I hope you enjoyed Christmas time and wish you all the best for this year. From my side, I will continue to reflect what I meet professionally, what surprises me, confirms what I thought to know or what confirms my ignorance. In 2017 I did that 41 times (a surprise to me!) and here are some trending topics that will most likely return in 2018.
Preprint publishing
In December 2017 I “promised” that I would put all new publications on a preprint server first. In fact, 9 did (which demonstrates that most of my PhD students and postdocs either don’t read my blogs, or don’t care of what I promise). I have not checked whether the preprints were cited, but two of them received responses. One will appear soon in Critical Care Medicine, describing the (rather disappointing) predictive power of a new RNA-based biomarker test for infection as a cause of sepsis in ICU patients. Compared to the first paper on this test (comparing obvious infectious and obvious non-infectious causes of sepsis) this study enrolled patients in which the cause of sepsis was not so clear (and where a test could be helpful).
ESBL predict
Four blogs advertised the ESBL-predict study: “a web-based subject raising initiative”. We aim to validate a very simple prediction rule in patients that start empiric treatment with antibiotics covering Gram-negative bacteria. We have passed the 5,000 subjects and hope to reach 10,000 in 2018. We’re still hoping for subjects from the UK and USA!
Bacteriophages
The blogs on the effects of bacteriophage therapy for treating infections caused by MDR bacteria appeared to have had the most reads. I guess a significant part of that readership came from the Netherlands, where this topic received considerable media attention. Even this week we had a news item showing how desperate patients try to collect 5000€ to travel to Tbilisi to get treatment. I have no idea how many parents and patients have already paid similar amounts to get phages, but know for sure of at least 2 patients. Insurance companies are not covering this treatment, because of lacking evidence of effect. Our research lab has been using bacteriophages for years and we have now developed our “clinical phage library”, testing in vitro effectiveness for Pseudomonas aeruginosa and with a protocol ready for IRB submission next week. We aim to quantify the effectiveness of this approach for different pathogens and different infections in 2018. I’ll keep you posted.
Meta-analyses
There is hardly anything so convincing as a conclusion from a meta-analysis. Last year I have questioned some of these, and I hope that it stimulates some to also read the method and results sections. Bottom-line: a meta-analysis was intended to pool data from similar studies (interventions or not) performed in similar patient populations with results suggesting a similar effect, but lacking statistical power for each individual study. The pooling then creates a larger study, with more power to demonstrate (or rule out) a statistically (and clinically) significant effect. Yet, many meta-analyses lump highly different studies (high clinical heterogeneity) with widely varying outcomes (high statistical heterogeneity) and then provide an overarching effect estimate. It’s like your new PhD student telling you: “I did 10 experiments, but each time I used different controls, different reagents, different concentrations, and different pathogens. As the results went into all directions, I pooled everything, and look: there is a 25% reduction.” Some might say: garbage in is garbage out”. I am sure/afraid that we will see more of this in 2018. Happy to discuss some of them.