An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Respiratory Tract Infections - Antibiotic Prescribing: Prescribing of Antibiotics for Self-Limiting Respiratory Tract Infections in Adults and Children in Primary Care. London: National Institute for Health and Clinical Excellence (NICE); 2008 Jul. (NICE Clinical Guidelines, No. 69.)
This short clinical guideline seeks to optimise the use of antibiotic prescribing for RTIs in adults and children presenting in primary care settings. The conditions included in the review are those common RTIs presenting in primary care where antibiotic prescribing is often considered for resolving symptoms and preventing complications. The five RTIs covered in this short clinical guideline are: acute otitis media (AOM), acute sore throat/acute pharyngitis/acute tonsillitis, the common cold, acute rhinosinusitis and acute cough/acute bronchitis. These are the five most common RTIs consulted for in UK general practice2.
The aim of this overview section is to summarise the evidence on antibiotic efficacy for the above five RTIs. It provides the rationale for the conduct of this short clinical guideline, which is to ascertain the clinical effectiveness and cost effectiveness of specific antibiotic management strategies for RTIs (see section 2).
The overview draws on recently published systematic reviews (the Cochrane Library) and other relevant studies. The identified evidence is summarised and presented narratively.
This overview is to demonstrate the efficacy of antibiotics in treating RTIs (acute otitis media [AOM], acute sore throat/acute pharyngitis/acute tonsilitis, the common cold, acute rhinosinusitis and acute cough/acute bronchitis) in adults and children presenting in primary care settings. The term ‘acute rhinosinusitis’ is used instead of ‘acute sinusitis’ for consistency throughout this guideline because acute rhinosinusitis is the terminology that is currently internationally accepted. However, in some medical literature, studies still refer to the condition as acute sinusitis.
One Cochrane systematic review on the efficacy of antibiotics for AOM was identified (Glasziou et al. 2004). This Cochrane review included 8 randomised controlled trials (RCTs) involving 2287 children (6 months to 15 years) of either gender without tympanostomy tubes, suffering from AOM, irrespective of the setting from which they were recruited. The type of intervention in the studies was any antibiotic therapy versus placebo. The studies were set in primary care (general practice) (4) and hospital (1) (randomised by hospital pharmacy). The settings of the other 3 studies were unknown.
This Cochrane review carried out meta-analyses using pooled relative risk (RR) on patient-relevant outcomes (symptoms or problems that are important to patients’ sense of wellbeing) and other key outcomes. The two key patient-relevant outcomes of the reviews were duration and severity of pain and hearing problems (mid- to long-term) caused by fluid in the middle ear. The other two key outcomes were adverse events (vomiting, diarrhoea, rash) and progression of symptoms (complication – contralateral otitis media).
In the meta-analyses, duration and severity of pain was not significantly reduced by antibiotics in the first 24 hours (RR = 1.02, 95% confidence interval [CI] 0.85 to 1.22, p = 0.91) (4 studies) but was significantly reduced by antibiotics on days 2 to 7 (pooled RR = 0.70, 95% CI 0.60 to 0.81, p < 0.00001, number needed to treat3 [NNT] = 15, 95% CI 11 to 24) (8 studies).
Within the 4 studies that reported the duration and severity of pain in the first 24 hours, only 1 was from primary care setting. The study sample in this investigation was 229 children. The result of this individual study was RR = 0.99, 95% CI 0.75 to 1.30, which was not significant. Within the 8 studies that had outcomes for days 2 to 7, only 3 studies were set in general practice. The total study sample of these 3 studies was 586 children. The results of these 3 individual primary care studies were RR = 0.89, 95% CI 0.41 to 1.93; RR = 0.71, 95% CI 0.43 to 1.18 and RR = 0.82, 95% CI 0.68 to 0.98, respectively; (2 of the 3 studies showed nonsignificant results).
For the outcome of hearing problems, no significant difference in the meta-analysis for tympanometry results was reported at 1 month (3 studies) or 3 months (2 studies) after the acute episode, suggesting no beneficial effect of antibiotics on hearing (1 month: pooled RR = 0.94, 95% CI 0.75 to 1.19, p = 0.6; 3 months: pooled RR = 0.80, 95% CI 0.55 to 1.16, p = 0.2).
Two studies from the analyses (at 1 month) were set in general practices with a total study sample of 323 children (RR = 0.74, 95% CI 0.49 to1.13 and RR = 1.05, 95% CI 0.74 to 1.48, respectively), and 1 study (out of 2) from the 3-month analysis was set in primary care, with a study sample of 221 children (RR = 0.65, 95% CI 0.40 to 1.07).
As well as patient-relevant outcomes, 4 studies in the Cochrane systematic review reported adverse events experienced by individual children (side effects of antibiotics such as nausea, diarrhoea and rash). When all 4 studies were combined, the results showed that children who took antibiotics were more at risk of having adverse events compared with the placebo group (pooled RR = 1.60, 95% CI 1.19 to 2.16, p = 0.002).
Out of these 4 studies, 2 were reported as primary care-based with a total study sample of 472 children. The RRs for the 2 individual studies were 1.52 (95% CI 1.09 to 2.13) and 1.75 (95% CI 0.90 to 3.42).
In terms of progression of symptoms, the meta-analysis showed no beneficial effect of antibiotics in reducing contralateral otitis (3 studies) (pooled RR = 0.48, 95% CI 0.17 to 1.33, p = 0.2).
Apart from the Cochrane systematic review (Glasziou et al. 2004), another meta-analysis with individual patient data on antibiotics for AOM (Rovers et al. 2006) was also identified for inclusion in this overview. Unlike the Cochrane review (Glasziou et al. 2004), which had a wide age range (6 months to 15 years), this IPDM identified subgroups of children who would and would not benefit more than others from treatment with antibiotics. A total of 6 randomised trials were included and individual patient data from 1643 children aged between 6 months and 12 years were validated and re-analysed. The primary outcome of the study was a protracted episode of AOM (consisting of pain, fever or both at 3 to 7 days).
The results showed that, relative to placebo, the overall RR for symptoms at 3 to 7 days with antibiotics was 0.83 (95% CI 0.78 to 0.89; NNT = 8). When pain and fever were analysed separately, results for both outcomes showed a modest effect of antibiotics in reducing pain at 3 to 7 days (RR = 0.86, 95% CI 0.81 to 0.91, NNT = 10) and reducing fever at 3 to 7 days (RR = 0.95, 95% CI 0.92 to 0.98, NNT = 20).
Further analyses for the primary outcome (pain, fever or both at 3 to 7 days) also showed that the effect of antibiotics was modified by age and bilateral AOM, and by otorrhoea. In children younger than 2 years with bilateral AOM, 30% of the antibiotics group and 55% of the control group still had pain, fever or both at 3 to 7 days, with RR = 0.64 (95% CI 0.62 to 0.80; NNT = 4). In contrast, in children aged 2 years or older, there was no significant difference between the two groups in pain, fever or both at 3 to 7 days (RR = 0.80, 95% CI 0.70 to 1.02). Pain, fever or both were still reported at 3 to 7 days in 24% of children with otorrhoea in the antibiotics group and 60% of children with otorrhoea in the control group, with RR = 0.52 (95% CI 0.37 to 0.73; NNT = 3). The risk difference, which was 36%, was much greater than the risk difference for children without otorrhoea in the two groups (14%). This suggests that children with otorrhoea seemed to benefit more from treatment with antibiotics irrespective of other characteristics.
The summary findings of the Cochrane review and the IPDM are as follows. Antibiotics for AOM are effective only in reducing duration of pain in children aged between 6 months and 15 years. The NNT in order to prevent one child from having some pain after 2 days was 15, and children who took antibiotics were more at risk of having adverse events. However, despite possible adverse reactions, antibiotics seem to be beneficial in children younger than 2 years with bilateral AOM (NNT = 4), and in children with both AOM and otorrhoea (NNT = 3). However, the pain on day 3 for those children who still have pain is mild and most parents used suboptimal doses of analgesics (Little et al. 2001). Hence, it is debatable whether it is worthwhile to treat children with antibiotics, particularly when analgesic use to relieve pain has not been optimised.
One Cochrane systematic review on the efficacy of antibiotics for acute bronchitis was identified (Fahey et al. 2004). The authors of this review included 9 RCTs involving 750 children and adults (aged 8 years and over) with acute bronchitis or acute productive cough without underlying pulmonary disease. Both smokers and non-smokers were included in the primary analysis and the duration of illness at entry was less than 30 days. The type of intervention in the studies was any antibiotic therapy versus placebo. The review excluded trials with patients diagnosed with pre-existing chronic bronchitis (that is, acute exacerbation of chronic bronchitis). The 9 studies were set in primary care (general practice) (5), a hospital ambulatory screening clinic (1) and hospital outpatient units (3).
This Cochrane review carried out meta-analyses using the pooled RR of having a cough, improvement on clinician’s global assessment, having an abnormal lung examination, duration of cough, duration of feeling ill and adverse events.
Overall, this Cochrane review showed that patients receiving antibiotics had better outcome for cough than the patients receiving placebo. Results from the meta-analyses showed that patients receiving antibiotics were less likely to have a cough 7 to 14 days after the initiation of treatment (4 studies) (pooled RR = 0.64, 95% CI 0.49 to 0.85, p = 0.002).
Results from the meta-analysis showed that patients receiving antibiotics were less likely to show no improvement on clinician’s global assessment (6 studies) (pooled RR = 0.52, 95% CI 0.31 to 0.87, p not provided) but the NNT was relatively high (NNT = 14, 95% CI 8 to 50).
Of the 6 studies, 4 were from primary care settings, with a total sample of 473. The 4 studies individually showed no differences between antibiotics and placebo (RR = 0.46, 95% CI 0.18 to 1.16; RR = 0.42, 95% CI 0.11 to 1.57; RR = 0.52, 95% CI 0.25 to 1.09; RR = 1.73, 95% CI 0.16 to 18.20, respectively).
For the outcome abnormal lung examination, the meta-analysis showed that patients receiving antibiotics were less likely to have an abnormal lung examination (5 studies) (pooled RR = 0.54, 95% CI 0.41 to 0.70, p < 0.00001; NNT = 11, 95% CI 6 to 50) compared with the placebo group.
Of the 5 studies, 4 were from primary care settings, with a total sample of 270. The pooled results of the meta-analysis of the 5 studies were heavily skewed by 1 large trial from a hospital setting that constituted 77.8% of the weight of the meta-analysis. When the 4 studies from primary care were examined separately, none showed any differences between antibiotics and placebo.
Further meta-analyses showed that patients receiving antibiotics had shorter durations of cough (5 studies) (weighted mean difference = −0.58 days, 95% CI −1.16 to −0.01 days), shorter productive cough (5 studies) (weighted mean difference = −0.52 days, 95% CI −1.03 to −0.01 days), and shorter duration of feeling ill (4 studies) (weighted mean difference = −0.58 days, 95% CI −1.16 to 0.00 days).
In total, 4 out of 5 studies on duration of cough and 4 out of 5 studies on duration of productive cough were from primary care settings, whereas all 4 studies on duration of feeling ill were from primary care settings. Although the results showed statistically significant reductions in illness durations, in practice the actual size of the reductions was small: all less than 1 day in duration.
The differences in adverse events (that is, adverse effects from antibiotics) (9 studies) were not statistically significant between the antibiotic group and the control group, with pooled RR = 1.22, 95% CI 0.94 to 1.58, p = 0.1.
Out of the 9 studies, 7 were from primary care settings with a total sample of 643 patients. None of the 7 studies showed any differences between the antibiotic group and the control group.
The summary findings of the Cochrane review are as follows. Patients receiving antibiotics are less likely to have a cough, with an NNT of 5. However, the NNTs for improvement on clinician’s global assessment and the likelihood of having an abnormal lung examination were considerably higher (14 and 11, respectively). Moreover, when those studies from primary care settings were examined individually, none showed significant effects of antibiotics in improving clinician’s global assessment and in reducing the likelihood of having an abnormal lung examination. Although there were significant effects of antibiotics on the durations of cough and productive cough, and on feeling ill, these were small – a fraction of 1 day in an illness lasting several weeks.
One Cochrane systematic review on the efficacy of antibiotics for sore throat was identified (Del Mar et al. 2006). The authors of this Cochrane review included 27 RCTs involving 2835 cases of sore throat (in adults and children). Of these RCTs, 17 did not distinguish between bacterial and viral aetiology (that is, the patients were only clinically judged by practitioners/researchers to have suspected group A beta-haemolytic Streptococcus pharyngitis (GABHS); no diagnostic investigations were carried out). However, 8 studies included only GABHS-positive patients, while 2 studies excluded patients who were GABHS-positive. The type of intervention in the studies was any antibiotic therapy versus placebo. The settings of the 27 studies were: US air force bases (8), general practices (10), paediatric clinics (4), hospitals (2) and not reported (3). Of the 10 studies set in primary care, 2 studies used a GABHS-positive result as an inclusion/exclusion criterion: 1 study excluded patients with a GABHS-negative throat swab and 1 study included only patients with a GABHS-negative throat swab. The remaining 8 studies set in primary care did not use GABHS as a strict inclusion or exclusion criterion; instead, patients were included if they were clinically judged by physicians to have simple sore throat/pharyngitis or if they had three or more Centor criteria. (Centor criteria have been developed to predict bacterial infection – presence of: tonsillar exudate, fever and cervical lymphadenopathy, and an absence of cough.) Some studies carried out throat swabs at follow-up visits to confirm the aetiology of sore throat.
