Predicting bacteriology in pleural infection using CT scans
Pleural infections are relatively common with 65,000 cases annually in the UK and the USA combined. After clinical diagnosis, the first step towards identifying the causative organism is to take a sample of fluid from the space between the pleural membranes and to culture the specimen in the laboratory. However, this may take several days and a large proportion of the samples do not grow any culture at all.
Microbiology results show that a wide range of bacteria cause pleural infection, for example pseudomonas, streptococcus and gram negative anaerobes. While culture results are awaited, patients are given a broad spectrum antibiotic with a more specific antibiotic prescribed later when the bacteriology results are known. The problem with broad spectrum antibiotics is that they increase the risk of antibiotic resistance and clostridium difficile infection. Therefore, if the appropriate antibiotic could be selected more rapidly, this would be a step forward in patient care.
A pilot study with 23 patients has suggested that some patterns identified on a CT scan, for example the location of pneumonia or lung destruction, may predict the type of infection. We are now looking at a greater number of variable characteristics in a larger sample of more than one hundred patients to confirm these findings.
