ORIGINAL ARTICLE

Fatma Tortum, Erdal Tekin, Emine Ozdal

Department of Emergency Medicine, Faculty of Medicine, Atatürk University Erzurum, Turkey

Use of the systemic immune-inflammation

index to predict the severity of pneumonia
in the emergency department
Corresponding author: ABSTRACT
Fatma Tortum, Assist. Prof., MD, Introduction: This study aimed to investigate the use of the systemic immune-inflammation index obtained
Atatürk University, Faculty
by hemogram parameters in determining the clinical severity of pneumonia.
of Medicine, Department of Emergency
Medicine, Erzurum, Turkey; Material and methods: This study was conducted retrospectively with a total of 6,802 patients diagnosed
e-mail: drcitirik@hotmail.com with viral and bacterial pneumonia from January 1, 2013, through January 1, 2023, at the emergency
department of a tertiary hospital. The patients’ age, gender, white blood cell, neutrophil, lymphocyte, and
platelet counts, clinical outcomes (mortality and discharge), and mechanical ventilator requirements during
treatment were obtained from the electronic patient files.
Results: The mean age of the patients was 62.3 ± 17.3 years, and 57.8% (n = 3,928) were male. The sys-
temic immune-inflammation index was found to predict mortality in patients with a sensitivity of 77.9% and a
specificity of 36.2% at a cut-off value of 114.72 [area under the curve (AUC): 0.654]. The receiver operating
characteristic (ROC) curve analysis showed that the systemic immune-inflammation index was statistically
significant in determining mortality among the patients (p < 0.001, 95% confidence interval: 0.639–0.669).
The systemic immune-inflammation index was found to predict the mechanical ventilator requirement
with a sensitivity of 70.0% and a specificity of 47.5% at a cut-off value of 137.88 (AUC: 0.629). According
Medical Research Journal 2023; to the ROC curve analysis, the systemic immune-inflammation index was also statistically significant in
Volume 8, Number 3, 186–191 determining the mechanical ventilator requirement among the patients (p < 0.001, 95% CI: 0.599–0.658).
10.5603/mrj.96312 Conclusions: The systemic immune-inflammation index was found to be valuable in determining clinical
Copyright © 2023 Via Medica
ISSN 2451-2591 severity in patients with pneumonia.
e-ISSN 2451-4101 Keywords: Pneumonia, systemic immune-inflammation index, SII, NLR, PLR, mortality
Med Res J 2023; 8 (3): 186–191

Introduction Studies investigating the severity of pneumonia also

assess various haematological or biochemical param-
Pneumonia is an inflammation of tissues in one eters in addition to scoring systems [3–4]. There are
or both lungs, usually caused by a bacterial agent. In ongoing studies on the testing of certain biomarkers,
the USA, more than 1 million people are admitted to such as C-reactive protein, procalcitonin, presepsin,
hospitals annually due to pneumonia, and 50,000 of adrenomedullin, and proenkephalin [5]. In a previous
these cases result in mortality [1]. Although pneumo- study addressing the advantages of different biomark-
nia is a treatable disease, it remains mortal; therefore, ers, it was reported that procalcitonin had the highest
patients presenting to the emergency department with accuracy for bacterial aetiologies and the presence of
pneumonia are evaluated for disease severity and hos- bacteraemia, lactate was a biomarker of hypoxia and
pitalization indication. During this evaluation, scoring tissue hypoperfusion, and proadrenomedullin was
systems, such as the Pneumonia Severity Index (PSI), successful in predicting mortality when used togeth-
CURB-65, and CRB-65 are generally used [2]. However, er with prognostic exposures [6]. However, most of
the performance of these scoring systems may vary due these biomarkers are not used during routine patient
to differences in the distribution of etiological agents, care in emergency departments. In addition, some
comorbidities, and the presence of social support [2]. require very expensive assays to obtain. Therefore,

This article is available in open access under Creative Common Attribution-Non-Commercial-No Derivatives 4.0 International (CC BY-NC-ND 4.0) license, allowing to
download articles and share them with others as long as they credit the authors and the publisher, but without permission to change them in any way or use them commercially.

