Introduction

Neutrophils and lymphocytes are immune system cells that are part of the pathophysiological process of many diseases, whose blood concentrations can be used to monitor hospitalized patients [1,2,3,4]. Since neutrophils produce several pro-inflammatory cytokines, high blood concentrations of neutrophils are indicative of increased oxidative stress mainly in more vulnerable patients [5,6,7,8], including critically ill patients with Coronavirus Disease 2019 (COVID-19), in which excessive levels of reactive oxygen species are responsible for lung tissue damage, thrombosis, and red blood cell dysfunction, thus resulting in the COVID-19 disease severity [9]. Regarding lymphocytes, the lower the concentration, the greater the oxidative stress, given that several anti-inflammatory cytokines (e.g., interleukin (IL)-4, IL-10, IL-13, and interferon-gamma) are derived from lymphocytes [10, 11]. More importantly, lymphopenia is widely recognized as a marker of poor survival outcomes and was used in the COVID-19 pandemic [12].

The neutrophil and lymphocyte ratio (NLR) has emerged through massive research as a complementary marker of critical care-related disorders [12,13,14,15]. However, NLR merits further investigation as a marker in severe COVID-19 due to the lack of cutoff values and therefore we performed a cross-sectional study in this regard. The primary aim of this research was to ascertain the association (or lack thereof) between elevated NLR values and intensive care unit (ICU) mortality in critically ill patients with COVID-19. In addition, associations between high NLR values with length of stay and need for mechanical ventilation were considered as secondary outcomes.

Methods

Study design and patients

A cross-sectional study was conducted in an ICU specialized in the treatment of COVID-19 at Hospital São Lucas (HSL), Rede D’OR – São Luiz, Aracaju/Sergipe, Brazil. This research is affiliated with the Federal University of Sergipe and was approved by the local Research Ethics Committee (nº 35128820.0.0000.5546.). Volunteers or their legal guardians who agreed to be evaluated signed a consent form.

Inclusion criteria were ≥ 18 years, patients of both sexes, COVID-19 diagnosis, and available neutrophil and lymphocyte count. One hundred ninety volunteers were enrolled, of which only one patient was excluded due to the lack of neutrophil and lymphocyte count. The low number of exclusions occurred mainly because neutrophil and lymphocyte counts are routine markers ordered by critical care physicians. Clinical and demographic variables were acquired directly by physicians or medical students under supervision using medical records.

Biochemical analyses

The diagnosis of COVID-19 was detected by reverse-transcription PCR (RT-PCR), which was collected non-randomly, without the patient contact, from electronic medical records. Peripheral blood collection with anticoagulants was performed by a phlebotomy specialist and then aspirated and injected into the analyzer.

More specifically, an automated hemocytometer (Sysmex XN-1500™) working on the flow cytometry principle was used to count and categorize the types of white blood cells by arranging these cells in a single file line, passing in front of a laser beam, scattered light and fluorescent light. Ultimately, NLR was calculated by dividing the absolute neutrophil count by the lymphocyte count.

Statistical analyses

We estimated the best NLR cutoff point for ICU death by calculating the sensitivity, specificity, and area under the Receiver Operating Characteristic (ROC) curve. The significance level was set at 5%. The cutoff value was chosen based on the area under the curve (AUC) and the balance between specificity and sensitivity values. NLR ≥ 10.6 was the cutoff value chosen employing an AUC equivalent to 0.6551 ± 0.05 (95% CI, 0.57–0.73) (Fig. 1).

Fig. 1
figure 1

Receive Operating Characteristic (ROC) curve analysis for neutrophil-lymphocyte rate (NLR) cutoff to discriminate ICU mortality risk in COVID-19 patients

Full size image

Sex was categorized as male or female, while comorbidities (obesity, diabetes, kidney failure, deep vein thrombosis, hypertension, respiratory failure, sepsis, etc.) and diarrhea were categorized as “no” or “yes”. Categorical variables were described through absolute frequency (N) and percentage (%), whereas continuous or discrete variables were defined with median and standard deviation (SD).

