Effect of Antibiotics for Otitis Media on Mastoiditis in Children: A Retrospective Cohort Study Using the United Kingdom General Practice Research Database
BACKGROUND. Information is needed on whether mastoiditis has increased in association with the decline in antibiotics prescribed to children by primary care physicians in the United Kingdom.
OBJECTIVE. To determine time trends in mastoiditis incidence, the frequency of antecedent otitis media, and the effect of antibiotics for otitis media on the risk of mastoiditis in children.
PATIENTS AND METHODS. We conducted a retrospective cohort study by using the UK General Practice Research Database. Children aged 3 months to 15 years between 1990 and 2006 were included. Risk of mastoiditis within 3 months after otitis media diagnosis and the protective effect of antibiotics were determined.
RESULTS. There were 2 622 348 children within the General Practice Research Database; 854 had mastoiditis, only one third of whom (35.7%) had antecedent otitis media. Mastoiditis incidence remained stable between 1990 and 2006 (∼1.2 per 10 000 child-years). Risk of mastoiditis, after otitis media, was 1.8 per 10 000 episodes (139 of 792 623) after antibiotics compared with 3.8 per 10 000 (149 of 389 649) without antibiotics, and increased with age. Antibiotics halved the risk of mastoiditis. General practitioners would need to treat 4831 otitis media episodes with antibiotics to prevent 1 child from developing mastoiditis. If antibiotics were no longer prescribed for otitis media, an extra 255 cases of childhood mastoiditis would occur, but there would be 738 775 fewer antibiotic prescriptions per year in the United Kingdom.
CONCLUSIONS. Most children with mastoiditis have not seen their general practitioner for otitis media. Antibiotics halve the risk of mastoiditis, but the high number of episodes needing treatment to prevent 1 case precludes the treatment of otitis media as a strategy for preventing mastoiditis. Although mastoiditis is a serious disease, most children make an uncomplicated recovery after mastoidectomy or intravenous antibiotics. Treating these additional otitis media episodes could pose a larger public health problem in terms of antibiotic resistance.
Otitis media is among the most common infections of childhood, the main reason for a child to visit their primary care physician (ie, general practitioner [GP])1,2 and one of the primary childhood conditions for which GPs prescribe antibiotics.3–5 Otitis media affects all age groups but is most frequent in the preschool years.6 Most episodes are self-limiting without antibiotic treatment,7 but otitis media is the major cause of childhood deafness8 and can lead to meningitis and mastoiditis, albeit rarely.9 Marked variation in antibiotic prescribing between practices reflects the different thresholds GPs use when weighing the need to avoid unnecessary prescribing with the benefits of preventing complications such as mastoiditis.10
Recent findings from a meta-analysis of randomized, controlled trials involving 1643 children showed that children under 2 years of age stand to benefit most from antibiotics for otitis media.11 The authors recommended observation rather than immediate antibiotic treatment for most children with acute otitis media, but this meta-analysis was not adequately powered to determine the protective effect of antibiotics for otitis media on the risk of mastoiditis. Findings from ecological analyses of routine health care databases have suggested that the reduction in GP prescribing of antibiotics to children may be associated with an increase in rare complications of bacterial infection, such as mastoiditis. One study associated the 23% decrease in overall antibiotic prescribing between 1996 and 2002 with an increase in hospital admissions for respiratory tract infections.12 A data linkage study of 96 health authorities of England concluded that higher overall use of penicillin was associated with significantly fewer hospital admissions for mastoiditis.13 Another study noted that the 37% decline in overall antibiotic prescribing between 1993 and 2003 coincided with an increase in hospital admissions for mastoiditis in children under 4 years of age.14 Large studies with treatment and outcome data on individual patients are now indicated to determine if these associations are explained by adverse effects of the decline in antibiotic use or changes in access to services, referral patterns, awareness of the diagnosis, and coding of routine data.
In the United Kingdom, >98% of the population is registered with a National Health Service GP15 and, on average, children (aged 0–15 years) consult their GP at least 3 times a year.16,17 We used individual patient data from the UK General Practice Research Database (GPRD) to investigate whether the occurrence of mastoiditis is a useful marker of the potential adverse effects of reduced antibiotic prescribing.
This study was conducted by the UK collaborative Improving Children's Antibiotic Prescribing Group.