186 www.journals.viamedica.pl/medical_research_journal
 Fatma Tortum et al., Predict the severity of pneumonia in the emergency department

their applicability in the emergency department does status of these patients were determined by screening
not seem feasible. the electronic patient files. Further excluded were pa-
The systemic immune-inflammation index (SII) tients diagnosed with COVID-19 based on a positive
is obtained by multiplying the platelet count and the COVID-19 test. Lastly, of the patients who were planned
neutrophil-to-lymphocyte ratio (NLR) and is used to to be included in the study, those with missing data and
simultaneously evaluate the inflammatory and immune those who wanted to be discharged from the hospital
status of patients [7]. Previous studies have determined during their clinical follow-up were also excluded. After
a relationship between SII and mortality or clinical sever- applying all inclusion and exclusion criteria, a total of
ity in cases such as malignancies, coronary diseases, 6,802 patients were included in the sample. The details
and heart failure [8–10]. Due to its very low cost and of patient selection are presented in Figure 1.
acquisition through a routine hemogram test, SII has
been evaluated in different disease groups. Data collection
This study aimed to evaluate whether SII could be
used in the evaluation of clinical severity in patients pre- Data on the patients’ age, gender, white blood cell
senting to the emergency department with pneumonia. (WBC), neutrophil count, lymphocyte count, platelet
To this end, the relationship was investigated between count​, clinical outcomes (mortality and discharge) and
SII ​​and hospitalization requirements, mortality, and MV requirements and the use of MV during treatment
mechanical ventilator (MV) requirements in patients di- were obtained from the electronic patient files.
agnosed with pneumonia at the emergency department. The WBC, neutrophil, lymphocyte, and platelet
counts obtained from the first hemogram test performed
at the time of presentation to the emergency department
Material and methods or in the inpatient ward were used. The NLR and plate-
let-to-lymphocyte ratio (PLR) was calculated by dividing
Study design the absolute neutrophil and platelet counts by the abso-
lute lymphocyte count, respectively. SII was calculated
This study was conducted retrospectively at the using the following formula: platelet count × neutrophil
emergency department of a tertiary hospital and count/lymphocyte count [11].
included patients diagnosed with viral and bacterial
pneumonia (the International Classification of Diseases Statistical analysis
diagnosis codes: J10.0, J11.0, J12, J12.8, J12.9, J13,
J14, J15.0, J15.3, J15.7, J15.8, J15.9, J16.0, and J18.1) In this study, statistical analyses were performed
from January 1, 2013, through January 1, 2023. The using the IBM SPSS package program v. 25.0. The
data were obtained by screening electronic patient Kolmogorov-Smirnov test was used to evaluate the
files from the hospital information management system. normality of the data distribution. Categorical vari-
Ethical approval for the study was obtained from the ables were given as frequency and percentage, and
local ethics committee, and the study was conduct- continuous variables as mean and standard deviation.
ed in accordance with the tenets of the Declaration Categorical variables were analysed using the chi-
of Helsinki. square test, while continuous variables were analysed
using Student’s t-test for normally distributed vari-
Study population ables and the Mann-Whitney U test for non-normally
distributed variables. The area under the receiver
Using the hospital management system, a total operating characteristic (ROC) curves of SII for the
of 25,756 patients who presented to the emergency prediction of hospitalization, MV, and mortality were
department and were diagnosed with pneumo- calculated. The Youden J index was used to estimate
nia were identified for the study period, and those the optimal cut-off points. Sensitivity and specificity
aged > 18 years whose data and patient files were were calculated at the 95% confidence interval (CI).
available in the electronic system were included in the For all analyses, p < 0.05 was considered statisti-
study. Patients with malignancies, immunodeficiency, cally significant.
known renal or hepatic dysfunction, inflammatory bowel
disease or other inflammatory conditions, and pregnant
women (considering that their systemic immune status Results
might be affected by their current state) were excluded
from the study. Pre-existing diseases (especially chron- Table 1 shows demographics (age and gender)
ic renal failure, chronic liver failure, and inflammatory and laboratory parameters. Accordingly, the mean
bowel disease) that would affect the systemic immune age of the patients was 62.3 ± 17.3 years, and 57.8%

www.journals.viamedica.pl/medical_research_journal 187
MEDICAL RESEARCH JOURNAL 2023, vol. 8, no. 3

