I attended a brilliant seminar at Imperial College last week on the role of machine learning and artificial intelligence (AI) in infectious management, and to a lesser extent, infection prevention and control. There’s so much potential for this exciting technology to revolutionise the way we identify, treat, and prevent the spread of infectious diseases. But, there’s also some risks – some are already asking whether the robots are taking over, and whether that is an entirely good thing!
Dogs are recognised to have the keenest of noses and have been used for detecting illicit drugs, early stage cancer and even C. difficile including an outbreak (possibly a cheaper option than PCR for screening – I should have used this in my debate with Jon). Now a new study finds that trained dogs can reliably detect significant bacteriuria.
A study in JAMA Internal Medicine suggests that we may be ‘overdiagnosising’ C. difficile in this era of molecular diagnostics. The researchers from California grouped the 1416 patients tested for C. difficile into three groups: Tox+/PCR+ (9%), Tox-/PCR+ (11%), and Tox-/PCR+ (79%) (see Figure). Perhaps unsurprisingly, compared with Tox+/PCR+ cases, Tox-/PCR+ cases had lower bacterial load, less prior antibiotic exposure, less faecal inflammation, a shorter duration of diarrhoea, were less likely to suffer complications, and were less likely to die within 30 days. Perhaps even more importantly, patients with Tox-/PCR+ were pretty much identical to patients with Tox-/PCR- specimens in all of these metrics. In short: these patients had C. difficile in their gut, but they did not have C. difficile infection. The key message here is that we should not be treating patients who are C. difficile “positive” by molecular tests only.
CRE are an emerging threat to healthcare systems worldwide. Most guidelines recommend screening and isolation of carriers. But relying on conventional agar-based culture presents a dual threat of poor sensitivity (depending on which method is used) and slow turnaround time, with a minimum overnight incubation before a presumptive positive result. PCR solves both of these problems – but at a cost. I gave a talk today at a BD seminar considering whether it’s time to switch to PCR diagnostics for the detection of CRE. You can download my slides here. (The title of the talk “Time to throw away those agar plates” was inspired by a talk by Dr Dan Diekema at a recent SHEA conference.)
There are a number of options for CRE screening, summarized in the flow chart below:
Flow chart: Overview of laboratory methods for the diagnosis of CRE. (To be precise, throughout the blog I really mean CPE most of the time, but I’m using CRE for consistency with other blogs…)