The role of antibiotics in the treatment of acute rhinosinusitis in children: a systematic review
- Department of Paediatrics, Gwynedd Hospital, Betsi Cadwaladr University Health Board, Bangor, North Wales, UK
- Correspondence to Dr Michael John Cronin, Department of Paediatrics, Gwynedd Hospital, Betsi Cadwaladr University Health Board, Bangor, North Wales LL57 2PW, UK;
- Received 18 September 2012
- Accepted 18 January 2013
- Published Online First 16 February 2013
Objective A systematic review of randomised controlled trials reporting the efficacy of antibiotics compared with placebo in the treatment of acute rhinosinusitis in children.
Design Systematic review and meta-analysis.
Data sources Cochrane Register of Controlled Trials, Medline, Embase and references obtained from retrieved articles.
Results Four studies fitted the selection criteria for inclusion. Risks for internal bias were thought to be small for each study, but external bias is potentially significant. The pooled OR for symptom improvement at 10–14 days favouring the use of antibiotics was 2.0 (95% CI 1.16 to 3.47; I2=14.8%).
Conclusions While the meta-analysis provides evidence to support the use of antibiotics for acute rhinosinusitis in children, it is the assessment of this review that such efficacy has not been adequately demonstrated. There remains a clear methodological challenge in the examination of this important clinical question; this challenge relates to difficulties in the application of appropriate diagnostic and inclusion criteria which are also consistent between studies.
What is already known on this topic
Acute rhinosinusitis is one of the leading reasons for antibiotic use in children.
UK guidelines recommend a conservative approach and avoidance of antibiotics where possible.
Recent systematic reviews suggest a marginal advantage for antibiotic use over placebo for treating acute rhinosinusitis in adults.
What this study adds
Meta-analysis of four trials studying acute rhinosinusitis in children demonstrated great uncertainty in any benefit provided by antibiotics.
The role of antibiotics in acute rhinosinusitis in children is not clear; current UK guidelines remain applicable.
Acute rhinosinusitis (defined here as rhinosinusitis lasting less than 30 days)1 results from inflammatory processes involving mucosal membranes and secretions within the paranasal sinuses, and is generally associated with some obstruction of sinus drainage. The inflammation may be caused by pathogens (bacterial, virus or fungal), by allergic reactions, chemical irritants, foreign bodies or a variety of other systemic disorders.2–5 The maxillary and ethmoid sinuses are present at birth, but small. The sphenoid and frontal sinuses pneumatise later between 5 and 8 years of age.1 Sinus infections may occur in children younger than 1 year, but this is rare. A synonymous term, sinusitis, is still used in the literature, however, sinus involvement is usually a complication of a viral upper respiratory infection, and given the contiguity of sinus and nasal mucosa, rhinosinusitis has become the internationally accepted term.6 The frequency of this condition remains unclear largely due to practical difficulties in identifying the extent and site of inflammation in any one instance.5 Children are estimated to experience between 6 and 8 upper respiratory infections each year (adults 2–3), of which 5–13% may be complicated by bacterial rhinosinusitis (adults 0.5–2%).1 ,7 Complications of acute bacterial rhinosinusitis (ABRS) can be serious and include orbital or periorbital cellulitis, subdural or epidural empyema, brain abscess and osteitis.8 Imaging of the sinus cavities (CT, MRI, x-ray, ultrasound) may provide evidence for the presence of inflammation, and direct aspiration of secretions from within sinuses may identify causative pathogens, but neither of these two forms of investigation are routinely requested, reliable or indeed available especially in the context of primary care.1 ,6 ,9–11 Symptoms and signs ascribed to sinusitis include headache, nasal discharge, nasal obstruction, sneezing, otalgia, daytime cough, pyrexia, localised tenderness, localised oedema, malodorous breath, sore throat and altered sense of smell. Symptoms in children tend to be a little different from those occurring in adults; nasal discharge and cough are common, whereas headache, localised oedema and tenderness are rare findings. While these symptoms are non-specific they are usually taken in combinations to suggest a diagnosis of acute rhinosinusitis, but these are not always consistently applied.1 ,10–14 The diagnosis of acute rhinosinusitis, especially ABRS, remains a clinical challenge.
