Abstract
Glucocorticoids have been shown to improve outcomes of patients with severe cases of Covid-19. However, criteria for prescribing glucocorticoids are currently limited. To identify potential for targeting, we perform an observational analysis of mortality of hospitalized patients. Our results agree with current clinical understanding that glucocorticoids benefit patients with severe cases of Covid-19, and that elevated Neutrophil/Lymphocyte Ratio (NLR) is associated with mortality. Furthermore, our results suggest that glucocorticoids could be targeted to patients with elevated NLR (especially in the range 6-25) at time of admission. Finally, we note there are also high-risk patients with low NLR, suggesting varying presentations of severe Covid-19.
Introduction
Early in the pandemic, clinical trials like RECOVERY established that glucocorticoids (GCs), such as dexamethasone, improved survival for patients with severe Covid-19. However, "severe" remains a broad clinical category, and prescribing criteria often rely on blunt metrics like oxygen requirements. This study aims to move toward precision medicine by identifying a specific biomarker—the Neutrophil-to-Lymphocyte Ratio (NLR)—that can predict which hospitalized patients will respond to steroids. NLR is a simple, low-cost metric derived from standard blood work reflecting the balance between innate inflammation and adaptive immune response.
Methodology
The researchers analyzed de-identified electronic health records from 3,108 patients hospitalized for Covid-19 between March and August 2020. To address "confounding by indication"—where sicker patients are more likely to receive steroids—they employed a two-stage machine learning procedure:
- Baseline Risk Modeling: A Generalized Additive Model (GAM) was trained on half the sample to predict mortality risk based on admission features before treatment. This model achieved a high AUC of 0.91.
- Excess Risk Estimation: On the remaining data, they calculated "excess risk" (actual mortality minus predicted risk).
- Treatment Effect Calculation: By comparing the average excess risk of patients receiving GCs within 24 hours of admission to those who did not, they estimated the Adjusted Risk Ratio (ARR).
Key Insights and Results
While the average treatment effect across the entire cohort was negligible (ARR = 0.96), the benefit of steroids proved highly heterogeneous when stratified by NLR. The study identified a specific "sweet spot" for treatment effectiveness.
The NLR "Sweet Spot" (6-25)
Patients with an NLR between 6 and 25 saw a significant survival benefit. In this range, the treatment resulted in a 26% reduction in the risk of death compared to similar untreated patients.
The Extremes
- Low NLR (<6): These patients likely do not experience the specific hyper-inflammation that steroids target, showing no significant benefit (ARR 0.99).
- Very High NLR (>25): While these patients are at extremely high risk, the data suggested steroids might be "too little, too late," or the sample size was insufficient to confirm a benefit.
As shown in the table above, the only statistically significant benefit (p=0.042) was observed in the 6-25 NLR group. Additionally, the researchers found high-mortality-risk patients with low NLR, suggesting that severe Covid-19 can manifest through pathways not primarily driven by neutrophil-heavy inflammatory storms.
Discussion and Limitations
This study suggests that NLR at admission is a powerful predictor of GC treatment success. However, as an observational study, it cannot perfectly replicate the gold standard of a randomized controlled trial (RCT). Rapidly changing treatment protocols during the first wave also introduced potential drift in clinical decision-making. Future randomized trials should consider stratifying patients by NLR to confirm these targeting criteria and refine precision treatment for severe Covid-19.
Summary updated