Data source
We performed a retrospective cohort study using data extracted from the Medical Information Mart for Intensive Care Ⅲ (MIMIC Ⅲ) database (v1.4) which integrated deidentified and comprehensive clinical data of patients admitted to the Beth Israel Deaconess Medical Center (BIDMC) in Boston, Massachusetts [14]. MIMIC III database contains over 58000 hospital admissions data for adult patients and neonates admitted to various critical care units between 2001 and 2012. The Institutional Review Board of the BIDMC (Boston, MA, USA) and Massachusetts Institute of Technology (Cambridge, MA, USA) have approved the use of MIMIC III database for authorized users. Wei Zhou was allowed to download data from the database, having completed the “Data or Specimens Only Research” course (record identity: 25222342).
External validation was collected from the First Affiliated Hospital of Wenzhou Medical University (Wenzhou, Zhejiang, China) after approval from that institution’s Ethical Committee.
Informed consents of all patients were not required because the present study neither contained any protected health information nor impacted clinical care.
Study cohort
A flowchart of the inclusion and exclusion procedure for the MIMIC III was depicted in Figure 1. We adopted the third international consensus definitions (Sepsis-3, a diagnosis flowchart was presented in Figure S1) to extract patients with sepsis and septic shock from the database [1]. Based on the Sepsis-3 criteria, patients with suspected infection and evidence of organ dysfunction [Sequential Organ Failure Assessment (SOFA) score ≥ 2] were identified as septic patients [1]. Suspected infection was defined as the concomitant administration of antibiotics and sampling of body fluid cultures (blood, urine, sputum, etc) [1]. In other words, if the culture was obtained, the antibiotic was required to be administered within 72 hours, whereas if the antibiotic was first, the culture was required within 24 hours [1]. Moreover, we defined the period of suspected infection as ranging between 24 hours before and 24 hours after admission to an ICU. Patients in the CareVue and MetaVision information systems of MIMIC III were admitted before and after 2008, respectively. Only patient data stored in the MetaVision system were collected for analysis. Antibiotic prescription data were only available after 2002, thus, there was a fraction (1/7) of the CareVue patients who had missing data for the suspected infection definition. It was the simplest option for us to limit the cohort to the MetaVision system, because the resulting sample size was sufficient. Additionally, the exclusion criteria for the initial sepsis cohort were as follows: (1) repeat hospitalization at ICU, (2) aged 16 years or younger, and (3) current service relating to cardiac, vascular or thoracic surgery. We assumed that these sub-populations had physiological abnormalities yet caused by factors unrelated to sepsis. Furthermore, we excluded the patients who had incomplete covariate data for further multivariate analysis.
External validation data were collected between September 15, 2018 and December 31, 2020 according to the same inclusion and exclusion criteria. The main diagnosis of these patients clearly met the Sepsis-3 criteria within 24 hours of ICU admission. The clinical outcomes were followed up for 90 days after admission (13 patients were excluded due to loss to follow-up).
Data extraction
The data were extracted from MIMIC III and our hospital system, including gender, age, laboratory data, vital statistics, comorbidities, ICU interventions, and hospital length of stay (LOS). Severity scores of illness including Simplified Acute Physiology Score Ⅱ (SAPS Ⅱ), Acute Physiology and Chronic Health Evaluation Ⅱ (APACHE Ⅱ) and SOFA were calculated on the basis of their predefined criteria [15,16,17]. The mean values of laboratory data and vital statistics during the first 24 hours of ICU stay were regarded as baseline data. The scores of Glasgow coma scale (GCS), SAPS Ⅱ, APACHE Ⅱ and SOFA as well as the necessity to perform interventions with vasopressor and mechanical ventilation were evaluated during the first 24 hours of ICU stay. Additionally, SAPS Ⅱ and APACHE Ⅱ were used for MIMIC III and external validation data analysis, respectively.
Predictor and outcome variables
We recorded the FFP transfusion status of each patient during the first 3 days of their ICU stays. To minimize the potential bias, the values of international normalized ratio (INR) and partial thromboplastin time (PTT) were obtained before FFP transfusion.
The primary end point was 28-day mortality. The secondary end points were 90-day and in-hospital mortality. Mortality information in the MIMIC III was calculated based on the dates of admission and death obtained from social security records.
Statistical analysis
Kolmogorov-Smirnov normality test was used to check the normality assumption for numerical variables. Differences in the normally and non-normally distributed variables were compared using the unpaired Student’s t‑test and Wilcoxon rank-sum test, respectively. Comparisons for categorical variables were performed by Pearson χ2 test and Fisher exact test. Normally distributed data were expressed as the means with standard deviations, and non-normally distributed data were expressed as the medians with inter-quartile ranges (IQRs). Categorical variables were expressed as frequencies with percentages.
We assessed the association of early FFP transfusion with survival in septic patients using logistic regression and Kaplan-Meier (K-M) analysis. The results were presented in form of odds ratios (ORs) with 95% confidence intervals (CIs) and survival curve, respectively.
For the cox regression analysis, 3 multivariate models were constructed as follows: Model 1, adjusting only for gender and age; Model 2, adjusting for gender, age, and scores of SAPS Ⅱ (APACHE Ⅱ for external validation) and SOFA; Model 3, adjusting for gender, age, laboratory data (white blood cell, platelet, hemoglobin, lactate and creatinine), vital statistics (heart rate, mean blood pressure, respiration rate, temperature, pulse oxygen saturation and glucose), scores of GCS, SOFA and SAPS Ⅱ (APACHE Ⅱ for external validation), ICU interventions (vasopressor, mechanical ventilation and renal replacement therapy), history of alcohol abuse, comorbidities, and hospital LOS. The hazard ratios (HRs) and 95% CIs were calculated for these models.
Sensitivity analysis was performed to further validate the effects of early FFP transfusion in septic patients with low coagulation and non-low coagulation status. Moreover, subset analysis was performed for patients with FFP transfusion (N = 288) to evaluate the relationship between infusion volume of FFP and survival. Subsequently, we performed an additional subset analysis to establish whether similar results also existed in septic shock cohort (N = 625). Finally, external validation was introduced to verify whether similar results can be observed in the East Asian population.
A two-sided P-value < 0.05 was regarded as representing statistical significance. Statistical analyses were performed using SPSS software 20.0 (SPSS, Chicago, IL, USA) and MedCalc software 19.0.5 (MedCalc, Ostend, Belgium).