The review carried out meta-analysis using pooled RR on two groups of outcome measures – incidence of complications (suppurative and non-suppurative), and symptoms of sore throat.
For non-suppurative complications, the findings from the meta-analysis (16 studies) showed that antibiotics reduced the incidence of acute rheumatic fever within 2 months (pooled RR = 0.29, 95% CI 0.18 to 0.44, p < 0.00001).
When the meta-analysis was further analysed, only 7 out of the 16 studies had recorded the incidence of rheumatic fever and these 7 studies were carried out between 1954 and 1961, when rheumatic fever was much more common than in later years. Moreover, only 6 out of the 16 studies were from primary care settings, with a total study sample of 2267 adults and children. None of the studies reported any cases of rheumatic fever.
Another non-suppurative complication in the meta-analysis was acute glomerulonephritis within 1 month (10 studies). The results showed antibiotic treatment did not reduce the incidence of acute glomerulonephritis (pooled RR = 0.22, 95% CI 0.02 to 2.02, p = 0.2). Again, only 2 studies out of the 10 recorded the incidence of acute glomerulonephritis (both studies were carried out before 1960).
Of the 8 studies that did not identify cases of acute glomerulonephritis, 4 were from primary care settings, with a total study sample of 2186 adults and children.
The incidence rates of rheumatic fever and acute glomerulonephritis have continued to decline in Western society. A recent retrospective cohort study using data from the UK General Practice Research Database between 1991 and 2001 (during which time there were 3.36 million episodes of RTI) (Petersen et al. 2007) claimed that it was difficult to examine rheumatic fever and acute glomerulonephritis as potential complications of sore throat because of the very small number of cases of these complications occurring after sore throat. Thus, any reported relative risk reduction in the efficacy trials must be viewed in the context of an extremely small absolute risk of developing both of these conditions in primary care settings after an episode of sore throat.
For suppurative complications, the findings showed that antibiotics reduced the incidence of AOM within 14 days (11 studies) (pooled RR = 0.28, 95% CI 0.15 to 0.52, p = 0.00005) and quinsy within 2 months (8 studies) (pooled RR = 0.14, 95% CI 0.05 to 0.39, p < 0.0002), but antibiotics did not reduce the incidence of acute rhinosinusitis (the study used the term acute sinusitis) within 14 days (8 studies) (pooled RR = 0.53, 95% CI 0.18 to 1.55, p = 0.2).
In the analysis of AOM, only 4 out of 11 studies were from primary care settings, with a total study sample of 1612 adults and children. Of the 1612 patients, only one case of AOM was recorded (in a control group). In the analysis of quinsy, 6 out of the 8 studies were from primary care settings, with a total study sample of 1810 adults and children. Of the 1810 patients, nine cases of quinsy were recorded (eight cases in control groups and one case in a treatment group). However, further analysis from the systematic review showed that the NNT for AOM was nearly 200. In the study by Little et al. (2002), the median annual incidence of hospital admission of quinsy (interquartile range) within the residents of the health authority who had acute uncomplicated RTIs was low, at 1.66 per 10,000.
Results from the meta-analysis showed that antibiotics reduced the symptom of throat soreness on day 3 (15 studies) (pooled RR = 0.72, 95% CI 0.68 to 0.76, p < 0.00001) and at 1 week (13 studies) (pooled RR = 0.65, 95% CI 0.55 to 0.76, p < 0.00001). Antibiotics also reduced the symptom of headache (3 studies) (pooled RR = 0.47, 95% CI 0.38 to 0.58, p < 0.00001) and fever on day 3 (7 studies) (pooled RR = 0.69, 95% C: 0.53 to 0.88, p = 0.003). No cases of fever (3 studies) were recorded at 1 week.
When further subgroup analyses were performed in the meta-analysis (GABHS-positive compared with GABHS-negative compared with untested/inseparable), the findings showed a different picture. For instance, for the symptom of throat soreness on day 3, all three subgroups (GABHS-positive, GABHS-negative and untested/inseparable) showed beneficial effect of antibiotics over placebo (RR = 0.59, 95% CI 0.54 to 0.64, RR = 0.79, 95% CI 0.71 to 0.88, RR = 0.89, 95% CI 0.80 to 0.99, respectively). Of the 11 studies from the GABHS-positive subgroup, 4 were from primary care settings; of the 6 studies from the GABHS-negative subgroup, 3 were from primary care settings; and all 3 studies from the untested/inseparable subgroup were from primary care settings.
However, for the symptom of throat soreness at 1 week, only the GABHS-positive subgroup showed a beneficial effect of antibiotics over placebo (RR = 0.28, 95% CI 0.17 to 0.44) but not the GABHS-negative subgroup and the untested/inseparable subgroup. Out of the 7 studies, 3 were from primary care settings.
Studies that used three or four of the Centor criteria for bacterial infection to determine eligibility (Dagnelie et al. 1996; Zwart et al. 2000) in the Cochrane review showed a little more benefit from antibiotics both for symptom resolution (of the order of 1 to 2 days at a time when symptoms are milder) and for the prevention of complications (NNT = 60). However, caution is required in generalising from these studies, which are from a setting where the level of antibiotic prescribing has traditionally been very low. A low level of antibiotic prescribing is likely to reduce consultation rates and result in a more severe illness spectrum among patients presenting in primary care.
As well as the Cochrane review (Del Mar et al. 2006), a systematic review on appropriate antibiotic use for acute pharyngitis in adults has been carried out (Cooper et al. 2001). In this review, the findings showed that treatment with antibiotics within 2 to 3 days of symptom onset hastened symptomatic improvement by 1 to 2 days in patients with three or more Centor Criteria (Centor et al. 1981) (where throat cultures of a significant proportion of these patients ultimately grew GABHS). However, antibiotics did not have this beneficial effect in patients with a negative GABHS culture. The Centor criteria include presence of tonsillar exudate, tender anterior cervical lymphadenopathy or lymphadenitis, history of fever and an absence of cough.
The summary findings of the Cochrane review are as follows. There is evidence from studies in selected populations carried out in the 1950s and 1960s that suggests a beneficial effect of antibiotics in reducing the incidence of rheumatic fever and acute glomerulonephritis following an episode of sore throat. However, observational studies show that these two complications are now extremely rare in modern Western society. Thus, the absolute risk of developing these complications following sore throat is now extremely small. The evidence from the Cochrane review also suggests that antibiotics confer relative benefits in preventing AOM and peritonsillar abscess (quinsy) butthe NNTs are high.
The evidence from the Cochrane review also suggests that antibiotics appear to have a modest beneficial effect in reducing the symptoms of throat soreness, fever and headache. However, nearly half of the study population in the review were GABHS-positive. Since current UK general practice does not use throat swabs or rapid diagnostic tests to detect the presence or absence of GABHS, primary care clinicians rely on symptoms and signs to decide on initial treatment with antibiotics. Studies in settings where antibiotic use is low and that use three or four of the Centor criteria for bacterial infection to determine eligibility do show some benefits from antibiotics both for symptom resolution (of the order of 1 to 2 days at a time when symptoms are milder) and for the prevention of complications. The symptomatic benefits from antibiotics within this subgroup of patients (with three or more Centor criteria) were also supported by results from the systematic review (Cooper et al. 2001).
There is one Cochrane systematic review on the efficacy of antibiotics for the common cold and acute purulent rhinitis (Arroll and Kenealy 2005). This review included 13 RCTs involving 2467 adults and children (aged 2 months and older) who had been diagnosed with an upper respiratory tract infection with symptoms for 7 days or acute purulent rhinitis of less than 10 days’ duration. The type of intervention in the studies was any antibiotic therapy versus placebo. The review excluded patients diagnosed with pharyngitis and bronchitis, conforming to the 1986 International Classification of Health Problems in Primary Care (ICHPPC) definition.
The settings of the 13 studies were: general practice (4), military bases (4), hospital outpatient units (2), unspecified research unit (factory and office workers) (1), accident and emergency department (1) and unknown (1).
The review carried out a meta-analysis using pooled RR on a number of outcome measures. The key outcomes were lack of cure or persistence of symptoms of nasopharyngeal inflammation on days 1 to 7 (rhinitis, sore throat and sneezing), acute persisting purulent rhinitis and adverse events.
The results from the meta-analysis showed that there were no significant findings for lack of cure or persistence of symptoms of nasopharyngeal inflammation (6 studies) (pooled RR = 0.89, 95% CI 0.77 to 1.04, p = 0.1).
The results from the meta-analysis (5 studies) also showed no significant benefit from antibiotics (pooled RR = 0.62, 95% CI 0.38 to 1.01, p = 0.06) for acute persisting purulent rhinitis.
Out of the 5 studies, 3 were from primary care settings, with a total study sample of 554 adults and children.
In addition to persistence of symptoms and acute persisting purulent rhinitis, 6 studies in the systematic review also reported adverse events experienced by individual patients. When all 6 studies were combined, the results showed that patients who took antibiotics were more at risk of having adverse events (adverse side effects from antibiotics) compared with the control group (pooled RR = 1.80, 95% CI 1.01 to 3.21, p = 0.05). However, there was also a high level of heterogeneity. When subgroup analyses were performed for adults and children, only adult patients who took antibiotics were more at risk of having adverse events compared with the control group (4 studies) (pooled RR = 2.62, 95% CI 1.32 to 5.18, p < 0.00001); no difference was found for children (2 studies) (pooled RR = 0.91, 95% CI 0.51 to 1.63, p = 0.8).
In the adult subgroup analysis, 2 out of the 4 studies were from primary care settings, with a total study sample of 946. In the children subgroup analysis, 1 of the 2 studies was from a primary care setting, with a study sample of only 188.
The summary findings of this Cochrane systematic review are that antibiotics are not effective in reducing persistence of common cold symptoms and adult patients may experience adverse events from antibiotic use.
There is also 1 Cochrane systematic review4 that addresses the efficacy of antibiotics for persistent nasal discharge (Morris and Leach 2002). This review included 6 RCTs involving 562 children (aged between 0 months and 18 years) with persistent nasal discharge for at least 10 days. For inclusion in the review, nasal discharge had to be the primary condition requiring medical intervention. Trials that only compared or combined antibiotics with surgery or sinus puncture and lavage were excluded. Trials that only compared two or more antibiotics without a non-antibiotic comparison group were also excluded from the systematic review. The type of intervention in the 6 studies was any antibiotic therapy versus placebo or standard therapy (standard therapy included decongestants or nasal saline drops). The settings of the 6 studies were: hospital paediatric allergy clinic (1), allergy referral clinic (1), general hospital (2), paediatric primary care practice (1) and hospital ear, nose and throat clinic (1). The paediatric primary care studies were from the United States.
This Cochrane review carried out meta-analyses using pooled RR for a number of outcome measures. The two key outcomes of the reviews were overall clinical failure (proportions of patients with nasal discharge at follow-up, or those with no substantial improvement if failure to cure rates were not available) and adverse events.
Results from the meta-analysis of overall clinical failure (6 studies) showed that antibiotics are modestly effective in reducing the probability of symptom persistence in children with nasal discharge of more than 10 days’ duration (pooled RR = 0.75, 95% CI 0.61 to 0.92, p = 0.005; NNT = 8, 95% CI 5 to 29).
Out of the 6 studies, only 1 was from a primary care setting, with a study sample of 161 children. The results of this study showed no benefit of antibiotics in reducing nasal discharge (RR = 0.91, 95% CI 0.48 to 1.07). The fact that this result is at variance with the pooled RR could be because the patients who attended hospital or allergy clinics were more ill or had more severe symptoms than patients seeking help in primary care practices.
In terms of adverse events, results from the meta-analyses (4 studies) also showed that there were no significant harmful side effects of antibiotics in the intervention group compared with the control group (pooled RR = 1.75, 95% CI 0.63 to 4.82, p = 0.3).
Of the studies that investigated adverse events, 1 out of 4 was from a primary care setting, with a study sample of 157 children. The Cochrane review also attempted to carry out subgroup analysis of very young children. However, only 1 small study was limited to children younger than 8 years and hence there was insufficient evidence to determine whether age has an impact on the effectiveness of antibiotics in children with persistent nasal discharge.
The summary findings of this Cochrane review are that, for children with persistent nasal discharge, the evidence suggests that antibiotics are effective in reducing the probability of persistence in the short to medium term only in children with nasal discharge of more than 10 days’ duration. However, the benefits appear to be modest and around eight children must be treated in order to achieve one additional cure. No long-term benefits have been documented in the review. Since only 1 study out of the total of 6 was from a primary care setting, and this particular study showed no benefit of antibiotics, the generalisability of the results from meta-analysis to a primary care population of children is uncertain.
