Introduction

In today’s healthcare landscape, one of the paramount challenges facing emergency departments (EDs) is the accurate risk stratification of elderly patients. As the population ages, EDs are witnessing a growing influx of older adults seeking medical attention, often presenting with complex and multifaceted health issues. Effectively assessing the risk associated with these patients is crucial, as it directly impacts clinical decision-making, resource allocation, and the overall quality of care provided [1,2,3]. In-hospital mortality for patients older than 65 was found to be between 8.2% and 22.8% when admitted for acute conditions [4,5,6].

Currently the risk stratification is performed through clinical scores that are specific for each disease (i.e., SOFA score, Ranson’s criteria), but general criteria that can help the clinician identify patients at high risk of death are lacking. As the rapid identification of high-risk patients is crucial for the early initiation of evidence-based life-saving treatment and management, new scores and biomarkers are being tested with this aim.

One of those is the neutrophil-to-lymphocyte ratio (NLR), which is a simple and inexpensive score calculated as a ratio between the neutrophil and lymphocyte counts. NLR is a biomarker that takes into account two faces of the immune system: the innate immune response, mainly due to neutrophils, and adaptive immunity, supported by lymphocytes [7, 8]. Several studies reported the prognostic value of NLR and its role as a marker of systemic inflammations in numerous pathologies such as chronic obstructive pulmonary disease (COPD), atrial fibrillation, coronary artery disease (CAD), sepsis, and acute pancreatitis [9,10,11,12,13,14,15,16,17,18,19,20,21]. Furthermore, the kinetic of this biomarker is really fast, NLR is thought to change within 6 h of the onset of physiologic stress, far earlier than other markers such as white cell count or C-reactive protein (CRP)[22].

Given those premises and the association with poor outcomes in different conditions, NLR could be useful in risk stratification of patients presenting to the ED with non-surgical conditions.

The aim of this study was to assess the value in the risk stratification of the NLR, its capability to predict all-cause mortality in patients presenting to the ED and establish a cut-off for elderly acute medical patients.

Methods

Study design and population

This was a single centre, prospective, observational study which included consecutive patients presenting to the ED of the Policlinico Universitario Umberto I in Rome between the 28th of April and the 28th of May 2022. Adult patients older than 65 years old presenting to the ED were enrolled after providing written informed consent. Patients presenting for a traumatic injury or non-traumatic surgical diseases and SARS-CoV-2 positive patients were excluded from the study. The study was carried out according to the principles of the Declaration of Helsinki and approved by the Sapienza University of Rome Ethics committee (Prot. 0405/2022). Reporting is in accordance with the STROBE statement. The authors designed the study, gathered, and analyzed the data anonymously, vouched for the data and analysis, wrote the manuscript, and decided to submit it for publication.

Relative NLR calculation

The relative neutrophil–lymphocyte ratio (NLR) was calculated as follows: (% neutrophil count, cells/μL)/(% lymphocyte count, cells/μL).

Routine clinical assessment

All patients underwent clinical assessment that included medical history, physical examination, standard blood tests, 12-lead electrocardiogram (ECG) and vital parameters at admission, everything was documented on the medical records. ED discharge diagnosis was prospectively documented. Management of patients was left to the discretion of the attending physician.

Triage and priority codes

At entry into the ED, patients are assigned a code with decremental priority ranging from 1 to 5 depending on their severity. In detail: Code 1: interruption or impairment of one or more vital functions with immediate access; Code 2: risk of impairment of vital functions. Condition with developmental risk or severe pain access within 15 min; Code 3: deferrable urgency stable condition without developmental risk with suffering and effect on the general state which usually requires complex services access within 60 min; Code 4: green minor urgency stable condition without developmental risk which usually requires simple single specialist diagnostic therapeutic services access within 120 min; Code 5: non-urgent problem not urgent or of minimum clinical relevance access within 240 min (source Italian Ministry of Health website: https://www.salute.gov.it/imgs/C_17_pubblicazioni_3145_allegato.pdf).

Follow-up and clinical endpoints

Information regarding death during follow-up were obtained from the patient’s hospital records and the national death registry. The primary endpoint was all-cause mortality at 30 days.