Differences between groups were tested with Pearson’s chi-squared test for categorical variables and Student’s t-test for continuous variables. We performed the analyses for crude and adjusted models. Model 1 included adjustments for age, sex, kidney dysfunction, diabetes, obesity, hypertension, and deep vein thrombosis. Model 2 included adjustments for model 1 variables plus drugs (antibiotics, anticoagulants, antivirals, corticoids, neuromuscular blockers, and vasoactive drugs).

Logistic regression was performed with a confidence interval of 95%, and the results were presented as Odds Ratio (OR). P-values < 0.05 were considered significant. We performed the analyses using Stata 15.1 (StataCorp, College Station, TX, USA).

Results

Demographic and clinical characteristics

Table 1 depicts demographic and clinical characteristics divided by NLR cutoff values.

Table 1 Demographic and clinical characteristics of the patients according to the neutrophils-lymphocytes ratio
Full size table

Regarding clinical data, participants with NLR ≥ 10.6 were older than those with NLR < 10.6 and had a high prevalence of kidney dysfunction. Moreover, a high need for vasoactive drugs and invasive mechanical ventilation, as well as a high mortality rate, was observed for an NLR ≥ 10.6.

As far the biochemistry parameters, an NLR ≥ 10.6 was associated with low lymphocyte levels, along with high C-reactive protein, leucocyte, neutrophil, NLR, and red blood cell distribution width levels.

Main outcomes

Table 2 shows the association between NLR and mortality as the primary outcome. Table 3, in turn, shows the association between NLR and secondary outcomes (i.e., length of stay and invasive mechanical ventilation).

Table 2 Association between NLR cutoffs with ICU mortality (n = 55)
Full size table
Table 3 Association between NLR cutoffs with ICU length of stay ≥ 14 days (86 patients) and need for mechanical ventilation (113 patients)
Full size table

Collectively, an NLR ≥ 10.6 was associated with higher odds of ICU mortality, length of stay, and invasive mechanical ventilation in both crude and fully adjusted models compared with an NLR < 10.6.

More specifically, patients with higher NLR had 2.77 (95% CI, 1.24–6.18), 3.56 (95% CI, 1.01–12.5), and 5.39 (95% CI, 1.96–14.81) more likely to have ICU mortality, length of stay ≥ 14 days, and need for invasive mechanical ventilation, respectively.

Discussion

In this study, higher NLR (≥ 10.6) was considered a marker associated with a high likelihood of ICU mortality, length of stay, and invasive mechanical ventilation in critically ill patients with COVID-19, compared to those with lower NLR (< 10.6). The higher mean NLR (20.6 ± 12.3 for NLR ≥ 10.6) found here seemingly is related to an exceedingly high inflammatory response to COVID-19-related critical illness triggered by a cytokine storm [16,17,18]. However, it is noteworthy to mention that even the lower NLR mean (5.9 ± 2.6 for NLR < 10.6) in our study is an alarming rate if extrapolated to other populations, such as inpatients, given the close link of this NLR level with chronic low-grade systemic inflammation (e.g., diabetes, obesity, and cardiovascular and renal diseases) [19,20,21,22].

Low lymphocyte count is a recognized predictor marker of mortality in intensive care [23,24,25]. That said, a higher NLR is expected in COVID-19 mortality due to low lymphocyte counts, and we confirmed this clinical link by comparing patients with NLR ≥ 10.6 vs. <10.6. Similarly, Ullah et al. considered NLR > 11 as a predictor of COVID-19 mortality compared to < 10, but did not perform specific analyses to find an ideal cutoff value [26]. Noteworthy, the higher NLR mean (20.9 ± 12.5 for patients with NLR ≥ 10.6) of our study was similar to the mean (20.7 ± 24.1) observed by Yang et al. in critically ill patients with COVID-19, [27] which presented a smaller sample size (n = 24) compared to our data. In contrast, some studies show low mean or median NLR values (3.7 (2.0, 6.7) n = 28 [28]; 4.24 (3.00-10.87) n = 16 [29]; 6.29 ± 3.72, n = 16 [30]) for patients with severe COVID-19 compared to our findings, but their sample size is insufficient to draw a reliable cutoff value.

