Neutrophil-to-lymphocyte ratio in predicting mortality in sepsis and its correlation with sofa score and C-reactive protein values

ABSTRACT

Introduction:

Sepsis is one of the leading causes of death worldwide in the critically ill patients. Early diagnosis and stratification of severity play a vital role in management of sepsis. Different biomarkers like C-reactive protein (CRP) and presepsin are available as markers of inflammation. Neutrophil-to-lymphocyte ratio (NLR) is used as a prognostic marker in many inflammatory conditions and cancers, which is cost-effective.

Aim and Objective:

To evaluate NLR as a predictor of mortality in sepsis and correlating it with Sequential Organ Failure Assessment (SOFA) score and CRP values.

Materials and Methods:

This is a prospective observational study including 100 patients of sepsis. Apart from routine investigations, NLR, SOFA score, and CRP (Q) were determined on days 1, 3, and 5 of hospital admission. Patients were followed up during hospital stay for primary outcome as either death or discharged.

Observations and Results:

The present study included 100 patients, out of which 53% were females and 47% were males. 54% survived in the female group, whereas it was 48.9% in the males. Mortality was the highest in the age group of 41–50 years (63.3%), followed by > 60 years (58.13%). Pneumonia (58%) was the most common cause of sepsis, followed by urinary tract infection (24%). Diabetes (52%) was the most common risk factor, followed by hypertension (36%) and chronic kidney disease (13%). The area under curve (95% confidence interval) of NLR on days 3 and 5 for the prediction of mortality is 75.5% (65.9%–83.5%) and 94.7% (88.3%–98.2%, P = 0.001), respectively, that is, very highly significant. The median of CRP (Q) on day 3 for surviving patients was 57.65 and that for nonsurviving patients was 92.10 (P = 0.005), that is, highly significant. The median of CRP (Q) on day 5 for surviving patients was 38.0 and that for nonsurviving patients was 106.0 (P = 0.001), that is, very highly significant. The median of SOFA score on day 3 for surviving patients was 3 and that for nonsurviving patients was 8 (P = 0.001), that is, very highly significant. The median of SOFA score on day 5 for surviving patients was 2 and that for nonsurviving patients was 11 (P = 0.001), that is, very highly significant.

Conclusion:

A high value of NLR, SOFA score, and CRP on days 3 and 5 of admission in sepsis patients is associated with increased mortality. A serial increase in the value of NLR during the initial period of sepsis is a better marker for prediction of mortality, rather than a single point value, and the same is true for SOFA score and CRP values. NLR can be used as a simple, cheap, and easily available biomarker for sepsis during the initial period to predict severity and mortality.

Introduction

As per the Third International Consensus Definition, sepsis is defined as a dysregulated host response to infection that leads to acute organ dysfunction.[1] Septic shock is defined as a subset of sepsis cases in which underlying circulatory and cellular/metabolic abnormalities are profound enough to substantially increase mortality risk.[1] Sepsis is still one of the leading causes of death worldwide in the critically ill patients.[1] Hence, early diagnosis and stratification of severity play a vital role in management of sepsis. CRP level is widely used as a relatively nonspecific marker of inflammation. Different biomarkers like presepsin, PTX 3, IL 6, and IL 10 are used. But these biomarkers are very costly and have limited availability. NLR (neutrophil-to-lymphocyte ratio) is an inflammatory biomarker calculated by dividing absolute neutrophil count by absolute lymphocyte count. In sepsis, there is an increase in number of neutrophils, that is, cells of innate immunity, and a decrease in number of lymphocytes, that is, cells of adaptive immunity, resulting in a high NLR. NLR is used as a prognostic marker in many inflammatory conditions and cancers[2]; moreover, it is cost-effective. There are different scoring systems that are used to predict mortality in sepsis, for example, SOFA (Sequential Organ Failure Assessment), APACHE II(Acute Physiology and Chronic Health Evaluation), REMS (Rapid Emergency Medicine Score), and NEWS (National Early Warning Score).[3] Among them, SOFA score is widely used to assess organ failure and to predict mortality in sepsis patients. In the present study, NLR, CRP values, and SOFA scores are calculated on days 1, 3, and 5 for all the sepsis patients included in the study, and the values are correlated with the outcome of the patients in the form of death or improved during the period of hospital stay.

