Red cell distribution width (RDW) is an automatically measured complete blood count (CBC) parameter that reflects the degree of heterogeneity in erythrocyte size (4). The RDW is calculated from the distribution of red blood cell (RBC) volumes, measured by impedance, using the following formula:RDW=standarddeviationofRBCvolume/meancorpuscularvolume(MCV)×100

The “width” in RDW refers to the width of the volume distribution histogram, expressed as a coefficient of variation. In samples with mixtures of small and large RBCs, the histogram is wider; thus, the RDW value is higher. In contrast, in samples with a homogeneous red cell size, the histogram is narrower; thus, the RDW is lower. The normal reference range for RDW among healthy adults used in most laboratories is 11% (lower reference interval limit) to 15% (upper reference interval limit) (5). In 1987, Dr. Robert Novak reported that the normal range of RDW in infants ranged from 11.5% to 14.5% (6). A recent study of 144 infants reported that the lower reference limit for RDW in premature infants at <72 h of age is 14.2%, while the upper reference limit is 21.8% (4). However, due both to the difference in methods of measuring RBC size, instruments, researchers, laboratory standards, and statistical approaches between laboratories, as well as the ethical and practical challenges unique to pediatric care, thus far, no universal reference range has yet been established for pediatric patients (7).

Recently, RDW levels have been shown to be associated with neonatal mortality and/or significant morbidity, including neonatal sepsis, intrauterine growth restriction (IUGR), and bronchopulmonary dysplasia (BPD), etc., (8–10). The RDW has also been indicated as a useful tool for evaluating the medical conditions of newborns, especially preterm infants (11, 12).

Regional cerebral tissue oxygen saturation (rcSO₂) can be measured by using near-infrared spectroscopy (NIRS), a frequently used noninvasive technique (13). RcSO₂ primarily reflects the brain venous compartment, as it is larger than other compartments; however, it is also sensitive to changes in arterial parameters. RcSO₂ increases with time after birth, and tends to reach a plateau at 8–10 min after birth (14, 15). However, great individual variability can be seen among patients, and depending on the location. A prior publication provided an rcSO₂ range in preterm infants of 55%–85% (16). An rScO₂ value of 55% is close to or below the thresholds reported to be associated with neuronal damage and adverse neurodevelopmental outcome (17).

Conventional pulse oximeter measures arterial oxygen saturation around the limb, meaning that pulse oxygen saturation (SpO₂) does not include information on local tissue oxygen utilization (18). Cerebral oxygenation is influenced by three main components: the blood oxygen content, cerebral perfusion, and oxygen consumption. Each of these factors are themselves influenced by several other variables, including SpO₂, blood glucose level, partial pressure of carbon dioxide, blood pressure, and hemoglobin level (19). NIRS measures composite arterial, capillary, and venous blood saturation in the brain tissue; as such, rcSO₂ can reflect real-time subtle changes in tissue oxygenation in the brain (20). Studies have found that the neonates with adverse outcome (including intraventricular haemorrhage) had significantly lower rcSO₂ values during the immediate transition, although there was no difference concerning SpO₂ (21, 22). Therefore, potential cerebral hypoxia could not be identified with SpO₂ monitoring. Although in a research involving extremely preterm infants, treatment guided by cerebral oximetry monitoring in the first 72 h after birth did not lead to a lower incidence of death or severe brain injury at 36 weeks' postmenstrual age than usual care (23), a previously conducted interventional trial performed by the SafeBoosCII study group demonstrated that the combination of cerebral oxygenation monitoring by NIRS combined with an evidence-based treatment guideline significantly reduced the burden of cerebral hypoxia (24, 25). Other studies have reported high sensitivity of NIRS-detected variables of rcSO₂ to hypoxemic and ischemic events occurring in preterm infants (26).

Due to their ability to deform and flow in the microvascular network, RBCs have many vital physiological functions in our body, including carrying oxygen and carbon dioxide and gas exchange between blood and tissues (27). However, Patel et al. found that when the RDW exceeds 14% in microvascular disorders, the deformability of RBCs decreases (28), which may affect these aforementioned functions. In newborns, ineffective erythropoiesis due to prematurity (4), hemolysis, erythrocyte transfusions, and alterations in osmolality under some pathophysiological conditions can all decrease the ability of RBCs to deform, triggering an increase in RDW, thereby resulting in low microvascular perfusion (29, 30). We hypothesized that a high RDW value would be associated with an increased risk of low rcSO₂ based on analyses of existing clinical data.

This study was therefore conducted to investigate the relationship between RDW and rcSO₂ in preterm neonates.

The study population consisted of 110 preterm infants born at <32 weeks or with birth weight (BW) <1,500 g between January and June 2022 at the neonatal intensive care unit (NICU) of the West China Second University Hospital, Sichuan University. The exclusion criteria were as follows: advanced resuscitation in the delivery room, cardiovascular instability requiring treatment with inotropes, anemia, and major congenital or chromosomal anomalies. This study was approved by the Ethics Committee of the West China Second University Hospital, Sichuan University (2022–67). Written and verbal informed consent was obtained from the infants' parents, and the study was registered on ChineseClinicalTrials.gov (09/04/2022, ChiCTR2200058482).