25,756 patients

7,965 patients with malignancies were excluded

17,791 patients

1,358 pregnant patients were excluded

16,433 patients

4,782 patients diagnosed with COVID-19 were excluded

11,651 patients

2,358 patients diagnosed with kidney failure, liver

failure, or inflammatory bowel disease were excluded

9,293 patients

2,491 patients with missing data were excluded

6,802 patients constituted the sample

Figure 1. Flow chart of patient selection

Table 1. Demographic characteristics and laboratory (n = 3,928) were male. Of the patients, 7.9% (n = 540)
parameters of the patients required MV, and 7.5% (n = 508) died.
Variable Mean ± SD (min–max); n (%) Table 2 shows the comparison of the demographic
Age (years) 62.3 ± 17.3 (18–102) characteristics, laboratory values, ​​and treatments of
the patients according to mortality. Gender did not
Gender
statistically significantly differ between the mortality and
Female 2,873 (% 42.2)
Male 3,928 (% 57.8) survivor groups (p > 0.05), but there were statistically
significant differences in relation to the remaining pa-
Neutrophil count (×103 μL) 7.16 ± 4.77 (0–67.08)
rameters (p < 0.001).
Platelet count (×103 μL) 274.40 ± 10,333 (1–1215) SII was found to predict inpatient treatment with
Lymphocyte count (×103 μL) 2.22 ± 7.52 (0.03–407.23) a sensitivity of 77.9% and a specificity of 36.2% at
WBC (μL) 10.34 ± 9.49 (0.11–440.51) a cut-off value of 114.73 [area under the curve (AUC):
0.654]. The ROC curve analysis showed that SII was
NLR 5.81 ± 8.30 (0–181.07)
statistically significant in determining inpatient treatment
PLR 1,553.54 ± 2,411.56 (0–41,102.13)
(p < 0.001, 95% CI: 0.639–0.669) (Fig. 2).
Type of treatment SII was found to predict the MV requirement of the
Outpatient 4,686 (68.9%) patients with a 70.0% sensitivity and 47.5% specificity
Inpatient 2,115 (31.1%)
at a cut-off value of 137.88 (AUC: 0.629). According to
MV requirement the ROC curve analysis, SII was statistically significant
Patients requiring MV 540 (7.9%)
Patients not requiring MV 6,261 (92.1%)
in determining the MV requirement (p < 0.001, 95% CI:
0.599–0.658) (Fig. 3).
Outcome
It was determined that at a cut-off value of
Mortality 508 (7.5%)
Discharge 6,293 (92.5%) 137.99 (AUC: 0.626), SII had a sensitivity of 70.1%
and a specificity of 47.4% in the prediction of mortality
SII 191.70 ± 164.68 (0.15–2.240)
among the patients. The ROC curve analysis revealed
SD — standard deviation; WBC — white blood cell; NLR — neutro-
that SII was statistically significant in determining mor-
phil-to-lymphocyte ratio; PLR — platelet-to-lymphocyte ratio; MV —
mechanical ventilation; SII — systemic immune-inflammation index tality (p < 0.001, 95% CI: 0.596-0.657) (Fig. 4).

