Diagnostic and prognostic value of red blood cell distribution width in sepsis: A narrative review

Highlights

• RDW can be used as a prognostic and diagnostic index severe infections and sepsis.

• In suspected sepsis, RDW has modest value for predicting positive blood culture.

• Increased RDW on admission is associated with unfavorable outcomes.

• RDW could be used as a prognostic index in septic patients.

Abstract

Previous studies showed that red blood cell distribution width (RDW) can be used as a prognostic and diagnostic index in various non-hematological diseases, including severe infections and sepsis. Here, we provide a narrative review to summarize the findings of available studies investigating the relationship between RDW and sepsis. Current evidence suggests that increased RDW on admission, both in adults and neonates, may be associated with unfavorable outcomes on the short- and long-term. In patients with suspected sepsis, RDW has modest value for predicting positive blood culture. Accordingly, its diagnostic value for sepsis seems limited, whilts dynamic changes of RDW are associated with outcome of sepsis. Taken together, these results suggest that RDW could be used as a prognostic index in septic patients.

Introduction

Red blood cell distribution width (RDW) is a red blood cell (RBC) index which is rapidly and automatically calculated by all modern hematological analyzers [1], [2]. The reference interval of RDW varies with age, gender, race and instrumental used [3], [4], [5], [6], [7], with the lower limit around 11.5% and upper limit around 15%. Decreased RDW has no clinical implication; however, increased RDW, which reflects large size variation of RBC, is clinically meaningful. For a long time RDW has been regarded as a diagnostic index for distinguishing thalassemia and iron deficiency anemia [8]. However, several studies have indicated that it may have meaningful clinical value also in many non-hematological conditions, such as autoimmune disorders [9], cardiovascular disease [10] and critical illness [11]. To date, several studies have investigated the clinical implications of RDW in sepsis. Here, we provide a narrative review to summarize major findings of available studies.

Several studies have measured RDW value on admission for predicting short-term mortality. The endpoint in these studies typically includes in-hospital mortality [12], [13], [14], [15], 28-day mortality [16], [17], [18], [19], intensive care unit (ICU) mortality [15], as well as 30-day mortality [20], [21], [22], [23].

Two retrospective studies have explored the prognostic value of RDW in severe sepsis and septic shock patients aged 65 years or older [12], [21]. In one study including 117 subjects, the prognostic value of RDW was analyzed using a multivariable regression model. Increased RDW on admission was hence found to be an independent risk factor for in-hospital mortality, displaying a hazard ratio (HR) of 1.18 (95% CI: 1.03 – 1.35) per 1% increase of RDW [12]. In another study, including 417 subjects, the authors found that the HR of per 1% RDW was 1.10 (95% CI: 1.04 – 1.17) for 30-day mortality [21]. Notably, these studies lack adjustment for some confounding factors such as comorbidities and treatment approaches. Therefore, other investigations with full-adjusted model are needed to further investigating the prognostic value of RDW in older sepsis patients.

Some studies also investigated baseline RDW and short-term outcomes in patients with severe sepsis and septic shock, regardless of their age [16], [18], [19]. In a prospective study, involving 329 subjects, increased RDW (both on admission and 72 h afterwards), was associated with higher 28-day and 90-day mortality rates [19]. In a retrospective study, based on 566 severe sepsis and septic shock patients, the 28-day mortality was found to be 13.1% in subjects with RDW < 14%, but increased to 44.9% in the group of subjects with RDW higher than 15.8% [16]. Moreover, in multivariable Cox regression model, subjects with RDW > 15.8% had a nearly 2.6 higher risk of mortality compared to those with lower values (HR, 2.57; 95% CI: 1.53 – 4.34) [16]. The third study which investigated the prognostic value of RDW in patients with severe sepsis and septic shock was published in 2019 [18]. A total number of 730 subjects were used to generate a scoring system aimed at predicting 28-day mortality. The predictive power of a model including RDW, delta neutrophil index, and platelet count, was found to be better than each single test, and was also found to be an independent predictor of 28-day mortality, displaying an area under the receiver operating characteristic (ROC) curve (AUC) of 0.79 (95% CI: 0.74 – 0.83) [18].

In a study based on 279 septic shock patients, RDW remained independently associated with in-hospital and ICU mortality after adjusting for age, gender, body mass index (BMI), acute physiology and chronic health evaluation II (APACHE II) and sequential organ failure assessment (SOFA) scores, comorbidities and organ failure [15]. RDW showed good discriminative efficiency for in-hospital mortality, displaying an AUC of 0.74, which was found to be even higher than that of APACHE II and SOFA scores (AUC = 0.69 for both). The combination of RDW and APACHE II score was also found to improve the predictive accuracy of the APACHE II score alone, with an AUC increased to 0.77 [15].