The term ‘acute rhinosinusitis’ is used instead of ‘acute sinusitis’ for consistency throughout this guideline because acute rhinosinusitis is the terminology that is currently internationally accepted. However, in some medical literature, studies still refer to the condition as acute sinusitis. One Cochrane systematic review on the efficacy of antibiotics for acute rhinosinusitis (the study used the term acute maxillary sinusitis) was identified (Williams Jr et al. 2003). The review included 49 studies involving 13,660 patients. However, only 3 studies (out of 49) compared antibiotics with placebo (whereas the other studies compared one antibiotic with another). Hence, only these 3 studies are discussed in this overview. The 3 studies were RCTs and involved 416 adults (aged 18 years and older) with acute rhinosinusitis confirmed radiographically or by aspiration. An additional inclusion criterion was that trials must have a sample size of at least 30 participants with acute rhinosinusitis. The type of intervention in the meta-analysis was any antibiotic therapy versus placebo or a topical decongestant. The treatment duration ranged from 3 to 15 days and the settings of the 3 studies were: primary care (2), not reported (1).
Results from the meta-analysis showed that patients treated with amoxicillin were more likely to be clinically cured (2 studies) (pooled RR = 1.49, 95% CI 1.18 to 1.88, p = 0.001) or more likely to be clinically cured/much improved (2 studies) (pooled RR = 1.20, 95% CI 1.05 to 1.37, p = 0.007). Similarly, the meta-analysis showed that patients treated with penicillin V were more likely to be clinically cured (2 studies) (pooled RR = 1.79, 95% CI 1.05 to 3.05, p = 0.03) or more likely to be clinically cured/much improved (2 studies) (pooled RR = 1.25, 95% CI 1.01 to 1.54, p = 0.04).
Both studies using amoxicillin were based in primary care settings, with a total study sample of 303 adult patients. However, only 1 of the 2 studies with penicillin was set in primary care, with a study sample of 85 adult patients. Although there were significant results for both amoxicillin and penicillin treatment, they need to be interpreted very cautiously because neither X-ray nor aspiration are routinely performed or indicated in primary care settings. Thus, these results cannot be generalised to patients presenting with sinusitis-like complaints in primary care settings where the effect of antibiotics is likely to be less. The study sample in this review was also relatively small, involving only 375 adult patients.
In order to address the small sample and generalisability issues in the Cochrane review (Williams Jr et al. 2003), a current primary care-based IPDM (Young et al. 2008) of 2547 patients aged 12 years and older (from 9 RCTs) with clinical signs and symptoms of rhinosinusitis was also identified. In this IPDM, trials were excluded if patients were recruited partly on the basis of results of imaging or laboratory tests or bacterial culture because in a primary care setting such methods are not routinely used or recommended.
The results from this IPDM showed that 15 patients would have to be given antibiotics before an additional patient was cured (95% CI NNT [benefit] 7 to NNT [harm] 190). In this analysis of individual patients’ data, the estimated OR of the overall treatment effect for antibiotics relative to placebo was 1.37 (95% CI 1.13 to 1.66). Further subgroup analyses also showed that patients with the symptom of purulent discharge in the pharynx had a longer duration of illness with an NNT of 8 (95% CI NNT [benefit] 4 to NNT [harm] 47). The multiplicative of individual baseline signs or symptoms on the odds of cure if a patient remained untreated and on the OR for cure if treated were also analysed. The analyses showed that patients who were older, reported symptoms for longer, or reported more severe symptoms also took longer to cure but were no more likely to benefit from antibiotics than other patients. [age: odds of cure if untreated = 0.88 (95% CI: 0.81–0.96), OR for cure if treated = 1.04 (95% CI: 0.92–1.18); duration of symptoms: odds of cure if untreated = 0.90 (95% CI: 0.81–0.99), OR for cure if treated = 0.95 (95% CI: 0.82–1.10); symptom severity: odds of cure if untreated = 0.93 (95% CI: 0.90–0.91), OR for cure if treated = 0.99 (95% CI: 0.93–1.05)]. This IPDM showed that antibiotics are not justified for adult patients with rhinosinusitis-like complaints even if the patient reports symptoms for longer than 7–10 days. Although purulent discharge in the pharynx had some prognostic value, eight patients with this symptom still needed to be treated before one additional patient benefited.
The summary findings of this Cochrane review are that for acute rhinosinusitis (termed as acute maxillary sinusitis in William Jr et al. 2003) confirmed by radiography or aspiration, there is evidence that antibiotics make clinical cure more likely. However, these results cannot be generalised to patients in UK primary care settings where neither radiography nor aspiration is in use or recommended. This was supported by the IPDM (Young J et al. 2008), (which had inclusion criteria that reflected primary care settings), where the results showed that antibiotics are not justified for adult patients with rhinosinusitis-like complaints, even if the patient reports symptoms for longer than 7–10 days.
The previous section summarised the evidence underpinning the rationale for developing this short clinical guideline. There is good evidence that antibiotics are of limited efficacy in treating a large proportion of RTIs seen in adults and children in primary care.
The use of a no antibiotic or a delayed antibiotic prescribing strategy to reduce the inappropriate prescribing of antibiotics for RTIs has been advocated since the late 1990s (Little 2005). A potential advantage of a delayed prescribing strategy is that it offers a rapid ‘safety net’ for the small proportion of patients who develop complications or whose symptoms worsen significantly. A patient expecting antibiotics may also be more likely to agree with this course of action rather than with a no prescribing strategy, and this could help to maintain the doctor-patient relationship.
There is therefore a need to determine whether the use of a no prescribing strategy or a delayed prescribing strategy is clinically and cost effective compared with the use of an immediate antibiotic prescribing strategy. It is also important to consider whether there are benefits from using a printed information leaflet or structured verbal information to deliver the chosen antibiotic management strategy. A delayed antibiotic prescribing strategy may be delivered in primary care settings in a number of ways: patients may be issued with a prescription at the consultation but advised to use it only if symptoms persist or worsen, or they may be asked to re-attend to collect the prescription from the general practice (surgery) reception. It is important to determine which of these delivery methods is the most effective.
In order to make the recommendations as useful as possible for clinicians it has been necessary to include specific information on likely illness duration for each of the five reviewed conditions. Expected duration of illness is a factor in the decision about when to start a delayed prescription. It is outside the scope of this short clinical guideline to conduct a systematic review of illness duration. However, the evidence from the included clinical trials together with other relevant identified studies has been used to support the consensus recommendations made by the GDG in this area.
We identified 2 systematic reviews and 12 published studies on the effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing RTIs compared with an immediate antibiotic prescribing strategy. No meta-analyses were carried out in the 2 systematic reviews because of significant heterogeneity across studies. Heterogeneity included variations in the treatment and symptoms of different RTIs and in the methods and duration of delayed prescribing. Both of these reviews therefore provide a narrative systematic review. A Cochrane review (Spurling et al. 2007) included 9 RCTs and 1 review. (Arroll et al. 2003) included 4 RCTs and 1 before-and-after controlled trial. All 4 RCTs in the review by Arroll et al. (2003) were also included in the Cochrane review.
Apart from the 2 systematic reviews, 29 published individual studies were also identified based on study abstracts. Out of these 29 studies, only 12 were included in the evidence review. (12 studies were not relevant, 4 were excluded as they were non-RCT studies and 1 lacked generalisability because it was carried out in a developing country). Of the 12 RCTs included, 8 were the same 8 RCTs presented in the Cochrane systematic review; 1 extra study that was not included in the Cochrane review was identified and a further 3 studies looked at the use of specific information leaflets or structured explanations in antibiotic management strategies for RTIs. All 12 included studies were appraised individually and presented in the evidence tables and GRADE (Grading or Recommendations Assessment, Development and Evaluation) profiles. For the methodology of GRADE, see section 4.2.7.
Of the 12 included RCTs, 3 were on AOM (1 from UK general practice, 1 from a USA paediatric emergency department and 1 from a USA university paediatric clinic); 2 were on cough (both from UK general practice); 3 were on sore throat (1 from UK general practice and 2 from USA private paediatric practice) and 1 was on common cold (from New Zealand general practice). No studies were identified on acute rhinosinusitis. Out of these 9 studies, 4 were open pragmatic RCTs. Open pragmatic trials lack internal validity but seek to maximise external validity to ensure that the results reflect everyday practice more closely and are more generalisable (Fransen et al. 2007; Godwin et al. 2003). Open pragmatic trials are appropriate for answering questions on effectiveness and for assessing outcomes in situations where perceptions and behaviour in everyday practice and patients’ knowledge of treatment are important factors. The remaining 3 RCTs (out of 12) were on the use of specific information leaflets or structured explanations in antibiotic management strategies for RTIs (2 from UK general practice and 1 from a primary care clinic in Israel).
The natural history or usual course of illness duration of the five RTIs was also identified from various sources. The average duration of AOM is about 4 days (Little et al. 2001); the average duration of acute cough/acute bronchitis is about 3 weeks (Little et al. 2005); the symptoms accompanying acute sore throat/acute pharyngitis/acute tonsillitis last on average for 1 week (Little et al. 1997); the average duration of symptoms of the common cold is around 1.5 weeks (Heikkinen and Jarvinen 2003); and the average duration of acute rhinosinusitis is around 2.5 weeks (Williamson et al. 2007).
Overall, the quality of the evidence was good and the studies provided the evidence statements that form the basis of the guideline recommendations. There were particular challenges in summarising and presenting the evidence on the effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing self-limiting RTIs. This was because of significant variations in factors such as patient populations, methods of delaying antibiotic prescription, duration of delays in antibiotic prescribing and outcome measures. The use of the GRADE approach to summarising the evidence was found to be helpful in addressing these challenges. For full GRADE evidence profiles see appendix 4.
At the first face-to-face contact in primary care, including walk-in centres and emergency departments, adults and children (3 months and older) presenting with a history suggestive of the following conditions should be offered a clinical assessment:
The clinical assessment should include a history (presenting symptoms, use of over-the-counter or self medication, previous medical history, relevant risk factors, relevant comorbidities) and, if indicated, an examination to identify relevant clinical signs.
Patients’ or parents’/carers’ concerns and expectations should be determined and addressed when agreeing the use of the three antibiotic prescribing strategies (no prescribing, delayed prescribing and immediate prescribing).
A no antibiotic prescribing strategy or a delayed antibiotic prescribing strategy should be agreed for patients with the following conditions:
Depending on clinical assessment of severity, patients in the following subgroups can also be considered for an immediate antibiotic prescribing strategy (in addition to a no antibiotic or a delayed antibiotic prescribing strategy):
Centor criteria are: presence of tonsillar exudate, tender anterior cervical lymphadenopathy or lymphadenitis, history of fever and an absence of cough.
For all antibiotic prescribing strategies, patients should be given:
When the no antibiotic prescribing strategy is adopted, patients should be offered:
When the delayed antibiotic prescribing strategy is adopted, patients should be offered:
A delayed prescription with instructions can either be given to the patient or left at an agreed location to be collected at a later date.
Three studies were included in the review of AOM (Little et al. 2001; McCormick et al. 2005; Spiro et al. 2006). The patient population for the Spiro study was children diagnosed with AOM (aged between 6 months and 12 years) and the patient population for the study by Little was children aged between 6 months and 10 years presenting with AOM. The patient population for the study by McCormick was children aged between 6 months and 12 years with AOM (screened using an AOM severity screening index).
The 3 studies had different settings: Spiro’s study was carried out in a paediatric emergency department in the United States, McCormick’s study was carried out at the University of Texas Medical Branch paediatric clinic and the Little study was conducted in 42 general practices in southwest England.
The studies differed in terms of inclusion criteria. In Spiro’s study, children were included if they were clinically diagnosed with AOM in an emergency department; in Little’s study, children with acute earache (otalgia) and otoscopic evidence of acute inflammation of the eardrum (dullness or cloudiness with erythema, bulging or perforation) were included. However, if children were too young for earache to be documented, then otoscopic evidence alone was a sufficient entry criterion. In McCormick’s study children were included if they had symptoms of ear infection, otoscopic evidence of AOM including middle ear effusion, and non-severe AOM.
