Summary of the evidence

Self-care

Oral analgesia

  • Overall aspirin, paracetamol and diclofenac potassium (not available to buy over the counter) were all more effective than placebo at reducing pain and fever in adults with sore throat associated with an upper respiratory tract infection. This was based on low to moderate quality evidence from 3 randomised controlled trials (RCTs) (Eccles et al. 2003, Gehanno et al. 2003 and Voelker et al. 2016).

  • Overall, adverse events for aspirin, paracetamol and diclofenac potassium in the 3 RCTs did not appear to be significantly different from placebo (very low to low quality evidence), although adverse events were poorly reported. Another RCT that assessed safety outcomes (Moore et al. 2002) found significantly higher rates of adverse events with aspirin compared with ibuprofen (low quality evidence).

  • Diclofenac is associated with higher cardiovascular risk than other non-selective non-steroidal anti-inflammatory drugs (NSAIDs). Risk is similar to that with selective COX‑2 inhibitors. Naproxen and low-dose ibuprofen (1,200 mg daily or less in adults) are considered to have the most favourable cardiovascular safety profiles (Medicines and Healthcare products Regulatory Agency [MHRA] Drug Safety Update on NSAIDs, October 2012). Of the non‑selective NSAIDs, low-dose ibuprofen has the lowest gastrointestinal risk (MHRA Drug Safety Update on NSAIDS and coxibs, December 2007).

Medicated lozenges

  • Medicated lozenges containing benzocaine, hexylresorcinol or flurbiprofen may help to reduce pain compared with placebo in adults. This was based on low to moderate quality evidence from 6 RCTs (Watson et al. 2000, Benrimoj et al. 2001, Blagden et al. 2001, Chrubasik et al. 2012, McNally et al. 2012 and Schachtel et al. 2014). However, the absolute improvements in pain score were small.

  • Few adverse events were reported in the RCTs with benzocaine lozenges or hexylresorcinol lozenges. Adverse events occurred in 30% to 50% of people using flurbiprofen lozenges, including taste disturbances, numbness, dry mouth and nausea.

Throat sprays

  • Chlorhexidine plus benzydamine combination throat spray (not available in the UK) significantly reduced pain symptoms by day 7 compared with placebo in adults who were also taking phenoxymethylpenicillin. This was based on high quality evidence from an RCT (Cingi et al. 2011). The absolute improvements in symptom score were small and the clinical relevance is not clear.

  • Local adverse events, including numbness and taste disturbances, were common (moderate quality evidence).

  • No systematic reviews or RCTs of local anaesthetic mouth sprays (without an antiseptic) were identified.

Other interventions

  • No systematic reviews or RCTs of non-medicated lozenges or mouthwashes were identified.

Committee discussions on self-care
  • Based on evidence, experience and safety data the committee agreed that it was reasonable to consider paracetamol (first-line) or ibuprofen for self-care of pain or fever associated with acute sore throat. Although no studies were identified on paracetamol and ibuprofen in children with sore throat, the committee noted that these medicines have well-established efficacy and safety profiles for managing pain and fever in children.

  • Based on evidence and experience, the committee agreed that people may wish to try self-care with medicated lozenges (containing a local anaesthetic, an NSAID or an antiseptic agent) to help reduce pain in acute sore throat, but should be told that the benefit is likely to be small.

  • Based on evidence and experience, the committee agreed that it is unclear whether throat sprays containing an antiseptic plus a local anaesthetic help symptoms. Furthermore, the combination product used in the study is not available in the UK.

  • The committee agreed that prescribers should be aware that no evidence was found on non-medicated lozenges, mouthwashes or local anaesthetic mouth sprays (without an antiseptic).

  • The committee was aware of the potential benefits of avoiding GP appointments if people access self-care and seek advice from other health professionals, particularly their community pharmacist, rather than making an appointment to see their GP. The committee agreed that community pharmacists are often more accessible to people than GPs to offer advice.

Corticosteroids

  • Corticosteroids (oral or intramuscular) significantly increased the number of adults and children with no pain at 24 and 48 hours (number needed to treat [NNT] 4 [range 3 to 6]; high quality evidence) and significantly reduced the time to pain resolution by about 14 hours (low quality evidence), compared with placebo. This was based on a systematic review of RCTs (Hayward et al. 2012). There were no significant differences between corticosteroids and placebo in recurrence or relapse of symptoms, or in the number of days missed from work or school (low to moderate quality evidence). All people in the studies were also treated with antibiotics.

