The association between RDW-to-platelet ratio and in-hospital mortality in critically ill stroke patients: A retrospective cohort study based on the eICU database.

The association between RDW-to-platelet ratio and in-hospital mortality in critically ill stroke patients: A retrospective cohort study based on the eICU database - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice PLoS One . 2026 Apr 17;21(4):e0344361. doi: 10.1371/journal.pone.0344361 Search in PMC Search in PubMed View in NLM Catalog Add to search The association between RDW-to-platelet ratio and in-hospital mortality in critically ill stroke patients: A retrospective cohort study based on the eICU database Yu Chen Yu Chen 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Data curation Find articles by Yu Chen 1, * , Xiangrong Yang Xiangrong Yang 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Methodology Find articles by Xiangrong Yang 1, # , Minmin Lan Minmin Lan 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Conceptualization, Resources, Software Find articles by Minmin Lan 1, # , Xinghua Qin Xinghua Qin 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Conceptualization, Formal analysis, Validation Find articles by Xinghua Qin 1, # , Dongmei Yi Dongmei Yi 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Conceptualization, Data curation, Project administration Find articles by Dongmei Yi 1, # , Lu Chen Lu Chen 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China Conceptualization, Formal analysis Find articles by Lu Chen 1, # Editor: Marwan Salih Al-Nimer 2 Author information Article notes Copyright and License information 1 Department of Neurology, Guangxi Hydroelectric Hospital, Guilin, Guangxi, China 2 University of Diyala College of Medicine, IRAQ ✉ * E-mail: [email protected] Competing Interests: The authors have declared that no competing interests exist. # ☯ These authors contributed equally to this work. # Contributed equally. Roles Yu Chen : Data curation Xiangrong Yang : Methodology Minmin Lan : Conceptualization, Resources, Software Xinghua Qin : Conceptualization, Formal analysis, Validation Dongmei Yi : Conceptualization, Data curation, Project administration Lu Chen : Conceptualization, Formal analysis Marwan Salih Al-Nimer : Editor Received 2025 Dec 8; Accepted 2026 Feb 19; Collection date 2026. © 2026 Chen et al This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. PMC Copyright notice PMCID: PMC13089741  PMID: 41996310 This article has been corrected. See PLoS One. 2026 Jun 30;21(6):e0352777 . Abstract Objective The red cell distribution width-to-platelet ratio (RDW-to-platelet ratio, RPR) is a potential biomarker of inflammation and bone marrow function. This study aimed to investigate the association between RPR and in-hospital mortality in critically ill stroke patients in the eICU, and to evaluate mediating role of APACHE-IV score. Method This retrospective cohort study utilized data from the 2014–2015 US multicenter eICU database. Of 200,859 initially included patients, 9,736 critically ill stroke patients were analyzed after excluding non-stroke cases and those with missing key variables. Multivariable logistic regression assessed the relationship between RPR and in-hospital mortality, adjusting for age, sex, comorbidities, and laboratory parameters, with multiple imputation for missing data. Subgroup analyses, mediation analysis, and restricted cubic spline modeling were performed. Results Patients with higher RPR exhibited significantly elevated in-hospital mortality (17.7% vs. 11.6%, P < 0.001). After adjustment, each unit increase in RPR was associated with 4.6% higher odds of death (adjusted OR=1.046, 95% CI: 1.032–1.061, P < 0.001), with consistent findings across subgroups. Restricted cubic spline analysis indicated a linear relationship. Mediation analysis showed that APACHE-IV score mediated 20.15% of the total effect (P < 0.001). Conclusions Elevated RPR is independently associated in-hospital mortality in critically ill stroke patients, partly mediated by disease severity. As a simple and accessible marker, RPR shows promise for clinical prognostic risk stratification. Introduction In-hospital mortality among stroke patients in the intensive care unit (ICU) reaches 25%–35%, escalating to 40%–50% for hemorrhagic stroke, and exceeding 50% when complicated by heart failure or acute myocardial infarction (AMI) [ 1 ]. In the United States, ICU-admitted stroke patients comprise 5%–10% of all hospitalized stroke cases, with over 70% comorbid with hypertension, representing a predominant clinical subgroup [ 2 ]. These patients impose a substantial disease burden, with average hospital stays of 7–14 days and high resource utilization, underscoring the need for reliable early prognostic markers to inform clinical decisions. Red cell distribution width (RDW), a routine complete blood count parameter reflecting erythrocyte volume heterogeneity, has been linked to systemic inflammation, oxidative stress, and endothelial dysfunction [ 3 , 4 ]. Mechanistically, elevated RDW indicates dysregulated erythropoiesis in bone marrow under the influence of inflammatory cytokines such as IL-6 and TNF-α, resulting in impaired oxygen-carrying capacity, microcirculatory dysfunction, and tissue hypoxia [ 5 ]. Furthermore, RDW correlates with hypoalbuminemia, malnutrition, and immunosuppression [ 6 ]. The RDW-to-platelet ratio (RPR), integrating RDW with platelet count, captures dual pathophysiological processes of inflammation and coagulopathy, providing a comprehensive reflection of stress in critically ill patients [ 7 ]. This study, utilizing the eICU database, conducts a retrospective cohort analysis to examine the association between RPR and in-hospital mortality in ICU stroke patients, adjusting for confounders including age, sex, APACHE-IV score, baseline comorbidities (e.g., diabetes, malignancy), laboratory parameters (blood urea nitrogen, albumin, hemoglobin, creatinine), and therapeutic interventions (anticoagulation/antiplatelet therapy). We aim to establish RPR as an independent association, supporting its application in clinical risk stratification. Methods Study design and data source This multicenter retrospective cohort study was conducted using the eICU Collaborative Research Database (eICU-CRD) version 2.0, a publicly available multi-center critical care database comprising de-identified electronic health records from over 200,000 admissions to 208 intensive care units (ICUs) across the continental United States, spanning 2014, to 2015. The study population consisted of adult critically ill patients with a confirmed diagnosis of stroke (ischemic or hemorrhagic) during their ICU stay. Inclusion criteria were: (1) age ≥ 18 years; (2) stroke diagnosis confirmed during ICU admission; and (3) availability of complete laboratory data within the first 24 hours of admission. Exclusion criteria included: (1) admissions not primarily related to