PATIENTS AND METHODS
We conducted a case review and retrospective cohort study by using the GPRD (Fig 1). All children aged 3 months to 15 years who were registered with an “up-to-standard” general practice (indicating that data recording by that practice has been verified to meet the required data quality criteria)18 between January 1, 1990 and December 31, 2006 were included. Children temporarily registered with a general practice were excluded to avoid duplication of data, because these children were most probably also permanently registered elsewhere.
Source Data Set
The GPRD is one of the world's largest computerized databases of anonymized longitudinal general practice patient records,19 comprising ∼6% of children in the UK census population. Practices contributing to the GPRD are representative of practices in the total United Kingdom and are under contract to record all diagnoses, prescriptions, immunizations, hospital referrals, and test results of all active patients. Validation studies show quality and completeness of the GPRD data are high,20 and it has been used to investigate child health issues.21,22
Identification of Children With Mastoiditis and Exposures of Interest
Diagnoses in the GPRD are classified by using either Oxford Medical Information Systems or Read codes (see Tables 3 and 4, which are published as supporting information on www.pediatrics.org/content/full/123/2/424). We identified children with mastoiditis by searching the GPRD for codes relating to mastoiditis or mastoidectomy. Codes for mastoidectomy were used in addition to mastoiditis diagnosis codes to ensure that all children with mastoiditis were identified. We defined the date of diagnosis as the first record of mastoiditis and censored any subsequent episodes. To determine the prevalence of antecedent otitis media or comorbidities that might predispose to mastoiditis, we reviewed all GP attendances preceding the date of mastoiditis diagnosis and recorded any otitis media or related middle ear problems, antibiotic prescriptions, and immunologic, craniofacial, or neurologic abnormalities. Our use of a comprehensive approach to capture all diagnoses indicative of otitis media minimized the problems associated with variability in diagnostic coding between GPs. Any mention of otitis media within the same 14-day period was defined as 1 episode. All codes were identified from the GPRD medical/product dictionaries and validated by a pediatrician. Analyses of otitis media and antibiotic prescribing focused on events in the 3 months preceding mastoiditis, because mastoiditis is an acute complication of otitis media,23–25 whereas analyses of predisposing factors such as immunologic, craniofacial, or neurologic abnormalities were based on the entire medical history preceding mastoiditis.
Age- and calendar year-specific incidence rates for mastoiditis (with and without otitis media in the preceding 3 months), for otitis media, and for antibiotic prescribing for otitis media were calculated per 1000 child-years at risk in the GPRD database. Ninety-five percent confidence intervals (CIs) were generated by using Poisson approximation, and tests for linear trend were conducted.
Analyses of the effect of antibiotic treatment and other covariates on the risk of mastoiditis were confined to episodes of otitis media without immunologic, craniofacial, or neurologic abnormalities. We estimated the risk of mastoiditis in children within 3 months of a treated and untreated otitis media episode. The negligible number of children (2.7%) with <3 months of follow-up time after an otitis media episode were included in these calculations. Children were classified as “treated” if they had an antibiotic prescribed in the same GP consultation as otitis media was diagnosed. If a child had multiple episodes of otitis media in the 3 months preceding mastoiditis, the episode closest to the date of mastoiditis was used. Because of potential miscoding of mastoiditis, we performed sensitivity analyses of treatment efficacy restricted to children who had surgery for mastoiditis. We did not take into account clusters of repeated episodes of otitis media in the same child. The risk difference and number needed to treat (NNT) (1/risk difference) were computed. The effect of antibiotic treatment was determined by using logistic regression to calculate the crude odds ratios (OR) for mastoiditis for the following covariates: gender, age group, otitis media diagnosis, and number of otitis media episodes in the 3 months preceding mastoiditis. We retained covariates in the model that gave a P value of <.05 by using the likelihood ratio statistic. Data management and analysis were performed by using Stata/SE 9.2 software (Stata Corp, College Station, TX).26
Ethics approval for this research was obtained from the Scientific and Ethical Advisory Group for the General Practice Research Database.
