Role of hemogram-derived ratios in predicting intensive care unit admission in COVID-19 patients: a multicenter study.

Purpose: As hyperinflammation is recognized as a driver of severe COVID-19 disease, checking markers of inflammation is gaining more attention. Our study aimed to evaluate the utility of cost-effective hemogram-derived ratios in predicting intensive care unit (ICU) admission in COVID-19 patients.
Methods: This multicenter retrospective study included hospitalized COVID-19 patients from four dedicated COVID-19 hospitals in Sylhet, Bangladesh. Data on demographics, clinical characteristics, laboratory parameters and survival outcomes were analyzed. Logistic regression analysis was used to identify the significance of each hemogram-derived ratio in predicting ICU admission.
Results: Of 442 included patients, 98 (22.17%) required ICU admission. At the time of admission, patients requiring ICU had a higher neutrophil count and lower lymphocyte and platelet counts than patients not requiring ICU. Peripheral capillary oxygen saturation at admission was significantly lower in those who subsequently required ICU admission. Neutrophil-to-lymphocyte ratio, derived neutrophil-to-lymphocyte ratio, neutrophil-to-platelet ratio, and systemic immune-inflammation index were significant predictors of ICU admission.
Conclusion: Hemogram-derived ratios can be an effective tool in facilitating the early categorization of at-risk patients, enabling timely measures to be taken early in the disease course.

Introduction

The ongoing COVID-19 pandemic poses a major global threat to population health and places a huge strain on the health care delivery system worldwide (Legido-Quigley et al., 2020). Even before the COVID-19 pandemic, health care systems in low-and middle-income countries faced considerable challenges in providing high-quality, affordable and universally accessible care (Agampodi, T. et al., 2015; McGregor, S. et al., 2014). The pandemic has challenged the already weak health systems in these countries (Okereke, M. et al., 2021). Epidemiological studies have shown that the majority of COVID-19 infected patients (>80%) are asymptomatic or have mild symptoms, whereas approximately 14% of infected patients have severe disease and need to be hospitalized (Wu, Z. and McGoogan, J., 2020; Guan, W. et al., 2020). Depending on ethnicity and geographical area, intensive care unit (ICU) admission rates vary between 3.1% and 26%, and the mortality of patients admitted to ICU ranges from 5.8% to 41.6% (Gottlieb, M. et al., 2020; Chang, R. et al., 2021; Zhou, F. et al., 2020; Armstrong, R. et al., 2021; Ali, H. et al., 2021).

While COVID-19 can directly damage epithelial tissues through epithelial cell injury and necrosis, evidence indicates that immune system activation/perturbation is the major cause of organ/tissue damage (Xu, Z. et al., 2020). The activation of multiple complement pathways, dysregulated neutrophil responses, endothelial injury, and hypercoagulability appear to be interlinked with SARS-CoV-2 infection and drive disease severity (Java, A. et al., 2020). It is also clear that hyperinflammation and coagulopathy contribute to disease severity and death (Merad, M. and Martin, J., 2020).

High levels of inflammatory markers, including C-reactive protein (CRP), ferritin and D-dimer and increased levels of inflammatory cytokines and chemokines (Herold, T. et al., 2020; Zhang, X. et al., 2020; Mehta, P. et al., 2020) have been observed in patients with severe COVID-19 diseases. White blood cells, neutrophils, lymphocytes and monocytes are directly involved in this systemic inflammatory response, while platelets are the primary mediators of hemostasis. Neutrophils constitute the majority of the leukocytes and are primarily responsible for activating the immune system by migrating from the venous system. Free oxygen radicals that can damage the cell's nuclear material are thereby released (Kral, J. B. et al., 2016; Koupenova, M. et al., 2018). The rapid spread and potential lethality of COVID-19 has generated an urgent need to identify indicators that could be used to predict disease severity and risk associated with infection. Such indicators can help identify patients at high risk of developing severe disease, allowing for better allocation of limited human and technical resources, preventing unnecessary hospitalization and mitigating other impacts.

