22.6 — Antibiotics

Alexander Fleming used his 1945 Nobel Prize lecture to warn about the thing he had just been given a prize for.

He described how easy it was to make bacteria resistant to penicillin in the laboratory by exposing them to doses too low to kill them, and said the same thing would happen in people.

He was right, and it took less than a decade.

Antibiotic resistance now causes an estimated 1.3 million deaths a year directly, and contributes to several million more.

How antibiotics work

They exploit differences between bacterial cells and ours (Chapter 17.2) — selective toxicity, which is the whole basis of the field.

Bacteria have a cell wall; we do not. Bacterial ribosomes differ from ours. Bacterial enzymes for making folate and copying DNA differ from ours.

Every antibiotic class attacks one of those differences:

Cell wall synthesis — penicillins, cephalosporins, carbapenems and glycopeptides. The bacterium cannot build or maintain its wall, so it bursts under its own internal pressure. We have no cell wall at all, which is why this class is among the safest.

Protein synthesis — macrolides, tetracyclines, aminoglycosides, clindamycin. They bind the bacterial ribosome, which differs enough from ours to be targeted.

DNA replication — quinolones, blocking the enzymes that unwind DNA.

Folate synthesis — trimethoprim and sulfonamides. Bacteria make their own folate; we get ours from food, so we do not have the enzyme to block.

Cell membrane — colistin, used as a last resort because it is toxic to us as well.

Bactericidal antibiotics kill; bacteriostatic ones stop growth and leave the immune system to finish the job. The distinction matters mainly in serious infections and in people with weak immune systems.

The common ones

Penicillins — amoxicillin, flucloxacillin, benzylpenicillin, and combinations such as co-amoxiclav.

Co-amoxiclav is amoxicillin plus clavulanic acid, which blocks the bacterial enzyme that destroys penicillins — a drug whose only job is to protect another drug.

Cephalosporins — broader spectrum, in generations of increasing breadth.

Macrolides — clarithromycin, azithromycin, erythromycin. Used in penicillin allergy. And they are strong CYP3A4 inhibitors, so they cause many interactions (Chapter 22.2).

Tetracyclines — doxycycline. Avoided in children under 12 and in pregnancy, because they stain developing teeth and affect bone growth. Causes photosensitivity, so sunburn is easier. Take with plenty of water and stay upright, because it can irritate the oesophagus.

Quinolones — ciprofloxacin, levofloxacin.

And their use has been restricted in recent years, because of tendon rupture — particularly the Achilles, particularly in older people and those on steroids — plus nerve damage that can be lasting, aortic aneurysm risk, and central nervous system effects. Now reserved for infections where alternatives will not do.

Trimethoprim and nitrofurantoin — urinary infections (Chapter 21.3). Nitrofurantoin concentrates in the urine and does not treat kidney infection well, and it needs adequate kidney function to work at all.

Metronidazole — anaerobic bacteria and some parasites. Causes an unpleasant reaction with alcohol, so alcohol is avoided during and for 48 hours after.

Aminoglycosides — gentamicin. Powerful, and toxic to the kidneys and to hearing, which can be permanent. Requires blood level monitoring, which is exactly the narrow therapeutic index situation from Chapter 22.1.

Vancomycin — for resistant Gram-positive organisms including MRSA. Also monitored.

And the reserve agents — carbapenems, linezolid, daptomycin, colistin — kept for resistant organisms, because using them widely is how you lose them.

Resistance

And it is worth being precise about how it happens, because the common understanding is slightly wrong.

People do not become resistant. Bacteria do.

Resistance is evolution in fast motion (Chapter 3.2). In any large bacterial population, a few individuals carry a mutation that happens to protect them. Antibiotic exposure kills the rest, so the survivors have the field to themselves and multiply.

The mechanisms bacteria use:

Enzymes that destroy the drug — beta-lactamases chopping penicillins apart.

Pumps that push it out.

Changing the target so the drug no longer binds.

Reducing permeability so it cannot get in.

Alternative pathways that bypass the blocked step.

And the part that makes it move so fast: horizontal gene transfer. Bacteria exchange resistance genes directly, on plasmids, including between different species. Which means resistance developed in one organism, in one place, can appear in an unrelated organism elsewhere. This is why resistance spreads far faster than ordinary inheritance would allow.

What drives it:

Antibiotics for viral infections, which is enormously common — colds, most sore throats, most coughs, most sinusitis. Antibiotics do nothing at all against viruses, and every course exposes the person's whole bacterial population to selection.

Agricultural use. A large fraction of global antibiotic use is in animals, much of it for growth promotion rather than treatment.

Incomplete or inadequate treatment.

Poor infection control in hospitals.

Over-the-counter availability without prescription in many countries.

The organisms of concern: MRSA; extended-spectrum beta-lactamase producers; carbapenem-resistant organisms, which are close to untreatable; multidrug-resistant tuberculosis (Chapter 17.8); and resistant gonorrhoea, which is approaching the end of its treatment options.

