Association between Fibrinogen-to-Albumin Ratio and Prognosis in Patients Admitted to an Intensive Care Unit

The objective of this study was to investigate the usefulness of fibrinogen-to-albumin ratio (FAR) as a prognostic marker in patients admitted to an intensive care unit (ICU) compared with Sequential Organ Failure Assessment (SOFA) score, a widely used prognostic scoring system. An inverse probability weighting (IPW) was used to control for selection bias and confounding factors. After IPW adjustment, the high FAR group showed significantly higher risk of 1-year compared with low FAR group (36.4% vs. 12.4%, adjust hazard ratio = 1.72; 95% confidence interval (CI): 1.59–1.86; p < 0.001). In the receiver-operating characteristic curve analysis associated with the prediction of 1-year mortality, there was no significant difference between the area under the curve of FAR on ICU admission (C-statistic: 0.684, 95% CI: 0.673–0.694) and that of SOFA score on ICU admission (C-statistic: 0.679, 95% CI: 0.669–0.688) (p = 0.532). In this study, FAR and SOFA score at ICU admission were associated with 1-year mortality in patients admitted to an ICU. Especially, FAR was easier to obtain in critically ill patients than SOFA score. Therefore, FAR is feasible and might help predict long-term mortality in these patients.


Introduction
Predicting the prognosis of critically ill patients can help maintain adequate management and prevent futile treatment in the intensive care unit (ICU) [1]. However, it is not easy to predict the prognosis of patients admitted to an ICU at the early stage [2]. Generally, prognostic scoring systems and biomarkers have been commonly used for predicting clinical outcomes in critically ill patients. They can help us assess the severity of disease, cost-benefit analysis, and clinical decision-making [3,4]. However, most scoring systems are difficult to obtain and complex in critically ill patients. Sequential Organ Failure Assessment (SOFA) score is a well-known scoring system. SOFA score can help us assess organ dysfunction or failure. They are useful for predicting clinical outcomes in patients admitted to an ICU [5].
Serum biomarkers are easy to obtain and known to be helpful for making early decisions in ICU treatment [6,7]. Among numerous biomarkers, fibrinogen and serum albumin can be used as predictors of morbidity and mortality in critically ill patients [8][9][10][11]. Fibrinogen, a human plasma glycoprotein synthesized in the liver, plays a role in the coagulation pathway as a clotting factor. Fibrinogen is also involved in tissue injury and inflammation [8,[11][12][13]. In addition, fibrinogen is associated with prognosis of several 2 of 10 diseases such as sepsis, coronary artery disease, diabetes, and chronic kidney disease [14,15]. In addition, serum albumin reflects the nutritional status, degree of inflammation, and therapeutic effect [16]. Fibrinogen-to-albumin ratio (FAR) has been shown to be effective in reflecting coagulation and nutritional status as well as inflammatory status [16]. Recently, some studies have shown its usefulness and significance as a prognostic factor in several diseases including cancer, coronary artery disease, stroke-associated pneumonia, and sepsis [8,[11][12][13]. However, studies about the relationship between initial FAR and clinical outcome of patients who are admitted to ICU are limited. Therefore, the objective of this study was to investigate the usefulness of FAR as a prognostic marker in patients admitted to an ICU for various causes compared with Sequential Organ Failure Assessment (SOFA) score, a widely used prognostic scoring system.

Materials and Methods
The Institutional Review Board of Samsung Medical Center approved this study (SMC 2022-07-050). Written informed consent from an individual patient was not required because this study used a de-identified registry. This study was conducted according to the Declaration of Helsinki. Results are reported following the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Data Curation & Study Population
The registry had medical data of 65,654 consecutive adult patients who were treated at the ICU with SOFA score evaluated upon admission in Samsung Medical Center, Seoul, Korea between June 2013 and May 2022. In this study, the SOFA score was reported in the medical record for all included patients. This large, single-center cohort was generated in a de-identified form using data extracted by the institutional electronic archive system. The "Clinical Data Warehouse Darwin-C" is an electronic system built for investigators to search and retrieve data from institutional electronic medical records for over 4 million patients with more than 900 million laboratory findings and 200 million prescriptions. For mortality data outside our institution, this system uses a unique personal identification number updated from the National Population Registry of the Korea National Statistical Office. Using an extracted raw medical record, independent investigators who were blinded to mortality organized relevant variables of demographic data and underlying diseases. Results of blood laboratory tests and SOFA score were automatically extracted.
For this study, we selected patients with available fibrinogen and albumin levels at admission to ICU. The decision to measure fibrinogen levels was based on the individual patient's clinical situation. Fibrinogen was routinely checked on ICU admission in patients with sepsis, suspicious disseminated intravascular coagulation, solid and hematologic malignancies, post-operative care, suspicious bleeding tendency, etc. We initially stratified study patients into three groups according to tertile values of FAR upon ICU admission. After estimating an optimal threshold associated with 1-year mortality, we divided patients accordingly.