Biomarkers of inflammation derived from the peripheral blood, such as white blood cell (WBC) count, neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), and serum CRP levels have been investigated as independent predictors for prognosis of systematic inflammatory diseases (Guthrie, G. et al., 2013; Hu, H. et al., 2016). They have also been widely investigated in several other conditions such as malignancies (including hematological malignancies) and respiratory, gastrointestinal, cardiovascular (including acute coronary syndrome and intracerebral hemorrhage) and systemic diseases. Several studies have tested these biomarkers as a marker of disease severity and prognosis in COVID-19 (Seyit, M. et al., 2021; Liu, J. et al., 2020, Jimeno, S. et al., 2020; Erdogan, A. et al., 2021). As these biomarkers are part of routinely evaluated blood tests and are inexpensive, easily measurable and widely available, they have potential (particularly in low-resource countries) as cost-effective predictors of progression to a severe disease requiring ICU admission. Our study aimed to investigate the role of different hemogram-derived ratios in predicting subsequent ICU admission.

Methods

We conducted a cross-sectional study with patients selected from COVID-19 patients visiting four hospitals in Sylhet City, Bangladesh, from October 2020 to January 2021. COVID-19 diagnosis was carried out by a specialist based on a polymerase chain reaction test, computed tomographic scan or suggestive clinical features. Data on clinical characteristics, results of laboratory tests and clinical outcomes of the enrolled patients were collected from hospital records. The inclusion criteria were as follows: age ≥18 years and a diagnosis of COVID-19 requiring hospitalization. The exclusion criteria were: age <18 years, pregnancy and lack of required data. The final study included 442 patients.

Enrolled patients were assessed at the emergency department, where a blood sample was drawn. Laboratory assessments consisted of complete blood count (including WBC count, leukocyte subtypes, hemoglobin count and platelet count) and biochemical parameters (random blood sugar (RBS), serum ferritin (S. ferritin), D-dimer).

Definitions

WBC count (× 109 cells/L), neutrophil (× 109 cells/L), lymphocytes (× 109 cells/L) and platelets (× 1011 cells/L) were used to define the hemogram-derived ratios. NLR is the ratio between neutrophil and lymphocytes; d-NLR is derived NLR and calculated as d-NLR = ANC/(WBC-ANC), where ANC is the absolute neutrophil count. NPR is the ratio between neutrophil and platelets, PLR is the ratio between platelets and lymphocytes, and finally, systemic immune-inflammation index (SII) is defined as neutrophil multiplied by platelets and divided by lymphocytes.

Study variables

The outcome variable was ICU admission (Yes or No). Clinical data included age; sex; clinical features; presence of comorbidities such as hypertension, chronic kidney disease (CKD), chronic obstructive pulmonary disease (COPD), diabetes mellitus (DM), ischemic heart disease (IHD) and cerebrovascular accident (CVA); peripheral capillary oxygen saturation (SpO2) at admission; and length of hospital stay (in days). Laboratory parameters included complete blood count, D-dimer, S. ferritin and RBS. The radiographic findings included chest CT scan reports.

Statistical analysis

We used descriptive statistics to describe the data. Shapiro-Wilk test was used to assess the normality of continuous variables. We presented continuous measurements by mean and SD for data that followed a normal distribution and by the median and interquartile range for data that were skewed. The mean difference between the two groups (ICU vs non-ICU) in a continuous variable was assessed using the two-sample t-test for the normally distributed data and the non-parametric Mann-Whitney U test for the non-normally distributed data. Categorical variables were presented using frequencies and percentages (%). The Chi-Square test (χ2 test) of independence was used to determine the association (difference) among categorical variables. Differences in the hemogram-derived ratios due to comorbidities were investigated using the multivariate analysis of variance (MANOVA) test.

Multiple logistic regression models were used to identify the predictors of ICU admission. The candidate predictors for the adjusted model were selected based on clinical relevance. Initially, simple logistic regression models were fitted for each candidate predictor. Variables that were highly correlated or associated were excluded from the model due to multicollinearity. Model A included age, DM, CKD and COPD. Models B–D included the previous model and RBS, D-dimer, S. ferritin and admission SpO2. Each hemogram-derived ratio was added separately to each model and their significance tested. Model findings were presented using odds ratio (OR) and 95% CI. A P-value of <0.05 was considered statistically significant. We used the receiver operating characteristic (ROC) curve to detect optimal cut-off values of the hemogram-derived ratios in predicting ICU admission. The effects of comorbidities on the hemogram-derived ratios and ICU admission were determined using the multiple linear regression model and multiple logistic regression model, respectively. Data analysis was performed using R software. The study is reported following the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) (von Elm, E. et al., 2008) statements.

