Strengths and limitations
The main limitation of this study relates to coding within the CPRD database. Diagnoses cannot be captured by the CPRD if individuals do not present to health services or if they were recorded using free-text entries. In addition, otorrhoea is a complication of AOM; coding the episode as AOM would not be captured in the results. It is expected this work underestimates the exact disease burden.
Otorrhoea in the paediatric population most frequently presents because of AOM. It can also, less frequently, be caused by otitis externa in teenagers. The authors have taken a pragmatic approach by collecting both diagnostic codes and clinical signs (for example, otorrhoea), which inevitably will identify a number of CYP with otitis externa. In this study, conditions causing otorrhoea have been grouped as ‘paediatric otorrhoea’ to represent real-life practice. The frequency of coded events in CYP aged 8–16 years, when otitis externa may present, is a small proportion of the total cohort, but will skew the results. Prescribing for otitis externa may account for the increased use of aminoglycoside topical antibiotics.
It is not possible to ascertain whether antibiotic prescribing was intended for immediate or delayed use. NICE guidelines suggest both are practical treatment options.9 Delayed antibiotics can reduce unnecessary use for those whose symptoms will self-resolve. Smith et al demonstrated that the majority of CYP with PO in primary care received immediate antibiotics.1 It was suggested this was because those who had otorrhoea were more unwell than those without.
The decision to prescribe antibiotics in primary care has been shown to be multifactorial.18 Factors are known to include comorbidities (for example, immunosuppression), prior otological surgery, prescribing experience, guidelines, time pressures, and professional and patient preferences.19 Within the CPRD the exact reason for a prescription cannot be determined, as the CPRD does not link prescriptions to diagnoses. Consequently, it is not possible to elucidate why prescribers took certain prescribing decisions. The decision-making process is complex, and therefore it was decided it was outside the scope of this study.
Monitoring antimicrobial resistance patterns is essential for antimicrobial stewardship.9 It was not possible to collect sufficient microbiological data from the CPRD because of insufficient coding. It is known that approximately 15% of microorganisms found in PO show resistance to amoxicillin.20 Co-amoxiclav has lower (5%) resistance.
Comparison with existing literature
Only one other study, to the authors’ knowledge, has been performed to investigate the incidence of PO, which was performed over a single-year period with 67 000 patients in the Netherlands.5 The presentation rate identified in the current study is much lower in comparison (3.15 versus 12.6 per 1000 patient-years).5 Reasons for this could include coding variation, study population size, population characteristics, and health-seeking behaviour differences between countries.
A reduction in AOM incidence has been linked with the introduction of the pneumococcal conjugate vaccine.2 The gradual reduction of PO incidence in the current study’s cohort may reflect that in PO Streptococcus pneumoniae is found in only 6% of microorganisms isolated.21
PO places a financial burden on NHS healthcare resources, costing primary care alone around £2 million per year. It is not currently possible, because of coding limitations, to determine the cost to secondary care services, which involves emergency department attendances, specialist appointments, audiology, and surgical management. It is likely there is a sizable personal cost associated with PO. Hollinghurst et al identified the cost to caregivers of children with acute cough in the NHS setting.22 Translated to PO, the cost to caregivers per year is approximately £608 000 in the UK.
The current results demonstrate the prescribing practices of primary care professionals who are able to prescribe within the NHS, the majority of which are from GPs (82.9% [47 991/57 887]). Private prescriptions are not included. The current results are generalisable to primary care services in the UK. The authors of the current study acknowledge that there is significant antibiotic prescribing variation within primary care.23 Factors that can contribute are local deprivation, health professional type, and experience.
Younger age and male sex had a significant effect on the incidence of PO (Table 1). AOM, which occurs before perforation and resultant otorrhoea, is known to particularly affect children in the first 2 years of life.24 Males have been previously shown to experience viral infections, such as bronchiolitis, more frequently than females.25 The reason for this is uncertain, although some have speculated it could be because of sex hormone effects on T-helper cells.26
Children from low socioeconomic backgrounds are known to more frequently experience recurrent ear infections and associated complications.6,27–29 The data in the current study show the converse of what would be expected in relation to disease incidence relating to socioeconomic background. The reason for this finding requires further research but may suggest there is a barrier or alternative route to healthcare access. This also suggests that the true rate of PO is likely to be much higher than described. In comparison with data from the US, the current data showed no prescribing disparities related to deprivation.29
The WHO states that antimicrobial resistance is one of the biggest threats to global public health, data collection on antibiotic use is required, and that evidence-based prescribing should be the standard of care.10 In 2010, Smith et al showed in a cohort of 38 CYP with acute otorrhoea that 92.1% (35/38) received antibiotics.1 The current study’s results demonstrate a slightly reduced rate: 71.9% (57 885/80 454) of CYP with otorrhoea received antibiotics, the majority being oral antibiotics. Of the oral antibiotics prescribed, 79% were amoxicillin in accordance with NICE guidance.9 Macrolides were the second and fourth most frequently prescribed oral antibiotic likely used, in accordance with NICE recommendations, for those with a perceived penicillin allergy.9 The rate of macrolide prescription (10.5%) mirrors the estimated penicillin allergy prevalence in the UK.30 Antibiotic prescribing data collected in this study relate to the initial presentation.
The data in this study show that 19.4% (15 608/80 454) of CYP represent with PO. Previous data has shown that 44% of CYP with PO have at least one further infection within 3 months.1 This demonstrates the treatment-resistant nature of PO. Approximately 15% of microorganisms from PO are resistant to amoxicillin and this may correlate with re-attendances.21 Resistance to both erythromycin and co-amoxiclav, which are second-line antibiotics outlined by NICE, are 36% and 5%, respectively. The collection of microbial samples in the form of ear swabs is not outlined in NICE guidance for AOM but may play a role to ensure effective antibiotics are administered.
The use of topical antibiotic drops for patients with a tympanic membrane perforation is debated because of the potential ototoxic effects of aminoglycoside antibiotics.11,31 The variation in prescribing practice for topical antibiotics could be related to the lack of guidelines, medication availability, and education. It is interesting that aminoglycoside antibiotics comprise the top three most frequently prescribed topical drops for otorrhoea in primary care (Table 2). Professionals prescribing topical antibiotic preparations may not be aware that they are prescribing aminoglycosides as many drops have numerous drug constituents (for example, Otomize and Sofradex). There is, however, a trend that ciprofloxacin-based topical drops are becoming more favoured over time (Figure 3). Microorganisms causing PO have been shown to have low resistance to ciprofloxacin (4–8%).22
Implications for research and practice
In conclusion, this longitudinal population-based study shows that approximately 41 000 primary care appointments are required for PO each year. Antibiotic treatment most frequently follows NICE recommendations to prescribe oral amoxicillin.
The medical community needs to be cognisant that when otorrhoea is present, especially in younger children, the tympanic membrane is likely to be perforated. This can provide an alternative route for antibiotic delivery in a topical form. Practice is changing from potentially ototoxic aminoglycoside topical antibiotics to ciprofloxacin.
To help improve future primary care research, coding of CYP with PO needs to be distinguished from AOM without otorrhoea. Future research should look to determine the best method of treating PO in primary care, which prevents repeat infective episodes, in turn informing future guidance.