Perioperative blood loss and allogeneic blood transfusion are major complications of
cardiac surgery, which increase perioperative complications, perioperative mortality and
medical costs.This study is a non-inferiority, randomized controlled trial, in order to
determine whether PCC is no worse than FFP in reducing perioperative blood loss and
allogeneic blood transfusion in patients undergoing cardiac surgery under CPB. Patients
signed the informed consent, aged 18 to 80 years, receiving elective CABG or valve
replacement or valvuloplasty under CPB will be included. 796 subjects will be randomly
divided into 2 groups (group PCC and group FFP), with 398 cases in each group.
Preoperative management, anesthetic and surgical techniques will be standardized for both
groups. When APTT is prolonged (> 45 s), patients will be given 815 IU/kg PCC in group
PCC and 610 ml/kg FFP in group FFP. All the patients will be followed up respectively at
24 hours, 48 hours, 72 hours and 7 days after the surgery to record observations relevant
to the study and the results of laboratory testing. The laboratory tests include
hemoglobin concentration, hematocrit, platelet count, INR, PT, APTT, fibrinogen levels
and blood biochemistry parameters. The primary outcome is the volume of blood loss within
24 hours after surgery. The secondary outcomes include (1) the total units of allogeneic
red blood cells (RBCs) transfused within 7 days after surgery and (2) length of intensive
care unit (ICU) stay. Adverse events and serious adverse events will be monitored as
safety outcomes. Exploratory outcomes include re-exploration due to postoperative
bleeding within 7 days after surgery and length of hospital stay. Modified
intent-to-treat analysis will be used for all valid variables. All randomised subjects in
the study, regardless of adherence to the study process or whether an adverse event
occurs before or after the intervention, should be included in the analysis. Sensitivity
analysis will be performed to assess the bias that may be introduced due to nonadherence
to the protocol or missing data. Baseline characteristics will be tabulated and compared
between the PCC and FFP groups using absolute standardised differences, and a value
larger than 0.1 will be regarded as a clinically relevant difference between groups.
Unbalanced baseline factors will be further adjusted by multivariable regression models.
The primary outcome, the volume of blood loss within 24 hours after surgery, will be
compared using the t-test with log transformation of the variable. Continuous secondary
outcomes and the total units of allogeneic RBCs transfused during and within 7 days after
surgery will be compared using a t-test with log transformation of the variable. The rate
of re-exploration due to bleeding within 7 days after surgery will be compared using the
chi-square test. Treatment effect will be measured by risk ratio and mean difference for
binary and continuous outcomes with corresponding 95% confidence intervals. Bonferroni's
correction for multiple comparisons will be conducted in the analysis of the secondary
outcomes. If there exists unbalanced baseline characteristics, the primary outcome and
secondary outcomes will be regressed against the group allocation and the unbalanced
factors using liner regression and Cox regression models. Proportional hazard assumption
will be checked by the Schoenfeld's residual plot. For safety outcomes, we will only
describe the incidence of overall adverse events, SAEs, and component adverse events
without statistical tests between two groups. A two-sided P-value < 0.05 was considered
indicative of statistical significance.
Our independent Data Safety Monitoring Board (DSMB), which includes epidemiologists,
cardiovascular surgeons, anesthesiologists, and blood transfusion specialists, conducted
an interim review of our study. Considering the shortage of blood resources and the
difficulties in large-sample recruitment, the DSMB recommended adding an interim
analysis. Therefore, we decided to conduct an interim analysis after recruiting 50% of
the patients, applying Bonferroni correction for alpha spending with a two-sided alpha of
0.025. And the final analysis will be at 100% patients recruited with a two-sided alpha
of 0.025. With the same standard deviation (824.38mL) and the superior margin (200mL) is,
as well as the 90% power and 10% dropout rate, we recalculated the sample size and found
out that 938 patients in total should be recruited. Thus our interim analysis will be
conducted when 469 patients recruited.