Results

Demographics and baseline characteristics of patients

The final analysis included 442 patients. The clinical characteristics of patients are summarized (overall and by ICU admission status) in Table 1. ICU admission was required in 98 (22.17%) patients. Of the total study patients, 55 (12.44%) died in hospital; 4 (1.2%) were from the non-ICU group . The mean age of study patients was 60.2 ± 13.7 years. The mean age in the ICU group was higher than in the non-ICU group (65.3 ± 14.9 vs 58.8 ± 13.1 years). Men comprised two-thirds of the study sample (65.8% vs 34.2%). Patients requiring ICU had significantly lower SpO2 at admission than those not requiring ICU (82.7± 11.4 vs 92.0 ± 7.0; P=<.001). The length of hospital stay was higher in ICU patients (P=<.001).

Table 1

Clinical characteristics of all patients, overall and by survivor status.

					Univariable analysis
Variables	Total	Non-ICU	ICU	p value	OR (95% CI)	p value
Age	60.2±13.7	58.8 ±13.1	65.3 ±14.9	<.001	1.04 (1.02-1.05)	0.001
Sex		113 (32.8%)	38 (38.8%)	0.332
Male	291 (65.8%)	231 (67.2%)	60 (61.2%)		0.77(0.48-1.23)	0.276
Female	151 (34.2%)	113 (32.8%)	38 (38.8%)
Co-morbidities
Hypertension	311 (70.4%)	241 (70.1%)	70(71.4%)	0.891	1.06(0.66-1.77)	0.793
DM	281 (63.6%)	212 (61.6%)	69(70.4%)	0.14	1.48(0.92-2.43)	0.112
CKD	78 (17.6%)	53 (15.4%)	25(25.5%)	0.03	1.88(1.08-3.20)	0.022
COPD	54 (12.2%)	33 (9.6%)	21(21.4%)	0.003	2.57(1.39-4.66)	0.002
IHD	98 (22.2%)	57 (16.6%)	41(41.8%)	<.001	3.62(2.21-5.92)	0.001
CVA	20 (4.5%)	17 (4.9%)	3 (3.1%)	0.607	0.61(0.14-1.85)	0.434
Admission SpO2	89.9 ±9.0	92.0 ±7.0	82.7±11.4	<.001	0.88 (0.85-0.91)	0.001
LOS	8.7 ± 4.5	7.8 ± 3.2	11.5 ± 6.6	<.001	1.19 (1.13-1.26)	0.001
In-hospital mortality	55(12.44%)	4 (1.2%)	51 (52%)	<.001	NA	NA

Abbreviations: ICU, Intensive Care Unit; DM, Diabetes Mellitus; CKD, Chronic kidney disease; COPD, Chronic obstructive pulmonary disease; IHD, Ischemic heart disease; CVA, Cerebrovascular disease; BP, Blood pressure; SpO2, peripheral capillary oxygen saturation.

Laboratory results are shown in Table 2. In those requiring ICU admission, median WBC and neutrophil count were significantly higher (10.45 vs 7.8 × 109/L; P=<.001 and 8.86 vs 5.7 × 109/L; P=<.001, respectively), while lymphocyte and platelet count were lower (1.16 vs 1.48 × 109/L; P=0.006 and 220 vs 230 × 109/L; P=0.449, respectively). Leukocytosis and lymphocytopenia were more prevalent in ICU patients (52% vs 29.4%; P=<.001 and 74.5% vs 48.8%; P=<.001, respectively). Compared with patients not requiring ICU, the median values of D-dimer (895 vs 505; P=0.005), S. ferritin (466 vs 325; P=0.018) and RBS (12 vs 9.2; P=0.003) were significantly higher in ICU patients.

Table 2

Lab findings on admission.

