The RDW is a simple, rapid, and easily available assessment measure of erythrocyte volume variability and heterogeneity (1). Recently, besides hematological disorders, RDW has emerged as a potential biomarker in many critical clinical cases, mainly in cardiovascular and cerebrovascular diseases (2, 3).

Globally, the aging of the population and the accumulation of risk factors lead to an increased lifetime risk of stroke (4). According to a report from the American Heart Association, the mortality rate after stroke was 10.5, 21.2, 39.8, and 58.4% at 30 days after stroke, 1 year after stroke, and 5 years after stroke or even longer, respectively (5, 6). Given complicated pathophysiological mechanisms of stroke, outcome prognostication in patients with stroke remains challenging. The ability of existing stroke prognostic prediction models was limited. Blood-based biomarkers might provide additional information to established prognostic factors (7, 8).

As far as existing research was concerned, no uniform conclusions had been drawn thus far about the role of RDW in the prognosis of stroke (9). There were several studies investigating the connection between RDW and stroke. A meta-analysis including a total of 31 studies with 3,487,896 patients concluded that higher baseline RDW could be as a predictor of stroke occurrence and an unfavorable functional outcome; higher mortality was found in patients with increased RDW (OR/RR, 1.278; 95% CI = 1.221,1.337) (10). Zhao et al. (11) analyzed 4,314 patients with stroke, retrospectively, and the results showed that the hazard ratio (95%, CI) of mortality for the second (RDW: 13.4–14.3%) and third (>14.3%) tertiles was 1.15 (0.96, 1.37) and 1.40 (1.17, 1.68), respectively, compared with the reference group (RDW <13.4%). However, Ntaios et al. (12) retrospectively analyzed 1,504 patients with stroke with multivariate robust regression analysis, and the findings manifested that RDW was significantly associated with a functional outcome (OR = 10.73 for a poor outcome, p < 0.001) at univariate analysis but not multivariate. The clinical significance of RDW in stroke has not been comprehensively investigated, owing to the differences in study population, study design, adjustment for covariates. Therefore, to provide additional evidence on the potential association of baseline RDW with prognosis following acute stroke, we performed a retrospective cohort study using the eICU Collaborative Research Database (eICU) (13), which is an available multi-center database, including 58 hospitals in the USA.

The retrospective cohort study was designed to find the association of RDW with in-hospital mortality in patients with stroke based on eICU. Our results indicated that a higher RDW, in the short term, was significantly and positively connected with increased all-cause mortality in patients with stroke. In addition, our research showed that this would make patients with the complication of AMI face a greater risk of death compared with people without. A few previous cohort studies indicated that RDW was independently associated with an adverse functional outcome or death after stroke (8, 10, 15–20). Their findings were consistent with ours and provided support for our present study. A study of 847 consecutive patients with first-ever acute cerebral infarction within 7 days of the symptom onset reported that RDW was relevant to a poor functional outcome (modified Rankin Scale >2) and all-cause mortality at 3 months, and 1 year after the stroke onset (8). Cohort analysis of 274 adult patients with subarachnoid hemorrhage (SAH) admitted to ICU demonstrated that RDW was a useful index to evaluate hospital mortality and 1-year mortality (20). However, most of these studies were all limited by the relatively small sample size and did not include all stroke populations but only included ischemic stroke or SAH. Additionally, present reports had indicated that RDW is related to the prognosis of patients with AMI or acute coronary syndrome (21–25), but these studies had not clearly clarified the effect of the comorbidities on the relationship between RDW and death in patients with stroke. In contrast, there were two previous studies that had pointed out opposite results with ours as follows. A cohort study of 1,504 participants with ischemic stroke admitted in the Center Hospitalier Universitaire Vaudois (CHUV) Stroke Unit, Lausanne cleared that RDW did not predict a functional outcome. In the study, patients with intracerebral hemorrhage and SAH were excluded (11). Another population-based cohort study of 1,152 subjects found no association between RDW and risk of all-cause mortality among patients with incident stroke (the median follow-up time was 15.8 years) (26). It is worth noting that RDW was measured, on average, 7.4 years prior to the cerebrovascular event in the study; therefore, the timing between RDW measurement, stroke, and death could have influenced the findings. A retrospective cohort study of 180 patients who underwent CEA reported that increased RDW-CV was associated with increased long-term mortality (adjusted hazard ratio [aHR] 2.455, 95% CI, 1.231–4.894, p = 0.011) (27). A secondary analysis of the Medical Information Mart for Intensive Care III (MIMIC-III) database study showed RDW as an independent risk factor of all-cause mortality of patients with AKI (HR, 1.219, 95% CI, 1.211 to 1.228); RDW is positively correlated with survival time of a 4-year follow-up in critically ill patients with AKI (28). Therefore, we speculated that RDW could predict both short-term and long-term prognoses, and this speculation still needed more evidence to confirm.