Between 1990 and 2006 there were 2 622 348 children aged 3 months to 15 years in the GPRD, contributing 7 119 677 child-years of data (Fig 1) from 423 general practices; 854 children with mastoiditis were identified, all of which were incident cases. The median age at diagnosis was 9.7 years (interquartile range [IQR]: 6.2–12.9 years); incidence of mastoiditis peaked in children aged <1 year, with 10.3% (88 of 854) of all mastoiditis occurring in children <2 years old. Just more than half (57.6% [492 of 854]) were male.
Only one third (35.7% [305 of 854]) of children with mastoiditis had visited their GP and been diagnosed with otitis media in the preceding 3 months. The average time between an episode of otitis media and mastoiditis was 21.5 (IQR: 6.0–51.5) days. Almost half (47.5% [145 of 305]) of the children with antecedent otitis media had been prescribed antibiotics for this condition, of whom 76% received amoxicillin and 15% erythromycin. This is in accordance with the treatment recommendations for otitis media.27 There were no differences in the choice of antibiotic therapy between children who developed mastoiditis after treated otitis media and those who did not. In total, 40.9% (349 of 854) of all children with mastoiditis had an antibiotic prescribed for any reason within the preceding 3 months; 53.6% (458 of 854) proceeded to mastoidectomy. Of the children with mastoiditis and antecedent otitis media, 5.6% (17 of 305) had an immunologic, craniofacial, or neurologic abnormality and were, therefore, excluded from the cohort analyses.
We identified 1 182 272 episodes of otitis media in 462 904 children. On average, children had 0.4 (SD: ± 0.80) episodes of otitis media per year, decreasing with increasing age. The median age at otitis media diagnosis was 5.4 (IQR: 2.3–7.7) years and approximately half (51.1% [604 113 of 1 182 272]) of the episodes occurred in boys. We found contrasting trends with age for rates of mastoiditis and otitis media diagnoses. The incidence of mastoiditis diagnoses was highest in infants, lowest in 2-year-olds, and increased steadily with age thereafter. The increasing linear trend between 2 and 15 years of age was significant at the 5% level only for children without recognized antecedent otitis media (Fig 2). In contrast, the incidence of otitis media diagnoses decreased with increasing age (from 523.0 [95% CI: 520.7–525.3]) to 44.4 [95% CI: 43.8–45.0]) diagnoses per 1000 child-years in 0- and 15-year-olds, respectively) (P < .01).
The incidence of mastoiditis diagnoses remained stable between 1990 and 2006, with an average incidence of 0.12 (95% CI: 0.11–0.13) diagnoses per 1000 child-years. There was no evidence for a linear trend over time (P = .45) (Fig 3). In contrast, the incidence of otitis media diagnoses fell by 34% between 1990 and 2006 from 221.4 (95% CI: 219.5–223.4) to 147.0 (95% CI: 146.1–148.0) diagnoses per 1000 child-years (P < .01), whereas the incidence of antibiotic prescribing for otitis media declined by 49.6% from 170.4 (95% CI: 168.7–172.2) to 85.9 (95% CI: 85.2–86.7) prescriptions per 1000 child-years (P < .01). The proportion of otitis media episodes treated with antibiotics decreased significantly over the study period, from 77% in 1990 to 58% in 2006 (P < .01).
The prescription of antibiotics for otitis media significantly reduced the risk of developing mastoiditis within the following 3 months (OR: 0.56 [95% CI: 0.44–0.71]; adjusted for gender, age group, and otitis media diagnosis) (Table 1). In sensitivity analyses restricted to children with mastoiditis requiring surgery, the adjusted OR was 0.39 (95% CI: 0.28–0.55; 55 children in the treated group and 92 in the untreated group). The overall risk of developing mastoiditis within 3 months of an otitis media episode was 2.4 per 10000 (288 of 1 182 272) otitis media episodes. Among children treated with antibiotics for otitis media, the risk of mastoiditis was 1.8 per 10 000 (139 of 792 623) compared with 3.8 per 10 000 (149 of 389 649) for untreated children (risk difference: 2.0 per 10 000 otitis media episodes) (Table 2). The likelihood of a child receiving an antibiotic for otitis media was significantly associated with age (P < .01), with older children being less likely to receive a prescription.