stroke; (2) missing data on red cell distribution width (RDW)-to-platelet ratio; and (3) missing in-hospital mortality data. From the initial cohort of 200,859 patients, 9,736 met the eligibility criteria and were included in the analyses ( Fig 1 ). Clinical variables were extracted automatically from standardized electronic health records and verified independently by two trained researchers to ensure data integrity. Stroke diagnoses were ascertained using International Classification of Diseases, Ninth Revision (ICD-9) codes supplemented by neuroimaging reports. Fig 1. Study population selection flowchart. Open in a new tab The figure illustrates the sequential patient selection process from the initial eICU database (N = 200,859) through application of inclusion and exclusion criteria to the final analytic cohort (N = 9,736). The flowchart shows: (1) Initial screening of all eICU patients, with exclusion of non-stroke admissions (n = 189,752); (2) Identification of stroke patients (n = 11,107); (3) Exclusion of those with missing key variables including RDW-to-platelet ratio or in-hospital mortality data (n = 1,371); and (4) Final classification of patients into outcome categories (survivors: n = 8,375; in-hospital deaths: n = 1,361). Abbreviation: eICU, eICU Collaborative Research Database; RDW, red cell distribution width. Exposure, outcome, and covariates The primary exposure variable was the RDW-to-platelet ratio (RPR), defined as the initial Red Cell Distribution Width (RDW) value (expressed as a percentage, %) obtained within 24 hours of ICU admission divided by the platelet count (expressed as ×10⁹/L). For example, if a patient’s RDW was 15.5% and platelet count was 250 × 10⁹/L, the RPR would be calculated as 15.5 ÷ 250 = 0.062. Data were recorded as continuous numerical values. The primary outcome was in-hospital mortality from any cause during the index hospitalization, determined from electronic hospital death registries and structured discharge status fields (coded as alive versus deceased). Adjudication was based on objective, structured clinical data without blinding, thereby reducing potential subjective bias. Covariate selection was guided by the following systematic principles: (1) Demographic variables (age, sex, race/ethnicity) were included as well-established independent prognostic factors for stroke patients; (2) Disease severity indicators (Acute Physiology and Chronic Health Evaluation IV [APACHE-IV] score) were selected for their recognized association in critically ill patients for assessing acute physiological status; (3) Comorbidities (congestive heart failure, hypertension, diabetes mellitus, malignancy, sepsis) were incorporated based on extensive evidence linking these chronic conditions to adverse outcomes in stroke populations; (4) Therapeutic interventions (mechanical ventilation, anticoagulation, antiplatelet therapy) were included to capture clinical management intensity and underlying physiological derangement; (5) Laboratory parameters (albumin, blood urea nitrogen, hemoglobin, creatinine) were selected as they reflect nutritional status, renal function, anemia, and inflammation—each independently associated with mortality in critically ill patients across multiple studies in the MIMIC and eICU databases. All covariates were simultaneously included in the model to comprehensively adjust for potential confounding while minimizing the risk of over-adjustment, as verified by variance inflation factors (VIF < 5 for all variables). The selected covariates were chosen based on three principal considerations: First, substantial evidence from clinical literature—we systematically reviewed prognostic studies in critically ill stroke patients and confirmed that all included variables have been established as independent association of mortality in multiple investigations. Second, biological and clinical relevance—each variable either directly reflects disease biology (e.g., RDW as a marker of inflammation and bone marrow dysfunction) or represents an important clinical characteristic (e.g., APACHE-IV score quantifying illness severity). Third, data completeness and statistical feasibility with a sample size of 9,736 patients, the number of covariates was balanced to adequately control for confounding while maintaining statistical power. For variables with ≥5% missing values, we employed the missforest algorithm for multiple imputation, which outperforms traditional methods in handling complex missing data patterns and was validated through sensitivity analyses demonstrating robustness of results. Ethical considerations The study protocol was approved by the Institutional Review Board (IRB approval No. IRB-2023-EICU-001). Informed consent was waived due to the use of de-identified retrospective data. Data access complied with PhysioNet credentialing requirements, with all records anonymized prior to analysis. Secure, encrypted platforms restricted access to authorized investigators only. The investigation adhered to the principles of the Declaration of Helsinki and eICU-CRD data-sharing policies. Statistical analysis We described baseline characteristics using means with standard deviations for continuous variables and frequencies with percentages for categorical variables, stratified by tertiles of the RDW-to-platelet ratio. Continuous variables across tertiles were compared using one-way analysis of variance (ANOVA), and categorical variables using the χ² test; trends were assessed with the Cochran-Armitage test for categorical variables and Jonckheere-Terpstra test for continuous variables (all P values two-sided). In-hospital mortality served as the primary outcome. We first performed univariate logistic regression to assess the crude association between RDW-to-platelet ratio (modeled as a continuous variable) and in-hospital mortality, reporting odds ratios (ORs) with 95% confidence intervals (CIs). Multivariable logistic regression models were constructed in a stepwise manner, adjusting for age, sex, comorbidities (e.g., congestive heart failure, malignancy, diabetes), laboratory values (e.g., albumin, hemoglobin, blood urea nitrogen, creatinine), and clinical variables (e.g., Glasgow Coma Scale score, mechanical ventilation, sepsis, APACHE-IV score). Adjusted ORs (aORs) with 95% CIs were reported per unit increase in RDW-to-platelet ratio. To address missing data (primarily albumin [29.21%] and APACHE-IV score [14.50%]; < 5% for others) ( S1 Table ), we used multiple imputation by chained equations with five iterations via the missForest algorithm, incorporating all covariates under a random forest framework. Pooled estimates were obtained using Rubin’s rules. Variance inflation factors (VIFs) were calculated to assess multicollinearity (all VIFs < 5). Nonlinearity between RDW-to-platelet ratio and in-hospital mortality was examined using restricted cubic splines with three knots at the 10th, 50th, and 90th percentiles, fitted within a multivariable