This study had a cross-sectional design. An EGOS-600B Cerebral/Somatic Oximeter device was used for rcSO₂ monitoring during the study period, with neonatal NIRS sensors (NIRSensor, Enginmed, A1Z2104110037) placed on the frontoparietal regions of the infants. NIRS monitoring was performed thrice daily for the first 14 days after birth. All data were obtained when the infants were mostly quiet or sleeping in the supine position. No handling or intervention was performed during NIRS monitoring. Recorded values were stored and used for offline analyses. NIRS data analysis was performed after the data were transferred to a computer in Excel format using a customized programmer for the EGOS device. Those with markers likely related to an artifact (during handling, positioning, etc.) were excluded for data cleaning and were averaged afterward.

Blood samples were extracted from all preterm infants once a week on the first and second week of life. The volume for blood sampling was 0.50 ml. RDW values were extracted from the CBC, and the average was calculated. Neonatal and perinatal characteristics, including gestational age (GA), BW, sex, mode of delivery, IUGR, hypertension during pregnancy, chorioamnionitis (CA), gestational diabetes mellitus (GDM), premature rupture of membranes (PROM), multiple pregnancies, overall dose of antenatal corticosteroid treatment, invasive mechanical ventilation, breast feeding, and blood transfusion, were recorded prospectively. Neonatal morbidities included early-onset sepsis (EOS), late-onset sepsis (LOS), and intraventricular hemorrhage (IVH).

According to the existing literature and clinical judgment, we divided the study population into two groups, with a mean rcSO₂ value over the first 14 days of 55% as the cut-off limit. For descriptive analyses, continuous variables are described as the means±standard deviations (SD), and categorical data are presented as numbers and percentages (31). The distribution of covariates in the rcSO₂≥55% and rcSO₂<55% groups was compared using the t-test (normal distribution) or Kruskal–Wallis rank sum test (non-normal distribution) for continuous variables, and the chi-squared test for categorical data (Table 1).

We used a generalized additive model (GAM) to investigate the dose–response relationship between RDW values and rcSO₂ (Figure 1), and subsequently used a two-piece-wise linear regression model to examine the threshold effect of the RDW value on rcSO₂ (Table 2). The threshold level (i.e., inflection point) was determined using a trial-and-error approach. This involved selecting inflection points along a predefined interval, and choosing the inflection point that resulted in the maximum model likelihood. We further conducted a log-likelihood ratio test to compare the one-line linear regression model with the two-piecewise linear model, as previously described (32).

Next, univariate logistic regression (Table 3) and multivariate logistic regression models (Table 4) were applied to examine whether a high RDW was associated with low rcSO₂ levels in preterm neonates. The statistical results are displayed as odds ratio (OR), with the corresponding 95% CI. We adjusted for variables deemed to be confounders according to the existing literature and clinical judgment. All statistical analyses were performed using EmpowerStats (www.empowerstats.com, X&Y solutions, Inc., Boston, MA) and R version 4.2.0 (http://www.r-project.org).

This study enrolled 110 preterm neonates, of whom 53 and 57 had rcSO₂ levels of ≥55% and <55%, respectively. The general characteristics of the study population are summarized in Table 1. There was no noticeable difference between the two groups in terms of GA (31 ± 2 vs. 31 ± 2 weeks; p = 0.273), BW (1524 ± 384 vs. 1432 ± 463 g; p = 0.265), female (47% vs. 40%; p = 0.471), VD (26% vs. 35%; p = 0.325), CA (43% vs. 46%; p = 0.815), hypertension during pregnancy (17% vs. 19%; p = 0.753), GDM (28% vs. 33%; p = 0.568), IUGR (11% vs. 10%; p = 0.894), PROM (36% vs. 37%; p = 0.914), multiple pregnancies (40% vs. 46%; p = 0.526), overall dose of antenatal corticosteroid treatment (75% vs. 72%; p = 0.717), invasive mechanical ventilation (42% vs. 44%; p = 0.803), breast feeding (36% vs. 32%; p = 0.636), blood transfusion (25% vs. 40%; p = 0.077), IVH (13% vs. 19%; p = 0.388), EOS (8% vs. 14%; p = 0.275), or LOS (8% vs. 11%; p = 0.587).

To investigate the association between RDW values and rcSO₂, the participants were divided into two groups according to the RDW value. The piece-wise linear regression threshold effect analysis of RDW and rcSO₂ showed that the rcSO₂ decreased with RDW values >18% (β, −0.03; 95% CI, −0.04– −0.02; p < 0.0001) (Figure 1, Table 2). Univariate analysis (Table 3) revealed that the RDW was significantly correlated with preterm rcSO₂ values <55% over the first 14 days (OR, 1.31; 95% CI, 1.04–1.65; p = 0.020). After multivariable risk adjustment for potential confounding factors (Table 4), including GA and BW in the adjusted model, a high RDW was still positively associated with low rcSO₂ in preterm neonates. In addition, an RDW of ≥18% over the first 14 days could significantly increase the risk of preterm rcSO₂ values <55% over the first 14 days (adjusted model: OR, 3.31; 95% CI, 1.28–8.53, p = 0.013).