| Study | Spiro et al. (2006) | Little et al. (2001) | McCormick el al. (2005) |
|---|---|---|---|
| Antibiotic prescribing strategy | Delayed | Delayed | Delayed |
| Duration of delay | 2 days | 3 days | 2 days |
| Methods of delay | Prescription was given to parents during the consultation with the healthcare professional. | Parents were asked to come back to collect the prescription (prescription left at the reception). | Prescription was given to parents during the consultation with the healthcare professional. |
| Verbal advice | No | Parents were also advised to use the prescription if their child had a discharge for 10 days or more. GPs were supported by standardised advice sheets. Advice on antibiotics given: that antibiotics do not work very well and have disadvantages such as adverse events and development of antibiotic resistance. | Parents of children received an educational intervention on definition of ear infection, causes of ear infection, characteristics of non-severe and severe AOM, antibiotic resistance, costs of antibiotics, rate of symptom response to antibiotics, possible adverse outcomes associated with immediate antibiotics versus delayed including the risk of mastoiditis. |
| Use of information leaflet | No | No | No |
| Use of analgesics | All patients received ibuprofen (100 mg/5 ml) and otic analgesic drops (4 drops every 2 hours if needed). | Advice on full doses of paracetamol for relief of pain and fever. Ibuprofen as well if child already taking full doses of paracetamol and is aged over 1 year. | Symptom medication provided (ibuprofen). |
| Immediate | Immediate | Immediate | |
| Duration of delay | N/A | N/A | N/A |
| Methods of delay | N/A | N/A | N/A |
| Verbal advice | No | GPs were supported by standardised advice sheets. Advice on benefit of antibiotics in helping symptoms to settle and prevent complications; importance of taking the full course. | Parents of children received an educational intervention on definition of ear infection, causes of ear infection, characteristics of non-severe and severe AOM, antibiotic resistance, costs of antibiotics, rate of symptom response to antibiotics, possible adverse events associated with immediate antibiotic versus delayed antibiotic prescribing, including the risk of mastoiditis. |
| Use of information leaflet | No | No | No |
| Use of analgesics | All patients received ibuprofen (100 mg/5 ml) and otic analgesic drops (4 drops every 2 hours if needed; each ml contains 54 mg antipyrene, and 14 mg benzocaine). | Advice on full doses of paracetamol for relief of pain and fever. Ibuprofen as well if child already taking full doses of paracetamol is aged over 1 year. | Symptom medication provided (ibuprofen). |
The effectiveness of delayed antibiotic prescribing as a strategy for managing acute otitis media Summary of findings
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Use of antibiotics after consultation [S, L & M] | 3 (758) | RCT | Delayed 120/382 (31%) | Immediate 357/376 (94%) | 0.33 (0.29, 0.39) | High |
| Otalgiag [S & L] | 2 (550) | RCT | Delayed 130/282 (46%) | Immediate 108/268 (40%) | 1.18 (0.99, 1.40) | High |
| Daily pain score (1 to 10) – daily diary (severity) (over 1 week) [L] | 1 (285) | RCT | Delayed 150 | Immediate 135 | Mean difference = −0.16 (−0.42, 0.11) t = 1.18, p = 0.24 | High |
| Night disturbances – daily diary (over 1 week) [L] | 1 (285) | RCT | Delayed 150 | Immediate 135 | Mean difference = −0.72 (−0.30, −1.13) t = 3.41, p < 0.01 | High |
| Diarrhoea [S&L] | 2 (550) | RCT | Delayed 24/282 (9%) | Immediate 56/268 (21%) | 0.41 (0.26, 0.65) | High |
| Belief antibiotics are effective [L] | 1 (271) | RCT | Delayed 64/140 (46%) | Immediate 100/131 (76%) | 0.59 (0.48, 0.73) | High |
| Very satisfied with treatment approach (parents/carers) [L] | 1 (284) | RCT | Delayed 115/150 (77%) | Immediate 123/134 (91%) | 0.84 (0.75, 0.93) | High |
| Parents’/carers’ satisfactioni [M] | 1 (209) | RCT | Delayed 100 | Immediate 109 | Total satisfaction scores: On day-12: I = 44.0, C = 44.4 On day-30: I = 44.6, C = 44.6 (not significant; p value not reported) | Moderate |
intervention = delayed antibiotics
control = immediate antibiotics
presence of earache/otalgia: [S] data collected at follow-up (4 to 6 days); [L] data collected through daily diary (at 1 week).
total satisfaction scores (4-point-scale). Data on [L] and [M] were not pooled owing to different methods of measurements.
Three large trials provide good evidence supporting the effectiveness of delayed antibiotic prescribing as a strategy for managing suspected AOM.
The GDG acknowledged that the 3 included studies were of reasonably good quality but that the study population was limited to children. Based on the evidence statements presented above, the GDG came to the conclusion that, compared with a delayed prescribing strategy, an immediate prescribing strategy provided modest benefits in reducing earache and night disturbances but increased the consumption of antibiotics and potentially might medicalise a self-limiting illness. However, the GDG also considered that the benefits of an immediate antibiotic prescription were very limited because by day 3, the pain of those children who still had earache was mild and it is debatable whether these limited benefits would outweigh the likelihood of having diarrhoea if immediate antibiotics were used. The GDG also discussed the outcome of parents’/carers’ satisfaction and noted that the overall satisfaction rates were high for both strategies. However, the GDG considered that the high satisfaction rate for an immediate prescribing strategy compared with that for a delayed prescribing strategy in 1 trial (Spiro et al. 2006) could be a result of the method of delivery (in which parents were given the delayed prescription during the consultation instead of being asked to come back to collect it at the surgery reception). The GDG agreed that this conclusion is tentative and remains a point for speculation, and that further research needs to be carried out on the mode of delivery of delayed prescribing strategies in order to clarify advice about practice. Overall, by weighing both the modest benefits and the risk of diarrhoea, the GDG thought that a delayed or no prescribing strategy should be offered to children with AOM who are not at risk of developing complications. However, owing to the lack of trials among important subgroups comparing an immediate and/or delayed prescribing strategy with a no prescribing strategy, the GDG thought that a consensus recommendation on the likely symptomatic benefits of antibiotics for particular subgroups of patients should be made. The GDG agreed that, based on the meta-analysis with individual patient data (Rovers et al. 2006) presented in section 2.1.2, an immediate prescribing strategy may be considered for two subgroups of patients depending on clinical assessment of severity and patient preference. The two subgroups are: children younger than 2 years with bilateral acute otitis media, and children with acute otitis media and otorrhoea.
Two studies were included in the review of acute cough/acute bronchitis: Dowell et al. (2001) and Little et al. (2005). The population in Dowell’s study consisted of patients aged over 16 years presenting with acute cough as the primary complaint. The patient population in Little’s study consisted of children aged 3 years and older with uncomplicated acute LRTI (duration 21 days or less). Both studies were set in UK primary care (general practice): Dowell’s study was set in 22 general practices with 48 GPs in Scotland; Little’s study involved 37 GPs in southwest England.
The inclusion criteria differed in the studies. In Dowell’s study patients with acute cough with or without coryza, shortness of breath, sputum, fever, sore throat or chest tightness were included. In Little’s study patients with cough (with a duration of 21 days or less) as the main symptom and with at least one symptom or sign localising to the lower respiratory tract (sputum, chest pain, dyspnoea, wheeze) were included.
| Study | Dowell et al. (2006) | Little et al. (2005) | Little et al. (2005) |
|---|---|---|---|
| Antibiotic prescribing strategy | Delayed | Delayed | No |
| Duration of delay | 1 week | 2 weeks | N/A |
| Methods of delay | Patients were asked to come back to collect the prescription for antibiotic (prescription left at the surgery reception). | Patients were asked to come back to collect the prescription for antibiotic (prescription left at the surgery reception). | |
| Verbal advice | No | All patients, irrespective of whether they had the leaflet, were given brief verbal information about the likely course of the illness and supporting the proposed prescribing strategy. | All patients, irrespective of whether they had the leaflet, were given brief verbal information about the likely course of the illness and supporting the proposed prescribing strategy. |
| Use of information leaflet | Information (patient information sheet) was given at consultation during recruitment. Content not reported. | 50% of patients received information leaflet, 50% did not. Leaflet included information about natural history, addressed patients’ major worries and provided advice about when to seek further help (for example, if persistent fever, worsening shortness of breath). | 50% of patients received info leaflet, 50% did not. Leaflet included information about natural history, addressed patients’ major worries and provided advice about when to seek further help (for example, persistent fever, worsening shortness of breath). |
| Use of analgesics | No | Advice to take an analgesic | Advice to take an analgesic |
| Antibiotic prescribing strategy | Immediate | Immediate | |
| Duration of delay | N/A | N/A | |
| Methods of delay | N/A | N/A | |
| Verbal advice | No | All patients, irrespective of whether they had the leaflet, were given brief verbal information about the likely range of natural history of the illness and supporting the proposed prescribing strategy. | |
| Use of information leaflet | Information (patient information sheet) was given at consultation during recruitment. Content not reported. | 50% of patients received information leaflet, 50% did not. Leaflet included information about natural history, addressed patients’ major worries and provided advice about when to seek further help (for example, persistent fever, worsening shortness of breath). | |
| Use of analgesics | No | Advice to take an analgesic | |
The effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing acute cough/acute bronchitis Summary of findings
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Collection of antibiotic prescriptiona [D] | 1 (187) | RCT | Delayed 43/95 (45%) | Immediate 92/92 (100%) | 0.45 (0.36, 0.56) | High |
| Use of antibiotics [L] | 1 (390) | RCT | Delayed 39/197 (20%) | Immediate 185/193 (96%) | 0.20 (0.15, 0.27) | High |
| Use of antibiotics [L] | 1 (375) | RCT | No AB 29/182 (16%) | Immediate 185/193 (96%) | 0.16 (0.11, 0.23) | High |
| Use of antibiotics [L] | 1 (379) | RCT | No AB 29/182 (16%) | Delayed 39/197 (20%) | 0.80 (0.52, 1.24) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Mean difference | Quality |
|---|---|---|---|---|---|---|
| Symptom duratione (cough) [D] | 1 (148) | RCT | Delayed unknown | Immediate unknown | Log-rank [Mantel-Haenszel] test (result not reported), with p value > 0.4 | Moderate |
| Symptom durationg (cough) [L] | 1 (426) | RCT | Delayed 214 | No antibiotic 212 | Mean difference = 0.75 (−0.37, 1.88) p = 0.19 | High |
| Symptom durationg (cough) [L] | 1 (426) | RCT | Immediate 214 | No antibiotic 212 | Mean difference = 0.11 (−1.01, 1.24) p = 0.19 | High |
| Symptom durationg (cough) [L] | 1 (428) | RCT | Immediate 214 | Delayed 214 | Mean difference = −0.46 (−1.76, 0.48) p = 0.265 | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Mean difference | Quality |
|---|---|---|---|---|---|---|
| Adjusted severity of symptomsh[L] | 1 (426) | RCT | Delayed 214 | No antibiotic 212 | Adjusted mean difference = −0.02 p = 0.86 | High |
| Adjusted severity of symptomsh[L] | 1 (426) | RCT | Immediate 214 | No antibiotic 212 | Adjusted mean difference = −0.07 p = 0.49 | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Odds ratio | Quality |
|---|---|---|---|---|---|---|
| Diarrhoea [L] | 1 (426) | RCT | Delayed | No antibiotic | 0.17 (0.67, 2.03) | High |
| Diarrhoea [L] | 1 (426) | RCT | Immediate | No antibiotic | 1.22 (0.70, 2.12) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Re-attendance within 1 month [L] | 1 (389) | RCT | Delayed 24/199 (12%) | No antibiotic 41/190 (22%) | 0.55 (0.35, 0.88) | High |
| Re-attendance within 1 month [L] | 1 (386) | RCT | Immediate 26/196 (13%) | No antibiotic 41/190 (22%) | 0.61 (0.39, 0.96) | High |
| Re-attendance within 1 month [L] | 1 (395) | RCT | Delayed 24/199 (12%) | Immediate 26/196 (13%) | 0.90 (0.54, 1.52) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Belief antibiotics are effective [L] | 1 (306) | RCT | Delayed 57/141 (40%) | Immediate 123/165 (75%) | 0.54 (0.43, 0.67) | High |
| Belief antibiotics are effective [L] | 1 (296) | RCT | No AB 61/131 (47%) | Immediate 123/165 (75%) | 0.62 (0.50, 0.76) | High |
| Belief antibiotics are effective [L] | 1 (272) | RCT | No AB 61/131 (47%) | Delayed 57/141 (40%) | 1.15 (0.87, 1.51) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Patient satisfactioni[D] | 1 (148) | RCT | Delayed 40/73 (54%) | Immediate 55/75 (73%) | 0.74 (0.58, 0.95) | High |
| Patient satisfactionk[L] | 1 (384) | RCT | Delayed 147/190 (77%) | Immediate 166/194 (86%) | 0.90 (0.82, 0.99) | High |
| Patient satisfactionk [L] | 1 (375) | RCT | No antibiotic 130/181 (72%) | Immediate 166/194 (86%) | 0.83 (0.75, 0.93) | High |
| Patient satisfactionk [L] | 1 (371) | RCT | No antibiotic 130/181 (72%) | Delayed 147/190 (77%) | 0.92 (0.82, 1.04) | High |
rates of consumption unknown
intervention = delayed antibiotics
control = immediate antibiotics
probability of recovery from cough over days 1 to 13
duration of cough – days (until very little problem)
on a point scale 0 to 6 on six symptoms (adjusted on baseline variables): cough, dyspnoea, sputum production, wellbeing, sleep disturbance, activity disturbance
‘very satisfied’ with the consultation
‘very satisfied’ with overall management.
One large and one smaller trial provide good evidence on the effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing acute cough.
The GDG acknowledged that the 2 included studies were both of good quality. Based on the evidence statements presented above, the GDG concluded that delayed and no antibiotic prescribing strategies significantly reduced the consumption of antibiotics and lessened beliefs that antibiotics were effective in patients with acute cough. There were no significant differences in managing symptom duration/severity compared with an immediate prescribing strategy. The GDG also considered that the evidence statement on patient satisfaction showed that overall, patients with cough are satisfied with all three management strategies (all with satisfaction rates above 70%). The GDG thought that the differences in satisfaction rates between delayed/no prescribing and immediate prescribing could be confounded by the methods of delivery (such as ways of collecting delayed prescriptions, verbal advice provided or the amount of information provided on symptomatic treatment) rather than reflecting differences in the antibiotic management strategies per se. However, the GDG recognised that currently there are no specific studies that address the issue of the best way to deliver a delayed prescribing strategy. In conclusion, the GDG considered that a delayed or no prescribing strategy should be offered to patients with acute cough who are not at an increased risk of developing complications.