There was no conclusive conclusion about the mechanism of RDW and increased mortality after stroke. The following were plausible explanations about the underlying biological mechanisms. A recent study has confirmed the independent association between RDW-standard deviation (RDW-SD) and serum neuron-specific enolase (NSE) levels, and NSE was used as a biomarker for evaluating neuronal damage and predicting the prognosis of stroke. These results indicated that RDW-SD is related to neuronal damage in patients with stroke (1). In the process of ischemic stroke (IS), the death of vascular endothelial cells destroyed the blood-brain barrier, leading to neuronal degeneration, swelling, and necrosis of glial cells. Subsequently, the NSE released by the injured neurons crosses the blood-brain barrier and rapidly diffused into the bloodstream through the ischemic tissue (29). Secondly, it had been found that RDW was positively correlated with plasma inflammatory biomarkers, such as C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and interleukin (IL)-6 (30–34). A study reported that inflammation was related to the process of IS, from initial ischemia to infarction and secondary repair (35, 36). Inflammation and oxidative stress may interfere with red blood cell production, inhibit red blood cell maturation, change the deformability of red blood cell membranes and the half-life of red blood cells in circulation, leading to an increase in RDW (37). Therefore, the elevated RDW increased the morbidity and mortality of IS by becoming a marker that increased inflammation, induced stroke, and led to a post-stroke state, or marked a prethrombotic state. Thirdly, higher RDW may reduce blood oxygen saturation, thereby reducing physiological reserves to resist hypoxic stress (38). In addition, it may cause abnormal blood coagulation and increase the risk of thromboembolism (39), and finally lead to stroke recurrence, which may exacerbate brain damage or lead to a worse prognosis. In view of the predictive role of RDW in many other diseases, RDW is more likely to be a general marker rather than a specific predictor of stroke (6). Based on the relationship between RDW and the prognosis of patients with stroke, future research should continue to explore whether the application of RDW can help doctors stratify the risk factors of stroke and secondary prevention in clinical practice to make clinical decisions more beneficial to patients.

This study had following strengths: Firstly, the sample size is large. Secondly, the interaction test expands our findings, that is, in patients with AMI and non-HP, an increase in RDW means a higher risk of death in patients, which will provide a better reminder for clinical decision-making. Thirdly, in this study, we ensured the robustness of the results through a series of sensitivity analyses (conversion of target independent variable form, subgroup analysis). This made our results more reliable.

Several limitations should be considered in the interpretation of our results. First, our study is limited to the American population; therefore, the conclusions of this study need to be cautious when applied to populations in other countries. Second, although we tried to adjust for possible confounding factors, However, residual confounders still cannot be ruled out completely. Finally, we did not exclude patients with hemoglobinopathies, which have been shown to increase RDW values like thalassemia and sickle-cell anemia (40–43).

In summary, in population of ICU with stroke, we observed RDW levels were significantly associated with a poor prognosis in a positive linear fashion. Most importantly, we found that this association could be modified by AMI and HP. Our findings, if further confirmed, suggested that maintaining a lower level of myocardial injury markers, especially among patients with stroke or AMI, may offer a safer intervention strategy to reduce the mortality of stroke in ICU.

The data analyzed in this study was obtained from the eICU Collaborative Research Database, the following licenses/restrictions apply: Researchers seeking to use the database must formally request access. Requests to access these datasets should be directed to https://eicu-crd.mit.edu/gettingstarted/access/.

The studies involving human participants were reviewed and approved by Massachusetts Institute of Technology (Cambridge, MA, USA). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

TG contributed to conception and design of the study, drafted the manuscript, and performed the statistical analysis. ZQ organized the database. TG and DH reviewed the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.