Aims and Objectives

1. To study the relationship between NLR and mortality in patients of sepsis.

2. To study the correlation between NLR and SOFA score in the outcome of sepsis.

3. To study the relationship between NLR and CRP values in predicting mortality in sepsis.

Materials and Methods

This is a prospective observational study including 100 patients admitted to the department of medicine. The patients of age ≥18 years having signs and symptoms of infection with qSOFA score ≥2 were included in the study. Patients with any of the following were excluded: radiation therapy, any malignancy, immunodeficiency, steroid therapy, immunosuppressive therapy, and lactating and pregnant women. After the approval from Institutional Ethics Committee (IEC Application No 865), informed written consent was taken from all the patients participating in the study. The sample size was calculated to be 78 using the formula [n = Z2 P (1-P)/d2] taking prevalence of sepsis in India to be 28.3% as per INDICAPS study (P = 0.28) with 95% confidence interval (Z = 1.96) and 10% as margin of error (d = 0.1). So, an average sample size of 100 was decided for the study. A total number of 100 adult patients admitted to hospital wards and ICU of Department of Medicine were included in this study.

Statistical analysis

All the observational data were tabulated and analyzed by the SPSS version 25 statistical package for windows. Mann–Whitney U test and Chi-square test were used to report and compare the continuous and categorical variables. Medians (and percentile 25–75) and frequencies (and percentages) of different variables were also obtained. NLR was compared at days 1, 3, and 5 between surviving and nonsurviving patients. Receiver operating characteristic (ROC) curves were plotted using NLR on days 1, 3, and 5 of sepsis diagnosis and mortality during hospital stay; the area under curve (AUC) and 95% confidence interval were reported. Cutoff values of NLR were obtained (point selected on the basis of Youden J index). The curves were compared by log-rank test. Box whisker plot was obtained using the values of NLR, CRP, and SOFA scores on days 1, 3, and 5 and mortality during hospital stay. Significant differences were established with P < 0.05. A P value of <0.05 is considered as significant, <0.01 as highly significant, and ≤0.001 as very highly significant.

Observations and Results

In the present study, out of 100 cases, 52% cases survived and 48% cases did not survive. Among 100 cases, the number of males was 47 and the number of females was 53 (i.e., 47% and 53%, respectively). Out of 47 males, 23 cases (48.94%) survived and 24 (51.06%) cases did not survive. Out of 53 females, 29 (54.72%) cases survived and 24 (45.28%) cases did not survive. Out of 19 patients in the age group ≤40 years, 7 patients died, making a mortality of 36.84%. In the age group of 41–50 years, out of 11 patients, 7 patients died, making a mortality of 63.63%. In the age group of 51–60 years, out of 27 patients, 9 patients died, making a mortality of 33.33%. Out of 43 patients of age >60 years, 25 patients died, making a mortality of 58.13% [Table 1].

Table 1.

Agewise distribution and mortality

Age group	Total number of patients	Improved	Death	Mortality (%)
≤40	19	12	07	36.84
41-50	11	04	07	63.63
51-60	27	18	09	33.33
>60	43	18	25	58.13

Among 100 patients, 58 patients had pneumonia, 24 patients had UTI, and 18 patients had other sites of infection. Among 58 cases of pneumonia, 34 patients died, making a mortality of 58.62%. Among 24 cases of UTI, 7 cases died, making a mortality of 29.16%. Mortality in other cases was 38.88%.

The different risk factors of all the patients were also noted in our study. Out of 100 cases, 52 cases were known cases of T2DM, 36 cases of HTN, 13 cases of CKD, 6 cases of IHD, and 4 cases of COPD. Out of 52 cases of T2DM, 25 patients died, making a mortality of 48.07%. Out of 36 cases of HTN, 23 patients died, making a mortality of 63.88%. Out of 13 cases of CKD, 7 patients died, making a mortality of 46.1%. Out of 6 cases of IHD, 4 patients died, making a mortality of 66.66%. Out of 4 cases of COPD, 2 patients died, making a mortality of 50% [Table 2].

Table 2.

Mortality as per risk factors

Risk factors	Number of patients	Improved	Death	Mortality
T2 DM	52	27	25	48.07%
HTN	36	13	23	63.88%
CKD	13	07	06	46.1%
IHD	06	02	04	66.66%
COPD	04	02	02	50%

The median of NLR on day 1 for surviving patients was 13.88 and that for nonsurviving patients was 12.15 (P = 0.139), that is, not significant. The median of NLR on day 3 for surviving patients was 8.75 and that for nonsurviving patients was 17.94 (P = 0.001), that is, very highly significant. The median NLR on day 5 for surviving patients was 6.8 and that for nonsurviving patients was 18.2 (P = 0.001), that is, very highly significant [Table 3] [see appendix].

Table 3.

