Neutrophil‐to‐lymphocyte ratio at admission helps to predict the need for blood transfusion after vaginal delivery

Abstract

Objective

This study assesses the association between complete blood count (CBC) parameters, including the neutrophil‐to‐lymphocyte ratio (NLR) and the platelet‐to‐lymphocyte ratio (PLR) and predicts the need for postpartum packed red blood cell transfusion (pRBCT).

Methods

This retrospective cohort study was conducted at a tertiary, university‐affiliated medical center with approximately 12 500 annual deliveries (2012–2023). Women requiring postpartum pRBCT were identified based on criteria including severe hemorrhage, symptomatic anemia with hemoglobin (Hb) levels of 7–8 g/dL, or severe anemia (Hb <7 g/dL). Maternal demographics, admission complete blood count (CBC), and delivery outcomes were analyzed. Multivariable logistic regression identified independent predictors of pRBCT, and a risk score was developed and evaluated using receiver operating characteristic (ROC) analysis.

Results

Admission CBC data were available for 37 631 vaginal deliveries, of which 957 (2.5%) required pRBCT. Risk factors for pRBCT included nulliparity, previous cesarean, assisted reproductive technology conception and intrapartum fever. Protective factors included spontaneous labor onset, body mass index >30, and admission hematocrit >40%. Key CBC independent predictors included Hb <11 g/dL (adjusted odds ratio [aOR] 5.70, 95% confidence interval [CI] 4.79–6.79), and NLR >5 (aOR 1.28, 95% CI 1.02–1.60). The scoring model, with a clinical cutoff of 5, predicted pRBCT with an area under the curve of 0.77 (95% CI 0.75–0.79, P < 0.001).

Conclusion

Admission CBC parameters, particularly NLR and Hb, alongside maternal factors, might help predict pRBCT in vaginal deliveries.

1. INTRODUCTION

Blood component therapy, especially packed red blood cell transfusion (pRBCT), is essential in managing hemorrhage during or after labor. ¹ Delays in transfusion increase maternal morbidity and mortality. ² , ³ Although critical, preparing packed red blood cell (pRBC) units requires blood typing and antibody screening that might take up to an hour. ⁴

Predictive models for pRBCT after vaginal deliveries often incorporate clinical risk factors like twin pregnancy, ⁵ parity, ⁶ anemia, thrombocytopenia, ⁷ and hypertensive or placental complications, ⁸ , ⁹ but their accuracy and calibration remain inconsistent. ¹⁰

The neutrophil‐to‐lymphocyte ratio (NLR) and platelets lymphocytic ratio (PLR) have emerged as a valuable inflammatory marker associated with the need for transfusions in other conditions, such as gastrointestinal bleeding and trauma. ¹¹ , ¹² While obstetric studies have begun examining NLR in pregnancy‐related complications, ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ its role in predicting transfusion requirements for vaginal delivery patients remains largely unexamined, most models relying primarily on admission hemoglobin levels. ¹

We aimed to determine the association between CBC parameters, including NLR and PLR, and the need for pRBCT in vaginal deliveries and to develop a prediction model for this outcome.

2. MATERIALS AND METHODS

2.1. Study population

This retrospective cohort study included all women who delivered vaginally at ≥24 weeks of gestation at a single university‐affiliated tertiary medical center from 2012 to 2023.

If no recent CBC (within the past month) was available, CBC samples were collected upon admission to the delivery room. To ensure accurate data analysis, women with incomplete CBC data at admission were excluded. Additionally, women who underwent cesarean delivery, whether elective or urgent, were also excluded from the study. In this study, we selected pRBCT as our primary outcome instead of postpartum hemorrhage due to the very low accuracy of estimating blood loss, particularly in vaginal deliveries. ¹⁸

Participants were categorized based on the need for blood transfusion during the index hospitalization. In our department, the criteria for postpartum pRBCT included severe ongoing hemorrhage, symptomatic anemia with Hb of 7–8 g/dL, or postpartum Hb <7 g/dL. A comparative analysis of demographic and clinical characteristics was conducted between transfusion and non‐transfusion groups.