					Univariable analysis
Variables	Total	Non-ICU	ICU	p-value	OR (95% CI)	p-value
TC WBC (× 109/L)	4-10	7.8 (6-11)	10.45(6.80-14.38)	<.001	3.13 (1.89-5.18)	0.001
High	152 (34.4%)	101 (29.4%)	51 (52%)	<.001	2.61 (1.65-4.14)	0.001
Normal	279 (63.1%)	233 (67.7%)	46 (46.9%)	<.001	0.42 (0.27-0.66)	0.0002
Low	11 (2.5%)	10 (2.9%)	1 (1%)	0.49	0.34 (0.02-1.83)	0.312
Neutrophil (× 109/L)	2.0-7.0	5.7(3.89-8.59)	8.86(5.20-13.43)	<.001	3.10 (2.01-4.76)	0.001
Lymphocyte (× 109/L)	0.8–4.5	1.48 (1.08-2.07)	1.16(0.748-1.76)	0.006	0.48 (0.32-0.71)	0.001
Low	241 (54.5%)	168 (48.8%)	73 (74.5%)	<.001	3.06 (1.87-5.12)	0.001
Normal	201 (45.5%)	176 (51.2%)	25 (25.5%)	<.001	0.32 (0.19-0.53)	0.001
High	4 (0.9%)	3 (0.9%)	1 (1%)	1	1.17 (0.05 -9.26)	0.891
Platelet (× 109/L)	150-350	230 (180-300)	220(175-285)	0.449	0.72 (0.42 -1.25)	0.248
Low	49(11.1%)	35 (10.2%)	14 (14.3%)	0.336	1.47 (0.73-2.80)	0.255
Normal	320 (72.4%)	252 (73.3%)	68 (69.4%)	0.53	0.83 (0.51-1.36)	0.45
High	73 (16.5%)	57 (16.6%)	16 (16.3%)	1	0.98 (0.52-1.76)	0.954
D-dimer (ng/L)	0-500	505 (278-1100)	895(502-2209)	0.005	1.57 (1.27-1.93)	0.001
S. Ferritin	20-300	325 (164- 695)	466(192-981)	0.018	1.25 (1.02-1.54)	0.02
RBS	4.4-7.2	9.2 (7.4-12.7)	12(8.4- 15.2)	0.003	2.36 (1.35-4.12)	0.002

Abbreviations: TC WBC, total count of white blood cells; RBS, Random blood sugar.

The difference in hemogram-derived ratios between the ICU and the non-ICU group is shown in Table 3. The median values of NLR (7.08 vs 3.85; P=<.001), d-NLR (5.25 vs 3.16; P=<.001), NPR (3.52 vs 2.4; P=<.001), PLR (1.98 vs 1.51; P=0.04) and SII (14.67 vs 8.90; P=<.001) were significantly higher in ICU patients.

Table 3

Hemogram-derived ratios predicting ICU requirement.

					Univariable analysis
Variables	Total	Non-ICU	ICU	p-value	0R (95% CI)	p-value
NLR	4.27 (2.55-7.72)	3.85 (2.40-6.55)	7.08 (4.05-15)	<.001	2.66 (1.95-3.62)	0.001
d-NLR	3.34 (2.12-5.66)	3.16 (1.94-4.88)	5.25 (3.35- 10.11)	<.001	3.01 (2.13-4.23)	0.001
NPR	2.59 (1.82-4.07)	2.4 (1.72-3.71)	3.52 (2.42-5.36)	<.001	2.77 (1.76-4.35)	0.001
PLR	1.63 (1.06-2.45)	1.51 (1.02-2.36)	1.98 (1.15- 3.52)	0.04	1.45 (1.06-1.98)	0.019
SII	9.67 (5.33-18.80)	8.90 (4.91-16.22)	14.67 (8.11-35.67)	<.001	1.79 (1.41-2.28)	0.001

Abbreviations: NLR, neutrophil-to-lymphocyte ratio; d-NLR, derived neutrophil-to-lymphocyte ratio; NPR, neutrophil-to-platelet ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index.

Independent ICU admission prediction ability is shown for each hemogram-derived ratio in Table 4 and ROC curves in Figure 1. The result of multivariable regression models assessing the relationship of different hemogram-derived ratios with ICU admission is shown in Table 4. Model A adjusted the hemogram-derived ratio for age, DM, CKD and COPD. The subsequent models additionally adjusted for RBS, D-dimer, S. ferritin and admission SpO2. Except for PLR, all ratios remained significant predictors for ICU requirements in all other models.

Figure 1

ROC curve for the different hemogram-derived ratios and their respective area under the curves (AUC).

Table 4

Multivariable adjusted model for ICU requirements.

