Daily Sepsis Research Analysis
A mechanistic study identifies a glutamatergic circuit in the ventrolateral periaqueductal gray that drives hypothermia and cardiovascular depression during systemic inflammation, offering a neural basis for the hypometabolic phenotype seen in septic shock. A large prospective pediatric cohort from low-income settings shows hospital readmission is not a reliable surrogate for postdischarge mortality after sepsis. Finally, interpretable EHR-based models accurately reproduce risk-adjusted postoper
Summary
A mechanistic study identifies a glutamatergic circuit in the ventrolateral periaqueductal gray that drives hypothermia and cardiovascular depression during systemic inflammation, offering a neural basis for the hypometabolic phenotype seen in septic shock. A large prospective pediatric cohort from low-income settings shows hospital readmission is not a reliable surrogate for postdischarge mortality after sepsis. Finally, interpretable EHR-based models accurately reproduce risk-adjusted postoperative infection metrics, including sepsis/septic shock, enabling scalable surveillance.
Research Themes
- Neuroimmune circuits and thermometabolic control in sepsis
- Postdischarge outcomes and metrics in pediatric sepsis in low-resource settings
- Data-driven, interpretable surveillance for sepsis and postoperative infections
Selected Articles
1. An excitatory circuit in the ventrolateral periaqueductal gray drives hypometabolic state during acute systemic inflammation.
Using activity-dependent tagging in LPS and CLP models, the authors identified a glutamatergic neuron population in the vlPAG that causally drives hypothermia and cardiovascular depression during systemic inflammation. Optogenetic manipulation supported a functional role for this circuit in the hypometabolic state linked to septic shock mortality.
Impact: This study uncovers a defined central circuit controlling the hypometabolic response to systemic inflammation, a key determinant of septic shock outcomes. It opens avenues for neuromodulatory strategies targeting thermometabolic control in critical illness.
Clinical Implications: While preclinical, these findings suggest that targeted neuromodulation of the vlPAG or its downstream pathways could mitigate hypothermia and cardiovascular depression in septic shock. They also caution that indiscriminate induction of hypothermia may have neural circuit–level consequences.
Key Findings
- Activity-dependent genetic labeling in LPS and CLP models identified a discrete glutamatergic neuron population in the vlPAG activated by systemic inflammation.
- Optogenetic stimulation of the identified vlPAG neurons drove hypothermia and cardiovascular depression, indicating a causal role in the hypometabolic state.
- The work provides a central neural mechanism linking systemic inflammation to thermometabolic and cardiovascular suppression relevant to septic shock mortality.
Methodological Strengths
- Use of two complementary systemic inflammation models (LPS and CLP) to generalize findings.
- Activity-dependent genetic labeling and optogenetics enabling causal circuit interrogation.
Limitations
- Findings are limited to murine models; human translational relevance remains to be established.
- The abstract does not detail downstream targets or broader network interactions.
Future Directions: Map downstream projections and inputs to the vlPAG circuit, test neuromodulatory interventions in severe infection models, and explore biomarkers of circuit engagement for potential translation.
Hypometabolism, characterized by hypothermia and cardiovascular depression, is associated with higher mortality in patients with septic shock. However, the neural substrates underlying the hypometabolic state during systemic inflammation remain poorly understood. Here, using activity-dependent genetic labeling of neurons activated by lipopolysaccharide (LPS) administration and cecal ligation and puncture (CLP) in mice, we identified a discrete population of glutamatergic neurons in the ventrolateral periaqueductal gray (vlPAG) that drives hypothermia and cardiovascular depression. Optogenetic stimulation of vlPAG
2. Diverging pathways: exploring the interplay between hospital readmission and postdischarge mortality in paediatric sepsis in low-income settings.
In 6074 children with suspected sepsis discharged from hospitals in low-income settings, 6.2% died and 18.2% were readmitted, with deaths occurring earlier (median 28 days) than readmissions (79.5 days). Malnutrition, HIV, and unplanned discharge strongly predicted postdischarge mortality but not readmission, indicating readmission is a poor surrogate for mortality in this context.
Impact: By disentangling determinants of mortality versus readmission, this study challenges the use of readmission as a primary performance metric in pediatric sepsis programs in low-resource settings.
Clinical Implications: Programs should prioritize mortality reduction with targeted postdischarge care (e.g., addressing malnutrition, HIV, and discharge planning) rather than relying on readmission rates as a proxy outcome.
Key Findings
- Among 6074 discharged children with suspected sepsis, 6.2% died and 18.2% were readmitted; deaths occurred earlier (median 28 days) than readmissions (79.5 days).
- Malnutrition (aHR 5.58), HIV (aHR 1.89), and unplanned discharge (aHR 3.31) strongly predicted postdischarge mortality but not readmission (aSHR ≤ 0.81).
- Readmission-to-mortality rate ratio was 3.12 overall and increased over time due to declining mortality, underscoring that readmission is not a reliable surrogate for mortality.
Methodological Strengths
- Prospective, multisite cohort with large sample size and standardized analyses.
- Appropriate use of competing risks (Fine-Gray) and Cox models to disentangle outcomes.
Limitations
- Observational design limits causal inference and may be affected by unmeasured confounding.