2.2. Data collection

Medical records were obtained from the computerized delivery room logbooks, providing a comprehensive set of demographic, obstetric, and clinical characteristics. These included maternal age, pre‐gestational body mass index (BMI), parity, gravidity, history of previous cesarean deliveries, gestational age at delivery, multiple pregnancies, and conception method. Additional data on pre‐existing or gestational conditions such as pre‐eclampsia and diabetes were also extracted.

Complete blood count parameters were collected from admission laboratory results. NLR was calculated as the absolute neutrophil count divided by the absolute lymphocyte count. PLR was calculated as the absolute platelet count divided by the absolute lymphocyte count. Leukocytosis was defined as a leukocyte count above 16 K/μL, anemia as Hb below 11 g/dL, ¹⁹ and thrombocytopenia as a platelet count below 100 K/μL. ²⁰

The study was approved by the Institutional Review Board (IRB No. TLV‐0284‐08, July 10, 2024).

2.3. Statistical analysis

Descriptive statistics were first used to summarize maternal characteristics, labor variables, and hematologic parameters to compare women who required pRBCT with those who did not. Normality of continuous variables was assessed using Q–Q plots and visual inspection of histograms. Univariate analyses identified differences between groups, with two‐tailed unpaired Student's t‐tests or Mann–Whitney tests (for non‐normally distributed variables) used to assess continuous variables. For categorical variables, χ ² or Fisher's exact tests were applied as appropriate. A multivariable logistic regression model, adjusting for statistically significant variables between groups, determined the impact of independent variables on transfusion needs. Model calibration was assessed using the Hosmer–Lemeshow goodness‐of‐fit test. To evaluate model robustness and assess potential collinearity among hematologic variables, we repeated the multivariable logistic regression after removing one predictor at a time. Model estimates and effect sizes were compared with the primary model to evaluate stability. Multicollinearity among predictors was assessed using variance inflation factors (VIF) derived from an ordinary least squares regression including all covariates entered in the multivariable logistic model. All predictors demonstrated VIF values <2, indicating no concerning multicollinearity.

A risk prediction score was then developed based on the β‐coefficients from the multivariable model. Predictors were assigned integer point values roughly proportional to the magnitude and direction of their β‐coefficients (with larger coefficients receiving more points and protective factors assigned negative or zero points), while prioritizing simplicity and bedside usability. Score performance was evaluated using receiver operating characteristic (ROC) curves. The area under the curve (AUC) was calculated with 95% confidence intervals (CIs), and the P‐value assessed the model's prognostic accuracy against the null hypothesis (AUC = 0.5). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated at the optimal cutoff identified using the Youden index.

Data analysis was performed using SPSS software (version 21.0; SPSS, Chicago, IL, USA). A P‐value <0.05 was considered statistically significant. All analyses used anonymized data; therefore, informed consent was waived.

3. RESULTS

During the study period, 147 439 women delivered in our center. Of these, 37 631 vaginal deliveries with available admission CBC data were included in the analysis, with 957 (2.5%) women requiring pRBCT (Figure 1).

FIGURE 1.

Study population. CBC, complete blood count; pRBC, packed red blood cell.

Demographic and obstetric characteristics are detailed in Table 1. Women in the pRBCT group were older (32.5 years; standard deviation [SD] 5.3) compared to those who did not receive pRBC (32.2 [SD 4.9] years, P = 0.023) and had a lower body mass index (BMI) (21.7 kg/m², interquartile range [IQR 19.8, 24.5 kg/m²] vs. 22.1 kg/m² [IQR 20.08, 24.8] kg/m², P = 0.022, respectively). The pRBCT group also had a higher prevalence of nulliparity, previous cesarean deliveries, multiple gestations, and in vitro fertilization (IVF) pregnancies. Additionally, deliveries that required pRBCT showed a lower prevalence of spontaneous labor onset (64.9% vs. 77.1%) and a higher prevalence of oxytocin use during labor (68.9% vs. 51%) (P < 0.001 for all).

TABLE 1.

Demographic and obstetrical characteristics for the study population.

