Antibiotics do nothing for a cold, and about 16 hours for a sore throat

Antibiotics have no effect on the common cold. Pooling six randomised trials with 1,047 participants, a Cochrane review found that people given antibiotics for a cold did no better on lack of cure or persistence of symptoms than people given placebo, at a risk ratio of 0.95 (95% CI 0.59 to 1.51) — while the risk of adverse effects was 1.8 times higher (95% CI 1.01 to 3.21) [s1].

The same review looked at the specific belief that keeps the prescriptions flowing: that green or yellow nasal discharge signals a bacterial infection that needs treating. Across four studies with 723 participants, antibiotics for acute purulent rhinitis gave a pooled risk ratio for persisting symptoms of 0.73 (95% CI 0.47 to 1.13) — not statistically significant — alongside a clear increase in adverse effects (risk ratio 1.46, 95% CI 1.10 to 1.94) [s1]. Adults given antibiotics for a common cold had a significantly greater risk of adverse effects than those on placebo (risk ratio 2.62, 95% CI 1.32 to 5.18); in children the difference was not significant [s1].

The reviewers' conclusion: there is no evidence of benefit from antibiotics for the common cold or for persisting purulent rhinitis in children or adults, and routine use for these conditions is not recommended [s1].

Sore throat is the more interesting case

The cold result is straightforward because colds are viral. Sore throat is where the honest answer gets more complicated, and where "antibiotics are useless" overstates the evidence.

A Cochrane review of 27 trials covering 12,835 cases of sore throat found antibiotics roughly halved throat soreness and fever, with the largest difference at day three [s2]. The number needed to treat to prevent one sore throat at day three was fewer than six; by week one it was 21 [s2]. Overall, antibiotics shortened symptoms by about 16 hours [s2].

The effect was concentrated where you would expect. At day three the risk ratio for symptoms was 0.58 (95% CI 0.48 to 0.71) when a throat swab was positive for Streptococcus, against 0.78 (95% CI 0.63 to 0.97) when it was negative; at one week the split widened to 0.29 (95% CI 0.12 to 0.70) positive versus 0.73 (95% CI 0.50 to 1.07) negative [s2].

Complications also fell. Antibiotics reduced acute rheumatic fever by more than two-thirds within a month (risk ratio 0.27, 95% CI 0.12 to 0.60), acute otitis media within 14 days (risk ratio 0.30, 95% CI 0.15 to 0.58) and quinsy within two months (risk ratio 0.15, 95% CI 0.05 to 0.47) [s2]. But the reviewers stress that these are relative benefits against a small absolute risk: protecting sore throat sufferers against these complications in high-income countries requires treating a great many people for one to benefit, and the number needed to treat may be lower in low-income countries [s2].

So the accurate statement is not that antibiotics never work for a sore throat. It is that the average benefit is about 16 hours, that it is much larger in the minority with confirmed streptococcal infection, and that the complication argument depends heavily on the background rate of those complications where the patient lives [s2].

What resistance means, in numbers

The reason a 16-hour gain is weighed so carefully is the cost carried by everyone else, and that cost is now measured directly.

The Global Research on Antimicrobial Resistance study, published in The Lancet in 2024, estimated deaths and disability-adjusted life-years attributable to and associated with bacterial antimicrobial resistance for 22 pathogens, 84 pathogen-drug combinations and 11 infectious syndromes across 204 countries from 1990 to 2021, drawing on 520 million individual records or isolates [s3].

In 2021 it estimated 4.71 million deaths associated with bacterial antimicrobial resistance, including 1.14 million deaths attributable to it [s3]. The two figures answer different questions: "associated" counts deaths from infections that were drug-resistant, while "attributable" estimates the deaths that would have been avoided had those infections been drug-susceptible instead [s3].

The trend is not uniform. Between 1990 and 2021, deaths from resistance fell by more than 50% among children under five while rising by over 80% among adults aged 70 and older [s3]. Meticillin-resistant Staphylococcus aureus rose more than any other pathogen-drug combination globally, from 57,200 attributable deaths in 1990 to 130,000 in 2021 [s3]. Among Gram-negative bacteria, carbapenem resistance grew fastest, from 127,000 attributable deaths in 1990 to 216,000 in 2021 [s3].

The forecast to 2050 is 1.91 million deaths attributable to resistance and 8.22 million associated with it [s3]. The same modelling estimates that better care of severe infections and improved access to appropriate antibiotics could cumulatively avert 92.0 million deaths between 2025 and 2050 [s3].

What this does and does not tell an individual

The falling under-five mortality in that dataset is worth noticing, because it complicates the familiar apocalyptic framing. The study's authors read it as evidence that infection prevention works [s3]. The rising burden in over-70s reflects an ageing population as much as any change in the bacteria.

For a person with a runny nose and a sore throat, the evidence above describes what trials found on average; it does not diagnose anyone. Sore throats with a positive streptococcal swab, sore throats in settings with high rheumatic fever rates, and symptoms that persist or worsen are all situations where the calculation changes, and only a clinician who can examine and test can make it.