Establishment of a model to predict the prognosis of pregnancy-related acute kidney injury

BACKGROUND: Our aim was to investigate the prognostic factors of pregnancy related acute kidney injury (PR-AKI), establish the prognosis prediction model and evaluate predictability with the model. METHODS: Serum creatinine (SCr), placental growth factor (PLGF), endothelin-1 (ET-1), urinary kidney injury molecule-1 (KIM-1), and β2-microglobulin (β2-MG) levels in 122 patients with PR-AKI were measured at 1 month, 3 months, 6 months, and 1 year after delivery and the logistic backward stepwise regression method was used to screen the factors associated with the prognosis, establish the model to predict the prognosis of patients 1 year after delivery, and evaluate predictability with the model. RESULTS: 1) At 1 month, 3 months, 6 months, and 1 year after delivery, SCr was not significantly different in the poor prognosis group (P˃0.05); ET-1, KIM-1 and β2-MG were decreased significantly, and PLGF was significantly increased (P˂0.05); all testing indicators in the good prognosis group were significantly different and returned to normal within 1 year (P˂0.05). By comparing the two groups, there were no significant differences in SCr, PLGF and β2-MG levels at 1 month after delivery (P˃0.05), while there were significant differences in the remaining testing indicators at each time point (P˂0.05). 2) Logistic backward stepwise regression analysis showed that SCr, KIM-1, ET-1, and time were related to the prognosis of patients with PR-AKI, and the above screening indicators were used to fit the model Logit. CONCLUSIONS: The logistic regression model fitting-effect was good, its accuracy of predicting renal injury prognosis was high. SCr, KIM-1, and ET-1 can be used for PR-AKI screening and are important for monitoring the disease condition of patients, intervention, and evaluation of prognosis.