- Generalizability may be limited to similar low-income settings and health system contexts.
Future Directions: Develop and test targeted postdischarge intervention bundles for high-risk children (e.g., malnourished, HIV-positive, unplanned discharges) and evaluate mortality as the primary endpoint.
BACKGROUND: Mortality and readmission rates are high in low-income countries following hospital discharge; however, few studies have studied the relationship between these outcomes. Hospital readmission is a complex outcome as it reflects illness severity and health-seeking behaviour. This study aims to better understand the heterogeneous nature of hospital readmission, especially as it pertains to mortality. METHODS: Secondary analysis of a prospective, multisite, observational cohort study included children aged 0-60 months old admitted to hospital with suspected sepsis. We used Fine-Gray models and Cox proportional hazards regression to identify and contrast risk factors for readmission and postdischarge mortality. We also compared the risk ratio of the two outcomes across several domains, including diagnosis, postdischarge time period and study site. RESULTS: Of 6074 children discharged, 376 (6.2%) died, while 1106 (18.2%) were readmitted shortly after discharge. The median time to death and readmission was 28 (IQR: 9-74) and 79.5 (IQR: 30-130) days, respectively. A few patient characteristics, such as prior care seeking and hypoxaemia, were associated with both mortality and readmission. However, other characteristics, such as malnutrition (adjusted HR (aHR): 5.58 (95% CI: 4.20 to 7.43)), HIV (aHR: 1.89 (95% CI: 1.20 to 2.98)) and unplanned discharge (aHR: 3.31 (95% CI: 2.61 to 4.21)), were strongly predictive of postdischarge mortality but not readmission (aSHR: 0.67 (95% CI: 0.56 to 0.81), 0.64 (95% CI: 0.40 to 1.00) and 0.81 (95% CI: 0.67 to 0.98), respectively). The overall rate ratio of readmission to postdischarge mortality was 3.12 (95% CI: 2.77 to 3.50) and increased over time, mostly due to decreasing mortality. CONCLUSIONS: Readmission as an outcome measure reflects perceived illness severity, health system capacity and complex healthcare-seeking behaviour. Unlike mortality, readmission is not a reliable surrogate for recurrent illness and should not be used as a primary measure of impact for programmes aiming to improve postdischarge outcomes.
3. Estimation of risk-adjusted postoperative infection outcomes using interpretable machine learning and electronic health record data.
Across five hospitals and four infection types (including sepsis/septic shock), parsimonious EHR-based models produced hospital-level O/E ratios that closely matched ACS-NSQIP manual surveillance (r=0.77; mean absolute difference 0.13%). Agreement improved as infection incidence increased, supporting automated, interpretable surveillance for quality monitoring.
Impact: Demonstrates that interpretable, automated EHR models can accurately approximate gold-standard infection surveillance, lowering the burden of manual chart review while maintaining fidelity for sepsis/septic shock metrics.
Clinical Implications: Hospitals can deploy EHR-based, risk-adjusted infection surveillance to track sepsis/septic shock, SSI, UTI, and pneumonia outcomes, enabling near-real-time quality improvement and benchmarking with reduced resource requirements.
Key Findings
- EHR-based models yielded hospital-level observed-to-expected infection ratios closely aligned with ACS-NSQIP manual surveillance (Pearson r=0.77; mean absolute difference 0.13%).
- Agreement improved for infection types with higher incidence, suggesting robustness where monitoring burden is greatest.
- Findings generalized across five large hospitals, nine surgical specialties, and four infection types, including sepsis/septic shock.
Methodological Strengths
- Very large, multicenter dataset with 307,335 patients and 441,047 operations.
- External benchmarking against ACS-NSQIP with overlapping confidence intervals for O/E ratios.
Limitations
- Retrospective design and dependence on EHR data quality and coding accuracy.
- Generalizability beyond the five participating hospitals and specific EHR implementations may be limited.
Future Directions: Prospective deployment to drive real-time quality improvement, evaluation across diverse health systems, and refinement for low-incidence infections and bias mitigation.
BACKGROUND: This study compared risk-adjusted postoperative infection outcomes estimated by statistical models applied to electronic health record (EHR) data ("automated") to gold-standard manual chart review outcomes estimated by the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP). METHODS: 307,335 adult patients who underwent 441,047 operations in nine surgical specialties at five large hospitals between 2013-2019 were included. Records from 30,603 patients were linked to the local ACS-NSQIP database (97% linkage). Previously published models for estimating preoperative risk and occurrence of postoperative infections were used to estimate observed-to-expected event ratios (O/E) for surgical site infections, urinary tract infections, sepsis/septic shock, and pneumonia. RESULTS: O/E ratios were similar when comparing automated methods to ACS-NSQIP across 5 hospitals and 4 infection types. The Pearson correlation coefficient of the hospital O/E ratios was 0.77, mean absolute difference was 0.13%, and 100% of the confidence intervals were overlapping. The correlations and mean absolute differences for individual infection types improved as incidence rates increased. DISCUSSION: Parsimonious statistical models applied to EHR data can be used to accurately estimate hospital risk-adjusted postoperative infection outcomes. CONCLUSIONS: These models could be used to augment postoperative infection surveillance for hospital quality monitoring.