Association between corticosteroid use and 28-day mortality in septic shock patients with gram-negative bacterial infection: a retrospective study

Yi Dong, Gang Heng, Jianxin Zhang, Yanbing Shen, Zhen Lan, K. Wei, Weidong JinFrontiers in Medicine 2023 · DOI

TL;DR

While international guidelines cautiously recommend corticosteroids for septic shock, this retrospective study suggests they may actually be harmful for patients with gram-negative bacterial infections. The researchers found that steroid use was associated with increased 28-day mortality and a higher incidence of severe acute kidney injury (AKI), particularly in patients with lower baseline inflammatory markers.

What Problem Does This Paper Address?

Septic shock remains a leading cause of death in ICUs, with mortality rates hovering around 40%. The 2021 Surviving Sepsis Campaign (SSC) guidelines provide a "weak recommendation" for using low-dose corticosteroids to help reverse shock. However, clinical trials have yielded conflicting results regarding their impact on actual survival.

The authors hypothesize that the "one-size-fits-all" approach to steroids in sepsis is flawed. Specifically, they focused on gram-negative infections, which are characterized by the release of endotoxins (lipopolysaccharides). These endotoxins trigger a complex immune response that can lead to both hyper-inflammation and profound immunosuppression. The authors suspected that adding steroids to this specific biological environment might inadvertently worsen outcomes by over-suppressing the immune system or inducing organ damage like AKI.

Methods

The researchers conducted a retrospective cohort study using data from the MIMIC-IV database. To account for the fact that sicker patients are more likely to be prescribed steroids (confounding by indication), they employed several robust statistical techniques:

  • Propensity Score Matching (PSM): They matched 90 patients who received corticosteroids with 123 similar patients who did not, based on a wide array of baseline characteristics (age, SOFA score, comorbidities, etc.).
  • Causal Mediation Analysis: This is a sophisticated approach used to determine if an intermediate variable (in this case, AKI) actually "carries" the effect from the treatment (steroids) to the outcome (death).
  • Subgroup Analysis: They stratified patients by the Neutrophil-to-Lymphocyte Ratio (NLR), a common biomarker for systemic inflammation and immune status.

Datasets

The study utilized the Medical Information Mart for Intensive Care IV (MIMIC-IV version 2.0). This is a massive, de-identified dataset containing clinical records for over 40,000 patients admitted to the ICU at Beth Israel Deaconess Medical Center between 2008 and 2019. It is considered a gold standard for retrospective critical care research.

Key Insights and Ideas

  • The "Harm" Signal: In this specific subgroup of gram-negative septic shock, steroids were associated with a significant decrease in survival. This contrasts with some larger trials that showed no effect or slight benefits in general sepsis populations.
  • The NLR Threshold: The detrimental effect of steroids was most pronounced in patients with an NLR < 20. This suggests that if a patient's immune system isn't already in a state of extreme hyper-inflammation, steroids might push them into a "paralyzed" immune state, hindering their ability to clear the infection.
  • The AKI Connection: One of the most striking findings was that steroids were linked to a higher incidence of Stage 2 or 3 Acute Kidney Injury. The mediation analysis estimated that AKI accounted for approximately 21% of the association between steroid use and mortality. The authors suggest steroids might sensitize kidney cells to certain types of cell death (like ferroptosis) during sepsis.

Key Results

  • Mortality: In the matched cohort, patients receiving corticosteroids had significantly lower 28-day survival (Log-rank P = 0.028). In the raw data, mortality was 44% for the steroid group vs. 23% for the non-steroid group.
  • Hospital Mortality: Even after matching, hospital mortality remained higher in the steroid group (40% vs. 25%, P = 0.022).
  • Kidney Injury: The incidence of severe AKI (Stage 2/3) was significantly higher in the steroid group (78% vs. 58%, P = 0.003).
  • Mediation: The Average Causal Mediation Effect (ACME) for AKI was 2.4% (P = 0.018), confirming that kidney damage was a significant pathway through which steroids likely contributed to death.

Limitations and Future Work

The authors acknowledge several important caveats:

  • Retrospective Nature: As an observational study, it can show association but cannot definitively prove causation. There may be unmeasured confounders (e.g., specific resuscitation strategies) that influenced the results.
  • Single-Center Data: While MIMIC-IV is robust, it represents a single hospital system, which may limit the generalizability of the findings.
  • Sample Size: The subgroup analyses, while suggestive, involved relatively small numbers of patients.
  • Future Directions: The authors call for prospective, randomized controlled trials (RCTs) that specifically target gram-negative septic shock and use biomarkers like NLR to personalize corticosteroid administration. They emphasize that "precision medicine" in the ICU is necessary to identify who will benefit from steroids and who will be harmed.

Summary updated

Abstract

Systemic corticosteroid use could be harmful in septic shock patients with gram-negative bacterial infection, especially in patients with relatively low NLR.
BibTeX
@inproceedings{dong2023association,
  author = {Yi Dong and Gang Heng and Jianxin Zhang and Yanbing Shen and Zhen Lan and K. Wei and Weidong Jin},
  title = {{Association between corticosteroid use and 28-day mortality in septic shock patients with gram-negative bacterial infection: a retrospective study}},
  booktitle = {Frontiers in Medicine},
  year = {2023},
  doi = {10.3389/fmed.2023.1276181},
}
Figure