Antibiotics for common childhood infections

The question

Your child is miserable with an earache or a sore throat, and you want to do something. Antibiotics feel like doing something. But for most common childhood infections, the evidence says they change little — while side effects and resistance are real. This topic walks through when antibiotics help, when they don't, and what to do instead.

Short answer

Most childhood colds, coughs and flu are viral: antibiotics do not help them at all. For ear infections and sore throats, antibiotics help a minority of children and only modestly — most get better on their own within days. A "watch and wait" approach, or a delayed (back-up) prescription you only use if things don't improve, works well for most cases and roughly halves antibiotic use without reducing satisfaction or increasing complications.

The evidence rating is B: multiple Cochrane reviews of randomised trials give us absolute benefit and harm numbers, plus national guidelines — though decisions about rare complications and specific subgroups rest on older or observational evidence.

Key numbers at a glance

• Colds: antibiotics give no benefit; side effects were more common with antibiotics (child-only evidence too thin for a precise figure).
• Ear infections: ~8 treated for 1 to feel better sooner (low certainty); ~1 in 10 gets vomiting, diarrhoea or rash (high certainty).
• Sore throats: ~1 in 6 extra better by day 3; by day 7 almost everyone is better either way.
• Complications are rare: quinsy in ~4 in 1,000 sore throats (UK observational data); ~200 treated to prevent one.
• Delayed prescriptions: 30% vs 93% actually used antibiotics, with equal satisfaction.
• Resistance: a course raises your child's chance of carrying resistant bacteria, most strongly in the month after.

What the strongest evidence says

Colds: no benefit. A 2025 Cochrane review of 6 trials (1,147 children and adults) found antibiotics did not improve the persistence or cure of the common cold — not even for purulent (green) nasal discharge, which many parents take as a sign of bacterial infection. Side effects were more common with antibiotics, though the child-only evidence was too thin for a precise estimate [5 — B].

Ear infections (acute otitis media): small benefit, real harms. The current Cochrane review (2023) found that compared with watchful waiting, immediate antibiotics reduced pain at 2–3 days (risk ratio 0.53; number needed to treat of 8 — but this came from a single trial, so it's low-certainty), with no clear benefit at 3–7 or 11–14 days. Vomiting, diarrhoea or rash were nearly doubled (risk ratio 1.87 — about 1 extra harmed child for every 10 treated, high certainty). No serious complications occurred in either group, and the reviewers concluded that most mild cases in high-income countries remit without complications, so expectant observation is justified [1 — B]. The benefit was greatest in children under 2 with infection in both ears, or with ear discharge [1 — B].

Sore throats: modest benefit, complications rare. A 2021 Cochrane review found antibiotics made about 1 in 6 extra children symptom-free by day 3 (51% vs 34% — number needed to treat 6), but by day 7 most children were better either way (87% vs 82% — number needed to treat 21), and symptoms were shortened by only about 16 hours on average. Antibiotics reduced quinsy (a tonsil abscess: risk ratio 0.15) and middle-ear infection (risk ratio 0.30), but the estimate for rheumatic fever was extremely uncertain [2 — B].

How rare are complications? In a large UK observational cohort of about 13,000 primary-care sore-throat patients, 1.4% had complications — mostly minor — with quinsy in 0.4% (4 in 1,000), and no cases of rheumatic fever or kidney complications at all. Around 200 patients would need immediate antibiotics to prevent a single complication [4 — C]. (Observational data, mixed ages — confounding by severity is possible.)

Delayed prescriptions work. A 2023 Cochrane review found that delayed (back-up) prescribing cut actual antibiotic use from 93% to 30%, with barely any difference in satisfaction (86% vs 91%) and no identified difference in complications [6 — B]. A straight no-antibiotic strategy with safety-netting cut use further still — so the evidence-backed default is watchful waiting, with the delayed script as the compromise for cases where uncertainty is higher [6 — B].

NICE's sore-throat rule. NICE guideline NG84 uses the FeverPAIN score: 0–1 → no antibiotic; 2–3 → no antibiotic or a delayed prescription; 4–5 → consider immediate or back-up antibiotics. Group A streptococcus causes only about 20% of sore throats, and most sore throats last around a week with self-care [3 — B].

Why habits changed. A generation ago, antibiotics were given far more freely for these infections — partly from fear of rheumatic fever, which was once common and is now vanishingly rare in the UK, and partly because the old trials were done before pneumococcal vaccination changed the landscape. The evidence moved: complication rates fell, the trials got better, and the benefit–harm balance shifted toward restraint. If a grandparent says "we always got antibiotics for that," this is why the answer is different now [2, 4 — C].