Model	NLR OR (95% CI) p-value	d-NLR OR (95% CI) p-value	NPR OR (95% CI) p-value	PLR OR (95% CI) p-value	SII OR (95% CI) p-value
Model A	2.38(1.74- 3.27)0.001	2.63(1.85-3.74)0.001	2.86(1.90 -4.29)0.001	1.32(0.95 -1.81)0.08	1.64(1.28 -2.10)0.001
Model B	2.37(1.73-3.24)0.001	2.61(1.84-3.71)0.001	2.77(1.84 -4.16)0.001	1.34(0.97 -1.84)0.06	1.64(1.28 -2.10)0.001
Model C	2.23(1.62-3.08)0.001	2.45(1.71-3.50) 0.001	2.55 (1.68-3.87)0.001	1.31(0.96-1.81)0.08	1.56 (1.22- 2.01)0.0004
Model D	2.16 (1.57-2.98)0.001	2.36(1.65-3.38) 0.001	2.45(1.61-3.73)0.001	1.30 (0.95-1.79)0.09	1.55(1.21- 1.99)0.0005
Model E	1.80(1.28- 2.53)0.0007	1.91(1.30-2.80)0.0008	2.09(1.32- 3.31)0.0016	1.20(0.86-1.69)0.273	1.38(1.06-1.80)0.016

Model A: Age, DM, CKD, COPD, Model B: Model A + RBS, Model C: Model B + D-dimer, Model D: Model C + Ferritin, Model E: Model D + Admission SpO2.

NLR, neutrophil-to-lymphocyte ratio; d-NLR, derived neutrophil-to-lymphocyte ratio; NPR, neutrophil-to-platelet ratio; PLR, platelet-to-lymphocyte ratio, SII, systemic immune-inflammation index.

ROC curve to detect optimal cut-off values of the hemogram-derived ratios

We analyzed the optimal cut-off values of NLR, d-NLR, NPR, PLR and SII, calculated by the ROC analysis and presented in Figure 1. Areas under the curve (AUC) of NLR, d-NLR, NPR, PLR and SII were 0.704, 0.705, 0.679, 0.588 and 0.651, respectively. The optimal cut-off values were NLR 7.29, d-NLR 5.26, NPR 3.69, PLR 2.32 and SII 19.81. SII had the highest specificity (0.78), followed by d-NLR (0.63), then NLR (0.58). The most heightened sensitivity was in favor of NLR (0.74), then NPR (0.72) and d-NLR (0.69) (Table 5).

Table 5

Cut off value of hemogram-derived ratios in predicting ICU requirement.

Variable	Cut off	Sensitivity	Specificity
NLR	7.29	0.74	0.58
d-NLR	5.26	0.69	0.63
NPR	3.69	0.72	0.56
PLR	2.32	0.67	0.49
SII	19.81	0.49	0.78

Abbreviations: NLR, neutrophil-to-lymphocyte ratio; d-NLR, derived neutrophil-to-lymphocyte ratio; NPR, neutrophil-to-platelet ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index.

Effects of comorbidities on the hemogram-derived ratios and ICU admission

Figure 2 shows the prevalence of comorbidities among all study patients and patients admitted to the ICU. The most prevalent comorbidity was DM (63.3% of all study patients, 70.4% of ICU patients) followed by IHD (22.2% all, 41.8% ICU), CKD (17.6% all, 25.5% ICU) and COPD (12.2% all, 21.4% ICU). The least prevalent comorbidity was CVA (4.5% all, 3.1% ICU). The result of the MANOVA test is presented in Table A1 in the Appendix. The Pillai's trace test statistic demonstrated that the values of the hemogram-derived ratios significantly varied due to IHD, COPD and CVA.

Figure 2

Ranking of comrbidities for total and ICU patients.

The effects of comorbidities on the hemogram-derived ratios and ICU admission are presented in Table 6. The results of multiple linear regression models showed that the presence of IHD significantly increased NLR (P=<.001), d-NLR (P=<.001) and SII (P<.001); COPD significantly increased NLR (P=0.02), d-NLR (P=<.001) and NPR (P=<.001); and CVA significantly increased PLR (P=<.001) only. Furthermore, the multiple logistic regression model revealed that IHD and COPD were significantly associated with ICU admission. Patients with IHD and COPD were 3.1 times (P=<.001) and 2.1 times (P=0.03), respectively, more likely to require ICU admission than a patient without these conditions.