Definitions & Study Endpoints
FAR was defined as the ratio of fibrinogen concentration (mg/dL) to albumin concentration (g/L). SOFA score was estimated by assessing each component of respiratory, coagulation, liver, cardiovascular, central nervous system, and renal parameters as described in a previous study [17]. In the case of multiple measurements of SOFA score, the worst value during the initial 24 h after ICU admission was used for analysis. The primary endpoint was mortality during 1-year follow-up. The secondary endpoints were 30-day mortality, in-hospital mortality, and ICU mortality.

Statistical Analysis
Continuous variables are presented as means ± standard deviations and categorical variables are represented as numbers with subsequent percentages. Data were compared using Student's t-test and one-way analysis of variance for continuous variables and Chisquare test or Fisher's exact test for categorical variables. To estimate an optimal threshold of FAR associated with 1-year mortality, we generated receiver-operating characteristic (ROC) plots. We also determined the specificity and sensitivity of the threshold. We obtained optimal cut-off for FAR to predict mortality during 1-year follow-up based on the ROC curve and Youden index [18,19]. Mortalities were compared according to the estimated threshold using Cox regression analysis. Results are presented as hazard ratio (HR) with a 95% confidence interval (CI). A rigorous statistical adjustment was conducted to reduce bias and achieve a balance between groups. We chose an inverse probability weighting (IPW) using propensity score and adjusted for all relevant variables [20]. We compared the balance of baseline covariates between nutrition groups by calculating the absolute standardized difference (ASD) [21]. If IPW methods were effective for balancing exposure groups, the ASD should be close to zero [22]. The ASD under 10% was deemed as an adequate balance between groups. We also generated a Kaplan-Meier curve and compared the risk of mortality with log-rank test. The power of analysis was computed based on the sample size. The power of our analysis was 0.99 when HR was over 1.2. It was 0.78 when HR was 1.1 [23]. We assessed predictive performances of FAR and SOFA score in ICU admission using areas under the curve (AUCs) of the ROC curves for sensitivity vs. 1-specificity. We compared AUCs using the nonparametric approach published by DeLong et al. [24] for two correlated AUCs. All statistical analyses in this study were performed using R 4.2.0 (Vienna, Austria; http://www.R-project.org/ (accessed on 7 December 2021)).

Results
From 65,654 patients in the entire registry, we excluded 46,608 patients without available fibrinogen level. We then excluded 484 patients without available albumin level. A total of 18,562 (28.3%) patients were included for analysis. Tertile values of FAR were 6.41 and 10.59, respectively. Baseline characteristics of study patients according to the tertile values of FAR are summarized in Table 1. There were statistical differences in most variables between 3three groups, except severe trauma. Mortalities were higher in patients with higher FAR. Figure 1 shows changes of risk of 1-year mortality according to FAR upon ICU admission.

Low Group
High Group Before IPW After IPW Figure 1. Estimated risk of one-year mortality according to fibrinogen/albumin ratio (FAR) upon intensive care unit admission. The optimal cutoff threshold of FAR associated with 1-year mortality was estimated to be 10.79 with the area under the ROC curve of 0.68. The sensitivity and specificity of the estimated threshold were 58.6% and 74.1%, respectively. According to this threshold, 12,533 (67.5%) and 6029 (32.5%) patients were stratified into a low FAR group and a high FAR group, respectively. The mean value of FAR was 6.57 in the low FAR group and 17.74 in the high FAR group. The high FAR group tended to be older, with higher incidences of comorbidities (  Table 3). The Kaplan-Meier curve for 1-year mortality according to FAR on ICU admission is shown in Figure 2. We also found that the association between FAR and mortality was significant regardless of the type of ICU (Figure 3).    In the ROC curve analysis associated with the prediction of 1-year mortality, there was no significant difference between the AUC of FAR on ICU admission (C-statistic: 0.684, 95% CI: 0.673-0.694) and that of SOFA score on ICU admission (C-statistic: 0.679, 95% CI: 0.669-0.688) (p = 0.532). The performance of SOFA score for predicting 1-year mortality was better than that of FAR in medical, oncologic, and neurosurgical ICUs, whereas FAR performed better than SOFA score in surgical ICU (Figure 3). In the ROC curve analysis associated with the prediction of 1-year mortality, there was no significant difference between the AUC of FAR on ICU admission (C-statistic: 0.684, 95% CI: 0.673-0.694) and that of SOFA score on ICU admission (C-statistic: 0.679, 95% CI: 0.669-0.688) (p = 0.532). The performance of SOFA score for predicting 1-year mortality was better than that of FAR in medical, oncologic, and neurosurgical ICUs, whereas FAR performed better than SOFA score in surgical ICU (Figure 3).