188 www.journals.viamedica.pl/medical_research_journal
 Fatma Tortum et al., Predict the severity of pneumonia in the emergency department

Table 2. Comparison of demographic characteristics, laboratory values, ​​and treatments according to mortality
Variable Survivor group (n = 6,293) Mortality group (n = 508) P-value
Age (years) 61.5 ± 17.1 (18–102) 72.5 ± 15.6 (20–101) < 0.001
Gender 0.779
Female 3,638 (57.8%) 290 (57.1%)
Male 2,655 (42.2%) 218 (42.9%)
Neutrophil count (× 103 μL) 6.89 ± 4.37 (0–67.08) 10.45 ± 7.54 (0.01–48.32) < 0.001
Platelet count (×103 μL) 277.74 ± 100.78 (1–1215) 233.15 ± 123.68 (4–792) < 0.001
Lymphocyte count (×103 μL) 2.21 ± 5.88 (0.1–363.65) 2.35 ± 18.16 (0.03–407.23) < 0.001
WBC (μL) 10.05 ± 7.71 (0.20–375.88) 13.85 ± 21.39 (0.11–440.51) < 0.001
NLR 5.17 ± 7.13 (0–181.07) 13.75 ± 14.98 (0.02–102.29) < 0.001
PLR 1,408.80 ± 2,085.18 (0–41,102.13) 3,346.48 ± 4,534.25 (0.27–34,951.44) < 0.001
Type of treatment < 0.001
Outpatient 4,686 (74.5%) –
Inpatient 1,607 (25.5%) 508 (100%)
MV < 0.001
Patients requiring MV 34 (0.5%) 506 (99.6%)
Patients not requiring MV 6,259 (99.5%) 2 (0.4%)
SII 183.51 ± 149.08 (0.18–2,240) 293.22 ± 277.06 (0.15–2,231.82) < 0.001
WBC — white blood cell; NLR — neutrophil-to-lymphocyte ratio; PLR — platelet-to-lymphocyte ratio; MV — mechanical ventilation; SII — sys-
temic immune-inflammation index

ROC Curve ROC Curve

1.0 1.0

0.8 0.8
Sensitivity

Sensitivity

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
1- Specificity 1- Specificity

Figure 2. Receiver operating characteristic graph of the Figure 4. Receiver operating characteristic graph of the
systemic immune-inflammation index in the prediction of systemic immune-inflammation index in the prediction of
patients’ treatment types mortality among the patients

Discussion
ROC Curve
1.0
On completion of the study, it was determined
0.8
that SII was successful in distinguishing patients who
Sensitivity

0.6 required hospitalized treatment among those present-

ing to the emergency department with pneumonia. In
0.4
addition, it was observed that SII could assist in the
0.2 prediction of mortality and the MV requirement among
patients diagnosed with pneumonia.
0.0
0 0.2 0.4 0.6 0.8 1.0 A review of the literature shows that SII has been
1- Specificity frequently used in the evaluation of inflammatory con-
ditions and COVID-19 pneumonia. However, studies
Figure 3. Receiver operating characteristic graph of the on pneumonia cases other than COVID-19 pneumonia
systemic immune-inflammation index in the prediction of are limited. In one of these studies, Wang et al. used
patients’ mechanical ventilation requirements SII to predict the development of stroke-associated

www.journals.viamedica.pl/medical_research_journal 189
MEDICAL RESEARCH JOURNAL 2023, vol. 8, no. 3