Table 6

The effects of comorbidities on the hemogram-derived ratios and ICU admission.*

Comorbidities	NLR		d-NLR		NPR		PLR		SII		ICU admission
	Coeff.	p-value	Coeff.	p-value	Coeff.	p-value	Coeff.	p-value	Coeff.	p-value	OR	p-value
DM	0.67	0.305	0.63	0.113	0.08	0.779	0.34	0.224	2.88	0.224	1.32	0.307
IHD	3.44	<.001	2.21	<.001	0.17	0.608	0.42	0.198	13.06	<.001	3.12	<.001
CKD	-0.49	0.552	-0.18	0.725	-0.29	0.414	-0.23	0.511	-1.83	0.545	1.21	0.542
COPD	2.21	0.023	2.05	<.001	1.43	<.001	0.34	0.410	6.78	0.053	2.11	0.028
CVA	1.18	0.424	0.64	0.475	-0.04	0.956	2.45	<.001	2.80	0.598	0.53	0.335

Results from multiple logistic regression for ICU admission and multiple linear regressions for hemogram-derived ratios.

DISCUSSION

Statement of principal findings

Our study described the clinical characteristics and laboratory parameters of hospitalized COVID-19 patients and investigated the role of hemogram-derived ratios in predicting ICU admission. Compared with patients who did not require ICU, ICU patients were older with increased comorbidities such as hypertension, CKD, IHD, COPD, DM, and CVA. Patients with IHD and COPD were 3.1 times and 2.1 times, respectively, more likely to require ICU admission than those without these conditions. SpO2 measured at admission was significantly lower in patients who subsequently required ICU admission. The death rate was significantly higher in ICU patients. Adjusted multivariable models revealed that NLR, d-NLR, NPR and SII were significant predictors of ICU admission.

Strengths and limitations

Clinical and pathologic observation has found that hyperinflammation and immunothrombosis are critical pathogenic mechanisms in cell injury in COVID-19; addressing these processes is therefore essential in COVID-19 management. Current studies primarily focus on combating these processes. Early risk stratification of patients can facilitate effective intervention at the outset, which may substantially improve outcomes. The present study contributes tools for risk stratification that are within reach of all hospitals, even peripheral centers.

However, our study has some limitations. We did not consider the effect of other inflammatory markers such as CRP, lactate dehydrogenase, procalcitonin, troponin and interleukin-6 because ours is a retrospective study and these measures were not available in the dataset. We used only admission laboratory parameters and did not evaluate the dynamic change of these biomarkers. As a result, their effects on disease course may be underestimated here.

Interpretation in the context of the wider literature

The COVID-19 pandemic has threatened the global health system. Even developed countries such as the USA, with their very organized health care system, have seen a huge death toll (Bilinski, A and Emanuel, E. J., 2020). On the other hand, some developing countries such as Vietnam, India's Kerala state and South Korea have tackled the pandemic far more effectively with limited resources, at low cost and with impressive results (Tran, T. P. T. et al., 2020; Chathukulam, J., and Tharamangalam, J., 2021; You, J., 2020). Identifying at-risk patients at the earliest opportunity and prioritizing available resources to them can be a very effective strategy to manage the unprecedented challenge of COVID-19, particularly for developing countries. Allocating scarce resources to those with the highest probability of getting benefits from them will save more lives; this is particularly important in critical care services, given the unprecedented need for ICU beds during this COVID-19 pandemic.

The rates of ICU admission and death in our study participants were 22.17% and 12.44%, respectively. These figures are similar to findings in other studies (Chang, R. et al., 2021; Covino, M. et al., 2020). The presence of comorbidity has been found to increase the risk of becoming infected with COVID-19 (Yang, J. et al., 2020; Guan, W. et al., 2020) and be predictive of severe disease with resultant increased ICU admission and higher mortality (Jain, V. and Yuan, J., 2020; Thakur, B. et al., 2021; Honardoost, M. et al., 2021; Liu, B. et al., 2021). Our study aligns with these findings.

Various immune-inflammatory parameters have been studied to understand their role as predictors of disease severity and mortality (Lipworth, B. et al., 2020; Del Valle, D. et al., 2020; Satış, H. et al., 2021; Prasetya, I. et al., 2021). However, inflammatory biomarkers such as interleukin, lymphocyte subset, CRP, Ferritin and D-dimer are costly and not widely available in all health care facilities, particularly in rural areas. Therefore, there is an urgent need to identify predictors of adverse outcomes that are readily available and cost-effective. The present study was designed to evaluate the role of different hemogram-derived ratios in predicting ICU admission of COVID-19 patients. These ratios are available from routine laboratory tests.