	No pRBCT (N = 36 674)	pRBCT (N = 957)	P‐value
Maternal age (mean, SD)	32.2 (4.9)	32.5 (5.3)	0.023
Gestational week, median (IQR)	39.5 (38.6, 40.4)	39.6 (38.6, 40.5)	0.026
BMI, median (IQR)	22.1 (20.08, 24.8)	21.7 (19.8, 24.5)	0.022
Nulliparity, n (%)	17 569 (47.9)	654 (68.8)	<0.001
Previous CD, n (%)	1652 (4.5)	60 (6.3)	0.01
IVF pregnancy, n (%)	2288 (6.6)	148 (16.3)	<0.001
Spontaneous onset of delivery, n (%)	27 683 (77.1)	613 (64.9)	<0.001
Oxytocin use, n (%)	18 715 (51)	659 (68.9)	<0.001
Multiple pregnancy, n (%)	668 (1.8)	38 (4)	<0.001
GDM, n (%)	3834 (10.5)	107 (11.2)	0.45
PET, n (%)	2818 (1.9)	153 (8)	<0.001
Pre‐gestational DM, n (%)	189 (0.5)	4 (0.4)	1
Regional anesthesia, n (%)	27 945 (76.3)	740 (77.6)	0.335
Intra‐partum fever, n (%)	2137 (8.8)	102 (13.7)	<0.001
CBC parameters
HB <11 g/dL, n (%)	6138 (16.7)	530 (55.4)	<0.001
HCT >40%, n (%)	5590 (15.2)	48 (5.0)	<0.001
WBC >16 × 103/μL, n (%)	8788 (24.0)	382 (39.9)	<0.001
PLT <100 × 103/μL, n (%)	700 (1.9)	28 (2.9)	0.027
PLR >100, n (%)	21 366 (58.4)	565 (68.7)	<0.001
NLR >5, n (%)	19 858 (54.3)	685 (71.7)	<0.001

Abbreviations: BMI, body mass index; CD, cesarean delivery; DM, diabetes mellitus; GDM, gestational diabetes mellitus; Hb, hemoglobin; HCT, hematocrit; IQR, interquartile range; IVF, in vitro fertilization; NLR, neutrophilic‐lymphocytic ratio; PET, preeclampsia toxemia; PLR, platelets lymphocytic ratio; PLT, Platelet Count; pRBCT, packed red blood cell transfusion; PTD, preterm delivery; SD, standard deviation; WBC, white blood cell.

Regarding CBC parameters, women in the pRBCT group exhibited a higher prevalence of anemia, leukocytosis, and thrombocytopenia, alongside lower hematocrit levels. The NLR >5 and PLR >100 were both more prevalent in the pRBCT group compared to the non‐transfusion group (71.7% vs. 54.3% and 68.7% vs. 58.4%, respectively; P < 0.001) (Table 1).

In the multivariable logistic regression model (Table 2), controlling maternal age >35, PTD <34 weeks, multiple gestation, oxytocin use and thrombocytopenia, NLR >5 was identified as a significant risk factor associated with pRBCT (aOR 1.28, 95% CI 1.02–1.60). Other significant risk factors associated with the need for pRBCT included nulliparity (aOR 2.52, 95% CI 1.99–3.20), previous cesarean delivery (aOR 2.79, 95% CI 1.86–4.20), IVF pregnancy (aOR 2.11, 95% CI 1.67–2.66), and intrapartum fever (IPF) ≥38 °C (aOR 1.31, 95% CI 1.02–1.67). Maternal anemia (Hb <11 g/dL) remained the strongest predictor (aOR 5.70, 95% CI 4.79–6.79). Protective factors included spontaneous labor onset (aOR 0.70, 95% CI 0.59–0.85), BMI >30 (aOR 0.65, 95% CI 0.45–0.93), and hematocrit >40% (aOR 0.56, 95% CI 0.38–0.84). Notably, PLR >100 and leukocytosis were not significantly associated with pRBCT after full adjustment.

TABLE 2.