On average, 4831 episodes of otitis media would need to be treated with antibiotics to prevent 1 child from developing mastoiditis (Table 2). The NNT was lower in older age groups but still exceeded 2000. Based on data extracted from the GPRD for the year 2006, there were ∼1 273 750 episodes of otitis media per year in children in the United Kingdom, of which 58% received an antibiotic. Completely stopping the practice of prescribing antibiotics for otitis media would result in an additional 2 cases of mastoiditis per 10 000 otitis media episodes (risk difference). This would total 255 extra cases of childhood mastoiditis per year in the United Kingdom while reducing antibiotic usage by 738 775 prescriptions per year.
Summary of Main Findings
Mastoiditis is a rare condition, affecting 1.2 children per 10 000 years of follow-up. Only one third of affected children had a GP record for otitis media within the preceding 3 months. The risk of mastoiditis after an otitis media episode increased with age but was never >16 per 10 000 otitis media episodes. Antibiotic prescribing for otitis media decreased by 50% over the study period, but the rate of mastoiditis diagnoses remained relatively stable. Antibiotics halved the risk of mastoiditis in children with otitis media but, because of the low risk of mastoiditis, the number of otitis media episodes needing antibiotic treatment to prevent 1 case of mastoiditis (NNT) is very high.
The strengths of this study were the large sample size, which allows for examination of the risk of mastoiditis according to patient characteristic and antibiotic treatment, and the representiveness of practices in the UK primary care setting.
The major weakness, common to all studies based on routinely collected clinical data, was misclassification of the occurrence and timing of exposures. The incidence of mastoiditis may have been under-estimated as it is possible that not all hospital diagnoses of mastoiditis were entered into the GPRD. However, over-estimation is unlikely to be a problem as mastoiditis is an unusual and specific condition. Mastoiditis may have been miscoded, but restriction of analyses to mastoiditis requiring surgery did not significantly change the treatment effect. Under-reporting of otitis media could have occurred if GPs recorded episodes under other categories, such as upper respiratory tract infection. Finally, treatment efficacy could have been underestimated if GPs were more likely to record otitis media if they prescribed antibiotics than when no antibiotics were prescribed.
A limitation of the GPRD is that prescriptions are not directly linked to indication. Our linkage method assumed that antibiotics prescribed in the same GP consultation as otitis media was diagnosed were for the treatment of otitis media. However, this assumption should not have influenced the mastoiditis outcome analysis so long as the antibiotic would be appropriate for the treatment of otitis media. Compliance data are not available in the GPRD, thus, the treated group may have included patients who did not fill or take their antibiotic prescription. Lack of severity data meant that we could not examine whether children with severe otitis media were more likely to be treated with an antibiotic and were at greater risk of developing mastoiditis. Both poor compliance and indication bias would bias treatment efficacy toward the null effect.
Although the analyses did not account for clusters of repeated otitis media episodes in children, we found no effect of the number of otitis media episodes within the 3 months preceding mastoiditis (P > .05). Inclusion in the analyses of treatment efficacy of the small number of otitis media episodes with <3 months of follow-up is unlikely to have biased results as loss to follow-up was unlikely to have been conditional on whether mastoiditis developed or not.
Mastoiditis is not a useful marker of the potential adverse effects of reduced antibiotic prescribing. Firstly, it is a nonspecific marker as only a small proportion (one third) of mastoiditis cases are affected by changes in antibiotic prescribing practice for otitis media. Secondly, it is insensitive as mastoiditis is a rare disease.
Strengths and Weaknesses in Relation to Other Studies
This is the largest study to investigate the frequency of antecedent otitis media in children with mastoiditis and the first study to investigate the potential adverse effects of reduced antibiotic prescribing for otitis media on mastoiditis, specifically in children. The national representiveness of our study is confirmed by the consistency of findings from other UK-wide studies with our results. The UK incidence of mastoiditis in 0- to 14-year-olds between 1991 and 1998 was reported as 0.15 per 1000 child-years (Department of Health hospital discharge data),28 which is comparable with our finding of 0.12 per 1000 child-years in 3 months to 15-year-olds. The occurrence of mastoiditis without preceding otitis media has also been observed in other case-review studies,29–31 with approximately half of all affected children having no recent history of otitis media.