Three studies were included in the review of acute sore throat (suspected pharyngitis or tonsillitis): Gerber et al. (1990), Little et al. (1997) and; Pichichero et al. (1987). The 3 included studies had different patient populations. The study population in Little consisted of patients aged 4 years and older with sore throat and an abnormal physical sign in the throat (84% had tonsillitis or pharyngitis). In contrast, the other 2 studies included only patients who were culture positive for GABHS pharyngitis: in the Pichichero study patients were aged between 4 and 18 years, and in the Gerber study patients were aged between 2 and 22 years.
In terms of study setting, only 1 (Little) was based in UK primary care (general practice – 25 GPs). The other 2 studies were based in a single paediatric clinic in the USA There are also differences in study design among the 3 studies: in the delayed arm of 1 study, (Little et al. 1997), patients were asked to return after 3 days to collect the prescription, which had been left at the surgery reception. In the delayed arms of the other 2 studies, placebo tablets were used as a method of delay for the first 48 hours and followed by a 10-day course of antibiotics.
The inclusion criteria in Little’s study were sore throat, either as principal or subsidiary symptom, and an abnormal physical sign localising to the throat (inflamed tonsils or pharynx, purulent exudate, faucial or palatal inflammation or cervical adenopathy). For children younger than 12 years, who were less likely to complain of sore throat, abnormal signs in the throat were sufficient.
The inclusion criteria for the children in the study by Pichichero were three of the following signs or symptoms compatible with the diagnosis of GABHS pharyngitis:
The inclusion criteria in the study by Gerber were a positive Q test Streptococcus result and a positive throat culture.
| Study | Little et al. (1997) | Pichichero et al. (1987) | Gerber et al. (1990) | Little et al. (1997) |
|---|---|---|---|---|
| Antibiotic prescribing strategy | Delayed | Delayed | Delayed | No |
| Duration of delay | 3 days | 2 days | 2 days | N/A |
| Methods of delay | Patients were asked to return to collect the prescription for antibiotic (prescription left at the surgery). | Use of placebo tablets | Use of placebo tablets | N/A |
| Verbal advice | The advice package given to patients (in each group) had six or seven standard statements supporting the particular strategy. | No | No | The advice package given to patients (in each group) had six or seven standard statements supporting the particular strategy. |
| Use of information leaflet | No | No | No | No |
| Use of analgesics | Advice to take analgesics or antipyretics. | Encouraged to take aspirin or acetaminophen (paracetamol) ad libitum every 4 hours as needed to control fever and discomfort. | No | Advice to take analgesics or antipyretics. |
| Little (97) | Pichichero (87) | Gerber (90) | ||
| Immediate | Immediate | Immediate | ||
| Duration of delay | N/A | N/A | N/A | |
| Methods of delay | N/A | N/A | N/A | |
| Verbal advice | The advice package given to patients (in each group) had six or seven standard statements supporting the particular strategy | No | No | |
| Use of information leaflet | No | No | No | |
| Use of analgesics | Advice to take analgesics or antipyretics. | Encouraged to take aspirin or acetaminophen ad libitum every 4 hours as needed to control fever and discomfort. | No | |
The effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing acute sore throat/acute pharyngitis/acute tonsillitis
Summary of findings
| Outcome | No. of studies (total patients) | Design | Intervention | Control | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Use of antibiotics [L] | 1 (385) | RCT | No 23/174 (13%) | Immediate 210/211 (99%) | 0.13 (0.09, 0.19) | High |
| Use of antibiotics [L] | 1 (387) | RCT | Delayed 55/176 (31%) | Immediate 210/211 (99%) | 0.31 (0.25, 0.39) | High |
| Use of antibiotics [L] | 1 (350) | RCT | No 23/174 (13%) | Delayed 55/176 (31%) | 0.42 (0.27, 0.65) | High |
| Outcome | No. of studies (total patients) | Design | Kruskal-Wallis, X2 | Quality |
|---|---|---|---|---|
| Resolution of symptoms by 3 daysa [L] | 1 (561) | RCT | No = 35%; immediate = 37%; delayed = 30% X2 = 2.50, p = 0.28 | High |
| Outcome | No. of studies (total patients) | Design | Student t-test Kruskal-Wallis, X2 | Quality |
|---|---|---|---|---|
| Sore throatc (severity) [P] | 1 (114) | RCT | Mean score, student t-test Delayed = 1.6, Immediate = 1.3, p = 0.006 | Moderate |
| Sore throatd (duration) [L] | 1 (561) | RCT | Median (IQR), Kruskal-Wallis, X2 Delayed = 5(3–7), No AB = 5(3–7), Immediate = 4(3–6) X2 = 1.9, p = 0.39 | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Diarrhoea [L] | 1 (394) | RCT | Delayed 23/179 (13%) | Immediate 23/215 (11%) | 1.02 (0.69, 2.06) | High |
| Diarrhoea [L] | 1 (401) | RCT | No 16/186 (9%) | Immediate 23/215 (11%) | 0.80 (0.43, 1.47) | High |
| Diarrhoea [L] | 1 (365) | RCT | No 16/186 (9%) | Delayed 23/179 (13%) | 0.66 (0.36, 1.22) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Re-consultation with sore throat (within 1 month) [L] | 1 (484) | RCT | Delayed 12/238 (5%) | Immediate 22/246 (9%) | 0.56 (0.28, 1.11) | High |
| Re-consultation with sore throat (within 1 month) [L] | 1 (478) | RCT | No 22/232 (9%) | Immediate 22/246 (9%) | 1.06 (0.60, 1.86) | High |
| Re-consultation wit sore throat (within 1 month) [L] | 1 (470) | RCT | No 22/232 (9%) | Delayed 12/238 (5%) | 1.88 (0.95, 3.71) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Re-consultation with sore throat (within 12 months) [L] | 1 (317) | RCT | Delayed 50/169 (30%) | Immediate 90/148 (61%) | 0.48 (0.37, 0.63) | High |
| Re-consultation with sore throat with(in 12 months) [L] | 1 (297) | RCT | No 70/149 (47%) | Immediate 90/148 (61%) | 0.77 (0.62, 0.95) | High |
| Re-consultation with sore throat (within 12 months) [L] | 1 (318) | RCT | No 70/149 (47%) | Delayed 50/169 (30%) | 1.58 (1.19, 2.11) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Belief antibiotics are effective [L] | 1 (372) | RCT | Delayed 99/165 (60%) | Immediate 181/207 (87%) | 0.68 (0.59, 0.78) | High |
| Belief antibiotics are effective [L] | 1 (380) | RCT | No 95/173 (55%) | Immediate 181/207 (87%) | 0.62 (0.54, 0.72) | High |
| Belief AB are effective [L] | 1 (338) | RCT | No 95/173 (55%) | Delayed 99/165 (60%) | 0.91 (0.76, 1.09) | High |
| Outcome | No. of studies (total patients) | Design | Interventionb | Controlc | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Patient satisfactionk [L] | 1 (388) | RCT | Delayed 165/177 (93%) | Immediate 202/211 (96%) | 0.97 (0.92, 1.02) | High |
| Patient satisfactionk [L] | 1 (395) | RCT | No 166/184 (90%) | Immediate 202/211 (96%) | 0.94 (0.89, 0.99) | High |
| Patient satisfactionk [L] | 1 (361) | RCT | No 166/184 (90%) | Delayed 165/177 (93%) | 0.96 (0.90, 1.02) | High |
symptoms included sore throat, cough, headache, feeling unwell and fever
the presence and severity of symptom from checklist scale 1 to 3 (day 3).
median (interquartile range) duration of symptom (days) after 3 days
data were not pooled owing to different methods of measurements
satisfaction with consultation (scoring ‘very’ or ‘moderate’)
Two large trials and one small trial provide mixed qualities of evidence on the effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing acute sore throat. The evidence suggests the following.
The GDG acknowledged that the 3 included studies were of mixed quality. Based on the evidence statements presented above, the GDG concluded that in patients with acute sore throat, a delayed and a no prescribing strategy significantly reduced the consumption of antibiotics and lessened beliefs that antibiotics were effective. The GDG also reviewed the effectiveness of different antibiotic prescribing strategies and concluded that delayed and no prescribing strategies showed no significant differences in managing symptom duration or resolution compared with an immediate prescribing strategy. The GDG thought that the only study providing evidence of a modest beneficial effect of immediate antibiotics in reducing the severity of symptoms of acute sore throat related to a study population of patients with confirmed GABHS pharyngitis. The GDG thought that this could not be generalised to UK primary care settings because diagnostic tests to determine the cause of sore throat are not currently routinely used. Nevertheless, the GDG considered the results from the two-way sensitivity analysis undertaken as part of the economic evaluation (see section 4.2.9 and appendix 5). In that analysis both the baseline probability of developing quinsy and the efficacy of immediate antibiotic prescribing (as determined by the probability of symptoms resolving after 3 days) were varied simultaneously in the model. The GDG noted that the relative risk of developing complications remained constant (that is, at its baseline values) in the analysis. While the analysis indicated that there were situations in which immediate antibiotic prescribing could be considered cost effective, these situations depended on making arguably extreme assumptions. The GDG also considered that these results should be interpreted with caution for two main reasons. First, the lack of relevant utility estimates was an important limitation of the economic evaluation. Second, the absence of evidence on the rate of complications resulting from a strategy of delayed antibiotic prescribing made the interpretation of the results problematic. Consequently, the GDG thought that there could be exceptional scenarios in which immediate prescribing could be an option, in addition to strategies involving delayed or no antibiotic prescribing. In these situations, the choice of strategy should be based on a discussion between the healthcare professional and the patient/carer. Based on the two studies from the Cochrane review (Dagnelie et al. 1996), (Zwart et al. 2000) and the systematic review (Cooper et al. 2001) in section 2.1.2 that suggested symptomatic benefits of antibiotics for subgroups of patients with sore throat, the GDG considered that the Centor criteria could be a useful means of identifying individuals with acute sore throat who may benefit from immediate prescribing. At the same time the GDG acknowledged that this means of risk stratification was not explored in the economic model because of data limitations.
In conclusion, the GDG came to the consensus that a delayed or a no prescribing strategy should be offered to patients with acute sore throat who are not at an increased risk of developing complications. However, depending on patient preference and clinical assessment of severity, an immediate prescribing strategy may be considered for subgroups of patients with three or more Centor criteria in addition to the reasonable options of a no antibiotic strategy or a delayed prescribing strategy.
A literature review was conducted to identify cost-effectiveness evidence on the five relevant RTIs (see section 2 for details).
A number of potentially useful studies were identified (Anzai et al. 2007; Balk et al. 2001; Coco 2007; Davey 1994; de Bock et al. 2001; Dippel et al. 1992; Hillner and Centor 1987; Koskinen et al. 2006; Neuner et al. 2003; Singh et al. 2006; Tsevat and Kotagal 1999; Van Howe and Kusnier 2006). Three studies examined the cost effectiveness of management strategies for sinusitis (Anzai et al. 2007; Balk et al. 2001; de Bock et al. 2001) and 2 studies examined strategies for managing otitis media (Coco 2007 and Koskinen et al. 2006). Hillner and Centor (1987), Neuner et al. (2003) and Singh et al. (2006) examined the cost effectiveness of the diagnosis and management of adults with pharyngitis. Dippel et al. (1992), Tsevat and Kotagal (1999) and Van Howe and Kusnier (2006) looked at the diagnosis and management of children with pharyngitis.
Only 1 study specifically examined delayed prescribing versus no prescribing in a full cost-utility analysis (Coco 2007). This study was quality assessed and data extracted into evidence tables (see appendix 6). The majority of studies examined strategies for the diagnosis of RTIs and did not follow up patients after a result was obtained. No UK-based studies examining delayed versus immediate or no antibiotic prescribing for RTIs were identified and no studies were identified that examined cold or acute cough/acute bronchitis.
Coco (2007) examined the cost effectiveness of treatment options for AOM. The objective of this USA-based study was to evaluate the costs and utility of four treatment options for children with AOM aged from 6 months to 12 years.
The setting was primary care offices. Four intervention strategies were included: watchful waiting, delayed prescription, 5 days of immediate amoxicillin, and 7 to 10 days of immediate amoxicillin. A decision analytic model was used to evaluate the incremental cost effectiveness of the four strategies by comparing short-term outcomes and cost utilities. The analysis adopted a societal perspective and included non-healthcare costs associated with parental work loss and transportation. The time horizon of the analysis was 30 days. The authors state that this reflects the lack of evidence on long-term outcomes for otitis media such as recurrent AOM and tympanic membrane rupture.
Effectiveness estimates for the clinical parameters, including non-attendance, clinical failure with attendance, clinical failure without attendance, probability of complications (mastoiditis), probability of experiencing gastrointestinal adverse effects owing to amoxicillin and probability of experiencing dermatologic adverse effects were taken from various sources. Non-attendance rates were based on data from a cross-national study and a clinical trial. Clinical failure data were obtained from a RCT, a pragmatic RCT and a cross-national study. The probabilities of developing mastoiditis were based on national statistics and the probabilities of adverse events were derived from 4 studies, 3 of which were clinical trials. The design of the fourth study was unclear. The watchful waiting strategy considered current practice in the Netherlands and included estimates of the percentage of parents not seeking consultation and the probability of clinical failure based on studies conducted in the Netherlands.
