictive value (NPV) of 96.9%, and a negative likelihood ratio of 0.11 for the diagnosis of ICP. The addition of history of ICP resulted in a sensitivity of 97.2%, NPV of 98.3% and a negative likelihood ratio of 0.06. The AUC of the final model was 0.76 (95% CI:0.72–0.79). [Display omitted]