Colistin was shelved decades ago because it damaged kidneys. Now infectious-disease teams reach for it when carbapenems fail, and the arguments about how to use it have never really been settled. The revival of old antibiotics is one of the more awkward stories in modern medicine: the drugs are back, but the trials that would tell clinicians exactly how to prescribe them largely are not.

The Old Drug Comeback

Colistin, fosfomycin, and nitrofurantoin were largely abandoned in high-income systems for good reasons. Colistin carried a reputation for nephrotoxicity; fosfomycin was awkward to dose; nitrofurantoin was confined to uncomplicated urinary infections. Multidrug-resistant Gram-negative bacteria have changed the calculus. When a carbapenem-resistant Enterobacterales isolate appears on a ward, the formulary options narrow fast, and older agents move from the back of the shelf to the front.

Carbapenem-sparing regimens have gained ground for the same reason. Preserving the last-line drugs means using older, narrower agents where susceptibility allows, a strategy that stewardship pharmacists have pushed for years. In practice, wards often move faster than the guidelines that are supposed to govern them. A patient deteriorating at 2 a.m. does not wait for a consensus document.

The pattern is not confined to one country. European centres report rising use of colistin in intensive care, and US hospitals have similar stories. The drugs are cheap, familiar to older clinicians, and available. That combination explains much of the revival, independent of any formal recommendation.

Availability is itself a moving target. Several of these agents were never marketed in some countries, and supply has depended on a small number of manufacturers. When a single supplier pauses production, a pharmacy that relied on the drug for its carbapenem-sparing pathway has to rewrite the pathway. Clinicians who planned around a familiar old agent discover that familiarity is not the same as security of supply. Procurement teams now track these molecules the way they track any critical item, and shortages surface in ward rounds before they surface in any bulletin.

Why Newer Is Not Always Better

The pipeline for genuinely novel antibiotic classes remains thin. New agents arrive, but resistance follows, and the economics of antibiotic development discourage the kind of long-term investment that would change the picture. Against that backdrop, older drugs retain a real advantage: narrow-spectrum activity that spares the broader flora when the target is known.

Stewardship pressure favours targeted therapy. Broad-spectrum empiricism is easy to start and hard to stop, and every day of unnecessary carbapenem exposure adds selection pressure. An older agent with a narrower profile can be the more responsible choice once susceptibility results return. Cost and access shape formulary decisions too; a cheap generic is easier to stock than a branded novel agent with a restricted indication.

None of this makes older drugs automatically safer. It makes them the pragmatic option in a constrained system. The trade-off is that pragmatic use often outruns the evidence, and clinicians are left extrapolating from small observational series. A related piece on hospital consolidation and pricing makes a similar point about systems making do with what they have.

The Evidence Gap Experts Contest

Colistin combination trials have produced mixed results. Some suggest a mortality benefit from pairing colistin with a second active agent; others find no clear advantage over monotherapy, with added toxicity. The trials differ in population, dosing strategy, and the resistance mechanisms involved, which makes pooling them difficult. Clinicians reading the literature often find themselves choosing between studies rather than following a clear signal.

Fosfomycin dosing remains poorly standardised. Regimens vary by infection site, renal function, and local practice, and the pharmacokinetic data are thin for critically ill patients. Observational data dominate clinical decisions as a result. That is not a comfortable position for a drug being used in life-threatening infections, but it is the position clinicians occupy.

Guidance diverges. IDSA and ESCMID documents have taken different positions on sequencing and combination strategy, reflecting genuine disagreement about what the evidence supports. Where guidelines conflict, local resistance patterns and formulary constraints often decide the matter. Stewardship teams end up writing their own protocols, and those protocols vary from hospital to hospital.

Nitrofurantoin illustrates the same problem in a quieter register. It works well in uncomplicated cystitis and is generally well tolerated, but its place in more complicated urinary infection, and in men, is debated. Renal function thresholds for use have shifted over time, and clinicians working in settings without reliable creatinine measurement face a judgement call rather than a rule. The drug is old, cheap, and useful; the guidance around it still carries qualifications that reflect thin prospective data rather than settled consensus.

Nephrotoxicity and the Trade-Off

Colistin kidney injury rates reported in the literature vary widely, from modest to substantial, depending on definition, dose, and co-administered drugs. That variability frustrates clinicians trying to counsel patients or weigh risk. Therapeutic drug monitoring would help, but it is unevenly available; some centres run colistin assays routinely, and others cannot. Where assays exist, turnaround is often measured in days rather than hours, which limits their value in the first 48 hours of a deteriorating patient.

Fosfomycin resistance can emerge on monotherapy, particularly when dosing is suboptimal or duration is prolonged. The temptation to use a single old agent to avoid toxicity collides with the risk of selecting resistance during treatment. Clinicians weigh harm against resistance risk in real time, often with incomplete information about the isolate and the patient's renal trajectory.

The objection from some pharmacists is blunt: reviving nephrotoxic drugs without robust monitoring infrastructure shifts harm from the population to the individual patient. That is a defensible position, and it is not fully answered by arguments about resistance pressure. The counter-argument is equally concrete: the alternative may be a carbapenem that no longer works.

There is a middle path that some units have taken, and it is worth describing because it is unglamorous. It involves writing the monitoring into the order set itself: baseline creatinine, a stated interval for repeat measurement, a list of concomitant nephrotoxins to avoid or dose-adjust, and a named person responsible for reviewing the assay result. None of that requires new evidence. It requires the ward to agree in advance what will happen when the drug is started at an inconvenient hour, which is when most of these decisions are actually made.

What Clinicians Should Do Now

Check local resistance patterns before prescribing. Antibiograms vary by hospital and by unit, and an agent that works in one ICU may fail in another. Ask the microbiology lab for the current susceptibilities of the older agents, not just the newer ones.

Use combination therapy where the evidence supports it, and document the rationale. For colistin in carbapenem-resistant infections, pairing with a second active agent is common practice in many centres, though the trial evidence is mixed. Record why the combination was chosen so stewardship audits can review it later.

Push for susceptibility testing of older agents. Fosfomycin and colistin are not always included in routine panels, which means clinicians may be prescribing blind. Requesting the tests makes the data exist for the next patient.

Advocate for trials comparing old versus new. The evidence gap will not close on its own, and observational series will not settle the dosing and combination questions. A related piece on vaccine stockpile shortages shows how supply constraints shape clinical choices in ways guidelines rarely anticipate.

Finally, review the monitoring plan before the first dose. Renal function, concomitant nephrotoxins, and the availability of drug assays should be checked at the point of prescribing, not after creatinine rises. A short checklist embedded in the order set is more reliable than memory.

What the Revival Says About the System

The return of these drugs is often framed as a story about clever pharmacology. It is more accurately a story about scarcity. When the pipeline delivers few genuinely new options, the system reaches back into its own history and asks old molecules to do new work. That works, up to a point. The point arrives when a clinician has to choose a dose without pharmacokinetic data, or explain a creatinine rise without an assay result.

Stewardship has spent years telling prescribers to use narrower agents and to stop unnecessary courses. That message is correct, and it has made the older drugs more attractive. The next step is harder: building the laboratory capacity, the monitoring routines, and the trial infrastructure that would let those drugs be used with the same confidence as the newer ones they are replacing. Until then, wards will keep improvising, and most of the time they will get it right.

This article is informational and does not constitute personalised medical advice. Treatment decisions for resistant infections should be made with an infectious-disease specialist or clinical pharmacist familiar with local resistance patterns.