Some studies also investigated the prognostic value of RDW in sepsis patients, regardless of severity. In a retrospective study with a large sample size (n = 6973), the authors reported that the AUC of RDW for predicting in-hospital mortality was as high as 0.75 (95% CI: 0.72–0.77) [13]. In the sepsis patients group with RDW > 15.4% the mortality rate was as high as 16% compared to 1.6% in those with RDW < 13%. In a recently published study [17], where sepsis was diagnosed using the Sepsis-3 criteria [24], the authors found that the AUC of RDW for predicting 28-day mortality was 0.718, lower than that of SOFA (AUC: 0.840). However, in multivariable logistic regression model, RDW was found to be associated with 28-day mortality (OR, 1.49; 95% CI: 1.01 – 2.20) [17]. In another study, using the International Classification of Diseases Ninth Revision (ICD-9) code for sepsis, the authors extracted clinical data of 43,212 critical ill patients admitted to Brigham and Women’s Hospital (BWH) and Massachusetts General Hospital (MGH) and investigated the prognostic value of RDW at discharge [23]. The endpoint in this study included 30-day, 90-day and 1-year mortality. In a subgroup analysis of sepsis patients (n = 4513), RDW > 15.8% was found to be independently associated with 30-day mortality (OR, 5.43; 95% CI: 2.35 – 12.54) [23]. Interestingly, the rate of sepsis patients among critical illness increased in parallel with the RDW value, thus suggesting that RDW may represent a risk factor of sepsis in critical illness [23].

Although it may be argued that the prognostic significance of RDW may vary according to the presence or not of anemia, a retrospective study in 349 septic patients found that RDW remained independently associated with in-hospital mortality after adjustment for hemoglobin values [14].

In a retrospective study performed by Chen et al. [25], the authors used ICD-9 code to define suspected sepsis and extracted baseline data from medical records. After subjects were divided in a training and validation cohort, the authors found that chills, hypothermia, anemia, RDW and history of cancer were independent risk factors for in-hospital mortality in both cohorts. In addition, a score system containing these factors (termed CHARM) showed good performance to predict in-hospital mortality (AUC: 0.77; 95% CI: 0.75 – 0.79).

Some studies also investigated the accuracy of RDW for diagnosing sepsis [26], [27]. In a retrospective study in China, blood culture was used to diagnose sepsis in 120 patients [26]. Sixty of 120 subjects had blood culture positive and RDW was higher in these patients (13.6 vs 14.1; p = 0.02). In ROC curve analysis, the AUC of RDW was 0.621 (95% CI: 0.520 – 0.722). In another prospective study in Switzerland, RDW was measured in 1083 patients with suspected infection admitted to the emergency department (ED) [27]. All these subjects had blood culture sampling at ED admission and 104 of them (9.6%) had positive blood cultures. Although RDW was increased in positive blood culture patients, it did not provide significant results in ROC curve analysis (AUC: 0.61). Notably, the AUC of procalcitonin (PCT) in these two studies [26], [27] was > 0.80, thus showing that the predictive accuracy of RDW for positive blood culture is lower than that of PCT, which is currently considered the biochemical gold standard in the diagnostic approach to septic patients [28].

In a Korean study, sepsis patients (n = 130) were found to have significantly higher RDW than healthy controls (n = 280) and RDW displayed a very high diagnostic efficiency (AUC: 0.951; 95% CI: 0.925 – 0.970) [29]. Nevertheless, caution shall be used when interpreting these data, due to the case-control design of the study and the inclusion of healthy subjects as controls [30]. In another study, researchers compared the RDW level in sepsis (n = 27) and adult onset Still’s disease (AOSD) (n = 21), and found that RDW was significantly increased in AOSD (15.0% vs 13.3%; p = 0.001) [31], thus concluding that RDW is a conjugated marker for differentiate sepsis and AOSD. However, ROC curve analysis was not used to evaluate the diagnostic accuracy of RDW.

Taken together, the current evidence does not seemingly support RDW as a reliable marker for diagnosing sepsis. Further studies, characterized by a large sample size and a more appropriate study design are needed for further evaluating the diagnostic accuracy of RDW in sepsis.

A total number of four studies have investigated the prognostic value of RDW in extremely low birth neonatal weight (birth weight less than 1000 g) patients with early sepsis to the best of our knowledge. In the first of such investigations, RDW was measured in 46 newborns and its values were then compared between hospital survivors and non-survivors. No significant association was found between RDW and in-hospital mortality, and this has been possibly attributed to the small sample size and the ensuing low statistical power [32]. In a subsequent study including 500 neonatal sepsis cases, 190 newborns died within 30 days and RDW was found to be higher in non-survivors than in survivors (19.13% vs 15.56%, p < 0.01) [20]. The AUC of RDW for predicting 30-day mortality was 0.751 (95% CI: 0.707 – 0.795). In multivariable logistic regression analysis, including RDW, leukocytes, platelets and hemoglobin, RDW was found to be independently associated with 30-day mortality, displaying an OR of 1.317 (95% CI: 1.241 – 1.399) [20]. In the third prospective study, RDW was measured in 251 cases of neonatal sepsis and was then compared with matched healthy children [33]. Compared with the healthy pediatrics cohort, those with neonatal sepsis had significantly higher RDW (19.9% vs 18.9%, p < 0.001). Furthermore, the risk of 28-day mortality increased two times in subjects with RDW > 20% compared with patients with RDW < 20% (HR: 0.50; 95% CI: 0.28 – 0.89). Nevertheless, multivariable regression analysis is not available in this study, so that it cannot be excluded that some confounding factor may have interplayed with the observed association between RDW and early neonatal mortality.