Utility estimates were obtained from a cost-utility analysis of second-line antibiotics conducted in Canada by Oh et al. (1996). Utilities were derived from responses of physicians to a standardised scenario of AOM with combinations of adverse events measured on a visual analogue scale on which 1 represented perfect health and 0 represented death. Utilities from this paper represented 1 day of being in each particular health state. Lost quality-adjusted life days (QALDs) were presented separately for each pathway in the model by combining the utility weights in Oh et al. with the number of days spent in each health state. QALYs were also presented using the utilities presented by Oh et al.
Costs were estimated for antibiotics including amoxicillin, amoxicillin-clavulanate and ceftriaxone (for mastoiditis only). Resource use and costs were estimated for mastoiditis treatment and included hospitalisation, medication and outpatient costs. The cost of outpatient consultations was also included. Non-healthcare costs such as babysitting, day care, travel, parking and other expenses related to an episode of simple AOM were included.
The strategy with the highest benefit in terms of QALYs was 7–10 days of amoxicillin. This strategy had an incremental cost-utility ratio (ICUR) of $55,900 per QALY (£42,7006), compared with the least costly option, which was delayed prescribing. The watchful waiting strategy was extendedly dominated by the delayed antibiotic prescribing strategy and the 7–10-day antibiotic prescribing strategy. The 5–day amoxicillin strategy was dominated (more costly and less effective) by the 7–10-day antibiotic prescribing strategy. In one-way sensitivity analysis the 7–10-day antibiotic prescribing strategy was compared with the delayed antibiotic prescribing strategy; the costs that had the greatest effect on the ICUR were amoxicillin prescribing, non-healthcare items, office consultations and work loss. Other variables that had the greatest effect on the ICUR were probability of clinical failure, probability of gastrointestinal events, probability of non-attendance, probability of prescription redemption and the utility of a day of treatment failure. The authors reported that a probabilistic sensitivity analysis had been undertaken demonstrating that 7–10 days of amoxicillin was associated with a 61% probability of the ICUR being under $50,000 per QALY gained compared with a delayed antibiotic prescribing strategy. No cost-effectiveness acceptability curves were presented.
An important limitation of this study is that it did not consider the cost implications of antibiotic resistance. The authors concluded that delayed prescription is the least costly option. Adopting such a strategy, it was argued, would lead to substantial savings for payers and would promote a decrease in the use of antibiotics for a common, primarily self-limiting RTI, potentially reducing the impact of antibiotic resistance.
In summary, there is a clear lack of evidence on the cost effectiveness of delayed antibiotic prescribing strategies compared with immediate and no antibiotic prescribing strategies for all of the RTIs examined. In particular, there is a complete lack of evidence for sore throat, cough, sinusitis and cold.
Given the scarcity of economic evaluations of delayed versus no antibiotic prescribing strategies for RTIs in primary care, it was considered appropriate to carry out a de novo economic analysis. A model was developed to estimate the cost effectiveness of a delayed antibiotic prescribing strategy compared with immediate or no antibiotic prescribing strategies for the management of one of the RTIs covered in the guideline, acute sore throat. The decision to use sore throat as the basis of the economic analysis reflects the fact that sore throat has a high prevalence and that there is sufficient clinical evidence available.
The economic evaluation consisted of a decision-tree analysis incorporating a care pathway for the management of patients with sore throat. This was based on an open randomised trial by Little et al. (1997). This trial investigated three prescribing strategies for sore throat. Patients aged 4 years and older (no upper age limit was specified) were randomised to three groups: prescription for antibiotics, no prescription and prescription for antibiotics if symptoms were not starting to settle after 3 days. The decision tree was built and analysed using TreeAge Pro 2007 Suite (TreeAge Software, Inc) and adopts a 1-year time horizon. The study was conducted within a UK primary care setting (general practice) and so provides direct evidence on which to base the economic model. As differences in utility are likely to be very small owing to the acute nature of sore throat, the base-case analysis assumes that all antibiotic strategies were of equal effectiveness in terms of utility, and is therefore presented as a cost minimisation analysis. Full details of the modelling are presented in appendix 5.
The model suggests that the least costly option is to adopt a delayed antibiotic strategy. This strategy is associated with an expected cost of £14 per patient compared with £16 and £45.50 for the no antibiotic and immediate antibiotic prescribing strategies, respectively. The difference was mostly attributable to the reduced costs of prescribing antibiotics in the delayed strategy and the effectiveness of antibiotics at lowering the rate of complications. The probability of complications was assumed to be the same in the delayed and the immediate antibiotic prescribing strategies. In the base case, some patients in the no antibiotics arm received immediate antibiotics, as reflected in the trial outcomes on which the model was based. This was examined in the sensitivity analysis.
When utilities are considered in the model, incremental benefits realised between the strategies are small. The evidence on utilities for sore throat is poor and therefore the base-case analysis did not consider the impact of health-related quality of life. One sensitivity analysis applied the utilities used by Neuner et al. (2003) for pharyngitis. The results showed that there were no QALY differences above 0.0001 and therefore the results were not clinically significant. The ICER for an immediate antibiotic prescribing strategy over a delayed prescribing strategy was £3,628,772 per QALY gained. The delayed antibiotic strategy dominated the no antibiotic strategy (was less costly and more effective) in the base case.
In sensitivity analysis, the results are most sensitive to the baseline risk of developing quinsy. A one-way sensitivity analysis was carried out to assess the impact on model results of varying the underlying baseline risk of complications. This analysis shows that patients’ baseline risk of quinsy must be approximately 6 times higher before immediate antibiotics can be considered cost effective. A two-way analysis combining the underlying baseline risk of complications and the probability of symptoms resolving following a prescription of antibiotics shows that when symptom resolution at 3 days following antibiotic prescription is between 30% and 60%, the baseline probability for developing quinsy has to be greater than 0.12 (12%) for immediate antibiotic prescribing to become the optimal strategy (requires a sixfold increase in baseline risk of complications).
A separate analysis was carried out to look at the potential difference in cost effectiveness of each of the strategies in adult and child populations, as the probability of developing complications and the resulting cost implications are likely to differ between these groups.
The GDG considered that the presented cost-effectiveness analyses demonstrated that it was cost effective to offer a delayed prescribing strategy for adults and children presenting with acute sore throat. It was also noted that a no prescribing strategy is an acceptable alternative if the patient’s/carer’s preference is to have no antibiotics prescribed. The GDG considered that an immediate prescribing strategy may be considered cost effective for patients with a high baseline risk of quinsy.
Only 1 study was included in the review of the common cold (Arroll et al. 2002). The patient population was patients of any age presenting with the common cold who requested antibiotics or whose physicians thought they wanted them. The study was based in primary care: 15 family physicians (general practitioners) in a family practice in New Zealand.
The inclusion criterion for this particular study was diagnosis of the common cold (URTI) based on the ICHPPC-2 (International Classification of Health Problems in Primary Care): the presence of acute inflammation of the nasal or pharyngeal mucosa in the absence of other specifically defined respiratory infection.
| Study | Arroll et al. (2002) | Arroll et al. (2002) | ||||
|---|---|---|---|---|---|---|
| Antibiotic prescribing strategy | Delayed | Immediate | ||||
| Duration of delay | 3 days | N/A | ||||
| Methods of delay | Prescription was given at consultation. | N/A | ||||
| Verbal advice | Patients were advised to return to see their doctor if symptoms worsened. | No | ||||
| Use of information leaflet | No | No | ||||
| Use of analgesics | No | No | ||||
The effectiveness of delayed antibiotic prescribing and/or no prescribing as strategies for managing common cold
Summary of findings
| Outcome | No. of studies (total patients) | Design | Interventionb | Control c | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Use of antibiotics | 1 (123) | RCT | Delayed 27/62 (43%) | Immediate 54/61 (89%) | 0.49 (0.36, 0.66) | Moderate |
| Temperature (°C) (day 3) | 1 (129) | RCT | Mean score (°C): delayed = 36.7, immediate = 36.9 (analysis of comparison not provided) | Moderate | ||
| Symptom scorese (day 3) | 1 (129) | RCT | Mean score: delayed = 5.4, immediate = 5.1 (analysis of comparison not provided) | Moderate | ||
| Belief antibiotics are effective | 1 (129) | RCT | Delayed 51/67 (76%) | Immediate 47/62 (76%) | 1.00 (0.82, 1.21) | Moderate |
| Patient satisfactionf (day 3) | 1 (129) | RCT | Delayed 64/67 (96%) | Immediate 58/62 (94%) | 1.02 (0.93, 1.10) | Moderate |
intervention = delayed antibiotics
control = immediate antibiotics
1 point scored for each of 15 symptoms (dry cough, night cough, sneezing, sore throat, pain on inspiration, pain when coughing, hoarse voice, headache, staying home from work or unable to do normal daily tasks, unwell, diarrhoea, vomiting, nausea without vomiting, runny nose, blocked nose
patient satisfaction with the consultation measured on ‘very or moderately satisfied’
The evidence suggests that a delayed prescribing strategy reduces the consumption of antibiotics by 46% compared with an immediate prescribing strategy for adults and children with the common cold.
The evidence suggests that there are no clinically significant differences in temperature and incidence of common cold symptoms between adults and children offered a delayed prescribing strategy and adults and children offered an immediate prescribing strategy.
There are no differences in patient satisfaction and belief in antibiotics being effective between a delayed prescribing strategy and an immediate prescribing strategy for adults and parents/carers of children with the common cold.
The GDG acknowledged that the evidence was only of moderate quality. Based on the evidence statements presented above, the GDG concluded that antibiotics have no beneficial effect on the common cold and that the common cold is a self-limiting condition. Therefore, an immediate prescribing strategy should not be offered to patients with no increased risk of developing complications.
No studies addressing the clinical effectiveness of the three different antibiotic management strategies were identified for acute rhinosinusitis.
The GDG considered that although there was no evidence on the effectiveness of antibiotic management strategies for acute rhinosinusitis7, there is limited evidence for the efficacy of antibiotics for acute rhinosinusitis (termed acute maxillary sinusitis in the study) from a systematic review of randomised placebo-controlled trials. (Williams Jr et al. 2003). However, based on the individual patient data meta-analysis on rhinosinusitis (Young et al. 2008) (see section 2.1.2), the GDG reached a consensus opinion that this condition should be treated in the same way as the other four types of RTI included in this guideline, that is, a delayed or a no antibiotic prescribing strategy should be offered to patients with acute rhinosinusitis who are not at increased risk of developing complications.
Four studies were also included in the review of the use of specific information leaflets or structured explanations when delivering the antibiotic management strategies. The leaflets included information on the likely course of a chesty cough, what is meant by a ‘chesty cough’, when the patient should use the prescription, what patients should look out for and four ways to relieve a chesty cough (plenty of fluids, analgesics, cough linctus or lozenges and steam or vapour). For AOM, the structured explanation was short and included the likely course of AOM, reassurance that in most cases children would recover regardless of antibiotic prescription, the information that late complications may occur regardless of whether antibiotics are administered, and that parents are advised in cases of high fever or severe pain to administer paracetamol prescribed according to the child’s weight. The 5 studies were (Gerber et al. 1990; Little et al. 2005; Macfarlane et al. 2002; Macfarlane et al. 1997; Pshetizky et al. 2003)). The study by Little was on cough, Macfarlane’s (2002) was on acute bronchitis, Pshetizky looked at AOM, and Marfarlane (1997) studied LRTI.
The patient population in the Little study was children 3 years and older with uncomplicated acute LRTI (21 days or below in duration) who presented in primary care. The patient population in Macfarlane’s study was adults 16 years and older presenting with ‘acute bronchitis’ defined as a ‘new, acute lower respiratory tract illness in a previously well adult’ (including smokers). Pshetizky’s study included children aged between 3 months and 4 years visiting family practice clinics and diagnosed with AOM; Macfarlane’s (1997) study included previously well adults (16 years and older including smokers) presenting with an illness defined as an LRTI. Three studies (Little et al. 2005, Macfarlane et al. 2002 and Macfarlane et al. 1997) were based in primary care general practices in the UK. Pshetizky’s study was based in two primary care clinics in Israel.
The inclusion criteria in the study on cough by Little were cough (21 days or less in duration) as the main symptom and at least one symptom or sign localising to the lower respiratory tract (sputum, chest pain, dyspnoea or wheeze). The inclusion criteria for Macfarlane’s study (acute bronchitis) were age 16 years or older, previously well and not under supervision or management for an underlying disease (for example, no pre-existing asthma, COPD, heart disease or diabetes). Further requirements for inclusion were cough as the main symptom; at least one other lower respiratory tract symptom (sputum production, dyspnoea, wheeze, chest discomfort or pain) and no alternative explanation (for example, not sinusitis, pharyngitis or a new presentation of asthma).
The inclusion criteria for Macfarlane’s (1997) study on LRTI were previously well adults (not under supervision or treatment for an underlying disease) who consulted with an LRTI (defined as a new cough and at least one other lower respiratory tract symptom, including sputum production, dyspnoea, wheeze, or chest pain, for which there was no explanation). In the Pshetuzky study of AOM, the inclusion criteria were children aged between 3 months and 4 years diagnosed with AOM (for example, fever of 38°C or higher, purulent ear discharge and opacity or bulging of the eardrum).
