Predictive value of admission red cell distribution width-to-platelet ratio for 30-day death in patients with spontaneous intracerebral hemorrhage: an analysis of the MIMIC database

Aim Prognostic assessment plays an important role in the effective management of patients with spontaneous intracerebral hemorrhage (ICH). The study aimed to investigate whether elevated red cell distribution width-to-platelet ratio (RPR) at admission was related to 30-day death in patients with spontaneous intracerebral hemorrhage (ICH). Methods This retrospective cohort study included 2,823 adult patients with ICH from the Multiparameter Intelligent Monitoring in Intensive Care (MIMIC) III and IV databases between 2001 and 2019. The Cox proportional hazard model was utilized to evaluate the relationship between RPR levels and 30-day death risk. The area under receiver-operating characteristic curve (AUC) was used to assess the predictive ability of RPR for 30-day death in patients with ICH. Results At the end of the 30-day follow-up, 799 (28.30%) patients died, and the median RPR level was 0.066 (0.053, 0.087). After adjusting for confounders, the tertile 3 of RPR levels [hazard ratio (HR) = 1.37, 95% confidence interval (CI): 1.15–1.64] were associated with a higher risk of 30-day death in patients with ICH compared with tertile 1. In the stratified analyses, elevated RPR levels were found to be associated with an increased risk of 30-day death in patients aged <65 years (HR = 1.77, 95%CI: 1.29–2.43), aged ≥65 years (HR = 1.30, 95%CI: 1.05–1.61), with Glasgow Coma Score (GCS) <14 (HR = 1.65, 95%CI: 1.27–2.14), with Charlson comorbidity index (CCI) ≥4 (HR = 1.45, 95%CI: 1.17–1.80), with (HR = 1.66, 95%CI: 1.13–2.43) or without sepsis (HR = 1.32, 95%CI: 1.08–1.61), and female patients (HR = 1.75, 95%CI: 1.35–2.26) but not in male patients (P = 0.139) and patients with GCS ≥14 (P = 0.058) or CCI <4 (P = 0.188). The AUC for RPR to predict 30-day death in patients with ICH was 0.795 (95%CI: 0.763–0.828) in the testing set, indicating a good predictive ability. Conclusion Elevated RPR levels were correlated with an increased risk of 30-day death in patients with ICH, and RPP levels showed good predictive ability for 30-day death.

Predictive value of admission red cell distribution width-to-platelet ratio for -day death in patients with spontaneous intracerebral hemorrhage: an analysis of the MIMIC database Introduction Spontaneous intracerebral hemorrhage (ICH) is a global, life-threatening disease with a poor prognosis and few proven treatments (1).Due to the non-traumatic rupture of intracranial vessels, blood flows into the brain parenchyma or ventricles and forms hematomas, causing neuronal and glial damage, which in turn causes an inflammatory response (2).ICH accounts for approximately 30% of all acute strokes and is related to high mortality and morbidity (3).In addition, more than 30% of deaths in patients with ICH occur within 30 days of the hemorrhage (4).The effective management of ICH patients includes close monitoring and treatment of blood pressure, seizures, elevated intracranial pressure, and reversal of anticoagulation and antiplatelet medications (2,5).Reliable tools for patient prognostic assessment are essential for the treatment and management of the disease.
The inflammatory response is closely associated with intracerebral blood extravasation in patients with ICH (6,7).Inflammatory changes in the tissue surrounding the hematoma led to an immune response, followed by the activation of microglia and cytokine release (8)(9)(10).A systematic review summarized the prognostic role of serum biomarkers, such as neutrophil-lymphocyte ratio, S100 calcium-binding protein B, and thioredoxin, in patients with ICH (11).However, there is still a lack of routine blood indicators with sufficient clinical evidence to evaluate the prognosis of ICH patients.Several studies have reported that red cell distribution width (RDW)-to-platelet ratio (RPR) correlated with the prognosis of various diseases such as hepatic fibrosis (12), sepsis (13), and acute traumatic brain injury (14).RDW and platelet represent the heterogeneity of circulating red blood cells and the pathophysiology of hemostasis, respectively.Recently, Lehmann et al. investigated the association between RPR and 90-day mortality in patients with ICH (15).However, their study was limited by the small sample size, and the association between RPR and death in patients with ICH across age and disease severity was unknown.Furthermore, the predictive ability of RPR for death in ICH patients is unclear.
We aimed to investigate the association between RPR levels and 30-day death in patients with ICH based on a large sample of the database.In addition, the predictive ability of RPR for 30-day death in patients with ICH was evaluated.