The age distribution of our study does differ in comparison to existing literature, but previous research has been based on case-series, which could be biased by referral practice. For example, most studies cite the rate of mastoiditis as peaking among 1- to 4-year-olds, but these analyses are mainly based on patients presenting to children's ear, nose, and throat (ENT) specialists and may ignore mastoiditis in adolescents, who could be referred to adult ENT specialists.
A recent study used the GPRD to investigate the extent to which antibiotics reduce the risk of serious complications after common respiratory tract infections, including mastoiditis after otitis media.32 The study found that antibiotics halved the risk of mastoiditis (adjusted OR: 0.56 [95% CI: 0.37–0.86]), estimating that 4064 episodes of otitis media would need to be treated with antibiotics to prevent 1 case of mastoiditis, which is in agreement with our findings. However, our study provides additional analyses, examining the antecedents of mastoiditis and the effect of age and changes in antibiotic prescribing on mastoiditis over time, specifically in children. Our data set comprised over twice as many general practices, yielding more than double the number of otitis media episodes for follow-up. Our study period also included more up-to-date information.
Most children with mastoiditis have not previously seen their GP for otitis media. The use of antibiotics for otitis media halves the risk of mastoiditis, but the high number needing treatment to prevent 1 case precludes the treatment of otitis media as a strategy to prevent mastoiditis. Even in adolescents with otitis media, who are most at risk of developing mastoiditis, the NNT exceeds 2000. Although mastoiditis is a serious disease, most children make an uncomplicated recovery after mastoidectomy or intravenous antibiotics. Treating these additional otitis media episodes could pose a larger public health problem because of the possibility of future treatment failure as a result of antibiotic-resistant pathogens, increased exposure of children to the adverse reactions to antibiotic treatment, and unnecessary prescribing costs. Nevertheless, GPs and pediatricians working in primary care need to be able to recognize the signs of mastoiditis (postauricular swelling and protrusion of the auricle),31 particularly in older children, and to refer promptly to ENT services.
Recent data suggest that the pneumococcal conjugate vaccine Prevenar (introduced into the UK immunization schedule from September 2006)33 is effective in reducing otitis media-related consultations and antibiotic prescribing for such.34 This may result in a future decline in mastoiditis. Trends in antibiotic prescribing need to be formally monitored in association with the longitudinal follow-up of individual patient disease outcomes data, to ensure that any changes in antibiotic use are not causing harm.
Dr Wong's post was funded by a Department of Health Public Health Career Scientist Award. Dr Saxena was funded by a postdoctoral award from the Department of Health. The license for the GPRD was funded by the European Commission via the Taskforce European Drug Development for the Young (TEDDY) network of Excellence European Commission Framework 6 Programme, 2005–2010.
We thank the general practitioners who contributed data to the GPRD.
- Accepted May 12, 2008.
- Address correspondence to Ian Chi Kei Wong, BSc, MSc, PhD, University of London, Centre for Paediatric Pharmacy Research, School of Pharmacy, BMA House, Tavistock Square, London WC1H 9JP, England. E-mail:
Financial Disclosure: Ms Thompson and Drs Gilbert, Sharland, and Wong were members of the UK Department of Health's Specialist Advisory Committee on Antimicrobial Resistance (SACAR), pediatrics subgroup. Drs Long and Saxena have no financial relationships relevant to this article to disclose.
Drs Sharland and Wong had the original idea for the study; Ms Thompson extracted the relevant data from the GPRD; Ms Thompson and Dr Gilbert developed the analytical strategy, and all authors were involved in the interpretation of the data; the manuscript was prepared by Ms Thompson in consultation with all other authors; and Dr Wong is the article guarantor.
What's Known on This Subject
Ecological studies of routine health care databases have suggested that the reduction in antibiotic prescribing to children may be associated with an increase in rare complications of bacterial infection, particularly mastoiditis after otitis media. Individual patient-level analyses are now required.
What This Study Adds
Most children with mastoiditis have no recent history of otitis media. Antibiotics for otitis media halve the risk of mastoiditis, but the high NNT precludes the treatment of otitis media as a strategy for preventing mastoiditis.