Statistical analysis

Continuous variables are presented as average and standard deviation (SD) or median and interquartile range [IQR] and compared using the student t- test or the Mann–Whitney U test as appropriate. Categorical variables are presented as count and percentages (%) and compared with the Pearson χ2 test or Fisher exact test, as appropriate.

The descriptive analysis was performed by comparing patients with a NLR below or above 8 and according to the vital status at 30 days. To calculate the odds ratio (OR) and its 95% confidence intervals (95% CI) for all-cause mortality at 30 days a univariable and multivariable logistic regression analysis was performed. In the multivariable model, we inserted significant variables from univariable regression analysis.

In detail, we inserted the following variables into the multivariable model: relative NLR > 8, female sex, age, SBP, shock, respiratory rate, swollen legs, atrial fibrillation (AF), altered mental status, active cancer, blood glucose, GOT/AST, hemoglobin, creatinine, LDH, and mean platelet volume (MPV). Other significant variables were not inserted if collinear (as an example total WBC, neutrophil or lymphocytes, CRP with NLR, dyspnoea with respiratory rate, insulin use with blood glucose).

We also built the receiver operating characteristic (ROC) curve with the value of the Youden index to test the predictive value of the NLR in the whole population and in the subgroup of women.

All hypothesis testing was two-tailed and P values of less than 0.05 were considered to indicate statistical significance without adjustments for multiple testing. Statistical analyses were performed using IBM SPSS 25 and MedCalc.

Results

Study population and characteristics

We included 1074 consecutive patients. Of these, 121 were excluded due to lack of personal data or for not having performed blood tests before discharge. Therefore, 953 patients remained available for analysis. Out of those, 267 (28%) had an NLR > 8.

The characteristics of the population, stratified by NLR values above or below 8, are shown in Table 1.

Table 1 Characteristics of the study population according to the NLR values
Full size table

When admitted to the ED, patients with NLR > 8, compared to those with a NLR < 8, were older and characterized by a more serious triage code (only 1% had a triage code = 4, while 72.6% had a triage code ≤ 2) and by an increased heart rate and body temperature. They more often presented with dyspnea, abdominal pain, falls and vomiting. Regarding comorbidities, there were no significant differences between the two groups.

Patients with a NLR > 8 were more frequently on chronic therapy with diuretics; they are also characterized by an increase in urea, creatinine, white blood cells, neutrophils, fibrinogen, D-dimer, glycemia, CRP, lactate dehydrogenase (LDH) and by a decrease in the estimated glomerular filtration rate (eGFR), of lymphocytes and monocytes (Table 1).

Mortality

A total of 27 patients were lost during follow-up, therefore, the survival analysis was performed on a cohort of 926 patients. In total, 142 (15.3%) patients died in-hospital or during the 30-day follow-up. Those patients who died during the follow-up were characterized by a more severe triage code, they were older and presented with a higher respiratory rate. In terms of symptoms and signs, the occurrence of an adverse outcome was more often associated with shock, dyspnea, swollen legs, vertigo, altered mental status, vomiting and diarrhea at presentation (Table 2). In our cohort, COPD, atrial fibrillation, dementia and active cancer were the comorbidities more commonly found in those patients who died during follow-up. The remaining characteristics stratified by vital status are presented in Table 2.

Table 2 Population characteristics according to vital status
Full size table

Association of NLR with mortality

In our cohort, the NLR as a continuous value reached an AUC of 0.70 (95%CI 0.67–0.73 p < 0.001) for the outcome of all-cause mortality at 30 days (on 926 patients for whom survival data were available) (Fig. 1).

Fig. 1
figure 1

ROC analysis for all-cause mortality. NLR Neutrophil to lymphocyte ratio

Full size image

This value was consistent in the group of women (n = 456, 71 deaths) with an AUC of 0.71, 95%CI 0.67–0.75 and a best cut-off of > 4.47. The optimal calculated cut-off for this variable in our cohort was 8.

