Outcome Prediction and Alternative Interventions in Pleural Infection

Introduction
Pleural infection is common with considerable healthcare burden, requiring prolonged antibiotics and multiple interventions comprising chest tube drainage, intrapleural therapy or thoracic surgery. Patient outcomes remain poor and current treatment pathways are insufficient. Using mixed methods, this thesis evaluates outcome prediction markers and the potential for redrawing the existing treatment pathway toward earlier escalation of therapy, understanding patient priorities, and assessing the complications of existing therapies.

Methods
Analysis of prospectively collected biological pleural fluid samples with matched radiology and clinical outcome data from a large multicentre observational cohort study was conducted to explore radiological and biological outcome predictors. A multicentre randomised controlled trial was conducted to explore the feasibility of early use of combination intrapleural fibrinolytic and enzyme therapy (IET) or surgery. A qualitative study using semi-structured interviews was conducted to understand the participant experience in pleural infection trials and identify patient priorities. A retrospective analysis of a large IET treated cohort was performed to analyse bleeding complications.

Results
Plasminogen activator inhibitor-1 (PAI-1) was identified as the first biological predictor of mortality in pleural infection. PAI-1 plays an important role in the development of sonographic septations, but their presence does not predict clinically important outcomes. The MIST-3 study demonstrated feasibility and patient acceptability of early randomisation to IET or surgery, with modifications to the study protocol required to improve compliance. MIST-3 and its qualitative sub-study highlighted important insights into trial design and patient centred outcomes. IET carries low incidence of bleeding complications and predictors of increased bleeding risk were identified.

Conclusion
The treatment pathway in pleural infection has remained largely unaltered for almost two decades. This thesis has explored the potential role of radiological and biological outcome prediction to personalise therapy, and evaluated the potential for earlier intervention in the patient pathway improve outcomes relevant to patients and clinicians.