Abstract
BACKGROUND: Early diagnosis is important to lower the mortality rate of acute fatty liver of pregnancy (AFLP). The Swansea criteria is commonly used to diagnose AFLP, but some terms could only be reached when symptoms and signs have progressed, or are not efficient in clinical practice. Therefore, it is necessary to select cost effective tests to simplify and facilitate early suspicion of acute fatty liver of pregnancy. METHODS: This is a retrospective study of 28,800 medical records at the Second Xiangya Hospital from 2009 to 2015, including 41 patients with AFLP and 172 other diseases that could show similar symptoms to AFLP. The evaluated variables included past history of liver diseases, blood pressure, gastrointestinal symptoms, blood count, liver function test, coagulation function test and blood sugar test. The sensitivity, specificity, positive predict value and negative predict value were calculated for models in diagnosing AFLP. RESULTS: The significant variables associated with AFLP included gastrointestinal symptoms, blood pressure > 140/90 mmHg, aminotransferase> 42 IU/l, total bilirubin> 0.8 mg/dl, total bilirubin acid> 10.0 μmol/L, activated partial prothrombin time(APTT) > 34 s, prothrombin time(PT) > 14 s, white blood cells> 11 *10(6)/l and blood sugar< 72 mg/dl. Gastrointestinal symptoms +aminotransferase +bilirubin +bile acid +APTT/PT showed 97.6% sensitivity and 97.1% specificity to diagnose AFLP. Adding blood pressure, blood sugar or white blood cells decreased the accuracy of the statistical model. CONCLUSIONS: Application of a statistical model including maternal symptoms, biochemical and haematological parameters has high diagnostic accuracy for earlier identification of AFLP. However, this finding needs to be tested in another cohort to determine whether this statistical model has the same performance.