What remains uncertain

Who exactly benefits. The average effects hide real variation: under-2s with bilateral ear infections or discharge benefit more, but there is no reliable test or score that cleanly separates the children who need immediate antibiotics from those who don't. Clinicians use severity, age and risk factors — plus judgement [1, 3 — B].

Rare complications in the modern era. The rheumatic-fever benefit comes from old trials in very different settings; in contemporary UK practice the risk is vanishingly low, but it is not zero, and that residual uncertainty is why guidelines keep back-up prescriptions in the picture [2, 4 — C].

Why "just in case" prescribing persists. If the evidence favours restraint, why do prescriptions still flow? The research on prescribing dynamics points to a mix of diagnostic uncertainty (viral and bacterial infections look alike in a 10-minute appointment), time pressure, and a desire to preserve the doctor–parent relationship — a prescription can feel like a tangible outcome of the visit [4 — C]. Understanding this helps you have a better consultation: state what you're actually worried about, and ask what would change the decision.

Resistance in absolute terms. Antibiotic prescribing raises an individual's risk of carrying resistant bacteria, strongest in the month after treatment and potentially persisting up to a year — but no good absolute-risk estimate exists, so this stays a real-but-unquantified harm [7 — C].

What resistance means for your family. This is not abstract. A child who has had several recent courses is more likely to carry resistant bacteria, which means the next infection may not respond to the first-choice antibiotic — leading to a second prescription, a broader-spectrum drug, or a longer illness. Every course is a trade: a real chance of modest benefit now against a real but unmeasured cost later. That trade is clearly worth it for a child who needs antibiotics; it is not worth it for a cold, where the benefit is zero [5, 7 — B].

Benefits and risks in absolute terms

How to read these numbers: "6 treated for one to benefit" (NNT 6) means that out of 6 children given antibiotics, 1 extra recovers sooner than they would have without them — the other 5 get better on their own anyway, or don't improve either way. Small NNTs mean bigger effects. Harms work the same way in reverse (NNH).

• Ear infection pain (2–3 days): reduced with immediate antibiotics, roughly 8 treated for 1 to benefit — from a single low-certainty trial [1 — B].
• Ear infection side effects: vomiting, diarrhoea or rash in about 1 extra child per 10 treated (high certainty) [1 — B].
• Sore throat, day 3: 51% vs 34% symptom-free with antibiotics (about 6 treated per extra recovery); day 7: 87% vs 82% (about 21 treated per extra recovery) [2 — B].
• Sore throat complications (UK observational): 4 in 1,000 developed quinsy; ~200 treated to prevent one complication; no rheumatic fever observed [4 — C].
• Colds: no benefit from antibiotics; side effects more common (child-only evidence too thin for a precise figure) [5 — B].
• Delayed vs immediate prescribing: 30% vs 93% actually took antibiotics; satisfaction 86% vs 91%; no more complications [6 — B].

Practical considerations

• What to do instead: for pain and fever, use paracetamol or ibuprofen as in the fever topic; offer fluids; rest. Most ear infections and sore throats improve within 2–3 days and resolve within a week [1, 3 — B].
• The delayed prescription: if your GP offers one, the usual arrangement is to fill it only if your child isn't improving after 48–72 hours, or gets worse. This is not "doing nothing" — it's an evidence-backed strategy [6 — B].
• Use antibiotics as prescribed: don't save leftover doses for next time, don't share them between children, and don't press for antibiotics for a cold — they can't help and the side effects are real [5, 7 — B].
• Ask about the diagnosis, not just the prescription: a useful question is "what signs would tell us this needs antibiotics?" — it gives you a concrete safety net instead of a vague worry [3 — B].
• When antibiotics are clearly the right call: this topic is about restraint, not refusal. Children under 2 with severe bilateral ear infection or ear discharge benefit most; high FeverPAIN scores may warrant immediate treatment; and some infections (urinary tract infections, for example) are different clinical situations where antibiotics are the standard treatment. The skill is telling the two groups apart — which is the clinician's job, with your observations as input [1, 3 — B].
• If antibiotics are prescribed: give the full course as directed and at evenly spaced times — don't stop early because your child looks better, and don't save leftovers "for next time." Expect the possibility of loose stools or mild rash; contact your GP if diarrhoea is severe, a rash spreads, or your child seems worse rather than better [1 — B].
• Keep a simple illness log: date symptoms started, peak temperature, what you gave and when, and any prescription with its start date. When the same child gets their fourth ear infection of the winter, this log is what turns a vague "they're always ill" into something a GP can act on.
• An unused delayed prescription is a success, not waste. If the 48–72 hours pass and your child improves without it, simply don't fill the prescription — and don't keep the slip "just in case" for the next illness. The next illness needs its own assessment. If you did collect the medicine but didn't use it, return it to the pharmacy for safe disposal [6 — B].