To analyze the relationship between all-cause mortality and different predictors we built a logistic regression model. The OR of having a NLR > 8 was 2.563 (95%CI 1.595–4.118) with a p < 0.001. Another strong association with the outcome was found to be an altered mental status at presentation with an OR of 3.610 (95%CI 1.976–6.596). The results for the other predictors are shown in Table 3.

Table 3 Multivariable logistic regression analysis of factors associated with all-cause mortality
Full size table

Discussion

In this analysis including almost 1000 patients, we have described the characteristics of elderly patients presenting to the ED for acute medical conditions according to their NLR at presentation. We found that NLR is an independent predictor of mortality during 30 days of follow-up. In particular, we found an optimized cut-off of NLR > 8 specific for acute medical patients aged > 65 years, which differs from previous studies. Patients admitted with a NLR > 8 had specific characteristics with a generally more severe disease presentation.

In our cohort, the mortality rate at 30 days was 15.3%. In the literature, we found very heterogeneous reports about short-term mortality in elderly patients admitted to the ED, ranging from 8 to 25% [2, 23, 24].

Byrne DG et al. found in a cohort of 23,114 patients that 25.8% were over the age of 75 years. For this group, the 30-day mortality was 20.7% compared with 4.5% for younger patients [25]. In these studies, the main symptoms associated with adverse outcomes were delirium, hypoxemia, unconsciousness at presentation, female sex, age [23, 24]. Our results are in line with those previous findings, showing that respiratory rate, altered mental status, and systolic blood pressure were associated with an increased risk of mortality.

Our main finding relies on the clinically useful predictive value of the NLR in acute elderly medical patients. We used an optimized cut-off for NLR > 8, providing an AUC of 0.70. This value is in line with many other clinical scores currently used in clinical practice [26, 27].

In our cohort, 28% of patients had a NLR > 8. These patients had more severe symptoms and signs at presentation. Patients with a NLR > 8 showed a 2.5-fold higher risk of dying during the follow-up.

Different cut-off values for the NLR have been proposed in recent publications, depending on the clinical setting and underlying pathology [8]. Regolo M. et al. showed that in-hospital mortality in COVID-19 patients is predicted by a NLR > 11.38 (AUC = 0.77, p < 0.001) [26].

In patients older than 80 years hospitalized a NLR on admission > 2.97 (sensitivity of 92.6% and specificity of 52.5%, AUC = 0.714, p = 0.001) was an independent predictor of all-cause 3-month mortality [27]. Similarly, in a cohort of more than 5000 hospitalized geriatric patients an NLR higher than 7.95 was associated with in-hospital mortality with an AUC of 0.707 (0.686–0.728) [28]. This is in line with our results, showing that the NLR could be useful in risk stratification for all-cause mortality in elderly patients presenting to the ED for medical conditions. Furthermore, a test that could quickly classify nearly 30% of such patients as low-risk of short term mortality could be an additional support to emergency physicians for a safe discharge.

This laboratory value can be obtained quickly as part of the laboratory routine and is inexpensive. Given the lack of systematic risk stratification strategies, the proposal of the simple NLR in the setting of the emergency may be particularly useful.

Another interesting finding is the association of MPV with mortality. This is in line with previous studies in literature. In fact, an increased MPV seems to correlate with higher mortality in patients suffering from myocardial infarction, critically ill patients and in septic patients [29,30,31]. This could reflect a systemic inflammatory response, as inflammation, endothelial dysfunction and thrombotic conditions may alter platelet size [32].

Some limitations should be considered when interpreting these findings. First, patients with traumatic injuries were excluded from this study, therefore, we cannot generalize our findings to those patients. Then, the value of NLR can be falsely increased in some patients. As described by Karakonstantis et al., factors such as age, exogenous steroid intake, endogenous sexual hormones, active hematological disorders, such as leukemia, cytotoxic or granulocyte colony-stimulating factor chemotherapies, and HIV can determine an increase of the NLR [33].

In conclusion, NLR is a simple tool that is associated with the severity of presentation and all-cause mortality; therefore, elderly patients with a NLR > 8 should be early identified as being at higher risk of adverse outcomes and promptly managed.