Data source and study population
All data in this retrospective cohort study were extracted from the Multiparameter Intelligent Monitoring in Intensive Care III and IV (MIMIC-III and -IV) database between 2001 and 2019 (https://mimic.mit.edu/docs/iii/).MIMIC-III is a large, freely available database that includes non-identifiable health data related to more than 40,000 patients admitted to the intensive care unit (ICU) at Beth Israel Deaconess Medical Center between 2001 and 2012.MIMIC-IV is an updated version of MIMIC-III and currently contains data on patients admitted to the ICU at the Beth Israel Deaconess Medical Center between 2008 and 2019.The MIMIC database includes information on demographic, vital

Outcomes
The study outcome was 30-day death, which was defined as death within 30 days after the bleeding event.Complete survival data were recorded up to a 30-day follow-up.For patients with more than one ICU admission, only data from the first ICU admission were selected for analysis.RPR was calculated as RPR = red blood cell distribution width (%)/platelet count (K/uL).Red blood cell distribution width and platelet count were used as measured at the time of patient admission.RPR value was divided into three categories based on the tri-sectional quantile: tertile 1, <0.057; tertile 2, 0.057-0.078;and tertile 3, >0.078.

Data collection
Demographic information, vital sign measurements, laboratory testing data, medication and disease information, and scores were

Management of missing data
There were no missing data for 30-day survival and RPR value.Variables with more than 20% missing values were excluded.Variables with less than 20% missing data were interpolated using the random forest interpolation method (n_estimators = 500).
Sensitivity analysis was performed by analyzing the differences between the data before and after interpolation.

Statistical analysis
Continuous variables were expressed as mean and standard deviation (mean ± SD) or median and quartiles [M (Q1, Q3)] and compared using Student's t-test or the Wilcoxon rank-sum test.Categorical variables were expressed as number and percentage [n (%)] and compared using the chi-square test or Fisher's exact test.
The univariate Cox proportional hazard model was used to screen variables that may be associated with 30-day death in patients with ICH.Variables with a significant statistical difference in univariate analysis were screened by stepwise regression with bidirectional elimination, and the final screened variables were included as confounders in the multivariate Cox proportional hazard model.The association between RPR and the risk of 30day death in ICH patients was analyzed by a multivariate Cox proportional hazard model and presented as hazard risk (HR) with a 95% confidence interval (CI).The concordance index (C-index) was used to evaluate discriminative ability.Kaplan-Meier (K-M) survival curves were plotted for ICH patients with different RPR values.The association between RPR levels and 30-day death in ICH patients was further analyzed based on age (<65 and ≥65 years), sex (female and male), GCS score (<14 and ≥14), CCI score (<4 and ≥4), and sepsis (yes and no).To compare the predictive ability of RPR, SOFA, SAPS II, and qSOFA for 30-day

Association between RPR and the risk of -day death in patients with ICH
The results of univariate and multivariate Cox proportional hazard models for the relationship between RPR and the risk of 30-day death in patients with ICH are demonstrated in Table 3.Compared with tertile 1, the tertile 3 of RPR levels (HR = 1.58, 95%CI: 1.33-1.87)were associated with an increased risk of 30-day death in patients with ICH.After adjusting for age, ethnicity, ICU type, ventilation, vasopressor, renal replacement therapy, sepsis, heart rate, SPO 2 , hemoglobin, blood urea nitrogen, glucose, urine output, mannitol, anticoagulation, surgery, and neurodegeneration, the tertile 3 of RPR levels (HR = 1.37, 95%CI: 1.15-1.64)were still associated with a higher risk of 30-day death.
Furthermore, 281 (9.95%) patients experienced an ICU readmission.Factors associated with ICU readmission in patients with ICH were analyzed (Supplementary Table 2).The results showed that the tertile 1 (HR = 1.45, 95%CI: 1.07-1.97)and tertile 3 (HR = 1.56, 95%CI: 1.13-2.15) of RPR levels were associated with a higher risk of ICU readmissions in patients with ICH compared with tertile 2 of RPR levels.The K-M survival curves showed that patients with tertile 3 of RPR levels had a higher risk of 30-day death compared with tertile 1 and tertile 2 of RPR (P < 0.001) (Figure 2).Table 4 presents the comparison of different tools for the identification of 30-day death in patients with ICH.The C-index was 0.772 (95%CI: 0.757-0.786)for the RPR tool, 0.656 (95%CI: 0.636-0.675)for the SOFA tool, 0.712 (95%CI: 0.695-0.729)for the SAPS II tool, and 0.529 (95%CI: 0.511-0.548)for the qSOFA tool, and the C-index of the RPR was higher than that of the other tools (all P < 0.001).
. /fneur. .Predictive ability of RPR for -day death in patients with ICH Table 6 shows the predictive ability of RPR, SOFA, SAPS II, and qSOFA for 30-day and 15-day death in patients with ICH, and Figure 3 demonstrates the ROC curves for these tools.The AUCs of the RPR, SOFA, SAPS II, and qSOFA tools for predicting 30-day death in patients with ICH in the testing set were 0.795 (95%CI: 0.763-0.828),0.695 (95%CI: 0.655-0.735),0.745 (95%CI: 0.710-0.779),and 0.539 (95%CI: 0.500-0.578),respectively.The RPR tool had the highest AUC for predicting 30-day death in patients with ICH compared with other tools (all P < 0.001).The RPR tool also had the highest AUC for predicting 15-day death in patients with ICH, with an AUC of 0.805 (95%CI: 0.773-0.837).The calibration curves showed no deviation between the predicted and observed probability of RPR in predicting 30-day and 15-day death in patients with ICH (Figure 4).