More recently the diagnostic value of RDW was investigated for in early onset neonatal sepsis (EOS) cases [34]. A total number of 149 patients (67 with EOS and 92 healthy newborns) were studied. Notably, RDW was found to be higher in EOS than in the other neonates (19.2% vs 16.9%, p < 0.001). Interestingly, the calculation of RDW to platelet ratio (RPR) was also found to be a useful index for diagnosing neonatal sepsis, displaying an AUC of 0.786, although its diagnostic performance remained lower than that of C reactive protein (CRP) (AUC: 0.88) and PCT (AUC: 0.83).

To present, only one study has investigated the association between RDW and prognosis of sepsis [35]. A public accessible database called Medical Information Mart for Intensive Care III (MIMIC III) [36], [37] was used for extracting data. ICD-9 was used to definite sepsis according to the method proposed by Angus et al. [38]. A total number of 4264 sepsis patients were included and their RDW, cause of admission, comorbidity, SOFA, quick SOFA (qSOFA), simplified acute physiology score (SAPS II), oxford acute severity of illness score (OASIS), modified logistic organ dysfunction system (MLODS), systemic inflammatory response syndrome (SIRS), acute physiology and chronic health evaluation III (APS III) were extracted using the structured query language (SQL). It was finally found that increased RDW was a risk factor for 4-year all-cause mortality, displaying an HR of approximately 1.14 (95% CI: 1.12 – 1.16) per 1% increase in models adjusting for severity scores each time. RDW prognostic value was not significantly modified by correcting for anemia [35]. It was also found that RDW provided additional prognostic information beyond traditional scores.

Kim et al. prospectively enrolled 329 subjects with severe sepsis and septic shock for investigating the association between dynamic changes of RDW (on admission and 72 h after admission) and 28-day and 90-day mortality rates [19]. Subjects were categorized in four groups according to their baseline RDW value (normal or increased) and its dynamic changes (increased or not by more than 0.2% during the following 72 h). In multivariable Cox regression model both baseline RDW and its dynamic change > 0.2% were found to be independently associated with mortality. Patients with abnormal baseline RDW and increased dynamic changed RDW had approximately 10 times of mortality risk compared with patients with normal and decreased dynamic RDW (HR: 9.97; 95% CI: 1.99 – 49.91). In another study with a small sample size (n = 45), RDW was assessed in patients with septic shock aged 65 years or older [39]. Nineteen of the 45 subjects died during hospital stay and nonsurvivors had higher RDW levels at baseline (16.5% vs 16.0%, p = 0.550), at day 4 (16.7% vs 15.8%; p = 0.011) and 7 (17.7% vs 15.6%; p = 0.001) than survivors. Notably, the RDW values were also found to increase during hospital stay in nonsurvivors, whilst RDW was significantly decreased in survivors [39]. Nevertheless, these promising findings could not be replicated in another prospective and multicenter study, where the role of RDW measured at day 1, 4 and 8 was used for predicting 30-day mortality [40]. Overall, 297 severe sepsis patients were enrolled, 104 of whom died during hospitalization. RDW values at day 1 in nonsurvivors were found to be significantly higher than that in survivors (15.7% vs 14.7%; p = 0.001). However, RDW in survivors increased at day 4 and 8, while it remained relatively stable in nonsurvivors. The AUCs of RDW at day 1, 4 and 8 was found to be always around 0.63 (95% CI: 0.57 – 0.68). In multivariable regression analysis, RDW at day 1, 4 and 8 was independently associated with 30-day mortality, with HRs ranging between 1.10–1.13 per 1% increase (p < 0.05 for all).

In a retrospective study RDW was measured in 103 community-acquired intra-abdominal sepsis (C-IAS) patients, 52 of whom died during hospital stay. RDW value on admission was significantly higher in survivors than nonsurvivors (17.9% vs 15.2%; p = 0.009). The AUC of RDW for predicting hospital mortality was 0.867 (95% CI: 0.791 – 0.942). In multivariable Cox regression model, RDW was not found to be independently associated with in-hospital mortality (HR: 1.07; 95% CI: 0.92 – 1.25). Notably, RDW values decreased on the third and seventh day after admission in survivors, whilst it gradually increase in nonsurvivors [41].