What it means for the parents

The consultation-room dynamic matters here. Research on primary-care prescribing suggests clinicians often overestimate how much parents want antibiotics — while parents' actual priority is usually pain relief and reassurance that nothing serious is going on. If that's you, saying "I'm mainly worried about the pain" can get you better care than asking for (or expecting) a prescription [4 — C].

• The pressure to "do something": watching your child suffer while doing "nothing" feels wrong, but watchful waiting is an active, evidence-backed strategy — not neglect. Naming it that way ("we're doing the watch-and-wait approach the guidelines recommend") helps with grandparents, partners and your own guilt [1 — B].
• Sleep and mental health: night-time earache is brutal — screaming, no sleep, a child who can only settle upright on your chest. None of the trials measure parental sleep loss, but it is the unspoken cost of the watchful-waiting days. Plan shifts with your partner if you can.
• Work and childcare: ear infections and sore throats mean days off work and nursery exclusions, often repeatedly. Delayed prescriptions don't change this burden; only the illness course does. This is an evidence gap — trials measure the child's symptoms, not the parents' lost workdays.
• Money: antibiotics themselves are cheap or free on prescription in the UK; the cost to families is time, sleep and lost earnings, which no study in this review quantifies.

When to talk to your doctor, midwife, or pediatrician

• Ear discharge, severe ear pain in a child under 2 (especially both ears), or a very young or unwell-looking infant.
• Sore throat with difficulty swallowing, drooling, difficulty breathing, or a rash.
• Symptoms worsening after 48–72 hours, or not improving as expected — this is exactly what a delayed prescription is for.
• Signs of dehydration (few wet nappies, sunken eyes, no tears) or if you are simply worried.
• Your child has a weakened immune system or a significant underlying condition — the watchful-waiting defaults in this topic assume an otherwise healthy child, and your clinician's threshold for antibiotics will rightly be lower. When in doubt, say so upfront.

This topic is information, not medical advice. Antibiotic decisions for your child should be made with your GP or pharmacist.

References

1. Venekamp RP et al. Cochrane Database Syst Rev. 2023;(11):CD000219. DOI: 10.1002/14651858.CD000219.pub5 — [evidence rating B]
2. Spinks A et al. Cochrane Database Syst Rev. 2021;(12):CD000023. DOI: 10.1002/14651858.CD000023.pub5 — [evidence rating B]
3. NICE NG84. Sore throat (acute): antibiotic prescribing. https://www.nice.org.uk/guidance/ng84 — [evidence rating B]
4. Little P et al. (DESCARTE cohort). https://pmc.ncbi.nlm.nih.gov/articles/PMC8884437/ — [evidence rating C]
5. Kenealy T, Arroll B. Cochrane Database Syst Rev. 2025;(1):CD000247. DOI: 10.1002/14651858.CD000247.pub4 — [evidence rating B]
6. Spurling GKP et al. Cochrane Database Syst Rev. 2023;(11):CD004417. DOI: 10.1002/14651858.CD004417.pub6 — [evidence rating B]
7. Costelloe C et al. BMJ. 2010;340:c2096. PMID: 20483949 — [evidence rating C]

Changelog

• 2026-09-08: Topic created (draft). Awaiting independent clinical review. (Superseded 2026-09-09: author-review model, no external review before v1.)
• 2026-09-09: Full review (Guille via delegated agent review). Rating confirmed B. Changes: removed the superseded 2013 otitis-media absolute figures from reader prose (2023 Cochrane numbers stand alone; ref removed and list renumbered 1–7); removed the naked "side effects ~1.8× more common" for colds (reframed: more common, child-only evidence too thin for a precise figure); trimmed the duplicated clinician-demand survey sentence from the "why prescribing persists" section (actionable tip kept in the parents section); removed the unsupported "sibling ping-pong" and "decision-fatigue" paragraphs; corrected the delayed-prescription disposal advice (an unfilled prescription is simply not filled; only collected-but-unused medicine goes back to the pharmacy — verified against NHS pharmacy guidance); verified current NHS "finish the prescribed course" guidance (unchanged). Status: approved.