  • A single dose of dexamethasone given to adults who did not need an immediate antibiotic prescription did not increase the proportion of people with resolution of symptoms at 24 hours, although a significant difference was seen at 48 hours. This was based on moderate quality evidence from an RCT (Hayward et al. 2017).

  • There was no difference in adverse events for people taking corticosteroids compared with placebo, although reporting of adverse events was incomplete. The RCTs were not large enough to identify rare adverse events associated with corticosteroids.

Committee discussions on corticosteroids

  • The committee noted that most studies of corticosteroids were carried out in accident and emergency departments and included people with more severe symptoms.

  • The committee noted that the studies did not report on the long-term safety of corticosteroids and the risks of recurrent treatment. No studies compared corticosteroids with analgesia.

  • The committee agreed that sore throat is a self-limiting illness and there are concerns about the safety of corticosteroids and the risks of recurrent treatment. The committee noted that there are safer alternatives and agreed not to recommend corticosteroids for managing acute sore throat.

No antibiotic

  • In most cases, acute sore throat is a self-limiting infection, often caused by a viral infection, and most people will not need an antibiotic. Group A beta‑haemolytic streptococcus (GABHS) is the most common bacterial pathogen in sore throat (European Society for Clinical Microbiology and Infectious Diseases sore throat guideline [2012]), isolated in approximately 20% of cases (Kronman et al. 2014).

  • Complications of sore throat caused by a GABHS infection are generally rare in adults and children. Complications can be suppurative (including quinsy [peri-tonsillar abscess], acute otitis media and acute sinusitis) or non-suppurative (including acute rheumatic fever and acute glomerulonephritis; European Society for Clinical Microbiology and Infectious Diseases Sore Throat Guideline [2012]).

Efficacy of antibiotics

  • With antibiotics, significantly more people with acute sore throat were symptom free at days 3 and 7 compared with placebo. At day 3, 51% were symptom free with antibiotics compared with 34% with placebo (NNT 6 [range 5 to 7]). At day 7, most people in both groups were symptom free (87% versus 82%, NNT 21 [range 14 to 49]). This was based on low quality evidence from a systematic review and meta-analysis of RCTs and quasi-RCTs (Spinks et al. 2013). Overall, antibiotics shortened the duration of symptoms by about 16 hours over 7 days.

  • Subgroup analyses suggest antibiotics are more effective in people with a throat swab positive for GABHS. The NNT with antibiotics compared with placebo to prevent 1 person with a negative throat swab having a sore throat on day 3 was 7 (range 5 to 12), with an NNT of about 4 (range 4 to 5) for people with a throat swab positive for GABHS (low to moderate quality evidence).

  • The overall incidence of suppurative complications, including acute otitis media, acute sinusitis and quinsy (peri-tonsillar abscess), was low and based on data from older studies, mostly conducted in the 1950s. These studies found that antibiotics significantly reduced the incidence of acute otitis media and acute sinusitis within 14 days, and quinsy (peri-tonsillar abscess) within 2 months, compared with placebo (low to high quality evidence). Based on the complication rates from studies conducted after 1975, Spinks et al. (2013) estimated that 200 people would need to be treated with antibiotics to prevent 1 case of acute otitis media.

  • Rheumatic fever was reported only in RCTs published before 1961, and the authors noted that the incidence has continued to decline in western societies since then. Results from these early studies found that antibiotics reduced acute rheumatic fever by more than two-thirds compared with placebo (low quality evidence).

  • There was no statistically significant reduction in acute glomerulonephritis in people taking antibiotics, although it was difficult to detect a significant reduction because the absolute rates of this complication were low (less than 0.1%; very low quality evidence).

Safety of antibiotics

Back-up antibiotics

  • A back-up antibiotic prescription (either patient-led collection or delayed collection) or no antibiotic prescription was as effective as an immediate antibiotic prescription for reducing duration and severity of symptoms in people with pharyngitis. This was based on moderate quality evidence from 1 RCT in adults (de la Poza Abad et al. 2016).