pneumonia (SAP) and determine clinical severity in Conclusion

patients with intracranial haemorrhage (ICH) [11].
The authors argued that SII could predict the devel- According to the results of this study, SII was
opment of SAP and show the severity of the disease valuable in determining the MV requirement and mor-
in patients with ICH. However, an increased inflamma- tality among patients with pneumonia presenting to
tory response in cases of ICH may affect SII values; the emergency department. It is considered that SII
therefore, the presence of ICH in those patients may values ​​determined by a hemogram analysis, which is
have been a factor that increased SII [11]. Similarly, a low-cost and simple test, may be more feasible than
in a study by Jiang et al., SII was found to be signif- expensive, time-consuming tests for the determination
icant in predicting the development of postoperative of patients at the emergency department.
pneumonia in patients with non-small cell lung cancer
[12]. However, the presence of a history of malig- Limitations
nancy among their patients may have affected the
SII values. Examining long-term mortality in patients Our study has several limitations. Due to its retro-
with ischaemic stroke who developed SAP, Xie et spective nature, the data were obtained by screening
al. determined that SII values successfully predicted electronic patient files; therefore, patients who did not
long-term mortality [13]. In the present study, patients have a COVID-19 test result in their electronic files but
with conditions that could affect SII values, such as were infected with COVID-19 may have been over-
a history of malignancy, were excluded from the looked. In addition, the data of patients who were re-
study. In addition, pneumonia was the main reason ferred to the emergency department of the hospital from
for presentation to the emergency department among another hospital and received pneumonia treatment
the patients selected for this study. Therefore, this were not included in the electronic files. These factors
study differs from previous studies in which cases of may have affected the present SII results.
pneumonia that developed during another disease
were examined (11–13). The study’s patient selection
was similar to that of Acar et al., who investigated the Article information
use of SII values in determining clinical severity and
28-day mortality in patients with community-acquired Data availability statement: All data is available.
pneumonia [14]. Consistent with the study findings,
the authors concluded that SII values were​​ valuable Ethics statement: Ethical approval for the study was
in predicting mortality and clinical severity in patients obtained from the local ethics committee.
with pneumonia [14].
NLR and PLR, which are among the inflammatory Funding: None.
parameters used to evaluate patients with pneumonia
presenting to the emergency department in terms of
Acknowledgements: None.
clinical severity, mortality, and hospitalization, have
Conflict of interest: None.
also been the subject of many studies [15, 16]. Altas
et al. recommended the use of NLR and PLR together
Supplementary material: None.
with clinical scores in determining clinical severity
[15]. In a study by Enersen et al., NLR and PLR values
were reported to be associated with 90-day mortality
[16]. Although the evaluation of NLR and PLR was
References
not the primary objective of the present study, when
the effects of these parameters ​​on clinical severity 1. Ayan E, Unver H. Diagnosis of Pneumonia from Chest X-Ray Images
Using Deep Learning. 2019 Scientific Meeting on Electrical-Electronics
were evaluated, it was determined that they ​​were & Biomedical Engineering and Computer Science (EBBT). 2019, doi:
both also associated with mortality and the use of MV 10.1109/ebbt.2019.8741582.
2. Hincapié C, Ascuntar J, León A, et al. Community-acquired pneu-
during treatment. monia: comparison of three mortality prediction scores in the emer-
One of the factors that determines both the devel- gency department. Colomb Med (Cali). 2021; 52(4): e2044287, doi:
10.25100/cm.v52i4.4287, indexed in Pubmed: 35499040.
opment and clinical severity of pneumonia is advanced 3. Ko SH, Lee JS, Kim SK, et al. Serum cholesterol as a predictor of
age. Age is considered an important part of clinical mortality among the elderly patients with pneumonia in the emergency
department. Am J Emerg Med. 2021; 45: 404–409, doi: 10.1016/j.
scoring systems, such as PSI and CURB-65, used to ajem.2020.09.012, indexed in Pubmed: 33039214.
4. Kim MW, Lim JY, Oh SH. Mortality prediction using serum bio-
make the hospitalization decision in patients with pneu- markers and various clinical risk scales in community-acquired
monia [17]. In the current study, the patient group with pneumonia. Scand J Clin Lab Invest. 2017; 77(7): 486–492, doi:
10.1080/00365513.2017.1344298, indexed in Pubmed: 28678546.
a mortal course of pneumonia was found to be older
5. Papasidero ID, Valli G, Marin D, et al. Utility of Measuring Circulating
than the discharged patient group. Bio-Adrenomedullin and Proenkephalin for 30-Day Mortality Risk Pre-