The use of specific information leaflet or structured explanation in antibiotic management strategies for respiratory tract infections
Summary of findings
| Outcome | No. of studies (total patients) | Design | Intervention | Control | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Use of antibiotics (next 2 weeks) [M2] | 1 (205) | RCT | Delayed (leaflet) 49/104 (47%) | Delayed (no leaflet) 63/101 (62%) | 0.76 (0.59, 0.97) | High |
| Use of antibiotics (next 2 weeks) [M2] | 1 (150) | RCT | Delayed (leaflet) 49/104 (47%) | Immediate (no leaflet) 44/46 (96%) | 0.49 (0.39, 0.60) | High |
| Use of antibiotics (at 1 week) [P] | 1 (81) | RCT | Delayed (structured explanation) 18/44 (41%) | Delayed (no structured explanation) 32/37 (86%) | 0.47 (0.32, 0.68) | Moderate |
| Use of antibiotics (at 3 week) [L] | 1 (572) | RCT | Leafletd 160/281 (57%) | No leafletd 159/291 (55%) | 1.04 (0.90, 1.20) | High |
| Outcome | No. of studies (total patients) | Design | Intervention | Control | Relative risk | Quality |
|---|---|---|---|---|---|---|
| Re-consultation (within 4 weeks) [M2] | 1 (209) | RCT | Delayed (leaflet) 11/104 (11%) | Delayed (no leaflet) 14/105 (13%) | 0.79 (0.37, 1.66) | High |
| Re-consultation (within 4 weeks) [M1] | 1 (283) | RCT | No (leaflet) 15/136 (11%) | No (no leaflet) 26/147 (18%) | 0.62 (0.34, 1.12) | High |
| Re-consultation (within 4 weeks) [M1] | 1 (723) | RCT | Immediate (leaflet) 60/369 (16%) | Immediate (no leaflet) 81/354 (23%) | 0.71 (0.52, 0.95) | High |
| Re-attendance (within 1 month) [L] | 1 (572) | RCT | No leaflet as control vs. leaflet Incidence rate ratio estimate = 1.63 (95% CI 1.07–2.49), | High |
leaflet factor: both leaflet and no leaflet included all three groups – delayed, no antibiotic prescribing and immediate antibiotic prescribing
One large trial with a high quality of evidence suggested that the use of an information leaflet in general (when used with any of the three antibiotic management strategies) does not affect the consumption of antibiotics. Two smaller trials show that within a delayed prescribing strategy, the use of information leaflets and structured verbal explanations reduced the consumption of antibiotics.
The use of an information leaflet in an immediate prescribing strategy reduced repeat consultation rates in one trial but not in another larger trial where all patients received structured verbal information.
The GDG thought that the evidence on the use of an information leaflet or structured verbal explanation to deliver a chosen antibiotic management strategy remained inconclusive, since the included studies showed inconsistent findings across different strategies within various comparisons (that is, leaflet versus no leaflet across all three prescribing strategies; leaflet in delayed arm versus no leaflet in immediate arm; study of the effect of leaflet and verbal explanation only in the delayed arm but not others). The GDG decided that, owing to inconsistent evidence, no recommendation could be made regarding the efficacy of information leaflets as opposed to structured verbal explanations.
It is clear from the previous overview of antibiotic efficacy and the review of the effectiveness of antibiotic management strategies that antibiotics are, in general, ineffective in treating RTIs. However, antibiotics may still be beneficial for a subgroup of patients who present with an RTI in primary care settings and who are likely to be at risk of developing complications.
The first group is adults and children who present with a complicated infection such as pneumonia. The diagnosis and management of complicated RTIs is outside the scope of this short clinical guideline. However, it is important that this guideline clearly signposts that such complicated infections should not be managed using a delayed or no antibiotic prescribing strategy.
The second group is adults and children who present with an uncomplicated infection, but who are at a high risk of developing complications. For this group, the use of a delayed or a no antibiotic prescribing strategy may potentially lead to an increased risk of developing complications, although in the case of delayed prescribing this risk may be reduced by offering the patient advice on when the antibiotic should be started. It is therefore important that for each of the RTIs covered in this guideline evidence is sought as to whether specific clinical symptoms, signs and risk factors can predict which patients seen in primary care and other first-contact care settings are more likely to develop complications. For the purposes of this guideline, the following complications of RTIs were considered to lead to significant morbidity and were therefore the focus of the review.
We identified 24 published individual studies based on study abstracts. After further assessment, only 6 studies that provided evidence on clinical symptoms, signs and risk factors that predict which patients with RTIs are likely to develop complications were included in the evidence review (15 studies were not relevant, 1 study had an inappropriate study population and 1 study was excluded as statistical analysis was inappropriate). All 6 studies were appraised individually using the NICE prognostic study checklist (see appendix 4) and presented in the evidence tables and narrative summary.
Of the 6 included studies, 1 case control study was on acute sore throat/acute pharyngitis/acute tonsillitis (from UK primary care data) (level of evidence +); 2 prospective studies and 1 retrospective cohort study were on acute cough/acute bronchitis (2 from UK primary care settings with level of evidence + and ++ respectively; and 1 from a Netherlands primary care setting with level of evidence ++). One prospective cohort and 1 analysis of RCT cohort were on AOM (1 from a Netherlands primary care setting and 1 from a UK primary care setting, both with level of evidence +). No studies were identified on acute rhinosinusitis or the common cold.
Overall, the quality of the evidence was good. However, 3 out of the 6 included studies need cautious interpretation as the evidence of clinical prediction criteria reported in these 3 studies has not been validated in other primary care populations.
An immediate antibiotic prescription and/or further appropriate investigation and management should only be offered to patients (both adults and children) in the following situations:
hospitalisation in previous year
type 1 or type 2 diabetes
history of congestive heart failure
current use of oral glucocorticoids.
For these patients, the no antibiotic prescribing strategy and the delayed antibiotic prescribing strategy should not be considered.
One reasonably good quality retrospective case control study was included as the basis for recommendations (Dunn et al. 2007). It was based on UK-wide primary care data from the General Practice Research Database between 1995 and 1997. The aim of this study was to identify clinical symptoms, signs and risk factors that were associated with the development of quinsy after initial presentation of uncomplicated sore throat. The study identified 606 cases of quinsy within the study period, of which only 192 cases developed following initial uncomplicated sore throat. These 192 patients with quinsy formed the study group and another 198,124 patients of sore throat without quinsy formed the control group for the analysis. The prevalence of quinsy within the study period was 96 cases per 100,000 patients with sore throat (per annum between 1995 and 1997).
Logistic regression was used to calculate odds ratios (ORs) for the risk of quinsy following a sore throat for different variables such as age, sex, smoking status, type of diagnosis, exposure to antibiotics and lung disease. Results for the analysis showed that only age (21 to 40 years) (adjusted OR = 3.4, 95% CI 2.1 to 5.5), smoking (adjusted OR = 2.5, 95% CI 1.8 to 3.5) and male gender (adjusted OR = 1.6, 95% CI 1.1 to 2.2) were significantly associated with the development of quinsy following a sore throat.
Further analysis was also carried out based on different diagnoses of sore throat, such as tonsillitis and sore throat/pharyngitis (adjusted for age, sex, smoking status, lung disease at patient level and clustering at practice level). The interval between diagnosis of a sore throat and development of quinsy was a median of 2 days (interquartile range 1 to 6 days) for tonsillitis, and 3 days (interquartile range 2 to 5 days) for sore throat/pharyngitis. Results from this further analysis showed that prescription of antibiotics after recording a diagnosis of a sore throat generally did not seem to reduce the risk of developing quinsy (antibiotic given after all diagnoses [adjusted OR = 1.2, 95% CI 0.7 to 1.8]; antibiotics given after tonsillitis [adjusted OR = 0.6, 95% CI 0.3 to 1.3]; antibiotics given after sore throat/pharyngitis [adjusted OR = 1.2, 95% CI 0.7–2.2]). However, considerable caution is needed in estimating the effect of antibiotics in this study owing to confounding by indication in routine databases (individuals with more severe illness are more likely to be given antibiotics than individuals with less severe illness).
Patients aged between 21 and 40 years who are male and are smokers are significantly more likely to develop quinsy after initial presentation of uncomplicated sore throat in primary care settings.
The GDG noted both that quinsy is a rare complication of sore throat in the UK (with an annual incidence of 96 cases per 100,000 patients) and therefore the absolute risk of developing quinsy is low (Dunn et al. 2007), and that the predictive value of the risk factors for the development of quinsy was not sufficient to make a recommendation to prescribe immediate antibiotics. It was also noted that the included study did not offer a validated clinical prediction rule, although the study did document the same risk factors in those presenting with a prior RTI and those presenting with de novo quinsy. The GDG came to the conclusion that patients with sore throat should not be excluded from delayed or no prescribing strategies based on the three risk factors identified (aged 21 to 40 years, male and smoker). Hence, no recommendation on exclusion criteria for antibiotic management strategies for patients with sore throat was generated from the evidence statement. Nevertheless, the GDG acknowledged that quinsy is a serious complication and came to the consensus conclusion that immediate antibiotic prescription and/or further appropriate investigation and management should be offered to adults and children who appear unwell and with symptoms and signs suggestive of peritonsillar abscess (quinsy).
Three good quality studies were included as the basis of the recommendations. Two were prospective cohort studies from the same research team (a derivation study and the further validation study). The studies were based in UK primary care settings (Dunn et al. 2007; Hay 2004; Hay et al. 2007) and aimed at identifying and validating a clinical rule for predicting complications of acute cough in pre-school children. The third study was a retrospective cohort study based on patient data from the Netherlands General Practice Research Network and the second Dutch National Survey of General Practice (Bont 2007). The aim of this study was to identify and validate a prediction rule for complications of LRTIs in elderly primary care patients.
A derivation study and a further validation study (Hay 2004; Hay et al. 2007) on a clinical rule for predicting complications of acute cough in pre-school children (aged between 0 years and 4 years) were identified. Complications in these two studies were defined as new sign/symptoms/conditions identified after initial consultation, which were bronchiolitis, possible asthma, vomiting, bronchitis, viral illness, cough and wheeze, conjunctivitis, LRTI, baby asthma, chest infection, chicken pox, viral induced wheeze, pharyngitis and otitis media. Hospital admission was defined as hospital admission before cough resolution owing to bronchiolitis, pneumonia, whooping cough and viral-induced wheeze.
In the derivation study (Hay 2004), multivariate analysis showed that only the presence of a chest sign (OR = 2.78, 95% CI 1.04 to 7.35, p = 0.048) and the presence of fever (OR = 4.65, 95% CI 1.63 to 13.3, p = 0.007) were significant independent predictors of complications and hospital admission before cough resolution in pre-school children. Further logistic regression also showed that lack of fever and chest signs was a good predictor for ruling out complications in children with cough, with a likelihood ratio (LHR) of 0.56 (95% CI 0.35 to 0.91). Fever only or both fever and chest sign LHR = 3.54 (95% CI 1.62 to 7.68) and only fever and chest sign LHR = 5.39 (95% CI 0.95 to 30.6) were found to be good predictors for complications in children with cough. However, the discriminatory ability of this particular prediction model was weak, with an area under receiver operating characteristic (ROC) below 0.70 (ROC = 0.68). A further validation study by Hay (2007) of the earlier derivation study (Hay 2004) was also identified. In the further validation study, however, chest sign and fever were not found to be significant predictors of complications and hospital admission in children with cough. Instead, chest sign and fever were found to be protective against complications and hospital admission (post-test probability: neither fever nor chest sign = 13.7 [95% CI 7.5 to 22.3]; chest sign only = 13.8 [95% CI 3.9 to 32.0]; fever only = 9.1 [95% CI 0.0 to 41.0]; both fever and chest sign = 0.0 [95% CI 0.0 to 37.0]). A completely different set of variables were found to be significant independent predictors of complications and hospital admission: age (OR = 0.95, 95% CI 0.90 to 0.99, p = 0.03); deprivation (OR = 0.79, 95% CI 0.64 to 0.97, p = 0.02); number of GP visits in previous year (OR = 1.14, 95% CI 1.02 to 1.27, p = 0.02). The authors commented that the contradictory findings from the validation study compared with the derivation study could be a result of spectrum bias (that is, sociodemographic differences, possible reduced levels of circulating influenza-like illness between the derivation and validation cohorts) and confounding by indication (that is, clinicians’ antibiotic prescriptions tended to be targeted at children with chest signs or fever). Thus, the evidence provided by these two studies needs cautious interpretation.
Another retrospective cohort study (Bont 2007) that derived and validated a prediction rule for complications of LRTIs in elderly primary care patients was also identified. The derivation cohort of this study was from the Netherlands General Practice Research Network and the validation study cohort was from the second Dutch National Survey of General Practice. Patients included in this study were 65 years or older. Logistic regression in the derivation cohort showed that after initial diagnosis, the following variables were significant predictors of 30-day hospitalisation and death (table 10) and a scoring system was derived based on regression coefficients.
Significant predictors and scoring system.
The scoring system was separated into three risk groups: low risk (score ≤ 2), medium risk (score 3–5) and high risk (score ≥ 7). The discriminatory abilities of this prediction scoring system in the derivation cohort were: low risk – sensitivity = 0.82, specificity = 0.52, percentage of risk of endpoint 3.2%; medium risk – sensitivity/specificity = not reported, percentage of risk of endpoint = 9.9%; high risk – sensitivity = 0.35, specificity = 0.92, percentage of risk of endpoint = 30.9%, with good discriminatory power (area under ROC = 0.75 [95% CI 0.72 to 0.78]).
