A Confirmatory Meropenem Level Amid Rapidly Changing Renal Function
A single patient, nine days into antibiotic therapy that should have worked. The disagreement is whether to measure a drug level before adjusting the dose, or adjust the dose because there isn't time to wait for the measurement.
Delia R., a 71-year-old woman, directed community theater productions for over thirty years before retiring, and still keeps a shelf of programs from every show she ever staged — a detail she brought up unprompted from her hospital bed, as though naming it might will her back to that life faster. She came in eleven days ago with what looked at first like a minor cellulitis on her left shin, until it wasn't: by the next morning the skin had gone dusky and the pain had outpaced the exam, and she went to the operating room within hours for what turned out to be necrotizing fasciitis. Three debridements and a brief run on continuous renal replacement therapy later, her kidneys have recovered enough native function to come off CRRT, but that recovery has been anything but steady — her creatinine has swung between 1.1 and 2.3 over the past four days, tracking her volume status more than any clear trajectory.
She has remained on meropenem for a persistent Enterobacter cloacae bacteremia that source control was supposed to have resolved days ago, and it hasn't: blood cultures drawn on hospital day nine are still growing the same organism, fully susceptible on sensitivities, at a dose that should be more than adequate in a patient with stable renal function — the dose itself unchanged since her transition off CRRT despite her creatinine more than doubling and then dropping again over that same window. The team's working assumption, once mechanical failure (an undrained collection, a missed abscess) was reasonably excluded on repeat imaging, is that her fluctuating clearance may be clearing the drug faster than the dosing interval assumes on the days her kidneys are working closer to normal — the DALI study (Roberts et al., 2014) found this exact pattern, real pharmacokinetic variability wide enough to produce treatment failure in patients whose dose looks textbook-correct on paper, particularly once renal function stops being a fixed number and starts being a moving one.
On the ward, day nine of unchanged cultures
Send a meropenem level before we do anything else. Nine days of persistent, fully-susceptible bacteremia on a dose that should be working is exactly the pattern DALI (Roberts et al., 2014) described in critically ill patients with variable renal function — real, clinically significant PK variability that a textbook-correct dose on paper doesn't guarantee achieves target concentration in this specific patient. Her creatinine has moved by a factor of two in four days; whatever dose was ‘renal-adjusted’ on Monday may not describe Thursday's kidneys at all. A level tells us which problem we actually have — underexposure, or something we haven't found yet.
I don't disagree that variability is real, but a send-out meropenem level in this hospital runs 48 to 72 hours, and she has an active bacteremia today, not in three days. Waiting on a number to tell us what we can reasonably infer — that fluctuating renal function plus persistent growth on a standard dose points toward underdosing — costs her more bacteremic days than the level is worth. Increase the dose and shorten the interval now, based on her current creatinine, the same way we'd renally adjust for any other patient.
You're both right about different halves of this. The turnaround argument is real — we shouldn't sit on our hands for three days with active bacteremia — but so is the case for actually measuring rather than guessing twice on a kidney that's already fooled us once.
Increase the dose today, on the current creatinine, the way you're proposing. But send the level with the increase rather than waiting the three or four days it would take her to reach a formal steady state on the new dose — a concentration drawn at a defined point in the interval can be interpreted pharmacokinetically even before steady state, and waiting for one on a kidney that has moved twice this week risks the level describing a patient she has already stopped being. If she shifts again, we'd at least have a real measured number to model from instead of adjusting blind a third time.
Agreed same day: dose increased on current renal function, level drawn ahead of the next scheduled adjustment rather than after.
Not agreed: how much weight to give empiric dose-adjustment going forward if the level comes back confirming adequate exposure — the intensivist reads that outcome as vindication for adjusting by creatinine alone next time; the pharmacologist reads it as one confirmed data point in an otherwise still-unpredictable kidney, not evidence the level was unnecessary.