logistic regression model; the reference level was set at the median value. Mediation analysis was conducted using the mediation R package, following the general approach to causal mediation analysis. The RDW-to-platelet ratio was treated as the exposure, APACHE-IV score as the mediator, and in-hospital mortality as the outcome, with adjustment for age, ethnicity, sex, congestive heart failure, diabetes, sepsis, malignant tumor burden, albumin, creatinine, and mechanical ventilation status. Nonparametric bootstrap resampling (1000 iterations) with the percentile method was used to estimate confidence intervals for the total effect, mediation (indirect) effect, direct effect, and proportion mediated. Subgroup analyses stratified by sex, age groups (<65 vs. ≥ 65 years), race, mechanical ventilation status, comorbidities, and APACHE-IV tertiles were performed by including interaction terms in the multivariable logistic regression model; effect modification was tested using the likelihood ratio test (P < 0.05 indicating significance). All analyses were performed using R version 4.3.1 (packages: mice, missForest, rms, mediation). To handle missing values in our dataset, we employed the missForest multiple imputation algorithm, which is based on random forests. This algorithm iteratively imputes missing data by building a random forest model for each variable with missing values, using all other variables as predictors to estimate the missing entries. We set the maximum number of iterations to 5 (maxiter = 5) to ensure convergence. For each random forest model constructed, we used 100 decision trees (ntree = 100). The number of variables randomly sampled as candidates at each split was set to the square root of the total number of predictor variables (mtry = sqrt(p), where p is the total number of predictors). To further enhance the stability of the imputation process, variables were imputed in decreasing order of the number of missing values (decreasing = TRUE). The convergence of the imputation process was confirmed by monitoring the changes in Out-of-Bag (OOB) error, with stabilization indicating model convergence. Detailed parameter settings and convergence diagnostics of the missForest algorithm are provided in Supplemental S2 Table . The quality of the imputed data was assessed by comparing the distributions of key variables between the original and imputed datasets. As shown in Supplemental S3 Table , no significant differences were observed in the distributions of key variables before and after imputation, confirming the rationality and reliability of the imputation method. Results This study included 200,859 patients from the eICU database, yielding 9,736 adult stroke patients with critical illness after exclusions (study flowchart, Fig 1 ). Baseline characteristics stratified by RDW-to-platelet ratio tertiles are summarized in Table 1 . The cohort had a mean age of 67.06 ± 14.86 years, with 52.0% male and 14.0% in-hospital mortality. Across tertiles (low to high), significant trends were observed toward older age (63.87 vs. 67.29 vs. 70.01 years), higher APACHE-IV scores (51.97 vs. 52.82 vs. 59.48), increased mechanical ventilation use (21.9% vs. 22.2% vs. 26.7%), greater prevalence of congestive heart failure and malignancy, lower albumin and hemoglobin levels, and elevated blood urea nitrogen and creatinine (all P < 0.05). In-hospital mortality rose significantly from 11.6% in the low-ratio group to 17.7% in the high-ratio group (P < 0.001). Table 1. Baseline characteristics of study participants by RDW-to-platelet ratio. Characteristic Overall Low Medium High P Value Sample size n 9,736 3,246 [1.16, 5.67] 3,245 [5.67, 7.56] 3,245 [7.56, 55.00] In-hospital mortality <0.001 Alive 8,375 (86.0%) 2,870 (88.4%) 2,834 (87.3%) 2,671 (82.3%) Dead 1,361 (14.0%) 376 (11.6%) 411 (12.7%) 574 (17.7%) Total GCS score 12.54 ± 3.55 12.59 ± 3.57 12.65 ± 3.47 12.39 ± 3.60 0.009 Age, years 67.06 ± 14.86 63.87 ± 15.52 67.29 ± 14.65 70.01 ± 13.72 <0.001 Gender <0.001 Male 5,058 (52.0%) 1,396 (43.0%) 1,722 (53.1%) 1,940 (59.8%) Female 4,677 (48.0%) 1,850 (57.0%) 1,522 (46.9%) 1,305 (40.2%) APACHE-IV score 54.74 ± 25.17 51.97 ± 24.71 52.82 ± 24.50 59.48 ± 25.64 <0.001 Ethnicity 0.005 African American 1,188 (12.2%) 372 (11.5%) 389 (12.0%) 427 (13.2%) Asian 208 (2.1%) 71 (2.2%) 66 (2.0%) 71 (2.2%) Caucasian 7,326 (75.2%) 2,473 (76.2%) 2,406 (74.1%) 2,447 (75.4%) Hispanic 397 (4.1%) 110 (3.4%) 166 (5.1%) 121 (3.7%) Native American 45 (0.5%) 17 (0.5%) 14 (0.4%) 14 (0.4%) Other/Unknown 525 (5.4%) 190 (5.9%) 187 (5.8%) 148 (4.6%) Mechanical ventilation <0.001 No 7,436 (76.4%) 2,535 (78.1%) 2,523 (77.8%) 2,378 (73.3%) Yes 2,300 (23.6%) 711 (21.9%) 722 (22.2%) 867 (26.7%) Congestive heart failure <0.001 No 9,244 (94.9%) 3,137 (96.6%) 3,085 (95.1%) 3,022 (93.1%) Yes 492 (5.1%) 109 (3.4%) 160 (4.9%) 223 (6.9%) Hypertension 0.048 No 6,646 (68.3%) 2,246 (69.2%) 2,162 (66.6%) 2,238 (69.0%) Yes 3,090 (31.7%) 1,000 (30.8%) 1,083 (33.4%) 1,007 (31.0%) Diabetes mellitus 0.066 No 8,434 (86.6%) 2,837 (87.4%) 2,822 (87.0%) 2,775 (85.5%) Yes 1,302 (13.4%) 409 (12.6%) 423 (13.0%) 470 (14.5%) Malignant tumor <0.001 No 9,655 (99.2%) 3,235 (99.7%) 3,227 (99.4%) 3,193 (98.4%) Yes 81 (0.8%) 11 (0.3%) 18 (0.6%) 52 (1.6%) Sepsis <0.001 No 9,143 (93.9%) 3,077 (94.8%) 3,097 (95.4%) 2,969 (91.5%) Yes 593 (6.1%) 169 (5.2%) 148 (4.6%) 276 (8.5%) Blood urea nitrogen, mg/dL 18.00 (13.00, 25.00) 16.00 (12.00, 23.00) 17.00 (13.00, 24.00) 20.00 (14.00, 29.00) <0.001 Albumin, g/dL 3.41 ± 0.67 3.49 ± 0.67 3.49 ± 0.63 3.26 ± 0.67 <0.001 Hemoglobin, g/dL 12.77 ± 2.30 12.98 ± 2.25 13.05 ± 2.16 12.28 ± 2.42 <0.001 Creatinine, mg/dL 0.95 (0.74, 1.28) 0.88 (0.70, 1.14) 0.94 (0.75, 1.22) 1.05 (0.80, 1.47) <0.001 Anticoagulant therapy 0.112 No 9,548 (98.1%) 3,191 (98.3%) 3,169 (97.7%) 3,188 (98.2%) Yes 188 (1.9%) 55 (1.7%) 76 (2.3%) 57 (1.8%) Antiplatelet therapy 0.897 No 9,137 (93.8%) 3,046 (93.8%) 3,050 (94.0%) 3,041 (93.7%) Yes 599 (6.2%) 200 (6.2%) 195 (6.0%) 204 (6.3%) Open in a new tab Values are n (%) for categorical variables and mean ± SD or median (IQR) for continuous variables. Abbreviations: SD, standard deviation; IQR, interquartile range; GCS, Glasgow Coma Scale; APACHE, Acute Physiology and Chronic Health Evaluation. P values were derived from χ² tests for categorical variables, one-way ANOVA for normally distributed continuous variables, and Kruskal-Wallis H tests for non-normally distributed continuous variables. Univariate logistic regression revealed that each unit increase in RDW-to-platelet ratio was associated with a 7% higher risk of in-hospital mortality (OR=1.07, 95% CI: 1.056–1.080, P < 0.001), with an effect size exceeding that of age (OR=1.02) and diabetes (OR=1.27). Other significant variables included lower GCS scores, mechanical ventilation, sepsis, anemia, and renal impairment (all P < 0.05)) ( Table 2 ). In multivariable logistic regression adjusted for age, sex, comorbidities, laboratory values, and