The prediction scoring system was also validated in a separate cohort with similar results: low risk – sensitivity = 0.42, specificity = 0.81, percentage of risk of endpoint = 5.3%; medium risk – sensitivity/specificity = not reported, percentage of risk of endpoint = 14.5%; high risk – sensitivity = 0.06, specificity = 0.98, percentage of risk of endpoint = 22.0%, with good discriminatory power (area under ROC = 0.74 [95% CI 0.71 to 0.78]). However, the limitation of the validation study is that it did not include exacerbation of chronic obstructive pulmonary disease (COPD) among the predictors.
There is inconsistent evidence on the utility of clinical rules for predicting complications of acute cough in pre-school children.
The following clinical signs/symptoms and risk factors are significant predictors of the development of complications of LRTIs in elderly primary care patients:
congestive heart failure
diabetes
COPD or exacerbation of COPD
The GDG discussed the evidence on predicting complications in elderly primary care patients with LRTIs. The GDG agreed the evidence statement but questioned the validity of the full prediction model provided by the study since this model was based on a single study; moreover, a large proportion of the study population had comorbidities. In addition, the study was conducted in the Netherlands, where the level of antibiotic prescribing is low and thus patients are more likely to present with a more severe illness.
The GDG also recognised that there is inconsistent and inconclusive evidence on predicting which children with acute cough are likely to develop complications.
Two good quality studies were included as the basis of recommendations. One was a prospective cohort study (Damoiseaux et al. 2006) on long-term prognosis of AOM in infancy (6 months to 24 months) with a prediction model for complication (recurrent AOM). The setting of this study was family practices in the Netherlands. The other study was a follow-up secondary analysis study of an RCT cohort (Little et al. 2006). This was a UK primary care-based study looking for clinical predictors of complications (recurrent AOM and hearing impairment) from AOM in children (6 months to 10 years). No studies were identified regarding the complication mastoiditis. Based on Hospital Episode Statistics (2006–07) there were 952 finished consultant episodes of mastoiditis and in relation to GP-registered populations (GP Registered Populations 2007), there were 50,542,505 registered patients in England. These constituted a crude rate of 144 cases of mastoiditis per 1,000,000 patients per annum, indicating that mastoiditis is a rare complication. A large Dutch cohort study also showed that mastoiditis is likely to be very rare when using a 72-hour wait-and-see policy before prescribing antibiotics (van Buchem et al. 1985).
In the Damoiseaux’s (2006) study, logistic regression showed that the variables listed in table 11 were significant predictors of recurrent AOM within 6 months in infants. A scoring system was derived based on regression coefficients (table 11).
Significant predictors and scoring system.
The scoring system was then separated into three cut-off points: below −8, below −1 and below 5. The discriminatory abilities of this prediction scoring system were: below −8 – sensitivity = 93%, specificity = 23%, positive predictive value (PPV) = 54%, negative predictive value (NPV) = 77%); below −1 – sensitivity = 72%, specificity = 56%, PPV = 62%, NPV = 67%; below 5 – sensitivity = 51%, specificity = 76%, PPV = 68%, NPV = 61%. The discriminatory power of the model was weak, with an area under ROC of 0.69 (95% CI 0.62 to 0.76), and this particular model was not validated in different primary care populations.
In Little’s study, logistic regression showed that ear discharge (otorrhoea) (LHR = 7.04, p = 0.004) and bulging eardrum (LHR = 5.50, p = 0.019) were significant predictors of recurrent episodes of otalgia within 3 months in children aged between 6 months and 10 years, whereas past history or previous episodes of AOM (LHR = 8.04, p = 0.005) were the significant predictors of recurrent episodes of otalgia within 1 year.
Little (2006) also investigated predictors of functional hearing impairment following initial AOM in children in their study. Functional hearing impairment in this study was measured by a child function score (in which a score of 9 or above indicates hearing impairment) based on 14 descriptions of how hearing impairment with chronic secretory otitis media presents. Results from logistic regression showed that only past history or previous episodes of otitis media were significant predictors of functional hearing impairment in children aged between 6 months and 10 years within both 3 months (LHR = 4.95, p = 0.026) and 1 year (LHR = 4.56, p = 0.033) of initial presentation of AOM. Further analysis also showed that, compared with an immediate antibiotic prescribing strategy, a delayed antibiotic prescribing strategy did not significantly increase the risk of recurrent AOM after 3 months (OR = 0.89, 95% CI 0.48 to 1.65) or after 1 year (OR = 1.03, 95% CI 0.60 to 1.78). Additionally, there was no significant increase in the risk of functional hearing impairment in children after 3 months (OR = 1.37, 95% CI 0.72 to 2.60) or after 1 year (OR = 1.16, 95% CI 0.61 to 2.23). Moreover, the study showed that a delayed prescribing strategy did not significantly increase the risk of otalgia at 3 months (OR = 0.89, 95% CI 0.48 to 1.65) or at 1 year (OR = 1.03, 95% CI 0.60 to 1.78), nor did it significantly increase the risk of a poor child (hearing) function score at 3 months (OR = 1.37, 95% CI 0.72 to 2.60) or 1 year (OR = 1.16, 95% CI 0.61 to 2.23). However, as noted by the authors, this is a secondary analysis and there was no validation study. Moreover, since recurrent AOM or recurrent episodes of otalgia are not serious complications, the evidence requires cautious interpretation.
In children aged between 6 months and 10 years, ear discharge and bulging eardrum are significant predictors of recurrent episodes of otalgia within 3 months of the initial consultation. However, the predictors are no longer significant after 1 year.
In children aged between 6 months and 10 years, a history of previous episodes of AOM is a significant predictor of recurrent episodes of otalgia only 1 year after the initial consultation.
In infants aged between 6 months and 24 months, male gender, passive smoking, winter season and persistent symptoms are significant predictors of recurrent AOM within 6 months of the initial consultation.
Delayed prescribing does not significantly increase the risk of otalgia or poor child (hearing) function at 3 months or at 1 year
Mastoiditis was considered by the GDG to be a rare but potentially serious complication of AOM, but no mastoiditis studies were identified that met the inclusion criteria for the review. The GDG recognised that the outcome measures reported in the included studies (recurrent AOM and recurrent episodes of otalgia) were not considered to be serious complications of AOM. Moreover, the GDG considered that the evidence merited a cautious interpretation as it was a secondary analysis from a previous RCT. The GDG considered that these three factors precluded the use of this evidence as the basis for making recommendations. The GDG concluded that it was not possible to identify subgroups of patients presenting with AOM who should be excluded from the offer of a delayed or no prescribing strategy.
However, the GDG acknowledged that mastoiditis is a serious complication of AOM and came to the consensus conclusion that immediate antibiotic prescription and/or further appropriate investigation and management should be offered to adults and children who appear unwell and with symptoms and signs suggestive of mastoiditis.
No studies were identified for acute rhinosinusitis.
No evidence was identified for acute rhinosinusitis.
The GDG noted the lack of evidence in this area and concluded that it was not possible to identify subgroups of patients presenting with acute rhinosinusitis who should be excluded from the offer of a delayed or no prescribing strategy.
However, the GDG acknowledged that intraorbital and intracranial complications are serious complications of acute rhinosinusitis. Hence, the GDG came to the consensus conclusion that immediate antibiotic prescription and/or further appropriate investigation and management should be offered to adults and children who appear unwell and with symptoms and signs suggestive of intraorbital and intracranial complications.
No studies were identified for common cold.
No evidence was identified for common cold.
The GDG noted the lack of evidence in this area and concluded that it was not possible to identify subgroups of patients presenting with common cold who should be excluded from the offer of a delayed or no prescribing strategy.
A central task of the healthcare professional during the patient consultation is to address the patient’s ideas, concerns and expectations regarding treatment before agreeing a management plan (Fraser 1999). This is particularly important in consultations for RTIs, when there may be an expectation on the part of the patient that an antibiotic will be required, whereas the opinion of the healthcare professional is that an antibiotic prescription is not clinically indicated. Conversely, there may be an expectation on the part of the healthcare professional that the patient has attended specifically with a view to obtaining an antibiotic prescription whereas the patient is seeking only advice and/or reassurance (Butler et al. 1998). Indeed, there is evidence that GPs overestimate the proportion of patients who attend with RTIs expecting an antibiotic prescription (Altiner 2004). The perceived advantage of delayed prescribing as a strategy over no prescribing is that a patient expecting antibiotics may be more likely to agree with this course of action than with a no prescribing strategy.
The issue of patients’ preferences regarding the three antibiotic management strategies (immediate, delayed or no prescribing) is therefore extremely important. In the overview presented in section 2.4.2, the included RCTs assessed patients’ preferences using satisfaction rating scales and the results are presented in the relevant GRADE tables and evidence statements by condition. Patients reported a high level of satisfaction (above 70% overall) with the use of a delayed or a no antibiotic prescribing strategy. The included studies, however, did not report whether patient preferences regarding the three antibiotic management strategies differed across ethnic and socioeconomic groups.
There is a body of literature suggesting that variations in prescribing in primary care may be a result, at least in part, of patient ethnicity and socioeconomic status. A secondary analysis (Gill et al. 1996) of data from the General Household Survey that examined the association between being given a prescription and ethnicity found that people of Pakistani or Indian origin were significantly more likely to receive a prescription from their GP than people of white or West Indian origin. Another study (Gill and Roalfe 2001) found that patients from manual classes and patients from the most deprived areas received significantly more antibiotics during primary care consultations than patients from other socioeconomic classes. There is also evidence that people’s knowledge of and attitudes toward antibiotics may vary according to their ethnicity. A large-scale household survey in Britain (McNulty et al. 2007) of the public’s knowledge of and attitudes to antibiotics showed that people of Asian/Asian British or Caribbean/black British origin were less knowledgeable about and had different attitudes toward antibiotics than people of white British origin.
Given the above findings, it is important to determine whether there is any additional evidence that reports patient preferences for the three antibiotic management strategies, in particular whether there is evidence pertaining to specific black and minority ethnic and socioeconomic groups.
We identified 10 published individual qualitative studies based on study abstracts. After further assessment, only 2 studies that provided information on patients’ preferences regarding antibiotic management strategies for RTIs were included in the evidence review (8 excluded studies were not relevant). Both studies were appraised individually and presented in the evidence tables and narrative summary.
Both of the 2 included studies were survey questionnaire studies. One (Edwards et al. 2003) explored patients’ responses to delayed antibiotic prescribing for acute URTIs in a UK primary care setting. The other (Couchman et al. 2000) studied patients’ self-reported satisfaction with a delayed prescribing strategy for common respiratory symptoms. Qualitative studies including survey questionnaires were assigned evidence level 3 in accordance with NICE technical guidance.
See Recommendation number 1.1.2.
Patients/parents’ satisfaction and expectations
In Edwards’ survey questionnaire study that investigated patients’ responses to delayed antibiotic prescribing for acute URTIs, the results showed that of the 256 patients who received a delayed prescription, 92.5% were satisfied and would choose to receive a delayed prescription again in the future. Further analysis from Edwards’ study showed that of the 256 patients who received a delayed prescription, approximately two-thirds (65.2%) had expected to receive an immediate antibiotic prescription, 37% had expected advice, 2.0% had expected tests or a hospital referral and 4.7% had anticipated a sickness certificate. Patients’ expectations were not associated with whether they had consumed the delayed prescription or not.
The study by Couchman of 286 patients who received a delayed prescription for common respiratory symptoms found that patients’ self-reported satisfaction was 96.1%. The overall delayed prescription fill rate of this study was 50.2% and the fill rates did not differ significantly by patient characteristics or their self-reported satisfaction with the care received.
For patients who were expecting to receive immediate antibiotics during consultation, over 90% of those who then received a delayed prescription for acute URTIs were satisfied and would choose to receive a delayed prescription again in the future.
No studies were identified that reported patient preferences for the three antibiotic management strategies in black and minority ethnic and differing socioeconomic groups.
The GDG noted that the evidence presented here was consistent with that presented in the RCTs on different antibiotic management strategies (see section 2.2.2). They also noted that no specific evidence was identified that reported on patient preferences by specific black and minority ethnic and socioeconomic groups.
In view of the lack of evidence in this area, the GDG considered that a general recommendation should be made on the need for patient concerns and expectations regarding antibiotic use to be determined during healthcare consultations with adults and children with RTIs in primary care settings. This should apply for all ethnic and socioeconomic groups.
Since studies and practitioners use slightly different terms for RTIs, the terminology used in this guideline for RTIs provides covers a range of acute symptoms and also a suspected diagnosis if appropriate. For example:
For a more detailed definition of number need to treat (NNT), please refer to the glossary.
This review was withdrawn from The Cochrane Library, Issue 3, 2007. The authors agreed that they could no longer work towards updating the review, owing to other work demands.
Converted for clarity from 2001 US dollars to 2006/7 pounds sterling using a purchasing power parity (PPP) exchange rate of 0.626 (www
Acute rhinosinusitis can also be referred as acute sinusitis in some medical literature.
All rights reserved. This material may be freely reproduced for educational and not-for-profit purposes. No reproduction by or for commercial organisations, or for commercial purposes, is allowed without the express written permission of the Institute.
Your browsing activity is empty.
Activity recording is turned off.
See more...
