Population-Based Analysis of Incidence and Mortality of Septic Shock

Septic shock is the most severe manifestation of sepsis and remains a leading cause of death and healthcare utilisation. Outcomes vary by patient mix (age, comorbidities), acquisition setting, infection source, and hospital performance, underscoring the need for robust, population-level surveillance to guide policy and care pathways. RADAR-Septic Shock leverages routinely collected administrative data to provide standardised measures of incidence, resource use, and outcomes across an entire regional health system.

The study aims to quantify the population incidence of septic shock and describe patient demographics, comorbidity burden, infection sources, organ failures, and use of life-support therapies; 2) estimate in-hospital and short-term mortality and length of stay; and 3) benchmark hospitals using risk-adjusted observed-to-expected (O/E) mortality to identify variation and inform improvement strategies.

It is a population-based cohort using the Catalan Minimum Basic Data Set for Acute Care Hospitals (CMBD-HA), an audited, mandatory discharge registry covering all acute-care hospitals in the public system. The analytic period spans January 2018-December 2023.

It will include all hospital discharges containing a primary or secondary diagnosis code for septic shock (ICD-10-CM R65.21) or post-procedural septic shock (T81.12XA). This definition prioritises specificity and uniformity across centres. Episodes will be stratified by acquisition setting (community- vs hospital-acquired).

Extracted variables include: demographics (age, sex); comorbidities and morbidity burden (Adjusted Morbidity Groups [GMA] index); infection source; organ failures; advanced life-support interventions (mechanical ventilation, renal replacement therapy, ECMO); ICU admission; length of stay; and mortality.

Primary outcome: in-hospital mortality. Secondary outcomes: 30-day mortality (where available), ICU admission, length of stay, and use of life-support therapies. Preliminary registry summaries show >31,000 episodes across 2018-2023 and, for 2023, an incidence of ~78 per 100,000 with in-hospital mortality ~33% overall (higher in hospital-acquired episodes). These figures motivate the benchmarking component.

Descriptive statistics will summarise case mix and care processes, including incidence rates per 100,000 inhabitants. Multivariable logistic regression will estimate individual mortality risk; model performance will be assessed (e.g., AUROC) and used to generate expected mortality for each hospital. Comparative performance will be reported via O/E mortality ratios with confidence intervals. Length of stay will be modelled after appropriate transformation. Sensitivity analyses will explore the impact of acquisition setting, infection source, organ failures, and comorbidity burden.

Based on audited counts, the registry includes 31,480 septic shock episodes over 2018-2023 (6,182 in 2023), ensuring sufficient precision for risk modelling and hospital-level benchmarking across the regional network.

Inclusion criteria

• Patients with a diagnosis of septic shock admitted to the hospitals of the public hospital network of Catalonia, Spain. Septic shock will be defined using the methodology described by Angus et al., which is currently referenced for population-based studies, consisting of coding a diagnosis of infection with acute organ failure, or sepsis or septic shock.

Exclusion criteria

• None