Comparing the cutoff, AUC (area under curve), and P values of NLR on days 1, 3, and 5 in predicting mortality

Parameter	NLR D1	NLR D3	NLR D5
Cutoff	13	09	10
Youden index	0.183	0.411	0.805
P	0.072	0.001	0.001
AUC	0.585 (0.482-0.683)	0.755 (0.659-0.835)	0.947 (0.883-0.982)

The median of SOFA score on day 1 for surviving patients was 6.5 and that for nonsurviving patients was 5.5 (P = 0.144), that is, not significant. The median of SOFA score on day 3 for surviving patients was 3 and that for nonsurviving patients was 8 (P = 0.001), that is, very highly significant. The median of SOFA score on day 5 for surviving patients was 2 and that for nonsurviving patients was 11 (P = 0.001), that is, very highly significant [Table 4] [see appendix].

Table 4.

Comparing median of NLR, SOFA score, and CRP (Q) on days 1, 3, and 5 among improved and dead patients of sepsis with corresponding P values

	Death Median (IQR)	Improved Median (IQR)	P
DAY 1
NLR	12.15 (11.71)	13.88 (10.54)	0.139
SOFA	5.5 (3)	6.5 (5)	0.144
CRP	65.22 (65.75)	91.25 (95.95)	0.124
DAY 3
NLR	17.94 (18.93)	8.75 (7.67)	0.001
SOFA	8.0 (3)	3.0 (3)	0.001
CRP	92.10 (67.42)	57.65 (71.32)	0.005
DAY 3
NLR	18.20 (18.5)	6.8 (3.5)	0.001
SOFA	11 (7)	2 (2	0.001
CRP	106.0 (100.1)	38.0 (41.3)	0.001

The median of CRP (Q) on day 1 for surviving patients was 91.25 and that for nonsurviving patients was 65.22 (P = 0.124), that is, not significant. The median of CRP (Q) on day 3 for surviving patients was 57.65 and that for nonsurviving patients was 92.10 (P = 0.005), that is, highly significant. The median of CRP (Q) on day 5 for surviving patients was 38.0 and that for nonsurviving patients was 106.0 (P = 0.001), that is, very highly significant [Table 4].

Discussion

Out of 100 cases, 52% cases survived and 48% cases did not survive. Among 100 cases, the number of males was 47 and the number of females was 53 (i.e., 47% and 53%, respectively). Out of 47 males, 23 cases (48.94%) survived and 24 (51.06%) cases did not survive. Out of 53 females, 29 (54.72%) cases survived and 24 (45.28%) cases did not survive. In a study done at Spain in 2020, out of a total number of 203 patients of sepsis, the mortality among females was higher (34.92%) than among males (32.85%).[2] But in our study, the mortality is higher among males (51.06%).

Among 100 patients, 58 patients had pneumonia, 24 patients had UTI, and 18 patients had other sites of infection. Among 58 cases of pneumonia, 34 patients died, making a mortality of 58.62%. Among 24 cases of UTI, 7 cases died, making a mortality of 29.16%. Mortality in other cases was 38.88%. Previous studies, done at Spain, India, Pakistan, and United states, have shown the most common site of infection being lungs.[2,3,4,5] The different risk factors of all the patients were also noted in our study. Out of 100 cases, 52 cases were known cases of T2DM, 36 cases of HTN, 13 cases of CKD, 6 cases of IHD, and 4 cases of COPD. Out of 52 cases of T2DM, 25 patients died, making a mortality of 48.07%. Out of 36 cases of HTN, 23 patients died, making a mortality of 63.88%. Out of 13 cases of CKD, 7 patients died, making a mortality of 46.1%. Out of 6 cases of IHD, 4 patients died, making a mortality of 66.66%. Out of 4 cases of COPD, 2 patients died, making a mortality of 50%. In a study done at United States, T2DM is the most common comorbidity associated with sepsis in nonwhite populations, followed by renal failure.[5] In a study conducted at Spain, the most common comorbidity associated with sepsis and related mortality is T2DM.[2] In our study, T2DM is the most common comorbidity, leading to sepsis. However, the mortality is the highest among patients having IHD as a risk factor, followed by hypertension (HTN).