other clinical variables, the RDW-to-platelet ratio remained independently associated with in-hospital mortality (aOR=1.046 per unit increase, 95% CI: 1.032–1.061, P < 0.001) ( Table 3 ). Sensitivity analyses using multiple imputation (five iterations) yielded consistent results (pooled estimates in S4 Table ), confirming robustness. Variance inflation factors were all < 5, indicating no substantial multicollinearity. To assess nonlinearity, a smoothed curve was fitted for the association between RDW-to-platelet ratio and in-hospital mortality ( Fig 2 ). The relationship showed an approximately linear trend. Mediation analysis demonstrated that the APACHE-IV score partially mediated the association between RDW-to-platelet ratio and in-hospital mortality ( Table 4 , Fig 3 ). Comparing high- versus low-ratio groups (8.25 vs. 5.23), the total effect was an absolute risk difference of 2.35% (95% CI: 1.54%–3.21%), comprising a direct effect of 1.88% (95% CI: 1.05%–2.72%) and an indirect effect via APACHE-IV of 0.47% (95% CI: 0.25%–0.71%), accounting for 20.15% of the total effect (95% CI: 10.43%–34.89%; all paths P < 0.001). This indicates that disease severity contributes substantially to the underlying biological pathway. Table 2. Univariate analysis for in-hospital mortality. Variable OR 95% CI P Value RDW-to-platelet ratio 1.07 (1.056, 1.080) <0.001 Total GCS score 0.77 (0.762, 0.785) <0.001 Age, years 1.02 (1.012, 1.020) <0.001 Gender Female vs Male 1.02 (0.911, 1.145) 0.720 APACHE-IV score 1.05 (1.044, 1.049) <0.001 Ethnicity Asian vs African American 1.17 (0.769, 1.772) 0.468 Caucasian vs African American 1.08 (0.897, 1.289) 0.431 Hispanic vs African American 1.23 (0.896, 1.697) 0.198 Native American vs African American 0.48 (0.146, 1.554) 0.219 Other/Unknown vs African American 1.25 (0.937, 1.671) 0.128 Mechanical ventilation Yes vs No 7.94 (7.022, 8.988) <0.001 Congestive heart failure Yes vs No 1.77 (1.418, 2.212) <0.001 Hypertension Yes vs No 0.90 (0.792, 1.017) 0.091 Diabetes mellitus Yes vs No 1.27 (1.080, 1.484) 0.004 Malignant tumor Yes vs No 2.03 (1.223, 3.379) 0.006 Sepsis Yes vs No 2.66 (2.202, 3.210) <0.001 Blood urea nitrogen, mg/dL 1.01 (1.012, 1.018) <0.001 Albumin, g/dL 0.57 (0.515, 0.623) <0.001 Hemoglobin, g/dL 0.93 (0.910, 0.956) <0.001 Creatinine, mg/dL 1.11 (1.066, 1.147) <0.001 Anticoagulant therapy Yes vs No 0.90 (0.583, 1.385) 0.628 Antiplatelet therapy Yes vs No 0.90 (0.705, 1.154) 0.413 Open in a new tab Abbreviations: OR, odds ratio; CI, confidence interval; GCS, Glasgow Coma Scale; APACHE, Acute Physiology and Chronic Health Evaluation. Statistical significance was set at P < 0.05 (two-sided). Table 3. Multivariable regression analysis of RDW-to-platelet ratio associated with in-hospital mortality. Characteristic Unadjusted OR (95% CI) P Value Adjusted OR (95% CI)† P Value† RDW-to-platelet ratio 1.068 (1.056, 1.080) <0.001 1.046 (1.032, 1.061) <0.001 Open in a new tab Abbreviations: OR, odds ratio; CI, confidence interval. Adjusted for age, sex, comorbidities, laboratory parameters, and other clinically relevant variables. Variance inflation factor (VIF) for all variables was < 5, indicating absence of multicollinearity. † Values derived from multiple imputation for missing data. Fig 2. RDW to platelet ratio (%). Open in a new tab Table 4. Mediation analysis of APACHE-IV score in the association between RDW-to-platelet ratio and in-hospital mortality. Effect Estimate 95% CI P Value Total Effect 2.35% (1.54%, 3.21%) <0.001 Direct Effect 1.88% (1.05%, 2.72%) <0.001 Indirect Effect (Mediated) 0.47% (0.25%, 0.71%) <0.001 Proportion Mediated 20.15% (10.43%, 34.89%) <0.001 Open in a new tab Effects represent absolute risk differences comparing high vs low RDW-to-platelet ratio (8.25 vs 5.23). Analysis adjusted for age, ethnicity, gender, congestive heart failure, diabetes mellitus, sepsis, malignant tumor, albumin, creatinine, and mechanical ventilation. Confidence intervals calculated using nonparametric bootstrap with percentile method (1,000 resamples). Abbreviations: CI, confidence interval; RDW, red cell distribution width; APACHE, Acute Physiology and Chronic Health Evaluation. Fig 3. Mediation analysis pathway diagram. Open in a new tab Subgroup analyses confirmed the main effect’s consistency across most strata ( Table 5 ; all P < 0.05 for RDW-to-platelet ratio–mortality interactions), including sex, age groups, race, mechanical ventilation status, and comorbidities. Significant heterogeneity was noted by APACHE-IV tertiles (Pinteraction < 0.001), with stronger associations in lower-risk patients (OR=1.118, 95% CI: 1.074–1.165) than in higher-risk patients (effect attenuated but significant). Modest effect modification was also evident by sex (Pinteraction = 0.038) and mechanical ventilation (Pinteraction = 0.003), with slightly stronger associations in females and non-ventilated patients. Missing data analysis revealed primary deficits in albumin (29.21%) and APACHE-IV scores (14.50%), with other covariates <5% missing ( S1 Table ). Multiple imputation via the missForest algorithm produced imputed datasets comparable to originals across key variables (all P > 0.05; S2 Table ), indicating no systematic bias and high data integrity. Table 5. Stratified analysis of RDW-to-platelet ratio and in-hospital mortality association. Stratification Variable Stratum n OR (95% CI) P Value P for interaction Total GCS score 0.015 Low 4,826 1.062 (1.047, 1.077) <0.001 High 4,656 1.068 (1.044, 1.092) <0.001 Age, years 0.388 Low 3,305 1.053 (1.034, 1.072) <0.001 Medium 3,261 1.084 (1.063, 1.105) <0.001 High 3,170 1.067 (1.043, 1.092) <0.001 Gender 0.038 Male 5,058 1.057 (1.041, 1.073) <0.001 Female 4,677 1.083 (1.064, 1.103) <0.001 APACHE-IV score <0.001 Low 2,776 1.118 (1.074, 1.165) <0.001 Medium 2,862 1.047 (1.014, 1.080) 0.004 High 2,686 1.032 (1.016, 1.048) <0.001 Ethnicity 0.171 African American 1,188 1.085 (1.053, 1.117) <0.001 Asian 208 1.019 (0.918, 1.131) 0.719 Caucasian 7,326 1.060 (1.046, 1.074) <0.001 Hispanic 397 1.100 (1.043, 1.160) <0.001 Native American 45 0.934 (0.597, 1.461) 0.765 Other/Unknown 525 1.116 (1.058, 1.178) <0.001 Mechanical ventilation 0.003 No 7,436 1.077 (1.058, 1.095) <0.001 Yes 2,300 1.038 (1.022, 1.055) <0.001 Congestive heart failure 0.866 No 9,244 1.067 (1.055, 1.079) <0.001 Yes 492 1.062 (1.011, 1.117) 0.018 Hypertension 0.172 No 6,646 1.072 (1.058, 1.087) <0.001 Yes 3,090 1.053 (1.030, 1.077) <0.001 Diabetes mellitus 0.064 No 8,434 1.072 (1.059, 1.086) <0.001 Yes 1,302 1.040 (1.010, 1.072) 0.009 Malignant tumor 0.872 No 9,655 1.067 (1.055, 1.079) <0.001 Yes 81 1.061 (0.992, 1.135) 0.086 Sepsis 0.006 No 9,143 1.070 (1.057, 1.084) <0.001 Yes 593 1.030 (1.005, 1.054) 0.017 Blood urea nitrogen, mg/dL 0.502 Low 3,337 1.051 (1.028, 1.075) <0.001 Medium 3,366 1.044 (1.019, 1.070) 0.001 High 2,985 1.071 (1.053, 1.088) <0.001 Albumin, g/dL 0.067 Low 2,483 1.047 (1.032, 1.062) <0.001 Medium 2,559 1.074 (1.044, 1.105) <0.001 