Discussion
The present study investigated the association between RPR and 30-day death in patients with ICH and explored the predictive value of RPR for 30-day death.We found elevated RPR levels were correlated with an increased risk of 30-day death in patients with ICH.In addition, RPR had a good predictive ability for 30-day death in patients with ICH, with an AUC value of 0.795.
Inflammatory response plays an important role in the pathophysiological processes of brain injury after ICH (16, 17).Both RDW and RPR have been reported to be associated with mortality in patients with ICH (15,18).Pinho et al. found that elevated RDW levels were independently associated with increased 30-day mortality in patients with ICH (18).Lehmann et al. showed that ICH patients with elevated RPR in the admission laboratory were more likely to die within 90 days of bleeding (15).In addition, RPR was considered a strong predictor of prognosis for a variety of diseases such as breast cancer (19), glioblastoma (20), myocardial infarction (21), and acute traumatic brain injury (14).The use of RPR for prognosis assessment of  disease is related to the physiological function of RDW and platelets.RDW represents heterogeneity in erythrocyte size, with higher values indicating greater variability.Abnormally elevated RDW is associated with oxidative stress, chronic inflammatory responses, and impaired erythropoiesis (22, 23).Inflammatory cytokines can affect the survival of circulating erythrocytes, reduce deformation, inhibit maturation, and lead to an increase in RDW (24,25).Excessive oxidative stress reduces the activity of red blood cells (23).Moreover, elevated RDW levels were found to be associated with the risk of recurrence of small artery occlusion (26).Pathological changes of cerebral small vessels can cause small artery occlusion and also involve deep ICH.Platelets are also a common laboratory indicator, and thrombocytopenia may increase the risk of bleeding (27).The most common cause of thrombocytopenia in critically ill patients is severe infection and/or inflammation, which causes circulating thrombocytopenia primarily through abnormal platelet-vessel wall interactions and abnormal platelet activation (28).The present study found that elevated RPR levels were related to an increased risk of 30-day death in patients with ICH, which was consistent with the previous study (15).Compared with previous studies, our study provided stronger evidence based on a large sample of the database.Second, we not only analyzed the association between RPR and death risk in ICH patients but also further analyzed the predictive value of RPR for death in ICH patients.Furthermore, the outcome of our study was 30-day death in patients with ICH compared to 90-day death in the previous study, which may be a difference.Pinho et al. (18) showed that the prognostic value of the same indicator for patients with ICH may vary depending on the length of time to assess death (30-day death or more).
Our results showed that compared with SOFA, SAPS II, and qSOFA scores, RPR had the highest AUC for predicting 30-day death in ICH patients, with an AUC value of 0.806.The calibration curves demonstrated that no deviation was observed between the predicted and observed probability of RPR in predicting 30-day death.These results suggested that RPR had a good predictive ability for 30-day death in patients with ICH.RPR as an independent marker of inflammation may be used in the prognostic assessment of patients with ICH.In the subgroup analysis between RPR and 30-day death risk, elevated RPR levels were observed to be linked to an increased 30-day death risk in patients aged <65, aged ≥65 years, GCS score <9, GCS score ≥9, female patients, and CCI ≥3, whereas no association was observed in male patients.The association between RPR and 30-day death in patients with ICH varies by gender, which may require further study.Several studies have reported sex differences in outcomes for patients with ICH (29-31).However, these studies did not reach consistent results and did not provide a reasonable explanation for the sex differences.Umeano et al. suggested that there may be an interaction between sex and age that influences the outcomes of patients with ICH (29), but prospective studies are needed to investigate this hypothesis.
Our study provided evidence of the relationship between RPR and 30-day death in ICH patients based on a large sample of data from the MIMIC database.The association between RPR and 30-day death was then analyzed stratified by age, sex, GCS score, CCI score, and sepsis.In addition, we investigated the predictive value of RPR levels for 30-day death in patients with ICH and compared it with other tools such as SOFA, SAPS II, and qSOFA.However, several limitations should be taken into account when interpreting our results.First, this study was based on single-center data from the MIMIC database, and further studies may require prospective multicenter studies to provide more evidence.Second, retrospective study design is subject to reporting bias and thus has an impact on the results, and prospective studies are needed.Third, the MIMIC database lacks imagingrelated records to capture indicators affecting the prognosis of ICH such as hematoma volume, but we used the GCS score to reflect the organic status and severity of patients with ICH, and the corresponding subgroup analysis was performed.Fourth, both neutrophil-to-lymphocyte ratio (NLR) and neutrophil-toplatelet ratio (NPR) have been reported to be correlated with ICH.However, it was not possible to compare the effects of RPR, NLR, and NPR because of the high rate of missing data (>80%) for neutrophils and lymphocytes in this study population.Fifth, we were unable to analyze the relationship between RPR and different subtypes of ICH (e.g., deep ICH and lobar ICH) due to the limitations of the MIMIC database.Sixth, the present study focused on the association of RPR levels at patient admission with death, while the association between dynamic changes in RPR levels during treatment (e.g., RPR trajectory) and death may merit further study.Seventh, the predictive value of RPR in combination with other biomarkers such as S100 calcium-binding protein B and thioredoxin for death in patients with ICH needs to be further explored.Eighth, future studies could explore the association between RPR and disease processes such as hematoma expansion and neurological deterioration in patients with ICH, which could contribute to the understanding of the disease process of ICH.