In a study done in Spain, it was found that 30-day nonsurviving patients (n = 68) compared to surviving patients (n = 135) showed higher NLR on the first (P < 0.001), fourth (P < 0.001), and eighth (P < 0.001) days of sepsis diagnosis. It was also associated with a higher value of serum lactate and SOFA score in those days.[2]

Our study showed that the nonsurviving patients had higher NLR than surviving patients on day 3 (P = 0.001, i.e. very highly significant) and day 5 (P = 0.001, i.e. very highly significant). However, the NLR is not higher in nonsurviving patients than in surviving patients on day 1 (P = 0.072, i.e. not significant). NLR on days 1, 3, and 5 had an area under curve (95% confidence interval) for the prediction of mortality of 58.5% (48.2%–68.3%, P = 0.072), 75.5% (65.9%–83.5%, P = 0.001), and 94.7% (88.3%–98.2%, P = 0.001), respectively. The cutoff NLR for days 1, 3, and 5 was 13, 9, and 10, respectively. Hence, the patients having persistent high NLR on days 3 and 5 had higher mortality than those having lower NLR. However, day 1 NLR was not helpful in predicting mortality.

A high NLR on day 1 was not associated with increased mortality or poor prognosis. However, a high NLR on days 3 and 5 was associated with higher mortality and poor prognosis. A high SOFA score on days 3 and 5 was associated with increased mortality. A high SOFA score on day 1 was not associated with increased mortality. A high CRP value on days 3 and 5 was associated with increased mortality. A high CRP value on day 1 was not associated with increased mortality.

In the study done at Spain, it has been shown a high NLR during admission or on day 1 is associated with increased mortality in sepsis patients. Similarly, a high SOFA score during admission is associated with poor prognosis.[2] However, in our study, the day 1 NLR, SOFA score, and CRP values were not associated with increased mortality; rather, a persistent rise in NLR, SOFA score, and CRP values during days 3 and 5 was associated with increased mortality and poor prognosis.

As per a previous study done in Pakistan, the NLR showed significant association with CRP values (P = 0.02) and SOFA score (P = 0.01) in predicting mortality.[4] However, in a study conducted at Turkey, it was shown that there was no obvious correlation between CRP and the severity of sepsis at the initial stage.[6] But in our study, we found out that a definite correlation exists between NLR, SOFA score, and CRP values on days 3 and 5 in predicting severity and mortality in sepsis.

As per Liu et al Neutrophil-to-lymphocyte ratio and plasma lactate levels of the patients with sepsis were significantly higher than those of control subjects.[7]

Adda et al in 2020 showed that Higher NLR value was associated with the occurrence of sepsis and sepsis-related mortality.[8]

The main findings from our study were the association between NLR in the first 5 days of sepsis and mortality and also its association with SOFA score and CRP values. We found out that a serial increase in the value of NLR, SOFA score, and CRP during the initial period of sepsis was associated with increased mortality. The persistently higher value of NLR is a better prognostic indicator than the day 1 NLR value, and the same is true for SOFA score and CRP values.

Previous studies have shown an increased NLR in nonsurvivors than in survivors during the course of sepsis evolution. In a study conducted at Turkey including 591 patients of sepsis, it was found that the NLR and CRP on the first, third, and last days of ICU stay showed a similar increasing trend for mortality.[7] But in our study, the association between NLR during the initial period of sepsis and mortality is determined, which is an interesting finding. Another interesting finding in our study was the prognostic capability of NLR on days 3 and 5, which was not different to SOFA and CRP values in the same days.

The immune response is the reaction of our body to defend itself against foreign microorganisms. It is of two types, that is, innate immune response and adaptive immune response. The innate immune system is the first-line defense in our body, which consists of physical barrier (mucous membrane), immune cells (neutrophils, macrophages), and circulating factors (cytokines, complements). The adaptive immune system is the second-line defense mechanism consisting of B-cells, T-cells, and circulating antibodies.[9] In the early stage of sepsis, there is a rapid increase in the number of neutrophils (component of the innate immune system) and a decrease in the number of lymphocytes due to generalized lymphocyte apoptosis (which occurs in the spleen, thymus, and lymph nodes).[10] This leads to a high NLR during the initial period of sepsis. We think that the association between NLR in the first 5 days of sepsis and mortality that has been found in our study can be due to a higher state of inflammation, leading to organ dysfunction in patients who die. This is also significant in terms of high SOFA score and CRP values on the same days.

NLR, a simple laboratory parameter obtained from CBC reports, can be used for the prediction of mortality, and its values during the first 5 days can help to know the prognosis.

Limitations of the study

• Only the patients admitted to the Department of Medicine were included in the study. Other patients of sepsis were not included.

• The day of sepsis was calculated based on the day of admission, not on the day of onset of symptoms.

• We found that in our study, there were some limitations with the sample size, which precluded us from getting statistical significance with regard to certain variables with the severity of sepsis.

Conflicts of interest

There are no conflicts of interest.

Appendix

DAY 1 NLR vs OUTCOME

Receiver operating characteristic (ROC) analysis using NLR on day 1 for prediction of mortality in sepsis patients.

