When Cultures Lie
You've had your morning coffee. The clock hits 8am and you start your shift in the clinic. Consults don't start for a few minutes so you check your inbox and there's a urine culture result you've been asked to report as your colleague is off today. It looks pretty nasty. They've cultured Pseudomonas aeruginosa and it's only sensitive to gentamicin and amikacin but looks like intermediate susceptibility to fluoroquinolones.
Do you risk it with the fluoroquinolones or get the animal in and start IV aminoglycosides? It's a tricky situation...
Now what if I said that the culture was a free catch urine or taken from an indwelling urinary catheter. Is it actually clinically relevant? Or if it was a cystocentesis, would you interpret it differently if the animal was a from a young, male, castrated dog or an older, female, spayed dog?
Male dogs are less likely to get urinary tract infections (UTIs) after all, and castration makes prostatic disease less likely too. That said, it doesn't mean the dog doesn't have a UTI, but I'd be considering if there is a complicating risk factor that also needs to be investigated. On the other hand, while female dogs are more likely to get spontaneous bacterial cystitis, it really depends on whether the animal is actually showing clinical signs to make the decision that the cultured organism is causing a clinical infection. If the sampling was performed well, it could represent true bacteriuria but in the absence of clinical signs, subclinical bacteriuria may not warrant treatment.
And therein lies the dilemma of diagnostic stewardship, a key aspect of antimicrobial stewardship.
What is diagnostic stewardship, really?
The World Health Organisation defines diagnostic stewardship as:
"coordinated guidance and interventions to improve appropriate use of microbiological diagnostics to guide therapeutic decisions."
This may sound a little vague, but in essence, diagnostic stewardship aims to ensure that a microbiological test provides rapid, accurate, and ultimately, useful diagnostic information. This can be achieved through guidance and standards for sampling and testing infectious diseases.
Although a significant role is borne by the laboratory and microbiologists, it all starts with a clinician suspecting an infection and taking a sample from the patient. Therefore, ensuring that an appropriate test is performed and in the correct manner, is crucial to having a result that is relevant to the patient, unlike the scenarios I played out above.
How to actually request a test
This is where Tom Boyle's book, How to Request a Test comes in. In this remarkably short book, Tom dives deep into how we evaluate diagnostic tests and what that actually means for our patients and our diagnostic reasoning. I would consider this book essential reading for every clinicians and anyone performing and interpreting tests.
While Tom covers all aspects of evaluating diagnostic tests, I want to focus on how he describes interpreting them in the context of the population (i.e. your patient). In the book, he discusses Bayes' theorem, a mathematical theory explaining how probabilities change in the light of new information. When this comes to diagnostics, we unconsciously use this to guide our choice of diagnostic tests (based on the patient and examination) and we update the probability our patient has the disease based on the results.
Urine for a surprise
Revisiting our urine culture scenario, let's consider how good our test really is. So at the start, we knew nothing about our patient or our testing method. All we have to go off is a rather resistant Pseudomonas aeruginosa isolate. Starting with that, we can ask ourselves, how common is P. aeruginosa in canine urine cultures? Sure it's possible, but it's not the most common organism (<10%; often reported <5%), so the likelihood that this is a true infection right now is low and more context is needed to be certain.
The next piece of information we have is that it was actually a free catch urine. Immediately this should raise flags. Up to 30% of free catch urine samples in healthy dogs can have significant bacterial growth simply due to the fact it is a non-sterile sampling procedure. Unfortunately, pre-sampling cleaning of preputial or peri-vulvar regions has not yet been shown to reduce this significantly.
The same goes for samples from indwelling urinary catheters as it can be quite difficult to differentiate colonisation from catheter-associated urinary tract infections (CAUTI). To properly diagnose CAUTI, the patient must show clinical signs and collection should be either from the most proximal port if recently placed, or perhaps more preferentially, from a new catheter or cystocentesis.
If sampling was performed by cystocentesis, then the probability of a true infection shoots up although contamination may be possible if the patient was not prepped correctly. Both young and male dogs are less likely to get spontaneous bacterial cystitis though, so I would want to be sure the patient has compatible clinical signs and questioning on whether the patient has other risk factors that need investigating. Neutering has varied effects; knowing a male is castrated is useful to know regarding it's possibility of prostatic disease, while neutered females are reported to have higher incidences of urinary tract infections.
My point is that before we've even performed the test and looked at the result, just by knowing more about the patient, we can dramatically adjust our probability of the patient having a disease. Then the test itself, how it was performed and its results, each updated our probability in their own way.
Diagnostic stewardship as antimicrobial stewardship
To summarise, in order to be able to prescribe antimicrobials or rationalise them appropriately, you need accurate microbiological results in a timely manner. To ensure our diagnostics give us the most information, we need to ensure four things:
Right patient
How likely is this patient to have an infectious disease?
Would additional testing confirm an infection or provide useful additional information to our workup?
Right test
Thinking beyond cultures, is this the correct test for our suspected disease?
Right sampling procedure
Has the sampling been performed in such a manner as to reduce contamination and maximise useful results?
Right time
Has the test been performed at an appropriate time during the patients disease?
For cultures, this would be before antimicrobial treatment has begun.
Other examples include PCR during period of bacteraemia or bacterial shedding vs antibody tests later on once the patient has seroconverted
Altogether, this requires knowledge about a whole range of things from presenting clinical signs, suspected diseases and how they progress, potential diagnostic tests and how to perform tests appropriately. This is a lot to ask of a clinician, but the microbiologist can do a lot to help interpret the results as long as you give them enough information on the topics above.
Conclusion
Hopefully by this point you can see just how important patient and test selection are in the process of diagnosis and subsequent treatment with antimicrobials. You were probably doing these things already, consciously or not.
I want to be clear, diagnostic tests are extremely useful but they need to be used and interpreted carefully to avoid unnecessary prescribing of antimicrobials. If in doubt, ask yourself, "What does this test add to my diagnostic reasoning?", and consider whether you think the result is true based on the patient, test, sampling method and time sampled.