Discussion
In this study, we investigated the usefulness of FAR as a prognostic marker in patients admitted to an ICU and compared the predictive value with SOFA score. One-year mortality was higher in patients with higher FAR than in those with low FAR. After an adjustment with IPW technique, the association between increased risk of mortality and high FAR remained significant. There was no difference between prognostic value of FAR and that of SOFA score on ICU admission. However, FAR is easier and simpler to obtain than SOFA score in patients admitted to an ICU. Therefore, our results suggest that FAR on ICU admission is a feasible and useful marker to predict 1-year mortality in critically ill patients.
Predicting prognosis plays an important role in the management and decision-making of critically ill patients [25]. Appropriate treatment strategies might be optimized based on predicted prognosis of an individual patient [25]. Prognostic factors may prevent futile treatment and allow us to be more selective and focused on critically ill patients [26]. Additionally, patients and their families can be informed about the risk of recurrence or death. However, even a prognostic factor with an excellent predictive performance may not be feasible in a clinical practice if it is too difficult to obtain it. The SOFA score is a well-known prognostic tool that is widely used. In ICU stetting, it can predict the prognosis through multi-organ dysfunction. However, it is somewhat complex in that it is necessary to evaluate multiple organs. In contrast, FAR has the advantage in clinical practice because it is easily and readily measured through blood tests, so it may be useful for identifying patients who are at high risk of death and require aggressive treatment in the early stages of ICU admission.
Systemic inflammation and coagulation processes are closely related. Coagulation factors play an important role in the immune system [27]. Beside clotting and thrombosis, fibrinogen is closely associated with inflammatory reaction [14]. Especially, fibrinogen can modulate the immune system and act as a chemotactic factor for monocytes and neutrophils [28]. Therefore, fibrinogen can be an important factor in the cascade of inflammatory reactions. In addition, elevated fibrinogen levels would be more likely to lead to an increased risk of long-term mortality associated with underlying diseases than systemic organ dysfunctions [29]. A recent study has revealed that fibrinogen elevation is associated with excessive inflammation and disease severity in COVID-19 patients [30]. Serum albumin is also an indicator that reflects nutritional and inflammations [16]. While SOFA score is focused on organ dysfunction of critically ill patients, FAR could reflect inflammation, immune reaction, coagulation, and nutritional state related to underlying diseases which might be associated with long-term mortality. In this study, predictive performances of FAR and SOFA score on ICU admission for long-term outcomes of critically ill patients were found to be similar.
We also conducted separate analyses according to types of ICU. SOFA score was shown to outperform FAR in medical and oncologic ICU, which might be related to the fact that many patients were already diagnosed with sepsis during treatment of underlying diseases before ICU admission. Organ failure is more likely to be accompanied in critically ill patients with sepsis. Therefore, SOFA score on ICU admission might be useful for predicting prognosis in these patients [5]. In neuro-critically ill patients, FAR also showed a limited predictive value as an early prognostic factor. However, it is difficult to conclude this result owing to the small number of neurosurgical patients included in the study. FAR showed a significant association with 1-year mortality of surgical patients. This might be due to perioperative bleeding and transfusion that might be directly related to FAR [31].
In this study, only patients with fibrinogen available at ICU admission were included. This could raise concerns about selection bias and external validity of our findings. Additionally, they may have had more severe and complex disease, potentially leading to an overestimation of mortality. Nevertheless, our results showed consistent trend with previous studies, which demonstrated an association between high FAR and mortality risk in ICU patients [32,33]. Future studies with larger sample size, and with measurements of fibrinogen in all patients could provide further support for our findings.
Although we have suggested that FAR may be useful for ICU patients, indications for its use have not yet been established. It may not be necessary to measure fibrinogen and albumin in all patients upon ICU admission, as this would likely be resource-intensive and may not provide significant additional prognostic information for all patients. Instead, it may be more appropriate to use a risk stratification approach to identify patients at high risk of poor outcomes who would most benefit from measuring these biomarkers. Additionally, measuring fibrinogen and albumin in patients who have already been identified as being at high risk of poor outcomes based on other clinical or laboratory parameters may also be useful in predicting prognosis.
This study has several limitations. First, this was a retrospective review of medical records and data extracted from the Clinical Data Warehouse. The nonrandomized nature of registry data might have resulted in a selection bias. Second, laboratory tests including fibrinogen and albumin levels were not protocol-based for patients admitted ICU. Third, although fibrinogen and albumin levels were based on initially identified values to reduce the changes due to blood transfusions and albumin infusions after ICU admission, the values might have been affected in some patients. Fourth, in neuroscience ICU, FAR values could only be obtained for a few patients upon ICU admission. Finally, follow-up FAR data were obtained for a few patients. Therefore, it was unclear whether changes of FAR were associated with long-term mortality. Despite these limitations, this is the first study to evaluate the usefulness of FAR in the entire ICU. Additionally, our study differed from other studies in that FAR was analyzed relative to SOFA scores and analyzed separately for each ICU. In this context, the present study provides valuable insights, but large-scale prospective studies are needed to further confirm the usefulness of FAR in predicting clinical outcomes of critically ill patients with evidence-based conclusions.

Conclusions
In this study, FAR and SOFA score at ICU admission were associated with 1-year mortality in patients admitted ICU. However, obtaining FAR is easier than obtaining SOFA score in critically ill patients. Therefore, FAR is feasible and might help predict long-term mortality in these patients.