  • Immediate antibiotics were significantly more effective than back-up antibiotics for improving fever, pain and malaise in some RCTs from a systematic review (Spurling et al. 2013), whereas there was no difference between groups in other RCTs (low to moderate quality evidence).

  • Across the RCT and systematic review there was generally no difference in adverse events between an immediate antibiotic prescription strategy and a back-up antibiotic prescription or no prescription strategy (very low to moderate quality evidence).

Identifying people more likely to benefit from antibiotics

  • Targeted use of antibiotics using the FeverPAIN clinical scoring system improved symptoms on days 2 to 4 (to a statistically but possibly not clinically meaningful amount) and reduced antibiotic use, compared with a back-up antibiotic prescribing strategy alone. This was based on low to moderate quality evidence from an open-label RCT (Little et al. 2013). People in the group with a low FeverPAIN score (0 or 1 points) were not offered an antibiotic. People with a moderate FeverPAIN score (2 or 3 points) were offered a back-up prescription, and people with a high FeverPAIN score (4 points or more) were offered an immediate antibiotic prescription. The additional use of rapid antigen tests for GABHS in people with a high FeverPAIN score had no clear advantage over using FeverPAIN score alone.

Committee discussions on no antibiotics, back-up antibiotics and identifying people more likely to benefit from antibiotics

  • Based on evidence and experience, the committee agreed that acute sore throat is a self-limiting infection, and most people will get better within a week without antibiotic treatment. Based on evidence and experience, the committee agreed that complications are rare in adults and children, and the committee noted the adverse effects associated with antibiotic use.

  • The committee agreed that prescribers need to weigh up the small clinical benefits from antibiotics against their potential to cause adverse effects.

  • Based on evidence and experience, the committee agreed that no or back-up antibiotic prescribing was as effective as immediate antibiotic prescribing for people with acute sore throat. A back-up antibiotic prescription could be used if symptoms deteriorate rapidly or significantly, or do not improve within the next 3 to 5 days.

  • The committee discussed the clinical scoring systems available to help identify people with acute sore throat who may be more likely to benefit from antibiotics. The committee noted that FeverPAIN and Centor criteria have not been validated in a UK population.

  • The committee was aware that the FeverPAIN criteria were developed in a UK primary care setting in 2013 and have not been assessed in children under 3 years. External validation has not been carried out, but the criteria have been tested in an RCT setting.

  • The FeverPAIN scoring tool can help prescribers to determine if a person's sore throat is more likely to benefit from antibiotics. The scoring tool includes the FeverPAIN criteria plus additional parameters to help prescribers determine the severity of the sore throat. The additional parameters do not affect the overall FeverPAIN score. The committee was aware that the tool may help prescribers implement FeverPAIN criteria in practice and supports shared decision-making in consultations with people.

  • The committee noted that the Centor criteria were developed in the US in an emergency department setting in 1981 and has only been assessed in an adult population. The committee was aware that the previous NICE guideline on upper respiratory tract infections used Centor criteria. FeverPAIN criteria were not available at the time this guideline was published.

  • The committee noted that a FeverPAIN score of 4 or 5 is thought to be associated with a 62 to 65% probability of having a bacterial infection, which is slightly higher than the 32 to 56% probability associated with a Centor score of 3 or 4. The committee was aware that using FeverPAIN in preference to Centor may increase the use of back-up antibiotic prescribing. However, the committee discussed that if more back-up antibiotic prescribing strategies are implemented the overall use of antibiotics may reduce, assuming that around two‑thirds of people will not collect (and take) the antibiotics.

  • The committee acknowledged the recommendation in the previous NICE guideline on upper respiratory tract infections for a no or back-up antibiotic prescribing strategy in acute sore throat, with an immediate antibiotic prescribing strategy also an option for people with an acute sore throat when 3 or more Centor criteria are present.

  • The committee discussed FeverPAIN scores of 4 or 5, or Centor scores of 3 or 4. In some cases people may have these scores but may have milder symptoms. To ensure people with milder and improving symptoms are not issued an immediate antibiotic prescription the committee used its expertise and agreed that a back-up prescription may also be appropriate for this group of people. Withholding antibiotics is unlikely to lead to complications.

