The prospective testing of the accuracy of a prognosis prediction score in pleural infection

Pleural infection describes infected fluid collecting between the lung and chest wall. There are around 80,000 new cases each year (US+UK), of which 35% are either fatal or require chest surgery. Standard treatment involves antibiotics and drainage of infected fluid, achieved through a tube placed in the chest. If this fails, surgical drainage is required, but we currently do not have a reliable way of identifying which patients will need surgery. Prompt surgery may be life saving in needy cases, but it is also associated with major complications and results in a substantial amount of pain. Thus, targeting surgery to those who need it most is important. This study will test a prediction tool which may allow us to identify ?high risk? who may benefit most from surgery.
The proposed test is a risk prediction score calculated on admission to hospital. Previous data suggests it may be possible to predict patients who would die from their pleural infection from five test results which are routinely collected in clinical practice. While these predictors seem very powerful, they need to be tested in ?real world? use. In addition, other work suggests that the type of bacteria responsible for the infection may predict death, and these factors will also be explored as possible ?predictors?.
This study will collect information about 550 pleural infection patients, recruited from UK, Europe, USA and the Far East. All patients will receive the best available treatment, including antibiotics, chest drainage and surgery if needed. We will record which patients die or need surgery. At the end of the study, we will calculate whether the ?predictors? identify patients at high risk of death.
The team conducting this research have previously successfully completed many studies in this area. Together the research team holds #45m of clinical research funding. If the score tested here turns out to be a good way predict which patients will do badly, this can be immediately used to change how we treat patients with this disease.

This proposal is to test a potentially powerful, novel, low-cost, and readily translatable prediction model that could be applied at presentation to all patients with pleural infection. If it is successful in prospective clinical testing, this model would allow clinicians to reliably identify subjects with pleural infection at high risk of death and poor outcome to receive targeted extra supportive care, surgery, and newer complex and expensive intrapleural interventions. Successful completion of this study would result in an immediately applicable score, which could be rapidly disseminated into care through specialist care guidelines.
The core of the prediction model is a combination score based on the blood urea, albumin, blood pressure, age, and the source of the infection. These factors have been identified using regression analysis from a previous large cohort of patients with pleural infection, and now require prospective testing. This clinical model will be combined with bacterial markers of prognosis recordable at baseline to create a novel multimodality score.
If it is successful in prospective clinical testing, it would allow clinicians to reliably identify subjects at high risk of death and poor outcome to receive targeted extra supportive care, surgery, and potentially the complex and expensive intrapleural interventions that have recently shown efficacy in clinical trials conducted by our unit.