Whilst the majority of upper respiratory infections are caused by viral agents, and generally, self-limiting the use of antibiotics as a treatment remains high.15 The economic costs of these treatments are significant.16 A recent review of antibiotic prescribing trends in the USA over a 10-year period has shown little change in the frequency of antibiotics prescribed for children with acute rhinosinusitis (about 75% of children diagnosed were prescribed antibiotics).17 This practice, favouring the use of antimicrobials, is relatively more supported by published North American guidelines, and acute rhinosinusitis is reported as the fifth leading reason for antibiotic use in children.1 ,18 ,15 In the UK there is some evidence to suggest that antibiotic prescribing for acute rhinosinusitis is also relatively high.19 UK guidelines, however, advise that antibiotics should be preserved for those children who are considered to be more severely ill, but otherwise to be generally avoided, acute rhinosinusitis being perceived as a self-limiting condition.6 ,20 It should be noted that UK and North American guidelines differ fundamentally; the former provide recommendations for acute rhinosinusitis whatever the assumed aetiology, whereas, the latter specifically attempt to apply diagnostic criteria to define ABRS (see table 1). All guidelines have drawn attention to the emerging risk of antibiotic resistance.
A Cochrane review first published in 1995 and updated in 2002 concludes that there is no good evidence for the routine prescribing of antibiotics in children with acute rhinosinusitis.22 A further systematic review of antibiotic versus placebo for acute rhinosinusitis was published in 2008; this examined antibiotic efficacy principally in the adult population but did include three randomised controlled trial (RCTs) involving children.23 The reviewers concluded that there was some evidence, although limited, to support the use of antibiotics. Most recently, a Cochrane review of RCTs in adults (this included trials in which there were less than 20% of participants under the age of 18) produced a more cautious statement to the effect that antibiotics did help some people a little, but overall made very little difference to most.24 There is, thus, variation both in practice and in assessments of the evidence. It would be true, perhaps, to say that medical practitioners in Europe and the UK are more likely to be conservative in their management of children with perceived acute rhinosinusitis than their counterparts in North America.
The aim of this study is to provide a systematic review of the current evidence for the efficacy of antibiotics in the treatment of acute rhinosinusitis in children.
We searched Medline, Embase and the Cochrane controlled trials register up to October 2011 using the terms sinusitis, paranasal, rhinosinutis, purulent, rhinorrhea, sinus infection, randomised, randomised control trial, double blind method, random allocation, placebo, antibiotic, antimicrobial, animal, human, child, children and adolescent. No restriction was made based on language. MC, SK and SS each independently conducted a literature search and assessment for inclusion. We contacted authors where relevant data was not available in published sources.
Studies were included if the following criteria were all met: involving children between the ages of 1 and 18 years; a randomised control study involving patients diagnosed to have either acute sinusitis or acute rhinosinusitis; the efficacy of antibiotics compared with placebos; analytical data available for children less than 15 years of age (this was to capture those studies which included both adults and patients less than 18 years old, but also children in younger age groups). We chose 30 days as the maximum symptom duration to exclude patients who may be defined to have chronic rhinosinusitis.1 Studies included needed to describe a primary outcome of symptom improvement following the intervention.
Evaluation and quantitative analysis—two authors, MC and SS, independently assessed the included trials for quality. This assessment was based on the methodological approach outlined by the Cochrane collaboration.25 Any disagreements between the two reviewers was resolved by discussion and consensus. STATA 10.0 was used for the meta-analysis of outcome data.26 A random effects model was applied using ORs and 95% CIs. An I2 statistic for heterogeneity was applied to the pooled ORs.
Ninety-six articles were identified in the search for inclusion; following screening and eligibility assessment, 84 articles were excluded on the basis of title and abstract alone (reasons for exclusion included: repeat citation—24, not a RCT—30, children not included—8, not acute rhinosinusitis—26, not antibiotic vs placebo—29). Twelve studies were included for full text scrutiny.27–38
Of the 12 studies subject to full text scrutiny, only four fulfilled all the selection criteria.27–30 A further two publications were considered for inclusion, both were RCTs conducted principally involving adult patients, but included patients less than 15 years of age, 31 ,32 unfortunately, no replies have been received on writing to the lead authors requesting a breakdown of data for patients aged less than 18 years. Of the remaining six publications not fulfilling selection criteria; two included patients with chronic rhinosinusitis, one was a repeat publication of an earlier RCT, and three included adult participants only or subjects older than 15 years of age.
Study characteristics and appraisal
Two reviewers (SS and MC) independently assessed each of the four RCTs for quality and potential bias. The assessment was based on the methodological process outlined by the Cochrane collaboration.25 Internal bias was evaluated using the described quality rating scale (randomisation, allocation concealment, blinding, outcome reporting and other sources of bias). Sources for potential external bias were analysed. Data was extracted relating to diagnostic criteria, inclusion and exclusion criteria, definition of outcomes, treatment protocols and study results. This data is summarised in table 2.