High 1,850 1.076 (1.030, 1.124) 0.001 Hemoglobin, g/dL 0.035 Low 3,331 1.057 (1.042, 1.072) <0.001 Medium 3,267 1.087 (1.062, 1.114) <0.001 High 3,138 1.048 (1.014, 1.084) 0.006 Creatinine, mg/dL 0.264 Low 3,269 1.051 (1.027, 1.075) <0.001 Medium 3,220 1.054 (1.029, 1.079) <0.001 High 3,202 1.069 (1.052, 1.087) <0.001 Anticoagulant therapy 0.185 No 9,548 1.066 (1.054, 1.079) <0.001 Yes 188 1.123 (1.036, 1.217) 0.005 Antiplatelet therapy 0.921 No 9,137 1.068 (1.055, 1.080) <0.001 Yes 599 1.066 (1.021, 1.112) 0.003 Open in a new tab Abbreviations: OR, odds ratio; CI, confidence interval; GCS, Glasgow Coma Scale; APACHE, Acute Physiology and Chronic Health Evaluation. P interaction values indicate whether the effect modification by stratification variable is statistically significant. Statistical significance was set at P < 0.05 (two-sided). Discussion This retrospective cohort analysis, utilizing the eICU database, examined 9,736 critically ill stroke patients to explore the association between the red cell distribution width-to-platelet ratio (RDW-to-platelet ratio, RPR) and in-hospital mortality risk. Results revealed a significant positive correlation; in multivariable-adjusted models, each 1-unit increase in RPR conferred a 4.6% higher mortality risk (OR = 1.046, 95% CI: 1.032–1.061, P < 0.001). Stratified analyses affirmed consistency across subgroups, with amplified effects among females, those with elevated APACHE-IV scores, and patients without diabetes. Mediation analysis further disclosed that the APACHE-IV score partially mediated the RPR-mortality relationship, accounting for approximately 20.15% of the total effect, implying a pivotal intermediary role for disease severity in this biological pathway. As an index combining red cell distribution width (RDW) and platelet count (PLT), RPR may participate in the pathophysiology of stroke through multiple pathways. Elevated RDW is often associated with systemic inflammatory responses, oxidative stress, and endothelial dysfunction, all of which exacerbate brain injury and neuronal apoptosis after stroke. Concurrently, platelet count directly reflects coagulation function; stroke patients often present with hypercoagulable states or platelet dysfunction, and changes in RPR may indicate an imbalance in the coagulation-fibrinolysis system, thereby affecting thrombus formation, reperfusion injury, or bleeding risk. Furthermore, RPR, as an indicator of red blood cell volume heterogeneity, may be related to red blood cell deformability and insufficient microcirculatory perfusion, particularly in ischemic brain tissue, which further aggravates tissue hypoxia and damage. Our findings converge with multiple investigations leveraging MIMIC and eICU databases, supports the association between RPR and in-hospital mortality in stroke-critical care. Chen et al. reported that RPR independently forecasted in-hospital mortality in patients with acute exacerbations of chronic obstructive pulmonary disease (HR = 1.47, 95% CI: 1.29–1.68), persisting after adjustment for APACHE scores and inflammatory markers. Huang et al. [ 2 ] observed in hypertensive stroke patients that each 1% RDW elevation heightened 28-day mortality by 18% (OR = 1.18, 95% CI: 1.12–1.25), bolstering RDW’s prognostic utility as a systemic inflammation surrogate. Although these did not directly utilize RPR, mechanistic alignment is evident: RDW signifies chronic inflammation and erythrocytic anisocytosis, while platelet count denotes coagulopathy; their ratio holistically gauges systemic stress in critical illness. Moreover, Guo et al. [ 1 ] demonstrated that acute myocardial infarction augmented RDW’s mortality predictive capacity in stroke (interaction P = 0.03), underscoring comorbidity modulation. Although we eschewed stratification by acute myocardial infarction, confounder adjustments for malignancy and infection enhanced robustness. Contemporary evidence reinforces RPR’s role in stroke prognostication: Liang et al. [ 8 ] linked elevated admission RPR to heightened 30-day mortality in spontaneous intracerebral hemorrhage (highest tertile HR = 1.37, 95% CI: 1.15–1.64), albeit non-significant in low comorbidity index (CCI 3 mg/L) subgroups (OR = 3.6, 95% CI: 1.6–8.3), with no significant difference in low-CRP groups (P = 0.6), underscoring RPR’s context-dependent prognostic value. These convergences substantiate RPR’s validity associated with stroke-critical care, while highlighting its amplified utility in high-risk subgroups, aligning with our high-APACHE-IV effects. Methodologically, this study harbors several strengths. Foremost, sourcing from the multicenter eICU database encapsulates real-world ICU dynamics, augmenting external validity. Second, multiple imputation via the missForest algorithm mitigated missing data, with sensitivity analyses evincing negligible baseline disparities pre- and post-imputation (all P > 0.05) [ S2 Table ], thereby curtailing selection bias. Third, restricted cubic splines and smoothed curve fitting corroborated the linearity of the RPR-mortality nexus, forestalling overinterpretation inherent to nonlinear paradigms. Finally, mediation analysis quantified APACHE-IV’s partial intermediary function in the RPR-mortality pathway (proportion: 20.15%), furnishing rigorous statistical scaffolding for pathophysiological inference. This delineates the indirect causal sequence “RPR elevation → exacerbated disease severity → mortality,” transcending conventional regression confines to enrich systemic inflammation-prognosis insights. Notwithstanding compelling results, interpretive caution is warranted due to inherent limitations. Primarily, the retrospective observational design precludes causal inference regarding RPR and in-hospital mortality, with reverse causation (e.g., disease progression precipitating RDW escalation and platelet depletion) unexcluded [ 5 ]. Moreover, due to the 2014–2015 eICU data used, the data may not fully capture contemporary management of stroke in the ICU, potentially limiting generalizability to current clinical practice. Secondly, eICU’s omission of stroke subtypes (ischemic vs. hemorrhagic), lesion loci, NIHSS scores, and imaging precludes granular neurological impairment adjustment. Third, reliance on admission-singleton RPR measurements neglects temporal fluctuations; Ye et al. [ 9 ] and Li et al. [ 10 ] affirmed that longitudinal platelet or RPR trajectories surpass baselines in sepsis prognostication, advocating serial assessments henceforth. Additionally, Tong et al. [ 11 ] highlighted RPR’s independent predictivity in critically ill acute myocardial infarction yet cautioned comorbidity heterogeneity’s bias amplification. Our U.S.