Multivariate analysis assessing the risk factors for pRBC.

	aOR (95% CI)	p‐value
Maternal age >35	1.14 (0.94–1.39)	0.19
BMI >30	0.65 (0.45–0.93)	0.019
Nulliparity	2.52 (1.99–3.20)	<0.001
PTD <34	0.75 (0.43–1.30)	0.302
Previous CD	2.79 (1.86–4.20)	<0.001
IVF pregnancy	2.11 (1.67–2.66)	<0.001
Spontaneous onset of delivery	0.70 (0.59–0.85)	<0.001
Oxytocin use	1.25 (0.99–1.58)	0.060
Multiple pregnancy	1.16 (0.69–1.94)	0.579
PET	1.46 (0.93–2.27)	0.097
Intrapartum fever	1.31 (1.02–1.67)	0.033
HB <11 g/dL	5.70 (4.79–6.79)	<0.001
HCT >40%	0.56 (0.38–0.84)	0.005
NLR >5	1.28 (1.02–1.60)	0.035
PLR >100	0.94 (0.77–1.14)	0.52
WBC >16 × 103/μL	1.14 (0.94–1.38)	0.17
PLT <100 × 103/μL	1.27 (0.75–2.14)	0.37

Abbreviations: aOR, adjusted odd ratio; BMI, body mass index; CD, cesarean delivery; CI, confidence interval; HB, hemoglobin; HCT, hematocrit; IUFD, intrauterine fetal diminish; NLR, neutrophilic‐lymphocytic ratio; PET, preeclampsia toxemia; PLR, platelets lymphocytic ratio; PLT, Platelet Count; pRBC, packed red blood cell; PTD, preterm delivery.

Model calibration was excellent, as demonstrated by a non‐significant Hosmer–Lemeshow goodness‐of‐fit test (χ ² = 8.89, degrees of freedom = 8, P = 0.35). No evidence of multicollinearity was identified. All variables included in the multivariable model had VIF values <2 (range 1.01–1.57).

To evaluate the robustness of the multivariable model and the independence of hematologic predictors, we conducted a series of sensitivity analyses, each excluding one variable at a time (white blood cell [WBC], NLR, IPF, Hb, Hct, and PLR). The model remained highly stable across all analyses.

In our model that included intrapartum fever, WBC was no longer statistically significant, suggesting that fever partially attenuates the inflammatory signal captured by leukocytosis. However, when NLR was removed from the model, WBC regained statistical significance (aOR 1.23, 95% CI 1.03–1.47), and when fever was removed, WBC again became independently associated with pRBCT (aOR 1.24, 95% CI 1.05–1.47). Conversely, NLR remained independently associated even after exclusion of WBC (aOR 1.35, 95% CI 1.09–1.67), indicating that NLR provides additional, non‐overlapping inflammatory information beyond leukocytosis and fever.

Excluding Hb or Hct resulted in the expected reciprocal strengthening of the complementary erythrocyte parameter. When Hb was removed from the model, Hct became markedly stronger (aOR 0.31, 95% CI 0.21–0.45), whereas all other predictors remained stable. Conversely, when Hct was excluded, Hb became even more dominant (aOR 6.14, 95% CI 5.18–7.28), again without materially affecting the regression coefficients of the remaining variables.

A prediction score model was constructed based on the multivariable logistic regression analysis to quantify the risk of pRBCT among women undergoing vaginal delivery (Table 3). Using a clinically oriented threshold of ≥5 points, the score identified women at increased risk with a sensitivity of 60.3%, specificity of 80.5%, PPV of 8.7%, and NPV of 98.5%.

TABLE 3.

Risk prediction score for pRBC transfusion.

Maternal age >35
Yes	1
No	0
BMI >30
Yes	0
No	1
Nulliparity
Yes	2
No	0
Previous CD
Yes	3
No	0
IVF pregnancy
Yes	2
No	0
Spontaneous onset of delivery
Yes	0
No	1
Intra‐partum fever
Yes	1
No	0
Hb <11 g/dL
Yes	6
No	0
HCT >40%
Yes	0
No	2
NLR >5
Yes	1
No	0

Abbreviations: BMI, body mass index; CD, cesarean delivery; Hb, hemoglobin; HCT, hematocrit; IVF, invitro fertilization; NLR, neutrophilic‐lymphocytic ratio; pRBC, packed red blood cell.

