Dosing Blind: Antibiotics Across an Unpredictable CRRT Circuit
A single patient with necrotizing soft-tissue infection now on continuous renal replacement therapy. The disagreement is how to dose life-saving antibiotics through a clearance mechanism that no standard renal-dosing chart was built to describe.
W.H., a 54-year-old warehouse manager, cut his shin on a shipping pallet three days ago and thought little of it — a butterfly bandage, back on his feet the same shift, three teenage kids at home who needed picking up from three different practices that evening. By yesterday morning the leg had gone from tender to unrecognizable, and by the time surgery took him for debridement, necrotizing fasciitis had already tracked well above the knee. He is two days post-debridement, on broad-spectrum antibiotics, and now in worsening septic shock with anuric acute kidney injury — continuous renal replacement therapy started overnight, since he was too hemodynamically unstable for intermittent hemodialysis to be safely attempted. He has no diabetes, no vascular disease, and no prior infection of any kind requiring hospitalization — a healthy man whose entire clinical trajectory over three days has been driven by one organism finding one small break in the skin.
The infectious disease team wants to continue meropenem and vancomycin, both reasonable choices for his organism coverage, but CRRT changes what those doses actually mean once they are running through him. Unlike native renal clearance, which a creatinine-based equation at least approximates, CRRT clearance depends on the specific modality, blood and dialysate flow rates, filter type, and how long the current filter has actually been running before clotting reduces its efficiency — none of which a standard renal-dosing chart, built around residual kidney function, was ever designed to capture. His effluent rate is set at 25 mL/kg/hr and the filter running now went on six hours ago with no clotting yet — which is to say his clearance is at its highest right now and will fall, unannounced, as that filter ages, so any dose calculated tonight is being calculated against a moving number rather than a fixed one. Meropenem's bactericidal activity is time-dependent, meaning what matters is how much of the dosing interval the free drug concentration stays above the pathogen's MIC, not simply the peak level reached — a distinction that argues for extended infusion over standard bolus dosing specifically in a patient whose clearance is this unpredictable. The trials behind that argument, BLING III and MERCY, both studied general ICU sepsis rather than patients whose clearance is set by a machine, so the reasoning that reaches him is the pharmacodynamic one, not the trial result itself. Vancomycin, cleared substantially by convection through the CRRT filter, carries its own separate risk: underdosing him into subtherapeutic troughs during a life-threatening infection because the circuit is quietly removing more drug than a textbook dose assumes, at precisely the moment his organism can least afford a gap in coverage.
ICU bedside, dosing the antibiotics running through the filter
The organism coverage question is settled — meropenem and vancomycin are the right choices for this infection. What I need help with is dosing them correctly through a circuit that changes hour to hour, because a standard critically-ill dosing regimen assumes some baseline residual renal clearance, and he has essentially none of his own left.
For meropenem, I want to move to extended infusion — the same total daily dose delivered as a three-hour infusion rather than a thirty-minute bolus, every eight hours. Meropenem kills in a time-dependent way, so what actually matters for efficacy is fT>MIC, the fraction of the dosing interval the free drug stays above the organism's MIC, not the peak concentration reached. BLING III randomized over 7,000 ICU patients with sepsis to continuous versus intermittent beta-lactam infusion and did not reach significance on 90-day mortality — 24.9% against 26.8% — but clinical cure was higher, 55.7% against 50.0%, and MERCY found the same shape with meropenem specifically in 607 patients. I am citing those for the exposure principle, not as proof of a mortality benefit, and neither trial enrolled a patient whose clearance is CRRT-determined the way his is.
This is a delivery-method change, not a dose increase — the total daily meropenem exposure stays the same, only how it is spread across the interval changes.
For vancomycin I would not trust a fixed nomogram at all, extended-interval or otherwise — CRRT clears a meaningful fraction of it by convection, and that clearance moves with effluent rate and filter age, both of which are already shifting on his circuit tonight. I want a loading dose now, then troughs checked before every subsequent dose rather than at a fixed interval, and dosing adjusted directly off the measured level rather than off a population estimate.
The pharmacist's extended-infusion point for meropenem is well supported and I am not contesting it; vancomycin is the drug where CRRT's unpredictability actually changes the dosing strategy itself, not just the delivery method, since convective clearance through this specific filter is not something any population nomogram can see.
Agreed: meropenem converted to extended infusion at the same total daily dose, vancomycin loaded and then dosed by measured trough rather than a fixed interval, and both reassessed if the CRRT prescription (blood flow, effluent rate) changes materially.
All three explicitly agreed this plan requires re-evaluation, not a one-time decision — a filter change, a shift in effluent rate, or transition off CRRT would each independently change the clearance assumptions this regimen is built on.