-centric data may curtail generalizability amid racial and socioeconomic variances, particularly in non-US cohorts. Lastyly, the paucity of RPR-focused systematic reviews or meta-analyses necessitates prospective multicenter corroboration. In addition, although the results are robust, the study’s use of a single RPR measurement at admission remains a significant limitation. Future work should obtain multi-timepoint RPR data during hospitalization and apply time-varying analyses to evaluate the incremental prognostic value of RPR dynamics. RPR is independently associated with in-hospital mortality in critically ill stroke patients, harboring substantial translational promise. Derived from routine complete blood counts, it amalgamates inflammatory (RDW) and hemostatic (platelet) cues, surpassing singular biomarkers in biological coherence and prognostic acuity. Its economical, accessible profile is ideally suited for resource-constrained milieus, facilitating precocious risk stratification, high-acuity triage, and stewardship of surveillance assets—such as vigilant vital-sign oversight, interdisciplinary orchestration, or advancement to escalated care tiers. Mediation scrutiny posits APACHE-IV as a partial mediator, framing RPR as a holistic disease burden surrogate beyond primordial aberrations. Prospective inquiries might prioritize RPR-steered precision modalities, like subgroup-specific anti-inflammatory or antiplatelet regimens to ameliorate outcomes [ 8 , 12 , 13 ], alongside multimodal risk-model assimilation to refine stroke-ICU stewardship. Supporting information S1 Table. Missing data analysis. (DOCX) pone.0344361.s001.docx (14.1KB, docx) S2 Table. Convergence diagnostics of the missForest imputation algorithm. (DOCX) pone.0344361.s002.docx (13.1KB, docx) S3 Table. Comparison of original and imputed data quality assessment. (DOCX) pone.0344361.s003.docx (14.1KB, docx) S4 Table. Results from multiple imputation analysis. (DOCX) pone.0344361.s004.docx (12.3KB, docx) Data Availability The datasets used and analyzed during the current study are derived from the eICU Collaborative Research Database, a publicly available and ethically approved critical care database. Researchers can access this database upon signing a data use agreement and completing the required training. The specific data extraction of this study were uploaded. Funding Statement The author(s) received no specific funding for this work. References 1. Guo T, Qin Z, He D. Acute Myocardial Infarction (AMI) as the Effect Modifiers to Modify the Association Between Red Blood Cell Distribution Width (RDW) and Mortality in Critically Ill Patients With Stroke. Front Med (Lausanne). 2022;9:754979. doi: 10.3389/fmed.2022.754979 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Huang J, Chen H, Deng J, Liu X, Shu T, Yin C, et al. Interpretable machine learning for predicting 28-day all-cause in-hospital mortality for hypertensive ischemic or hemorrhagic stroke patients in the ICU: a multi-center retrospective cohort study with internal and external cross-validation. Front Neurol. 2023;14:1185447. doi: 10.3389/fneur.2023.1185447 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Li Y, Xu C, Qin Z, Ge L. Relationship Between the Hemoglobin-to-Red Cell Distribution Width Ratio and in-Hospital Mortality in Patients with Chronic Heart Failure. Vasc Health Risk Manag. 2024;20:553–65. doi: 10.2147/VHRM.S486075 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Huang S, Zhou Q, Guo N, Zhang Z, Luo L, Luo Y, et al. Association between red blood cell distribution width and in-hospital mortality in acute myocardial infarction. Medicine (Baltimore). 2021;100(15):e25404. doi: 10.1097/MD.0000000000025404 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Bellan M, Azzolina D, Hayden E. Simple parameters from complete blood count predict in-hospital mortality in COVID-19. Dis Markers. 2021;2021:8863053. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Galindo-Martín CA, Chong-Aviña PA, Godinez-Breacher V, Aportela-Vázquez VA, Bueno-Hernández G, Gante-García MFD, et al. Malnutrition: muscle wasting, inflammation, RDW, and their relation with adverse outcomes. Cir Cir. 2024;92(2):150–8. doi: 10.24875/CIRU.22000634 [ DOI ] [ PubMed ] [ Google Scholar ] 7. Chen S, Shi Y, Hu B, Huang J. A Prediction Model for In-Hospital Mortality of Acute Exacerbations of Chronic Obstructive Pulmonary Disease Patients Based on Red Cell Distribution Width-to-Platelet Ratio. Int J Chron Obstruct Pulmon Dis. 2023;18:2079–91. doi: 10.2147/COPD.S418162 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Liang H, Liu P, Guo L, Feng J, Yin C, Zhao D, et al. 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. Front Neurol. 2023;14:1221335. doi: 10.3389/fneur.2023.1221335 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Ye Q, Wang X, Xu X, Chen J, Christiani DC, Chen F, et al. Serial platelet count as a dynamic prediction marker of hospital mortality among septic patients. Burns Trauma. 2024;12:tkae016. doi: 10.1093/burnst/tkae016 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Li Z, Hong H, Xu J. Red Cell Distribution Width to Platelet Ratio Is Associated with Mortality in Patients with Sepsis. Dis Markers. 2022;2022:4802702. doi: 10.1155/2022/4802702 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Tong L, Liu Y-Q, Shen JH, B O M, Zhou Q, Duan X-J, et al. Relationship between the red cell distribution width-to-platelet ratio and in-hospital mortality among critically ill patients with acute myocardial infarction: a retrospective analysis of the MIMIC-IV database. BMJ Open. 2022;12(9):e062384. doi: 10.1136/bmjopen-2022-062384 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Xu N, Peng C. Association between red cell distribution width-to-platelet ratio and short-term and long-term mortality risk in patients with acute ischemic stroke. BMC Neurol. 2023;23(1):191. doi: 10.1186/s12883-023-03219-1 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. He K, Xie X, Duan X, Zhou Q, Wu J. Red Cell Distribution Width-to-Platelet Count Ratio: A Promising Predictor of In-Hospital All-Cause Mortality in Critically Ill Patients with Acute Ischemic Stroke. Cerebrovasc Dis. 2023;52(6):692–9. doi: 10.1159/000529184 [ DOI ] [ PubMed ] [ Google Scholar ] PLoS One. doi: 10.1371/journal.pone.0344361.r001 Decision Letter 0 Marwan Al-Nimer Marwan Al-Nimer Academic Editor Find articles by Marwan Al-Nimer Author information Copyright and License information Roles Marwan Al-Nimer : Academic Editor © 2026 Marwan Al-NimerMarwan Al-NimerMarwan Al-NimerMarwan Al-Nimer This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. PMC Copyright notice 4 Jan 2026 Dear Dr.  Chen, Please submit your revised manuscript by Feb 18 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at [email protected] . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file. A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'. A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'. An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'. If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter. If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .... We look forward to receiving your revised manuscript. Kind regards, Marwan Salih Al-Nimer, MD, PhD Academic Editor PLOS One Journal Requirements: When submitting your revision, we need you to address these additional requirements. 