The ROC curve (Figure 2) demonstrated good discriminatory performance, with an AUC of 0.77 (95% CI 0.75–0.79, P < 0.001). The optimal cutoff according to Youden's index corresponded to a threshold of 3.5, offering the best balance between sensitivity (66.2%) and specificity (75.5%). However, for clinical applicability and ease of bedside implementation, we selected the ≥5 threshold as the primary operating point, given its substantially higher NPV and its utility as a rule‐out tool for identifying women at very low risk of requiring transfusion.

FIGURE 2.

Receiver operating characteristic (ROC) curve.

4. DISCUSSION

4.1. Principal findings

The main findings of our study include:

1. Approximately 2% of women in our cohort required pRBCT.

2. NLR >5 was associated with an increased likelihood of transfusion, alongside other factors such as nulliparity, previous cesarean section, conception via IVF, intrapartum fever, and anemia. Conversely, protective factors included spontaneous labor onset, a BMI greater than 30, and a hematocrit level above 40.

3. PLR was not identified as a significant risk factor in the multivariate analysis when controlling for other confounders.

4. Our predictive model, which incorporated these 10 independent risk factors and employed a cutoff of 5, demonstrated the best clinical performance in screening and predicted the need for pRBCT during vaginal deliveries.

4.2. Findings in the context of other observations

Blood transfusion is a life‐saving therapy for bleeding patients, particularly in obstetric populations, with rates reaching up to 25% in developing countries. ²¹ The prevalence of transfusion in vaginal deliveries varies across studies and populations, ⁸ generally ranging from 0.86% to 2.4%. ²² , ²³ Our study aligns with this, demonstrating a transfusion prevalence of 2.5%.

Consistent with prior findings, this study identified several risk factors associated with an increased likelihood of receiving a blood transfusion therapy in the obstetric population, including nulliparity, ²⁴ previous cesarean section, ²⁵ and conception via IVF. ²⁶ Additionally, spontaneous onset of labor was identified as a protective factor against pRBC transfusion, ²⁷ aligning with our findings. Although obesity is recognized as a risk factor for antepartum and postpartum hemorrhage, ²⁸ our study found it to be a protective factor against pRBCT. This finding aligns with other studies, ²⁹ which suggest that individuals with obesity might have a higher estimated blood volume, allowing them to tolerate greater blood loss before requiring a transfusion. ³⁰

The primary aim of our study was to investigate predictive risk factors for pRBCT based on admission CBC parameters. We identified that an NLR above 5 and anemia are significant independent risk factors for requiring pRBCT, while a hematocrit level above 40 serves as a protective factor.

Due to physiological changes during pregnancy, third‐trimester anemia is defined as a hemoglobin level below 11 g/dL. ¹⁹ Our findings align with the risk assessment for transfusions outlined by Xing et al. ⁹ and support a recent study on patient blood management in postpartum hemorrhage, emphasizing the significance of hemoglobin levels. ³¹ Additionally, among women who delivered vaginally and received pRBCT, hematocrit levels were found to be lower, ³² compared to the control group, which is consistent with our study's findings.

Infection is a well‐known risk factor for blood transfusion. Current evidence supports an association between intrapartum fever and increased risk of blood cell transfusion following vaginal delivery, primarily mediated by higher rates of postpartum hemorrhage and maternal infection, although direct quantification of this relationship remains limited. ³³ A study by Linström et al. evaluated the hematological profiles and transfusion history of pregnant mothers presenting to a tertiary hospital and found that leukocytosis significantly increased the likelihood of transfusion, with an odds ratio of 2.35. ³⁴ This finding corresponds with our study and might be attributed to an increased risk of peripartum infections, which can necessitate blood transfusions due to complications such as sepsis and hemorrhage.

A key novel finding of our study is the independent association between elevated NLR and the risk of pRBCT in women who delivered vaginally. NLR, a marker of systemic inflammation, has been linked to adverse outcomes in various medical conditions and has also been associated with obstetric complications, including preeclampsia, gestational diabetes, and preterm labor. ³⁵ , ³⁶ , ³⁷ In bleeding patients, elevated NLR has been associated with increased in‐hospital mortality in trauma patients requiring massive transfusions. ¹² Additionally, in cases of upper gastrointestinal bleeding, combining NLR with other scoring systems has improved predictive accuracy for transfusion requirements. ¹¹ To our knowledge, our findings are the first to establish an association between pRBCT and NLR in obstetric patients.