190 www.journals.viamedica.pl/medical_research_journal
 Fatma Tortum et al., Predict the severity of pneumonia in the emergency department

diction in Patients with COVID-19 and Non-COVID-19 Interstitial Pneu- 2023; 14: 1115031, doi: 10.3389/fimmu.2023.1115031, indexed in
monia in the Emergency Department. Medicina (Kaunas). 2022; 58(12), Pubmed: 36860868.
doi: 10.3390/medicina58121852, indexed in Pubmed: 36557054. 12. Jiang R, Li P, Shen W, et al. The predictive value of the preoperative
6. Julián-Jiménez A, Castillo JG, Candel F. Usefulness and prognostic systemic immune-inflammation index in the occurrence of posto-
value of biomarkers in patients with community-acquired pneumonia in perative pneumonia in non-small cell lung cancer: A retrospective
the emergency department. Medicina Clínica (English Edition). 2017; study based on 1486 cases. Thorac Cancer. 2023; 14(1): 30–35, doi:
148(11): 501–510, doi: 10.1016/j.medcle.2017.04.033. 10.1111/1759-7714.14691, indexed in Pubmed: 36495040.
7. Öcal L, Keskin M, Cerşit S, et al. Systemic immune-inflammation index 13. Xie M, Yuan K, Zhu X, et al. Systemic Immune-Inflammation Index and
predicts in-hospital and long-term outcomes in patients with ST-segment Long-Term Mortality in Patients with Stroke-Associated Pneumonia.
elevation myocardial infarction. Coron Artery Dis. 2022; 33(4): 251–260, J Inflamm Res. 2023; 16: 1581–1593, doi: 10.2147/JIR.S399371,
doi: 10.1097/MCA.0000000000001117, indexed in Pubmed: 35044330. indexed in Pubmed: 37092129.
8. Jomrich G, Paireder M, Kristo I, et al. High Systemic Immune-Inflam- 14. Acar E, Gokcen H, Demir A, et al. Comparison of inflammation markers
mation Index is an Adverse Prognostic Factor for Patients With Gas- with prediction scores in patients with community-acquired pneumonia.
troesophageal Adenocarcinoma. Ann Surg. 2021; 273(3): 532–541, Bratisl Lek Listy. 2021; 122(6): 418–423, doi: 10.4149/BLL_2021_069,
doi: 10.1097/SLA.0000000000003370, indexed in Pubmed: 31425286. indexed in Pubmed: 34002616.
9. Seo M, Yamada T, Morita T, et al. P589Prognostic value of systemic 15. Altas OF, Kizilkaya M. The Effects of Neutrophil-Lymphocyte Ratio,
immune-inflammation index in patients with chronic heart failure. Platelet-Lymphocyte Ratio and Prognostic Markers in Determining
European Heart Journal. 2018; 39(suppl_1), doi: 10.1093/eurhe- the Mortality in Patients Diagnosed With Pneumonia in Intensive Care.
artj/ehy564.p589. Medeni Med J. 2021; 36(2): 130–137, doi: 10.5222/MMJ.2021.64160,
10. Candemir M, Kiziltunç E, Nurkoç S, et al. Relationship Between indexed in Pubmed: 34239765.
Systemic Immune-Inflammation Index (SII) and the Severity of Stable 16. Enersen CC, Egelund GB, Petersen PT, et al. The ratio of neutrophil-to-
Coronary Artery Disease. Angiology. 2021; 72(6): 575–581, doi: -lymphocyte and platelet-to-lymphocyte and association with mortality
10.1177/0003319720987743, indexed in Pubmed: 33685239. in community-acquired pneumonia: a derivation-validation cohort
11. Wang RH, Wen WX, Jiang ZP, et al. The clinical value of neutrophil- study. Infection. 2023 [Epub ahead of print], doi: 10.1007/s15010-
-to-lymphocyte ratio (NLR), systemic immune-inflammation index 023-01992-2, indexed in Pubmed: 36763284.
(SII), platelet-to-lymphocyte ratio (PLR) and systemic inflammation 17. Martin-Loeches I, Torres A. New guidelines for severe community-
response index (SIRI) for predicting the occurrence and severity of -acquired pneumonia. Curr Opin Pulm Med. 2021; 27(3): 210–215,
pneumonia in patients with intracerebral hemorrhage. Front Immunol. doi: 10.1097/MCP.0000000000000760, indexed in Pubmed: 33405483.

www.journals.viamedica.pl/medical_research_journal 191