Conclusion
The current study indicated that elevated RPR levels were associated with a higher 30-day death risk in patients with ICH.RPR levels showed good predictive ability for 30-day death in patients with ICH compared with other tools.Increased RPR levels may provide clinicians with a signal of an elevated risk of death in patients with ICH.

FIGURE
FIGUREFlow chart of the study population.ICH, spontaneous intracerebral hemorrhage; MIMIC, Multiparameter Intelligent Monitoring in Intensive Care; ICU, intensive care unit; RDW, red cell distribution width.

FIGURE
FIGUREKaplan-Meier survival curves for -day death in ICH patients with di erent RPR levels.ICH, spontaneous intracerebral hemorrhage; RPR, red cell distribution width to platelet ratio.

FIGURE
FIGUREReceiver operating characteristic (ROC) curves for RPR, SOFA, SAPS II, and qSOFA to predict -day and -day death in ICH patients (testing set).RPR, red cell distribution width to platelet ratio; SOFA, Sequential Organ Failure Assessment; SAPS II, Simplified Acute Physiology Score; qSOFA, quick SOFA; ICH, spontaneous intracerebral hemorrhage.

FIGURE
FIGURECalibration curves for RPR, SOFA, SAPS II, and qSOFA to predict -day and -day death in ICH patients (testing set).RPR, red cell distribution width to platelet ratio; SOFA, Sequential Organ Failure Assessment; SAPS II, Simplified Acute Physiology Score; qSOFA, quick SOFA; ICH, spontaneous intracerebral hemorrhage.
TABLE Baseline characteristics of the patients with spontaneous intracerebral hemorrhage (ICH).
TABLE Univariate Cox proportional hazard model of factors associated with the risk of -day death in patients with spontaneous intracerebral hemorrhage (ICH).

TABLE (
Frontiers in Neurology frontiersin.orgdeath in patients with ICH, all patients were randomly divided into a training set and a testing set with a ratio of 7:3.The characteristics of patients in the training set and testing set were shown in Supplementary Table 1.Receiver operating characteristic (ROC) curves for the prediction of death in patients with ICH by different tools were constructed, and the area under the curve (AUC) was calculated.The Delong test was utilized to compare the difference in AUC between tools.C-index and AUC values >0.7 indicate a reasonable estimate.All statistical analyses were performed using SAS 9.4 software (SAS Institute Inc., Cary, NC, USA) and R 4.2.0 software (Institute for Statistics and Mathematics, Vienna, Austria).A two-sided Pvalue <0.05 was considered to be statistically significant.

TABLE Association between
RPR and the risk of -day death in patients with ICH.

TABLE C -
index of di erent tools for the identification of -day death in patients with spontaneous intracerebral hemorrhage (ICH).
TABLE Association of RPR with the risk of -day death in ICH patients stratified by age, sex, GCS score, and CCI score.

TABLE (
TABLE Predictive ability of RPR, SOFA, SAPS II, and qSOFA for -day and -day death in patients with ICH.
ICH, spontaneous intracerebral hemorrhage; RPR, red cell distribution width to platelet ratio; SOFA, Sequential Organ Failure Assessment; SAPS II, Simplified Acute Physiology Score; qSOFA, quick SOFA; AUC, area under receiver-operating characteristic curve.