A recent meta-analysis of COVID-19 patients showed that elevated NLR levels on admission were associated with a 174% higher risk of mortality [31]. The authors emphasized the importance of establishing an optimal cutoff value for NLR, and our study helped to expand this background into clinical practice [31]. We considered the NLR cutoff ≥ 10.6 to maintain a balanced sensitivity (67%) and specificity (60%), avoiding very high or moderate levels that are ubiquitously associated with many diseases, primarily cardiovascular diseases and related metabolic problems.

Little is known about the association between NLR and COVID-19-associated diseases and mortality in South America, and thus our study provides an important finding consisting of a representative sample from a continental country such as Brazil. Most research evaluating this association has been conducted in China [27,28,29,30, 32,33,34,35,36], with a couple of studies in the US [26, 37] and Europe (e.g., Turkey and Italy) [38,39,40,41], in which high NLR was deemed a marker of COVID-19-related mortality and overall severity.

Our secondary outcomes (i.e., ICU length of stay and invasive mechanical ventilation) must be discussed in more detail. In different regions of England, the ICU length of stay was estimated at 12–19 days [42]. Compelling European and American data show a mean ICU length of stay between 7 and 21 days [43,44,45,46]. Interestingly, there are reports of longer ICU length of stay, as found in a French study, in which the mean time required for intensive care was one month (27 days for those who survived and 45 days for those who died) [47]. Here, we observed an ICU length of stay close to one month (36.2 ± 7.2 and 31.9 ± 2.7 days for NLR < 10.6 versus ≥ 10.6). Thus, we selected an ICU length of stay equal to or greater than two weeks to afford a relevant outcome.

Although invasive mechanical ventilation is used to allow adequate gas exchange, we consider it a critical outcome due to its relationship with high morbidity and mortality in patients with COVID-19 [48, 49]. On average, 59% of our population required invasive mechanical ventilation [50]. Nevertheless, higher rates can also be observed, as reported by a Brazilian study whereby ~ 84% of patients needed invasive mechanical ventilation at ICU admission [50].

The present work has limitations. The cross-sectional design does not infer a cause-effect relationship between the NLR and medical outcomes in COVID-19. Thus, due to the methodology employed, we did not perform survival analysis since it is more suitable for studies designed to analyze time-to-event data (longitudinal studies) [51]. In light of this, we used ORs as the measure of association rather than Hazard ratios, the latter of which analyzes the impact of an intervention on survival or time-to-event outcomes [52]. Moreover, although we evaluated 189 patients, this sample is from a single-center study and hence reproducibility ought to be interpreted with caution, even for critically ill patients with COVID-19.

Finally, the pros and cons of using NLR as a surrogate inflammatory marker in the clinical setting must be highlighted. Considering the benefits, NLR is a cheap parameter as it is obtained by two components of the blood cell count (neutrophils and lymphocytes), which is perhaps the most ordered laboratory test worldwide to monitor basic and advanced clinical conditions (e.g., anemias and cancers). Pathophysiologically, however, attention to NLR per se as a marker of inflammation lacks meticulousness, given the plethora of pro-inflammatory cytokines that are part of the COVID-19 disease. Taking into account that cytokines are not often analyzed in the real-life scenario due to the lack of laboratory tools (reagents, kits, and devices) and cutoff values for the general population, NLR deserves consideration to guide practitioners.

Conclusion

Elevated NLR values are associated with high rates of ICU mortality, length of stay, and need for invasive mechanical ventilation in critically ill patients with COVID-19. Therefore, further attention to NLR is crucial in clinical practice as an inexpensive and valuable complementary tool to monitor patients with COVID-19 in intensive care.