  • The committee discussed FeverPAIN scores of 0 or 1, or Centor scores of 0, 1 or 2. The committee was aware that the previous NICE guideline on upper respiratory tract infections recommended either a no antibiotic or a back-up antibiotic prescribing strategy for people with these Centor scores. However, based on evidence, experience and the principles of antimicrobial stewardship the committee recommended a no antibiotic prescribing strategy for this group.

  • The committee was aware that FeverPAIN criteria had not been tested in populations under 3 years and that the Centor criteria were developed in an adult population. However, the committee, using its experience, advised that young children (under 3 years) are unlikely to present with sore throat symptoms alone. Prescribers should follow the NICE guideline on fever in under 5s to assess and manage fever in this population.

  • The committee agreed that there is currently uncertainty about which scoring tool is more effective in a UK population. They noted that both criteria are used in clinical practice and that using a scoring tool is preferential to not using any tool. The committee concluded that either FeverPAIN or Centor criteria should be used to identify people with acute sore throat who may be more likely to benefit from antibiotics.

Antibiotic choice

  • There were no major differences in clinical effectiveness between classes of antibiotics, including penicillins, cephalosporins, macrolides, and sulfonamides in adults and children with GABHS-positive sore throat. This was based on very low to moderate quality evidence from 2 systematic reviews and meta-analyses of RCTs (Altamimi et al. 2012 and van Driel et al. 2016). Statistically significant differences were seen for some comparisons but the absolute differences between antibiotic classes was small.

  • There was no significant difference in adverse events for cephalosporins, macrolides or sulfonamides compared with phenoxymethylpenicillin in 1 systematic review (van Driel et al. 2016). The other systematic review Altamimi et al. 2012) found that a shorter course of late-generation (broader spectrum) antibiotics was associated with significantly more adverse events compared with a 10‑day course of phenoxymethylpenicillin.

Frequency of antibiotic dosing

  • Twice daily dosing of phenoxymethylpenicillin was as effective as 3 or 4 times daily dosing for microbiological cure in adults and children with GABHS-positive sore throat. This was based on low quality evidence from 1 systematic review and meta-analysis of RCTs (Lan and Colford 2000). Once-daily dosing was significantly less effective than 3 or 4 times daily dosing of phenoxymethylpenicillin (low quality evidence).

Committee discussions on antibiotic choice, dose and frequency of dosing
  • The committee discussed that, generally, if an antibiotic is needed to treat an infection that is not life threatening, narrow-spectrum antibiotics should be used as the first choice. Indiscriminate use of broad-spectrum antibiotics is undesirable because it creates a selective advantage for bacteria resistant even to these 'last-line' broad-spectrum agents, and also kills normal commensal flora leaving people susceptible to antibiotic-resistant harmful bacteria such as Clostridium difficile. For infections that are not life threatening, broad-spectrum antibiotics need to be reserved for second-choice treatment when narrow-spectrum antibiotics are ineffective. Based on evidence, clinical experience and resistance data, the committee agreed to recommend phenoxymethylpenicillin as the first-choice antibiotic. This is a narrow-spectrum penicillin with the lowest risk of causing resistance.

  • The committee discussed whether amoxicillin would be a suitable alternative to phenoxymethylpenicillin to support medicines adherence. However, it was aware of evidence that the risk of resistance to amoxicillin is significantly increased in urinary isolates of Escherichia coli following a course of amoxicillin. These effects are greatest in the first month after use, but are detectable for up to 12 months. Also, if the sore throat is due to glandular fever, the BNF states that erythematous rashes are common in people with glandular fever who take amoxicillin.

  • The committee discussed the systematic review by Lan and Colford (2000) that suggested twice-daily dosing was as effective as four times daily dosing. The committee noted that four times daily dosing was the standard dose frequency for phenoxymethylpenicillin and the dose used most frequently in the included studies. The committee noted that this is low quality evidence, using data from only 6 studies and used bacteriological cure at follow-up as an efficacy outcome (rather than a patient-oriented outcome).

