High neutrophil-to-lymphocyte ratio associated with progression to critical illness in older patients with COVID-19: a multicenter retrospective study

Demographic and epidemiologic characteristics

In this study, 232 older (≥60 years) patients with confirmed COVID-19 were enrolled from January 17, 2020 to March 3, 2020 in Zhejiang province. Patients’ clinical outcomes were followed-up until April 13, 2020. As shown in Table 1, the median ages in the mild, severe, and critical disease groups were 66 years (interquartile range [IQR]: 63-70), 66 years (IQR: 62-71) and 72 years (IQR: 68-81). The critical group showed a significantly higher age than the mild and severe groups (P<0.001). The proportions of hypertension and heart disease in the critical group were 72.41% and 55.17%, respectively, which were significantly higher than those noted in the mild and severe groups (P<0.001). One case (0.71%) with mild disease, two (3.14%) with severe disease, and six (20.69%) with critical disease had chronic obstructive pulmonary disease (COPD) (P<0.001). There were no significant differences in the other coexisting medical conditions across the three groups, including the rates of diabetes, asthma, cancer, chronic liver disease, chronic renal disease, and immunosuppression.

Clinical features and laboratory abnormalities

On admission, the majority of cases showed decreased or normal leucocyte levels in all subtypes, as shown in Table 2. The median neutrophil levels in the mild, severe, and critical groups were 3.22×10⁹ /L [IQR: (2.59-4.20) ×10⁹], 3.50×10⁹/L [IQR: (2.70-4.80) ×10⁹], and 6.65×10⁹/L [IQR: (3.51-9.70) ×10⁹], respectively; the critical group showed significantly higher values than the mild and severe groups (P<0.001). The median lymphocyte levels in the mild, severe, and critical groups were 1.26×10⁹ /L [IQR: (0.90-1.60) ×10⁹], 0.98×10⁹/L [IQR: (0.70-1.26) ×10⁹], and 0.54×10⁹/L [IQR: (0.45-0.80) ×10⁹], respectively. The critical group showed significantly lower values than the mild and severe groups (P<0.001). The platelet levels were lower in the critically group than the mild and severe groups, but were still within the normal range. The levels of lactate dehydrogenase, creatinine, C-reactive protein, and procalcitonin increased with increasing illness severity (P<0.05). There were no significant differences in the blood test results across the three groups, including the values of albumin, alanine aminotransferase, aspartate aminotransferase, total bilirubin, potassium, sodium, and blood urea nitrogen. Multiple mottling and ground-glass opacity were typical imaging manifestations noted in patients with COVID-19, and their prevalence rates in the mild, severe, and critical groups were 24.29%, 42.86%, and 68.97%, respectively (P<0.001).

Treatment and outcomes

All patients were isolated in designated hospitals and received supportive care as well as the currently recommended medications. As shown in Table 3, 135 cases (84.77%), 60 cases (95.24%), and 29 cases (100%) received antiviral treatment, including interferon-α sprays, arbidol hydrochloride capsules, and lopinavir and ritonavir tablets in the mild, severe, and critical groups, respectively (P=0.504). The durations from illness onset to antiviral therapy initiation were 4 days (IQR: 2.0-7.0), 5 days (IQR: 1.5-8.5), and 4 days (IQR: 2.0-8.0) in the mild, severe, and critical groups, respectively (P=0.390). With increases in the illness severity, the proportion of the use of glucocorticoids and intravenous immunoglobins rose (P<0.001). Ten patients received extracorporeal membrane oxygenation (ECMO) therapy, and six underwent continuous renal-replacement therapy (CRRT) in the critical group; none of the patients received ECMO therapy and only one underwent CRRT in the severe group. Three patients had shock in the critical group, while there were no cases with shock in the mild and severe groups (P<0.001). The viral RNA shedding durations were 16 days (IQR: 12-22), 17 days (IQR: 14-21), and 25 days (IQR: 17-30) in the mild, severe, and critical groups, respectively (P<0.001).

By April 13, one patient had died, two had received lung transplantation, and eight remained hospitalized in the critical group. By May 27, among the eight patients who were still hospitalized, two withdrew from the ECMO treatment and were transferred to the general ward, while the other six patients were still receiving the ECMO therapy. In the other two groups, all patients had survived and were discharged. The number of days of hospitalization were 18 days [IQR: 14-23], 22 days [IQR: 19-26], and 32 days [IQR: 21-68] in the mild, severe, and critical groups, respectively (P<0.001).

Risk factors associated with progression to critical illness

Univariate Cox regression was used to analyze the risk factors for critical illness in the older patients with COVID-19, as shown in Table 4. Older age was shown to increase the likelihood of critical illness even in older patients (≥60 years) (hazard ratio [HR] 1.107, confidence interval [CI] 1.065-1.151, P<0.001). Shortness of breath as a symptom (HR 11.328, CI 5.370-23.894, P<0.001), and comorbidities including hypertension (HR 3.563, CI 1.578-8.047, P=0.002), heart disease (HR 9.638, CI 4.626-20.081, P<0.001), and COPD (HR 7.108, CI 2.891-17.481, P<0.001) were predictive of critical illness. The increasing odds of critical illness development in patients with COVID-19 were associated with higher NLR values (HR 1.157, CI 1.117-1.199, P<0.001), lower albumin levels (HR 0.875, CI 0.807-0.950, P<0.001), higher C-reactive protein levels (HR 1.012, CI 1.005-1.020, P=0.002), and multiple mottling and ground-glass opacity (HR 4.573, CI 2.082-10.045, P<0.001). In the multivariate analysis, only older age (HR 1.121, CI 1.070-1.174, P<0.001), heart disease (HR 2.587, CI 1.156-5.787, P=0.021), higher NLRs (HR 1.136, CI 1.094-1.180, P < 0.001), and multiple mottling and ground-glass opacity (HR 4.518, CI 1.906-10.712, P<0.001) remained predictors of critical illness when the other factors in the model were kept constant.