Cutoff NLR = 13.0

Sensitivity of NLR = 62.5%

Specificity of NLR = 55.8%

Area under curve (95% confidence interval) of NLR on day 1 for the prediction of mortality is 55.5% (48.2%-68.3%, P = 0.072), i.e., not significant.

DAY 3 NLR vs OUTCOME

ROC analysis using NLR on day 3 for prediction of mortality in sepsis patients.

Cutoff NLR = 9.0

Sensitivity of NLR = 89.1%

Specificity of NLR = 51.9%

Area under curve (95% confidence interval) of NLR on day 3 for the prediction of mortality is 75.5% (65.9%-83.5%, P = 0.001), i.e. very highly significant.

DAY 5 NLR vs OUTCOME

ROC analysis using NLR on day 5 for prediction of mortality in sepsis patients.

Cut-off NLR = 10.0

Sensitivity of NLR = 91.3%

Specificity of NLR = 89.2%

Area under curve (95% confidence interval) of NLR on day 5 for the prediction of mortality is 94.7% (88.3%–98.2%, P = 0.001), i.e. very highly significant.

BOX AND WHISKER PLOT OF NLR DAY 1 VS OUTCOME

MEDIAN OF NLR ON DAY1 IN IMPROVED PATIENTS = 13.88

MEDIAN OF NLR ON DAY1 IN DEAD PATIENTS = 12.15

P-VALUE = 0.139 (NOT SIGNIFICANT)

BOX AND WHISKER PLOT OF CRP (Q) DAY 1 VS OUTCOME

MEDIAN OF CRP (Q) ON DAY1 IN IMPROVED PATIENTS = 91.25

MEDIAN OF CRP (Q) ON DAY1 IN DEAD PATIENTS = 65.22

P-VALUE = 0.124 (NOT SIGNIFICANT)

BOX AND WHISKER PLOT OF SOFA SCORE DAY 1 VS OUTCOME

MEDIAN OF SOFA SCORE ON DAY1 IN IMPROVED PATIENTS = 6.5

MEDIAN OF SOFA SCORE ON DAY1 IN DEAD PATIENTS = 5.5

P-VALUE = 0.144 (NOT SIGNIFICANT)

BOX AND WHISKER PLOT OF NLR DAY 3 VS OUTCOME

MEDIAN OF NLR ON DAY 3 IN IMPROVED PATIENTS = 8.75

MEDIAN OF NLR ON DAY 3 IN DEAD PATIENTS = 17.94

P-VALUE = 0.001 (VERY HIGHLY SIGNIFICANT)

BOX AND WHISKER PLOT OF CRP (Q) DAY 3 VS OUTCOME

MEDIAN OF CRP (Q) ON DAY 3 IN IMPROVED PATIENTS = 57.65

MEDIAN OF CRP (Q) ON DAY 3 IN DEAD PATIENTS = 92.10

P-VALUE = 0.005 (HIGHLY SIGNIFICANT)

BOX AND WHISKER PLOT OF SOFA SCORE DAY 3 VS OUTCOME

MEDIAN OF SOFA SCORE ON DAY 3 IN IMPROVED PATIENTS = 3.0

MEDIAN OF SOFA SCORE ON DAY 3 IN DEAD PATIENTS = 8.0

P-VALUE = 0.001 (VERY HIGHLY SIGNIFICANT)

BOX AND WHISKER PLOT OF NLR DAY 5 VS OUTCOME

MEDIAN OF NLR ON DAY 5 IN IMPROVED PATIENTS = 6.8

MEDIAN OF NLR ON DAY 5 IN DEAD PATIENTS = 18.20

P-VALUE = 0.001 (VERY HIGHLY SIGNIFICANT)

BOX AND WHISKER PLOT OF CRP (Q) DAY 5 VS OUTCOME

MEDIAN OF CRP (Q) ON DAY 5 IN IMPROVED PATIENTS = 38.0

MEDIAN OF CRP (Q) ON DAY 5 IN DEAD PATIENTS = 106.0

P-VALUE = 0.001 (VERY HIGHLY SIGNIFICANT)

BOX AND WHISKER PLOT OF SOFA SCORE DAY 5 VS OUTCOME

MEDIAN OF SOFA SCORE ON DAY 5 IN IMPROVED PATIENTS = 2.0

MEDIAN OF SOFA SCORE ON DAY 5 IN DEAD PATIENTS = 11.0

P-VALUE = 0.001 (VERY HIGHLY SIGNIFICANT)

Funding Statement

Nil.