  • The committee discussed the benefits and harms of using twice daily dosing of phenoxymethylpenicillin. Twice daily dosing would support medicines adherence in those people who may struggle to take 4 doses at 6‑hourly intervals before food, such as children at school. The committee was concerned that if a twice daily dose was used, phenoxymethylpenicillin levels may fall below the minimum inhibitory concentration. However, they also discussed that streptococci are highly sensitive to phenoxymethylpenicillin, and that antibiotic penetration in sore throat tissue is good, therefore even small concentrations of antibiotic will treat the infection.

  • Based on evidence and clinical experience, the committee agreed that if phenoxymethylpenicillin was prescribed, twice daily or four times a day dosing could be used, providing the same total daily dose was given.

  • Based on evidence, clinical experience and resistance data, the committee agreed to recommend clarithromycin as the alternative first-choice antibiotic for use in penicillin allergy or for phenoxymethylpenicillin intolerance. In pregnancy, erythromycin was recommended if there is true penicillin allergy. Both are macrolides, given at usual doses.

  • The committee discussed the MHRA Public Assessment Report on the safety of macrolide antibiotics in pregnancy. This found that the available evidence is insufficient to confirm with certainty whether there is a small increased risk of birth defects or miscarriage when macrolides are taken in early pregnancy. They agreed with the UK Teratology Information Service monograph on the use of macrolides in pregnancy. They decided that there should be an informed discussion of the potential benefits and harms of treatment. Then, after such a discussion, macrolides can be used if there is a compelling clinical need and there are no suitable alternatives with adequate pregnancy safety data. Erythromycin is the preferred choice if a macrolide is needed during pregnancy, for example, if there is true penicillin allergy and the benefits of antibiotic treatment outweigh the harms. This is because there is more documented experience of its use than for other macrolides.

  • Based on the evidence that there are no major differences in clinical effectiveness between classes of antibiotics, the committee used its experience to agree that the choice should largely be driven by minimising the risk of resistance.

Antibiotic course length

  • In people with GABHS-positive sore throat, treatment with phenoxymethylpenicillin for 5 to 7 days had significantly lower microbiological cure rates compared with 10 days treatment (NNT 13 [range 8 to 37]). This was based on low quality evidence from 1 systematic review and meta-analysis of RCTs (Falagas et al. 2008).

  • There were no significant differences between 5 to 7 days treatment with phenoxymethylpenicillin compared with 10 days treatment, in the rate of relapse or recurrence (very low quality evidence).

  • The studies that compared different course lengths of phenoxymethylpenicillin treatment did not report on adverse events.

Committee discussions on antibiotic course length

  • The committee agreed that, when an antibiotic is appropriate, the shortest course that is likely to be effective should be prescribed to reduce the risk of antimicrobial resistance and minimise the risk of adverse effects.

  • The committee noted that most studies involving clarithromycin or erythromycin used a 5‑day course, whereas, most studies involving phenoxymethylpenicillin used a 10‑day course.

  • The committee noted that no studies were identified that compared 10‑day and 5‑day courses of phenoxymethylpenicillin given at the current recommended dose (500 mg four times daily). However, the committee was aware from its experience that many people do not complete a 10‑day course.

  • Based on evidence, the committee recognised that microbiological cure may be better with a 10‑day course of phenoxymethylpenicillin compared with a 5‑ or 7‑day course, although there were no differences in relapse or recurrence. They agreed that, in situations where bacterial eradication is not specifically needed, and where symptomatic cure is the goal, if a decision to prescribe an antibiotic is made, a shorter course of phenoxymethylpenicillin may be sufficient. However, in situations where there is recurrent infection, a 10‑day course may increase the likelihood of microbiological cure.

  • Based on evidence, clinical experience and resistance data, the committee agreed that when an antibiotic was appropriate, a 5- to 10‑day course of phenoxymethylpenicillin was needed.

  • The committee was aware that bottles of phenoxymethylpenicillin suspension expire within 7 days once reconstituted and a second bottle would be needed to complete a 10‑day course. Prescribing a 7‑day course may help with medicines adherence.

  • A 5-day course of clarithromycin or erythromycin (which is preferred in pregnancy) is an alternative for people with penicillin allergy or intolerance. This course length takes into account the overall efficacy and safety evidence for antibiotics, and minimises the risk of resistance.

See the full evidence review for more information.

  • National Institute for Health and Care Excellence (NICE)