1.Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf 2. PLOS requires an ORCID iD for the corresponding author in Editorial Manager on papers submitted after December 6th, 2016. Please ensure that you have an ORCID iD and that it is validated in Editorial Manager. To do this, go to ‘Update my Information’ (in the upper left-hand corner of the main menu), and click on the Fetch/Validate link next to the ORCID field. This will take you to the ORCID site and allow you to create a new iD or authenticate a pre-existing iD in Editorial Manager. 3. We note that you have indicated that there are restrictions to data sharing for this study. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions . Before we proceed with your manuscript, please address the following prompts: a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., a Research Ethics Committee or Institutional Review Board, etc.). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent. b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of recommended repositories, please see https://journals.plos.org/plosone/s/recommended-repositories . You also have the option of uploading the data as Supporting Information files, but we would recommend depositing data directly to a data repository if possible. We will update your Data Availability statement on your behalf to reflect the information you provide. 4. When completing the data availability statement of the submission form, you indicated that you will make your data available on acceptance. We strongly recommend all authors decide on a data sharing plan before acceptance, as the process can be lengthy and hold up publication timelines. Please note that, though access restrictions are acceptable now, your entire data will need to be made freely accessible if your manuscript is accepted for publication. This policy applies to all data except where public deposition would breach compliance with the protocol approved by your research ethics board. If you are unable to adhere to our open data policy, please kindly revise your statement to explain your reasoning and we will seek the editor's input on an exemption. Please be assured that, once you have provided your new statement, the assessment of your exemption will not hold up the peer review process. 5. Please include your full ethics statement in the ‘Methods’ section of your manuscript file. In your statement, please include the full name of the IRB or ethics committee who approved or waived your study, as well as whether or not you obtained informed written or verbal consent. If consent was waived for your study, please include this information in your statement as well. 6. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information .... 7. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. Additional Editor Comments: Major revision [Note: HTML markup is below. Please do not edit.] Reviewer's Responses to Questions Comments to the Author 1. Is the manuscript technically sound, and do the data support the conclusions? Reviewer #1: Yes Reviewer #2: Partly ********** 2. Has the statistical analysis been performed appropriately and rigorously? -->?> Reviewer #1: Yes Reviewer #2: No ********** 3. Have the authors made all data underlying the findings in their manuscript fully available??> The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.--> Reviewer #1: Yes Reviewer #2: Yes ********** 4. Is the manuscript presented in an intelligible fashion and written in standard English??> Reviewer #1: Yes Reviewer #2: Yes ********** Reviewer #1: This manuscript investigates the relationship between the Red Cell Distribution Width-to-Platelet Ratio (RPR) and in-hospital mortality in critically ill stroke patients using the eICU database. The study addresses a clinically relevant question regarding accessible biomarkers. However, the article has some methodological issues. Specifically, the rationale for covariate selection in the regression model is not sufficiently justified, and the distinction between "association" and "prediction" is blurred. Furthermore, crucial information regarding the calculation of the primary exposure (RPR) and the visual data (Figure 1) is missing or unclear. Given the retrospective nature of the study and the lack of external validation, the authors should refrain from overemphasizing "prediction" and focus on "independent association." Reviewer #2: The study has the following shortcomings: This study adopts a retrospective cohort design, which could reveal the association between the red cell distribution width-to-platelet ratio (RPR) and in-hospital mortality in critically ill stroke patients, but cannot establish a causal relationship between them. There is a possibility of reverse causation: the progression of the patient's condition may lead to an increase in red cell distribution width (RDW) and a decrease in platelets, thereby elevating RPR, rather than elevated RPR directly increasing the risk of death. This potential logical relationship has not been effectively excluded. The study uses data from the 2014–2015 US multicenter eICU database, which is nearly a decade old. The research conclusions drawn based on outdated data have limited applicability in current clinical practice. The database lacks key neurological information such as stroke subtype (ischemic or hemorrhagic), lesion location, and National Institutes of Health Stroke Scale (NIHSS) score. This makes it impossible to conduct stratified analysis for different subtypes of stroke patients and difficult to fully adjust for the impact of the degree of neurological impairment on prognosis, which may lead to confounding bias in the study results. The data are only derived from the US population, lacking samples from other ethnic groups (such as Asians, Africans, etc.) and regions. Thus, the applicability of the findings to non-US populations remains unclear. The study only uses a single RPR measurement within 24 hours of admission as the exposure variable, without considering the dynamic changes of RPR during hospitalization. Relying solely on a single measurement may fail to fully reflect the patient's disease progression and real prognostic risk. The NIHSS score is an important indicator for evaluating the degree of neurological impairment and prognosis in stroke patients, but it was not included as a confounding factor in the adjustment, which may lead to overestimation or underestimation of the association between RPR and mortality. Ischemic stroke and hemorrhagic stroke differ significantly in pathophysiological mechanisms, treatment strategies, and prognosis, and the predictive value of RPR for mortality may vary between the two. However, the study did not perform subgroup analysis based on this key stratification factor, resulting in insufficient precision of the research