The quantitative analysis has assessed improvement in symptoms at 14 days from commencement of treatment (10 days for the study reported by Wald 1986). The results of the meta-analysis of outcome data (figure 1) suggest a benefit for those participants treated with antibiotics with the pooled statistic giving an OR of 2 (95% CI 1.16 to 3.47 and estimated predictive interval 0.43–9.26). The I2 statistic would suggest moderate to substantial heterogeneity. The number needed to treat (NNT) for the pooled data is 8 (varying between 3 and 50 for the individual trials); this is the NNT for an improvement in symptoms for those treated with antibiotics compared with placebo.
This review of four RCTs comparing antibiotic and placebo in the management of acute rhinosinusitis would suggest to us that there is some but nevertheless insufficient evidence to support the routine use of antibiotics.
Each of these RCTs was, in our opinion, well conducted, but not without small risk of internal bias. Two RCTs do show benefit over placebo in the use of antibiotics. Although both these two RCTs have been conducted by the same principle investigator, there is no evidence that this should detract from the validity of their findings. Adverse reactions, principally gastrointestinal, were nearly three times more common in those treated with antibiotics (a total of 35 reported in the antibiotic groups compared with 13 in those prescribed placebo). No children treated with placebo only are reported to have developed significant complications due to acute sinusitis.
The meta-analysis (figure 1) does suggest evidence for the efficacy of antibiotics in the treatment of acute rhinosinusitis. The NNT of 8 (range 3–50 for individual studies) can perhaps be placed in the context of the NNT data for acute otitis media (AOM) in children quoted in National Institute for Health and Clinical Excellence (NICE) recommendations for self-limiting respiratory tract infections.6 Here the NNT values for children less than 2 years with AOM and those children with both AOM and otorrhoea are 4 and 3, respectively (NICE suggest these may benefit from antibiotics). In other children with AOM, the NNT varies between 8 and 20 pending the measured outcome (NICE does not advise antibiotics in these cases). In our view, and despite the assessment that all four trials were well conducted, the analysis is weakened by the low number of included RCTs and the methodological differences between studies. It probably should not be used for the basis of any firm conclusion regarding the use of antibiotics. Our assessment would be supported by the broad predictive interval for the random-effects distribution.
External bias is potentially a significant factor affecting all four studies. The difficulties relating to diagnosis have already been emphasised. Each of the four RCTs considered here uses slightly different diagnostic protocols (see table 2) which is likely to contribute to variation in outcome. There are different criteria for inclusion or exclusion based on duration of symptoms. Two of the four studies include imaging as part of their diagnostic criteria. In only two is the possible use of ancillary drugs considered, and just one has included detailed use of these drugs as part of the participant breakdown. This is of importance as there is evidence for symptom improvement using douching, steroids, decongestants, antileukotrienes and antihistamines.39 In all four studies, there is exclusion of patients based around severity assessment, or in one case positive streptococcal cultures. One of the RCTs has been criticised for its overly strict criteria in which patients with slightly more severe symptoms were excluded, clearly this may have biased this trial against bacterial infections.11 ,30 Age range criteria are different for each of the four studies as are either the antibiotics or antibiotic doses used as the treatment option. It is the questionable grounds for external validity, therefore, that makes these four studies a poor source for evidence-based generalisation. Clearly, patient selection methodology has the potential to either favour or disfavour the percentage of those with bacterial causes for acute rhinosinusitis included in each study; the true effect of antibacterials cannot be identified until this source of bias is made consistent between studies.
At the pragmatic level, it would be useful to know whether the prescription of antibiotics as routine treatment for acute rhinosinusitis in children, irrespective of uncertainties as to the cause of sinus inflammation, is warranted in terms of safety and effectiveness. Our conclusion from this review would be that the evidence to support the routine use of antibiotics here remains unclear despite the positive findings of the statistical analysis. The evidence base is clearly inadequate and may be placed in the context of the larger systematic review reported for adult patients which does suggest some small benefit from the use of antibiotics. Future RCTs on this subject are faced with the difficulty of bringing further uniformity and accuracy to the application of diagnosis; this is a significant challenge as the introduction of any radiological or other diagnostic test is likely to detract from utility in primary care, yet diagnostic criteria that are too unrestrictive may lack the power of consistency between studies. Studies with more inclusive criteria are less likely to demonstrate antibiotic efficacy than those that favour the capture of participants with more severe symptoms. The authors would support current UK guidelines that promote a conservative approach to the treatment of acute rhinosinusitis in children with antibiotic prescribing only in selective cases.
Contributors MJC is the principle author, undertook the quantitative analysis and contributed to all sections of the review. SK assisted with the literature search, assessment of studies for inclusion and planning of the review. SS undertook an independent review both of the literature search and of the assessment of included studies for internal and external bias.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Prisma We have referred to the PRISMA statement for the reporting of this systematic review.