Predictive value of the neutrophil-to-lymphocyte ratio in the prognosis and risk of death for adult sepsis patients: a meta-analysis

Abstract

Background: The neutrophil-to-lymphocyte ratio (NLR) is a commonly used biomarker for acute inflammation that often rises during sepsis, making it a valuable diagnostic indicator for clinical practice. However, no consensus has been reached on the prognostic value of NLR for predicting the prognosis and mortality risk in adult sepsis patients. In light of this controversy, we conducted a meta-analysis to clarify the prognostic significance of NLR in adult sepsis patients. The meta-analysis was registered in the PROSPERO database (registration number CRD42023433143).

Methods: We performed a comprehensive literature search in PubMed, Cochrane Library, Ovid, and Springer databases, using retrieval terms "sepsis" or "septic shock" and "prognosis" or "mortality" for studies published between January 1, 2000, and May 31, 2023. Children and neonates with sepsis were excluded from our research. Two independent researchers conducted the literature search and data extraction. Consensus was reached when discrepancies occurred, and in case of persistent discrepancies, the final decision was made by the research supervisor. The hazard ratio (HR) and its corresponding 95% confidence interval (95% CI) were extracted from each study included in the analysis. A random-effects model was used to synthesize all HRs and their 95% CIs. Sensitivity analysis was performed to investigate heterogeneity. Sensitivity analysis was conducted to identify studies that had a significant impact on the overall results of the meta-analysis. Subgroup analysis and meta-regression were performed to explore sources of heterogeneity. Egger's test was also used to investigate publication bias in this meta-analysis.

Results: After a comprehensive literature search and screening, we included 12 studies comprising 10,811 patients for the meta-analysis. The pooled results indicated that patients with a higher NLR level were associated with a poor prognosis (Random-effects model, HR: 1.6273, 95% CI: 1.3951-1.8981). Heterogeneity testing showed significant heterogeneity (I^{2 =} 87.2%, 95% CI: 79.5-92, p<0.0001). Sensitivity analysis was performed to investigate the sources of heterogeneity, which revealed that the omission of one highly sensitive study significantly reduced the I² value. After removing this study, a strong association was found between a higher NLR level and poor prognosis and risk of death in adult sepsis patients (Random-effects model, HR: 1.6884, 95% CI: 1.4338-1.9882). Both subgroup analysis and meta-regression indicated that the study design and testing time of NLR were sources of heterogeneity. Egger's test showed no obvious publication bias in this meta-analysis.

Conclusion: NLR is a reliable and valuable biomarker for predicting prognosis and the risk of death in adult sepsis patients.

Systematic review registration: [https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023433143] PROSPERO, identifier [CRD42023433143].

Keywords: NLR; adult sepsis; meta-analysis; mortality; prognosis.

Figure 1

Flowchart for selection of studies included in this meta-analysis based on PRISMA guidelines.

Figure 2

Forest plot of preliminary studies synthesis. HR, Hazard Ratio; SE, Standard Error; NLR, neutrophil-to-lymphocyte ratio. Red squares represent the point estimates of the HR of each study, with 95% CI indicated by horizontal bars. Black diamond represent the summary estimate from the pooled studies with 95%CI.

Figure 3

Sensitivity analysis of this meta-analysis. Red square are the point estimates of the omitting HR, with 95% CI indicated by horizontal bars. While black diamond is the heterogeneity from the pooled studies with 95% CI base on random effects model.

Figure 4

Forest plot of re-synthesis after eliminating one study identified by sensitivity analysis. HR, Hazard Ratio; SE, Standard Error; NLR, neutrophil-to-lymphocyte ratio. Red squares represent the point estimates of the HR of each study, with 95% CI indicated by horizontal bars. Black diamond represent the summary estimate from the pooled studies with 95% CI.

Figure 5

Subgroup analysis sepsis patients’ of mean age (A) and cut-off value of the NLR (B). (A).Red squares represent the point estimates of the HR of each study according to mean age subgroup grouping, with 95% CI indicated by horizontal bars. Upper black diamond showed the summary estimate from the pooled studies with >60 years old population demographic characteristics, the middle black diamond showed the pooled studies with ≤60 years old, and the lower black diamond indicated the total summary estimate of the HR of all studies.(B). Red squares represent the point estimates of the HR of each study according to cut-off value of the NLR subgroup grouping, with 95% CI indicated by horizontal bars. Upper black diamond showed the summary estimate from the pooled studies with cut-off value of the NLR ≤10, middle black diamond showed the pooled studies with cut-off value of the NLR >10, and the lower black diamond indicated the total summary estimate of the HR of all studies.

Figure 6

Subgroup analysis of study design (A) and testing time of NLR (B). (A).Red squares represent the point estimates of the HR of each study according to study design subgroup grouping, with 95% CI indicated by horizontal bars. Upper black diamond showed the summary estimate of the retrospective studies, the middle black diamond showed the pooled studies of prospective studies, and the lower black diamond indicated the total summary estimate of the HR of all studies. (B). Red squares represent the point estimates of the HR of each study according to testing time of NLR subgroup grouping, with 95% CI indicated by horizontal bars. Upper black diamond showed the summary estimate of the studies with testing time of NLR not on the first day of hospitalization, the middle black diamond showed the pooled studies with testing time of NLR on the first day of hospitalization, and the lower black diamond indicated the total summary estimate of the HR of all studies.

Figure 7

Enhanced contour funnel plot and Egger test for publication bias. (A) Enhanced contour funnel plot indicated that all of the studies fell into the blue region, which mean that funnel plots was basically symmetrical; (B) Egger test with a result of p=0.1535 illustrated that the publication bias of this meta-analysis was not obvious.