Primary and secondary care patient health record data linked to examine risk factors, mortality and antibiotic use in sepsis

Student thesis: Phd


Background: Sepsis is a serious and potentially life-threatening condition that occurs as a result of a dysregulated immune response to an infection. It causes significant damage to organ systems and tissues around the body and is associated with high mortality and long-term health implications. Antibiotics are vital in the management of sepsis, but in the face of increasing resistance of microorganisms there is a need to reduce and optimise consumption. Aims: To use sources of primary and secondary care data to explore the role of antibiotics in sepsis patients. Specific aims were to (i) evaluate risk factors for developing sepsis, including demographics, comorbidities and prior antibiotic use, and to see if these risk factors differ between patients with community- and hospital-acquired sepsis. (ii) to estimate the burden of antimicrobials in sepsis patients and to relate it to short-term mortality. (iii) to evaluate the longer-term outcomes following a sepsis episode, including mortality, hospital readmission and antibiotic use, and to see if these differ between community- and hospital-acquired sepsis. Methods: The studies used anonymised patient-level routinely collected electronic health record data from a US intensive care unit (MIMIC-III) and linked data from UK primary and secondary care (CPRD & HES). Sepsis cases were identified using either diagnostic codes or the Sepsis-3 criteria and in two of the studies were matched to population or hospital controls. Multivariable conditional logistic regression modelling was used to assess the risk factors associated with developing sepsis and time-to-event analysis including Kaplan-Meier and Cox proportional hazards regression modelling to evaluate the short-and longer-term outcomes. Findings: The key findings of this work were that prior antibiotic use and comorbidity levels were associated with an increased risk of developing sepsis, but there were differences between patients with community- and hospital-acquired sepsis. During a sepsis ICU admission patients were exposed to an average of three different antimicrobials, with lower number of antibiotics associated with a lower risk of 30-day mortality. Patients who survived a sepsis episode experienced high rates of hospital readmission and mortality in the three years after hospital discharge, as well as increased rates of common infections and antibiotic use. There were also differences in outcomes after the sepsis episode between patients who developed sepsis in the community and hospital. Conclusion: These studies are the first to use linked sources of primary and secondary care data and found substantive differences in risk factors and outcomes between patients with community- and hospital-acquired sepsis as well as differences by level of antibiotic use. These findings could help inform tools to help identify patients at risk of sepsis and optimise antibiotic use in these patients, which could benefit individual patients and help reduce overall consumption of antibiotics.
Date of Award31 Dec 2023
Original languageEnglish
Awarding Institution
  • The University of Manchester
SupervisorPaul Dark (Supervisor), Timothy Felton (Supervisor) & Tjeerd Van Staa (Supervisor)


  • antibiotics
  • sepsis
  • electronic health records

Cite this