Infections acquired by patients in Intensive Care Units (ICUs) are bad news. These patients, many on life support, are already vulnerable. An infection will increase their chance of dying.
Approximately , most commonly pneumonia (in 20 per cent) or a bloodstream infection (in seven per cent). These infections arise either from the patient’s own bacterial community (their ‘microbiome’) or from the microbiome of the ICU (bacteria found in the ICU surrounds).
There are more than 300 randomised controlled trials that aim to reduce ICU-acquired infections.
The mortality rate reported in these studies has remained stubbornly high at around 22 per cent.
On a positive note, has shown that mortality has not increased even though ICU patients these days are on average more than ten years older than they were in the 1980s.
While there are some promising infection prevention interventions, my analysis shows worrying signs that those using ‘topical antibiotics’ may do more harm than good.
The promise of topical antibiotics: too good to be true?
The most intriguing infection prevention methods are those using topical antibiotics – antibiotics applied directly to the body – to alter the patient’s microbiome by removing bad bacteria. The antibiotics are not new, what is new is how they are applied.
They are smeared as a paste into the mouth and intestinal tract (which are patient microbiome hot spots) several times a day while the patient is in ICU.
Published trials of topical antibiotics a 50 per cent drop in pneumonia cases, a 30 per cent reduction in bloodstream infections and a 10 per cent decrease in mortality.
Non-antimicrobial interventions – like control of gastric acid, methods of tube and intravenous feeding, airway care, patient positioning, probiotics – show at best a 30 per cent reduction in pneumonia rates with no effect on mortality.
These are seemingly impressive results for topical antibiotics and they are now commonly used in some parts of Europe. In Australian and elsewhere, the critical care community are more cautious over concerns that this will exacerbate the antibiotic resistance problem.
But are the results using topical antibiotics too good to be true?
has found four puzzling aspects of these studies that should be a cause for concern.
These all relate to the how the study is designed.
In studies like these, we randomly assign the study participants into two groups and give one-half the new treatment (treatment group), and the other half gets standard care (control group).
Importantly, we assume that these groups are independent, that is the treatment should not have any effect on the control group.
The first puzzling aspect is that the infection rates among the topical antibiotic study control groups – those patients who do not receive the antibiotic – are unusually high. , versus 20 per cent expected in a typical ICU. Bloodstream infections are also .
Even the control group mortality rate, 30 per cent, versus the expected 25 per cent.
Secondly, even though the infection rates among the topical antibiotic study intervention groups are lower than the control groups, the results are surprisingly close to what we would expect without any intervention.
Thirdly, the infection rate among the antibiotic intervention groups was no better than in studies of interventions that were shown to be ineffective.
Finally, and tellingly, the results of the topical antibiotic studies were different if the patients receiving the intervention were in the same ICU or a different ICU to the control group patients.
Where the control and intervention group patients were in the same ICU, the studies of topical antibiotic intervention appeared to show a difference in mortality rates and infection rates, but where they were located in different ICU’s, .
So what is going on?
Herd protection or herd peril?
Could it be that the use of topical antibiotics in the ICU changes not only the microbiome of the patient but also the microbiome of the ICU?
Unfortunately, there is no simple way to measure the ICU microbiome, and any change to the ICU microbiome in these studies likely went unnoticed.
This relates to the broader concept of herd protection, which many people understand from population vaccination programs. Herd protection in that context means the benefit of vaccination extends beyond those vaccinated (by direct protection) to those unvaccinated (as indirect protection).
This occurs because those vaccinated are less likely to transmit the disease.
But it raises the question – by giving antibiotics to some ICU patients to lower their infection risk, , that is, protection across the whole ICU, including those that did not get the treatment?
Or, do the paradoxical results described above imply the opposite – a negative herd effect … ?
Studying herd peril
It would be logistically challenging to directly study if topical antibiotics worsen the situation in ICU and is likely “” in terms of risk of harm from such a study to these highly vulnerable patients.
Moreover, it is not required.
Instead, I have using the existing published results realigned as a single .
This simulation raises two findings of concern.
Firstly, if the intervention and control patients were in the same ICU, then the control patients have a 39 per cent higher mortality than if the intervention and control patients were in different ICUs.
Secondly, this higher mortality rate in the same ICU control groups perversely inflates the apparent benefit of the treatment compared to studies of topical antibiotics where the intervention and control patients were in different ICUs.
So, we have a situation where a treatment that appears to be very successful might actually be harming those concurrent control group patients who are not receiving the treatment.
This is the first demonstration that there might be harmful herd effects – or herd peril – resulting from an intervention within a vulnerable population.
We know that topical antibiotics are a problem and cause harm in the ICU population. The next step is to examine the other results reported in these studies – do they show the same effect of increased harm to non-treatment groups?
And we really need to find how this herd peril is mediated.