And the pipeline problem: very few genuinely new antibiotic classes have been developed in decades. The economics are poor — a drug taken for a week and then deliberately held in reserve does not repay development costs the way a drug taken daily for life does. This is a market failure rather than a scientific one, and it is being addressed with subscription-style payment models in some countries.

Should you finish the course?

And this is genuinely more complicated than the message most people grew up with.

The traditional reasoning was that stopping early leaves surviving bacteria to develop resistance.

The current understanding is that for most common infections, longer courses cause more resistance, not less — because every extra day exposes all the bacteria in your body, including the harmless ones, to selection pressure.

Which is why recommended course lengths have been getting shorter for many infections, with trials showing that three or five days works as well as seven or ten for several common conditions.

But — and this matters — some infections genuinely require long courses. Tuberculosis needs months, and stopping early is exactly how multidrug-resistant tuberculosis is created. Bone infections, endocarditis and abscesses all need prolonged treatment.

So the accurate advice is: take the course you were prescribed, and the prescriber should be choosing the shortest effective duration. Not "always finish", and not "stop when you feel better" either. If you feel completely well early in a course for a simple infection, that is a reasonable thing to ask about rather than to decide alone.

Side effects worth knowing

Diarrhoea — very common, because antibiotics kill the normal gut bacteria along with the target (Chapter 9.6).

Clostridioides difficile infection — the serious version. The normal gut flora is wiped out, and this organism takes over, producing toxins that inflame the colon.

Severe watery diarrhoea, abdominal pain and fever, during or up to weeks after antibiotics.

Highest risk with clindamycin, quinolones, cephalosporins and co-amoxiclav.

Treated with specific antibiotics, and for recurrent cases with faecal microbiota transplantation, which works remarkably well — restoring a normal bacterial population from a healthy donor. Cure rates over 80 to 90 percent, which is better than antibiotics for recurrent disease.

Thrush — for the same reason, in the mouth or vagina.

Allergy, and this is worth a section of its own.

Penicillin allergy

Around 10 percent of people report a penicillin allergy.

Fewer than 1 in 10 of those are genuinely allergic when tested.

Which means over 90 percent of people labelled penicillin-allergic could safely take penicillin — and this matters, because the label leads to broader, less effective and more expensive alternatives, with worse outcomes and more C. difficile.

What most of these labels actually are: a childhood rash during an illness that was probably viral; nausea or diarrhoea, which is a side effect and not an allergy; or a family member's allergy, which does not transfer.

True allergy is: hives, swelling, wheeze, or anaphylaxis, occurring within minutes to an hour.

Which makes getting the label checked genuinely worthwhile — allergy testing and supervised challenge are available and remove the label in most cases.

And the severe delayed reactions must be taken seriously and never re-challenged: Stevens–Johnson syndrome, toxic epidermal necrolysis, and DRESS — widespread rash with fever and organ involvement (Chapter 14.5).

Interactions

Macrolides and quinolones raise warfarin's effect substantially, which can cause serious bleeding.

Macrolides inhibit CYP3A4, raising levels of statins, some calcium channel blockers and immunosuppressants.

Rifampicin induces enzymes powerfully, reducing the effectiveness of many drugs including hormonal contraception.

Metronidazole and alcohol.

Trimethoprim raises potassium, particularly with ACE inhibitors or ARBs.

And several antibiotics prolong the QT interval on the ECG, which raises the risk of a dangerous heart rhythm, particularly when combined.

What actually helps

Do not ask for antibiotics for a cold, a cough or a sore throat that is likely viral. Green sputum and green nasal mucus do not indicate bacterial infection — the colour comes from immune cells, not bacteria, and it is a persistent and completely wrong piece of folk knowledge.

Take them as prescribed.

Do not keep leftovers or take someone else's.

Delayed prescriptions work well — a prescription to fill only if things are not improving after a few days, which reduces antibiotic use substantially without worse outcomes.

Vaccination reduces antibiotic need, both by preventing bacterial infections directly and by preventing the viral illnesses that lead to unnecessary prescriptions.

And the everyday things: hand washing, food hygiene, and staying home when infectious.

The hopeful part

Rapid diagnostics that identify the organism and its resistance pattern in hours rather than days, allowing narrow targeted treatment from the start.

Bacteriophage therapy — using viruses that infect bacteria specifically. Long used in parts of eastern Europe, and now being seriously investigated elsewhere, with documented successes in individual desperate cases.

New classes in development, including some found by machine learning searches of chemical space — an approach that has already produced genuinely novel candidates.

Antibodies, vaccines against resistant organisms, and drugs that block resistance mechanisms rather than killing bacteria directly.

And stewardship programmes work. Countries and hospitals that reduced unnecessary prescribing saw resistance rates fall, which is the encouraging part: this is a problem that responds to the obvious solution.

What the next page fixes

Chapter 22.7 covers heart and blood pressure drugs — the medicines most people over 50 end up taking, what each one is actually doing, and why they are worth taking when you feel perfectly fine.