4.3. Clinical implications

Our findings demonstrate that hematologic parameters available at admission, particularly NLR, hemoglobin, and hematocrit, provide clinically meaningful information for anticipating the need for pRBCT in women undergoing vaginal delivery. In an era of increasing maternal age, comorbidities, and obstetric complexity, ³⁸ early identification of women at elevated transfusion risk is essential for optimizing intrapartum preparedness and resource allocation. Integrating these laboratory markers into pre‐delivery risk assessments might assist clinicians in stratifying patients more accurately. For example, a nulliparous woman with anemia and an NLR 5 on admission would be classified as high risk in our model. Recognizing such a profile in real time allows the clinical team to ensure appropriate blood product availability, anticipate potential hemodynamic instability, and provide more informed counseling.

Importantly, the scoring system demonstrated a very high negative predictive value (98.5%), enabling confident identification of women at low likelihood of requiring pRBCT. This has meaningful clinical and organizational implications, supporting efficient use of resources, minimizing unnecessary cross‐matching, and enhancing overall delivery unit readiness.

Overall, our model underscores the value of combining inflammatory and erythrocyte indices with readily available clinical variables to improve anticipatory management and maternal safety.

4.4. Strengths and limitations

Our study has several notable strengths. Our study offers valuable insights into predicting the need for postpartum pRBCT in vaginal deliveries. By analyzing the largest sample size to date, we provide a comprehensive evaluation of key risk factors, with a particular focus on the novel inflammatory marker, NLR. Through both univariate and multivariate analyses, we identified significant predictors of pRBCT, enhancing the robustness of our predictive model. This study's findings contribute new, clinically relevant data that can improve risk assessment and inform decision‐making in obstetric care.

However, as a retrospective cohort study, it has inherent limitations. Our ability to control all potential confounding factors was restricted; for example, data on prior postpartum hemorrhage, a known risk factor for pRBCT in subsequent deliveries, was not included, as noted in the methods. Moreover, this study covers a long time span (2012–2023), which includes periods of epidemiologic variation such as the COVID‐19 pandemic. Although COVID‐19 status was not consistently available, the large sample size and the fact that most deliveries occurred outside the pandemic years make it unlikely that this materially affected the overall results. Further, reliance on computerized chart records introduces the risk of information bias, which could affect the integrity of our findings. To analyze current data, we included only women with CBC results from delivery room admission, which might introduce selection bias. This limitation might hinder our ability to provide comprehensive decision‐making tools for predicting pRBCT in vaginal deliveries. Finally, our prediction model was derived from a single cohort and was not externally validated, which might limit its generalizability. Future studies using independent populations are required to validate and refine the score.

5. CONCLUSION

In conclusion, our study highlights the significant role of CBC parameters, including the NLR, in predicting the need for pRBCT during vaginal deliveries. The findings suggest that nulliparity, previous cesarean section, conception via IVF, IPF, and anemia are critical risk factors for increased transfusion likelihood. Our study provides a promising predictive model with strong accuracy, offering a valuable tool for clinicians to identify women at higher risk for postpartum pRBCT in vaginal deliveries. However, it is important to note that the clinical utility of this model should be validated through a prospective study to confirm its effectiveness in real‐world settings.

AUTHOR CONTRIBUTIONS

Daniel Gabbai and Emmanuel Attali contributed equally, conducted the literature search, and drafted the manuscript. Yariv Yogev, Anat Lavie, and Itamar Gilboa helped conceptualize the study design and edited and revised the manuscript. All authors revised the article for important intellectual content and approved the final version submitted for publication.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare.

ETHICS STATEMENT

The trial was conducted in accordance with the Declaration of Helsinki (2000) for human studies (IRB number TLV‐0284‐08).

PAPER PRESENTATION INFORMATION

The paper will be presented as a poster at the SMFM Annual Meeting in January 2025.

DATA AVAILABILITY STATEMENT

Research data are not shared.