Saturday, July 18, 2015
The Wellcome Trust and Diagnostics
I had the honor and pleasure of attending a Wellcome Trust meeting for the last few days looking at the medical need and potential utility of approaches to the rapid diagnosis of bacterial infection. The Trust gathered a very diverse group of experts ranging from patient advocates to general practice physicians to emergency room physicians to infectious disease specialists. The general question was – could rapid diagnostic technologies be useful in the more rational and appropriate treatment of patients with various infections. I think the general hypothesis was – if we can provide a specific and reliable rapid diagnosis, we might be able to decrease the empiric use of broad-spectrum antibiotics or at least de-escalate to more specific therapy more quickly.
The first question that was posed was – what diagnostics do we need? One fascinating and eye-opening observation (for me) was a clear consensus that we need a rapid and reliable approach, probably focusing on the host response, to distinguish bacterial from viral infection or even between colonization and infection. The idea here would be to document that an antibiotic is not needed and prevent unnecessary therapy. For most participants it was clear that for relatively mild or self limited infections – e.g. sinusitis – we probably do not need such tools. The idea here is that most infections are viral and even those that are caused by bacteria, for the most part, seem to go away without therapy. Obviously, the question of whether treatment could actually shorten the disease if only we could determine who to treat remains open. But the consensus was that this was not the place where we should invest our diagnostic development efforts.
We then explored hospital-acquired pneumonia and ventilator-associated pneumonia (HAP, VAP). There was universal disbelief, condemnation or scorn for the current US approach to the reporting of VAP using the new CDC criteria. In the CDC approach, the diagnosis requires bronchoscopy and many centers, like the one where I work, do not routinely do this procedure to make the diagnosis. Therefore, we will never report a single case. In addition, there is an active disincentive to make the diagnosis given the potential for loss of reimbursement by payers. But VAP was an infection where the consensus was that, again, some measure of whether the patient was actually infected using host-response measures combined with a rapid bacterial diagnostic directly on respiratory secretions would be useful to guide initial therapy. Or, if it could not be rapid enough for a very sick patent, such a diagnostic could guide de-escalation over the ensuing 24 hours or so.
A side discussion emphasized the difficulty in carrying out trials in VAP. There was the suggestion that the current approach using non-inferiority is becoming infeasible and that new trial designs are desperately needed.
HAP was a completely different story, though. Here, the major problem was that, like community-acquired pneumonia, adequate specimens to be used for a bacteriological diagnosis are only very rarely obtained – the estimate was that such a specimen was available in only ~ 5% of cases. Therefore, a rapid diagnosis of bacterial etiology could also not be performed. We did agree that again, some sort of test looking at host response might be helpful to reassure us that at least the patient in question had an actual bacterial infection.
Perhaps the most straightforward question we tackled was complicated urinary tract infection. But even here, some measure of local urinary tract inflammatory response was deemed to be potentially useful to help sort out whether the bacteria and white cells in the urine of a catheterized (or even not catheterized) elderly patient with non-specific symptoms actually represent infection vs. colonization. Once this was established, then a rapid bacterial diagnosis including an indication of susceptibility was considered to be very useful in guiding initial therapy.
So the surprise, and perhaps I should not be surprised, was the importance placed on the potential utility of using the host response to distinguish bacterial infection from either colonization or from viral infection. But, to my knowledge, this would still require vast amounts of research to discover such biomarkers and validate them in various clinical settings. So, our greatest desire is for something that will not exist for years to come. The utility, and even the potential utility, of rapid diagnostics for limiting empiric therapy remains elusive for many infections treated in hospital.