Neutrophil is a major component of the leukocyte population that activates and migrates from the venous system to the immune organ or system (Rosales, C., 2018). On the other hand, human immune response triggered by viral infection mainly relies on lymphocytes (Rabinowich, H. et al., 1987), whereas systematic inflammation significantly depresses cellular immunity, which significantly decreases CD4+ T lymphocytes and increases CD8+ suppressor T lymphocyte (Menges, T. et al., 1999). Thus, virus-triggered inflammation leads to an increased NLR. The role of NLR has been extensively studied and shown to be associated with poor outcomes in infectious diseases, stroke, cancer and cardiovascular diseases (Furman, D. et al., 2019; Liu, Y. et al., 2020; Park, J. et al., 2018; Wei, Y. and Qian, W., 2014; Duan, J. et al., 2018). Current evidence suggests that high NLR and d-NLR are associated with disease progression and ICU admission in COVID-19 (Yang, A. et al., 2020; Núñez, I. et al., 2021; Liu, Y. et al., 2020; Alkhatip, A. et al., 2021). Our study also found NLR and d-NLR as significant predictors of ICU admission.

Activated platelets enhance lymphocyte adhesion to the endothelium, promoting lymphocyte homing in endothelial veins and migration to inflammatory sites (von Hundelshausen, P. and Weber, C., 2007). PLR as a marker of inflammation reflects both aggregation and inflammatory pathways and may be more valuable in predicting various inflammations than platelet or lymphocyte counts alone (Qu, R. et al., 2020). A systematic review and meta-analysis concluded that elevated PLR levels on admission could be utilized as a prognostic indicator of severity in COVID-19 patients, especially in resource-limited settings where there is an urgent need to effectively allocate medical resources and divert attention to patients with poorer prognosis (Simadibrata, D. et al., 2020). A few retrospective studies found that NLR was a dependent predictor associated with mortality while PLR was not (Açıksarı, G. et al., 2021; Wang, X. et al., 2020; Dávila-Collado, R. et al., 2021). Our study did not find any significant association of PLR with ICU admission.

Platelets and neutrophils interact during infection, inflammation and thrombosis and modulate each other's functions (Lisman, T., 2017). Ratios of these cells (NPR) have been investigated in COVID-19 and found to be a prognostic factor of disease severity and mortality (López-Escobar, A. et al., 2021; Zhang, N. et al., 2021; López-Escobar, A. et al., 2021). Consistent with these findings, our study found NPR as a significant predictor of ICU admission.

The SII index uses measures of platelets, inflammatory activators (neutrophils/monocytes), and regulators (lymphocytes), which are accepted as potential prognostic markers and sometimes a more powerful tool than NLR and PLR for predicting survival outcomes in different types of cancer (Liu, J. et al., 2019; Yang, R. et al., 2018; Chen, J. et al., 2017). Evidence suggests that higher SII is found in COVID-19 ICU patients compared with non-ICU patients and that SII is a potent marker for predicting the requirement for invasive ventilator support, disease severity and poor prognosis in COVID-19 patients (Muhammad, S. et al., 2021; Nalbant, A. et al., 2021; Fois, A. et al., 2020), all of which is consistent with our study.

Implications for policy, practice and future research

With the unforeseen and unprecedented challenges to the global health system imposed by COVID-19, there has been an urgent need to find strategies to mitigate the loss of human life within existing health care resources. Research directed at identifying at-risk patients at hospital admission using readily available, low-cost parameters is paramount. We focused on cost-effective and straightforward investigations studying several hemogram-derived parameters and analyzing their ability to predict ICU admission. Our findings can guide policymakers and clinicians to risk-stratify patients at admission. We suggest multinational studies be carried out to validate these predictors in different ethnic groups and geographic areas. Future studies should also investigate the dynamic changes of these markers in the clinical course of the disease. Our findings can help policymakers adopt appropriate strategies that are likely to better target at-risk patients and substantially decrease health expenditure.

Conclusion

Our study investigated the efficacy of inflammatory markers in predicting ICU admission in COVID-19 patients. These markers are easily measurable, widely available at low cost and can be calculated from routine blood tests. We found that higher NLR, d-NLR, NPR and SII at hospital admission are significant predictors of subsequent ICU admission for COVID-19 patients. Therefore, we recommend using these markers on admission for triaging patients at high risk of developing severe disease and requiring ICU admission and ensuring appropriate resources are allocated to them.

Declarations

Funding

This research received no external funding.