Read on Twitter
THREAD: My top six reasons not to be (too) afraid of inadequate empiric antiinfective therapy. PLEASE COMMENT! IMO for most cases narrow and targeted ABx should be sufficient and here are my reasons why: @WinfriedKern @BradSpellberg @FungalDoc @ABsteward @eliowa @khalideljaaly
(1) Studies showing higher mortality w inadequate empiric ABx are often retrospective and bug-specific (i.e. for GNB-BSI). Confounding between comorbidity and MDR-infection is likely to be present.
(2) MICs are in vitro surrogates: they are conservative (err on the side of caution) and aren’t predictive of clinical outcomes (see https://jcm.asm.org/content/49/9_Supplement/S11). Admittedly, there are counter-examples (most prominently the recently published, MERINO trial).
(3) The source of infection: empiric levo works for cUTI even if tested resistant (see https://www.ncbi.nlm.nih.gov/pubmed/25931244 ); in abdominal septic shock, source control is much more important than early or appropriate ABx (see https://ccforum.biomedcentral.com/articles/10.1186/cc13854)
Addition to (3): have a look at the supplemental tables of the recent IAI guidelines: RCTs for antibiotic choice show mostly no (and if, small) differences between empiric regimens! https://www.liebertpub.com/doi/suppl/10.1089/sur.2016.261
(4) The natural course of disease: mortality in pre-antibiotic era was ≈5% for ruptured appendicitis w surgery (see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1390996/) and ≈30% for CAP (according to FDA).
(5) Accuracy of infection diagnosis: sepsis is the main culprit here, wrong diagnosis – in absence of a goldstandard – ranges probably between 10-40%, see https://www.ncbi.nlm.nih.gov/pubmed/26346055
Addition to (5): a good example for a low evidence recommendation of very broad empiric therapies for a usually vague diagnosis are the current VAP/HAP guidelines (US, EU and german).
(6) How bad is “late” ABx for sepsis? The pendulum swings back, see editorial by M. Singer ( https://www.atsjournals.org/doi/full/10.1164/rccm.201703-0621ED) and this RCT: https://www.thelancet.com/journals/lanres/article/PIIS2213-2600(17)30469-1/fulltext . (We need a consensus when to start empiric ABx in suspected infection, see my take on that here: https://twitter.com/mfkuepp/status/1007287153066479618 )
Notes: most points above are not independent from each other; I also focussed on mortality only (i.e. there are potentially other benefits of adequate empiric ABx).
Maybe that’s an approach: narrow beta-lactam + AG. Dutch (and similar, Scottish?) guidelines recommend amox + AG for empiric treatment of urosepsis; but stopping AG early. https://adult.swabid.nl/nl/node/8442
And - the elephant in the room: does empiric narrower ABx usage help solve the growing resistance problem? Biologically plausible - but we don’t know yet (or do we?).