conclusions. The study mentions using the missForest algorithm for multiple imputation to handle missing data, but it does not detail the specific parameter settings of the algorithm, the convergence test results during the imputation process, nor fully compare the differences in data distribution characteristics before and after imputation (only stating that there were no significant differences in key variables). This makes it difficult to evaluate the rationality and reliability of the imputation method. The study uses a multivariable logistic regression model to analyze the association between RPR and in-hospital mortality, but only reports the odds ratio (OR) and confidence interval without assessing the model's calibration (e.g., Hosmer-Lemeshow test). Thus, the consistency between the model's predicted results and actual observed results cannot be determined. The study only briefly mentions that RPR integrates pathophysiological processes related to inflammation and coagulation function, but does not deeply analyze the molecular mechanisms and pathophysiological pathways through which RPR affects patient prognosis in combination with the specific pathological mechanisms of stroke, resulting in insufficient theoretical depth of the research. ********** what does this mean? ). If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files.). If published, this will include your full peer review and any attached files. If you choose “no”, your identity will remain anonymous but your review may still be made public. Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our For information about this choice, including consent withdrawal, please see our For information about this choice, including consent withdrawal, please see our For information about this choice, including consent withdrawal, please see our Privacy Policy ..--> Reviewer #1: No Reviewer #2: No ********** [NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.] To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation . NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications. Attachment Submitted filename: Peer Review Report1.docx pone.0344361.s005.docx (14.9KB, docx) PLoS One. 2026 Apr 17;21(4):e0344361. doi: 10.1371/journal.pone.0344361.r002 Author response to Decision Letter 1 Article notes Copyright and License information Collection date 2026. PMC Copyright notice 18 Feb 2026 Dear editors, Thank you for your letter dated Jan 04 2026 concerning our manuscript titled "The Association between RDW-to-Platelet Ratio and In-Hospital Mortality in Critically Ill Stroke Patients: A Retrospective Cohort Study Based on the eICU Database" (Manuscript ID: PONE-D-25-61250). We appreciate the opportunity to revise and resubmit our work. We are grateful to you and the reviewers for the insightful comments and constructive criticism, which have significantly helped us improve the quality and clarity of our manuscript. We have carefully considered all the points raised by the reviewers and have revised the manuscript accordingly. Our point-by-point responses to each comment are detailed below, with changes highlighted in the revised manuscript. Yu Chen Attachment Submitted filename: Response to Reviewers.docx pone.0344361.s007.docx (22.5KB, docx) PLoS One. doi: 10.1371/journal.pone.0344361.r003 Decision Letter 1 Marwan Al-Nimer Marwan Al-Nimer Academic Editor Find articles by Marwan Al-Nimer Author information Copyright and License information Roles Marwan Al-Nimer : Academic Editor © 2026 Marwan Al-NimerMarwan Al-NimerMarwan Al-NimerMarwan Al-Nimer This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. PMC Copyright notice 20 Feb 2026 The Association between RDW-to-Platelet Ratio and In-Hospital Mortality in Critically Ill Stroke Patients: A Retrospective Cohort Study Based on the eICU Database PONE-D-25-61250R1 Dear Dr. Yu Chen, We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements. Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication. An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact  and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact  and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact  and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support .... If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact [email protected]. Kind regards, Marwan Salih Al-Nimer, MD, PhD Academic Editor PLOS One Additional Editor Comments (optional): Reviewers' comments: PLoS One. doi: 10.1371/journal.pone.0344361.r004 Acceptance letter Marwan Al-Nimer Marwan Al-Nimer Academic Editor Find articles by Marwan Al-Nimer Author information Copyright and License information Roles Marwan Al-Nimer : Academic Editor © 2026 Marwan Al-NimerMarwan Al-NimerMarwan Al-NimerMarwan Al-Nimer This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. PMC Copyright notice PONE-D-25-61250R1 PLOS One Dear Dr. Chen, I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS One. Congratulations! Your manuscript is now being handed over to our production team. At this stage, our production department will prepare your paper for publication. This includes ensuring the following: * All references, tables, and figures are properly cited * All relevant supporting information is included in the manuscript submission, * There are no issues that prevent the paper from being properly typeset You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps. Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact [email protected]. You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing . If we can help with anything else, please email us at [email protected]. Thank you for submitting your work to PLOS ONE and supporting open access. Kind regards, PLOS ONE Editorial Office Staff on behalf of Professor Marwan Salih Al-Nimer Academic Editor PLOS One Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials S1 Table. Missing data analysis. (DOCX) pone.0344361.s001.docx (14.1KB, docx) S2 Table. Convergence diagnostics of the missForest imputation algorithm. (DOCX) pone.0344361.s002.docx (13.1KB, docx) S3 Table. Comparison of original and imputed data quality assessment. (DOCX) pone.0344361.s003.docx (14.1KB, docx) S4 Table. Results from multiple imputation analysis. (DOCX) pone.0344361.s004.docx (12.3KB, docx) Attachment Submitted filename: Peer Review Report1.docx pone.0344361.s005.docx (14.9KB, docx) Attachment Submitted filename: Response to Reviewers.docx pone.0344361.s007.docx (22.5KB, docx) Data Availability Statement The datasets used and analyzed during the current study are derived from the eICU Collaborative Research Database, a publicly available and ethically approved critical care database. Researchers can access this database upon signing a data use agreement and completing the required training. The specific data extraction of this study were uploaded. Articles from PLOS One are provided here courtesy of PLOS ACTIONS View on publisher site PDF (729.9 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top