Association of the NLR with progression to critical illness

Figure 1A shows the association between the NLR and progression to critical illness, as identified using a Cox proportional hazards model adjusted for the baseline covariates. For the sensitivity analysis, we converted the NLR from a continuous variable to a categorical variable (the quartile of NLR), and the P for trend of the NLR with categorical variables in the fully adjusted model (model II) was consistent with that obtained when the NLR was a continuous variable. The relationship between the NLR and progression was significant and graded (HR: 1.16 per unit; 95% CI: 1.10-1.22; P<0.001). When adjusted for sex and age, the ratio of the highest quartile of the NLR compared to the lowest quartile was 33.017 (95% CI 4.436-245.732, P <0.001), and in the fully adjusted model, the odds of the NLR as a clinical risk factor was 21.755 (95% CI 2.854-165.860, P<0.001) (Table 5). Figure 1B shows the Kaplan-Meier analyses graphs for progression to critical illness based on the quartiles of the NLR.

Figure 1. Association between the neutrophil-to-lymphocyte ratio (NLR) and progression to critical illness. (A) Adjusted hazard ratio (HR) for progression to critical illness according to the NLR. (B) Cumulative probability of progression to critical illness with increasing NLR values.

Patients

This retrospective study, focusing on the epidemiological and clinical characteristics of older (age≥60 years) patients with confirmed COVID-19, was conducted from January 17 to March 3, 2020. All the enrolled cases showed real-time reverse transcriptase polymerase chain reaction (RT-PCR) positivity for SARS-CoV-2, and were retested several times during their hospitalization. Data were collected uniformly by the Health Commission of Zhejiang Province, wherein all patients were assigned to specific hospitals for unified treatment according to Zhejiang Province’s emergency rule. The diagnosis of COVID-19 infection was based on the interim guidance of the World Health Organization (WHO) [33], and all data were shared with the WHO, with the primary analytic results reported to the authority of Zhejiang Province. Since the collection and analysis of all cases were determined by the Health Commission of Zhejiang Province under national authorization and considered as part of the continuing public health outbreak investigation, our retrospective study was exempt from institutional review board approval.

The subtype definition of COVID-19 patients was based on the diagnosis and treatment scheme for COVID-19 in China, based on a minor modification of WHO standards [34]. The degree of COVID-19 was categorized as mild, severe, or critical: the mild type included non-pneumonia and mild pneumonia cases, and the severe type was characterized by dyspnea, respiratory frequency ≥30 min, blood oxygen saturation ≤93%, PaO2/FiO2 ratio <300, and/or rate of lung infiltration >50% within 24–48 h. Critical cases were those that exhibited respiratory failure, septic shock, and/or multiple organ dysfunction/failure.

Procedures

We obtained epidemiological, demographic, laboratory, clinical, management, and outcome data from patients’ medical records. Data were retrieved and reviewed by two independent observers. Clinical outcomes were followed-up until April 13, 2020. Missing or unclear data were confirmed by direct communication with healthcare providers. Throat swab specimens obtained from the upper respiratory tract and sputum of all patients were collected at admission. Laboratory confirmation of COVID-19 was performed at the First Affiliated Hospital at Zhejiang University, under the authorization of the Centers for Disease Control and Prevention at the Zhejiang Province/city level, by previously reported RT-PCR methods. All patients underwent chest CT at admission. Patients with other common respiratory viruses, including respiratory syncytial virus, parainfluenza virus, influenza A and B virus, and adenovirus were excluded from this study.

Data collection

In this study, we collected data on epidemiology, anthropometrics, demographics, as well as symptoms and signs at the time of admission to the hospital. We analyzed the blood collected within 48 hours of admission. Additional data collected included those on the results of laboratory tests and chest CT, comorbidities, co-infection with other respiratory pathogens, treatment (including drugs, intensive care and mechanical ventilation), and other clinical outcomes.

Statistical analysis

Continuous variables are expressed as medians (range), and were compared using t tests or Mann-Whitney U tests, and categorical variables were compared using chi-squared tests or Fisher’s exact tests. Follow-up was initiated on the day of admission, and ended at the patient’s death or until the last follow-up. The Kaplan-Meier method was used to evaluate the cumulative rate of progression to critical illness, and a log-rank test was used to assess differences between groups. HRs were calculated using the Cox regression model. Variables with P < 0.05 in the univariate analysis were included in a stepwise Cox proportional hazards regression model. We performed tests for linear trend by entering the median value of each quartile of the NLR as a continuous variable in the models. The Cox proportional hazards model was used to estimate the HRs associated with the NLR for the risk of progression to critical illness with adjustment for pertinent variables. The HRs and 95% CIs of the progression to critical illness in each subgroup were estimated, and their interactions tested. Statistical analyses were conducted using SPSS version 26.0 (IBM Corporation, Armonk) and R version 3.4 (R Foundation). A two-sided P value < 0.05 was considered to indicate statistical significance.