Patient enrolment and clinical characteristics
During the study period, 298 patients with the diagnosis of AFLP, and 8 patients were excluded due to chronic hepatitis B, chronic hepatitis C and liver cirrhosis, 290 patients were included in the final analysis, the patients’ disposition was shown in Fig. 1.
The clinical characteristics of patients were shown in Table 1, of totally two hundreds and ninety cases, fifty-seven were multiparae (n = 57, 19.7%), two hundreds and thirty-three were male of fetal sex (n = 233, 80.3%), and 8.6% were multiplets (n = 25, 8.6%). The women were 27.21±4.95 years of age at their AFLP diagnosis, gave birth at a mean gestational age of 253.10±17.58 days. The majority comorbidities were fetal distress (n = 130, 44.82%), encephalopathy (n = 81, 27.93%), infection (n = 60, 20.68%), fetal death (n = 78, 26.89%). Cesarean section and CRRT were used in 267 (24.9%) and 107 (4.9%) patients, respectively. A total of 79 patients (27.24%) received Plasma exchange (PE) treatment, and 22 patients (7.59%) received bilirubin adsorption treatment. 34 patients took in the surgery of hysterectomy (11.72%).
Clinical characteristics and managements of AFLP patients associated to death
In 290 patients with AFLP in this study, survival group contained 240 patients (82.8%), and a total of 50 patients (17.2%) were dead. As shown in Table 1, the death group had longer days from symptom onset to hospitalization and days from symptom onset to parturition than the survival group (p < 0.01, respectively), and APACHE II score in death group were higher (p < 0.05). Compared with AFLP patients alive, the dead of patients were more likely to be combined encephalopathy (p < 0.01), postpartum hemorrhage (p < 0.01), and other complications (p = 0.03). In terms of fetal/infant complications, a significantly higher frequency of fetal distress (p = 0.04), fetal death (p < 0.01) was found in death group than in survival group. There was no significant difference between dead and survival patients regards to laboratory examination and pregnancy information (gravidity, multiparae, fetal sex male, gestational age, and multiplets) except the total bilirubin (TBil) and alanine transaminase (ALT) (p < 0.05, respectively) (Table 1).
In addition, patients in death group were more likely to receive treatments of Bilirubin adsorption and CRRT, these differences had statistical significance (p < 0.01, respectively). Interestingly, there was no significant difference between the two groups with regards to management of plasma exchange, cesarean section, and hysterectomy (Table 1).
Based on above data which p value < 0.05, we developed a predicted probability value and using a binary logistic regression model with an area under the receiver operating characteristics curve of 0.94 (95%CI 0.87 to 1.00) (Fig. 2), indicating good discrimination of AFLP patients’ death.
Patients with PE and development of propensity score
The clinical characteristics of the AFLP patients received PE treatment or not were summarized in Table 2, a total of 79 patients (27.2%) were treated with PE (Table 2). As is show in Table 2, there were a significantly higher or lower frequency of maternal information and obstetrical complications in PE group (group A) than group B: multiparae (10.13% vs. 23.22%, p = 0.01), fetal sex male (68.35% vs. 84.83%, p < 0.01), multiplets (24.05% vs.2.84%, p < 0.01), HELLP syndrome (20.25% vs.8.53%, p = 0.01), and placenta previa (21.52% vs. 7.58%, p < 0.01). No difference was found of gravidity (1.71±0.76 vs. 1.68±0.84, p = 0.72), gestational age (241.43±16.56 vs. 256.82±16.54, p = 0.78), APACHEII score (p = 0.78) and Swansea score (p = 0.71) between two groups (Table 2).
When compared the laboratory examination we found that higher mean values of prothrombin time (PT) (34.57±38.01 vs. 18.65±6.25, p = 0.01) and white blood cells (17.45±9.11 vs. 15.28±5.07, p = 0.003) in Group A. Besides, more patients in Group A received bilirubin adsorption (25.32% vs.0.96%, p < 0.01), CRRT (83.54% vs.19.43%, p < 0.01) and cesarean section (97.47% vs.90.05%, p = 0.04) other than hysterectomy (p = 0.44) (Table 2).
In propensity score-matched cohort study, 79 AFLP patients treated with PE were successfully matched with 79 AFLP patients without PE (Table 2), including 46 cases for exact matches and 33 cases for fuzzy matches. There were no significant differences of the baseline characteristics between the matched patients with and without PE.
Plasma exchange as a protective factor for clinical outcome
There was no significant difference of maternal crude 60-day mortality rates (p = 0.65) and hospital-free days at day 28 (p = 0.85) between Group A and Group B (Table 3). After propensity score matching, PE group had a significantly lower 60-day mortality rate (OR 0.42, 95% CI 0.29 to 2.64, p = 0.04) (Table 3 and Fig. 3), and shorter hospital-free days at day 28 (SD 13.57, 95% CI 4.02 to 20.11, p = 0.01) (Table 3).