- ↵Congeni BL. Acute otitis media. Curr Treat Options Infect Dis.2003;5 :151– 157
- ↵Freid VM, Mukuc DM, Rooks RN. Ambulatory health care visits by children: principal diagnosis and place of visit. Vital Health Stat.1998;13 (137):1– 23
- ↵Froom J, Culpepper L, Grob P, Bartelds A, Bowers P, Bridge-Webb C. Diagnosis and antibiotic treatment of acute otitis media: report from International Primary Care Network. Br Med J.1990;300 (6724):582– 586
- Gonzales R, Malone DC, Maselli JH, Sande MA. Excessive antibiotic use for acute respiratory infections in the United States. Clin Infect Dis.2001;33 (6):757– 762
- ↵Casselbrant ML, Mandel EM. Epidemiology. In: Rosenfeld RM, Bluestone CD, eds. Evidence-based Otitis Media. Hamilton, ON: B.C. Decker; 1999:117–136
- ↵Glasziou PP, Del Mar CB, Sanders SL, Hayem M. Antibiotics for acute otitis media in children. Cochrane Database Syst Rev.1997;(1):CD000219
- ↵Bielicka A, Debska M, Brozek E, Chmielik M. Complications of otitis media in children: a continuing problem. New Med.2003;6 (2):18– 21
- ↵Wrigley T, Tinto A, Majeed A. Age- and sex-specific antibiotic prescribing in general practice in England and Wales, 1994 to 1998. Health Stat Q.2002;14 :14– 20
- ↵Majeed A, Williams S, Jarman B, Aylin P. Prescribing of antibiotics and admissions for respiratory tract infections in England. Br Med J.2004;329 (7471):879
- ↵Little P, Watson L, Morgan S, Williamson I. Antibiotic prescribing and admissions with major suppurative complications of respiratory tract infections: a data linkage study. Br J Gen Pract.2002;52 (476):187– 193
- ↵Sharland M, Kendall H, Yeates D, et al. Changes in antibiotic prescribing in general practice and hospital admissions for quinsy, mastoiditis and rheumatic fever, in children. Br Med J.2005;331 (7512):328– 329
- ↵Royal College of General Practitioners. General Practitioner Workload; 2004. Available at: www.rcgp.org.uk/pdf/ISS_INFO_03_APRIL04.pdf. Accessed April 3, 2008
- ↵Saxena S, Majeed A, Jones M. Socioeconomic differences in childhood consultation rates in general practice in England and Wales: prospective cohort study. BMJ.1999;318 (7184):642– 646
- ↵Office for National Statistics. General Household Survey 2004/05. London, United Kingdom: ONS; 2005
- ↵Card T, Logan RFA, Rodrigues LC, Wheeler JG. Antibiotic use and the development of Crohn's disease. Gut.2004;53 (2):246– 250
- ↵Rani F, Murray ML, Byrne PJ, Wong IC. Epidemiologic features of antipsychotic prescribing to children and adolescents in primary care in the United Kingdom. Pediatrics.2008;121 (5):1002– 1009
- ↵Murray ML, de Vries CS, Wong ICK. A drug utilisation study of antidepressants in children and adolescents using the General Practice Research Database. Arch Dis Child.2004;89 (12):1098– 1102
- ↵Ginsburg CM, Rudoy R, Nelson JD. Acute mastoiditis in infants and children. Clin Pediatr (Phila).1980;19 (8):549– 553
- ↵Stata Data Analysis and Statistical Software. Available at: www.stata.com. Accessed December 14, 2008
- ↵British National Formulary for Children 2007. London, United Kingdom: BMJ Publishing Group Ltd; 2007:295
- Harley EH, Sdralis T, Berkowitz RG. Acute mastoiditis in children: a 12-year retrospective study. Otolaryngol Head Neck Surg.1997;116 (1):26– 30
- ↵Petersen I, Johnson AM, Islam A, Duckworth G, Livermore DM, Hayward AC. Protective effect of antibiotics against serious complications of common respiratory tract infections: retrospective cohort study with the UK General Practice Research Database. Br Med J.2007;335 (7627):982
- ↵Chief medical officer letter. 2006. Available at: www.dh.gov.uk/assetRoot/04/13/71/75/04137175.pdf. Accessed December 14, 2008
- ↵Zhou F, Shefer A, Kong Y, Nuorti JP. Trends in acute otitis media-related health care utilization by privately insured young children in the United States, 1997–2004. Pediatrics.2008;121 (2):253– 260
- Copyright © 2009 by the American Academy of Pediatrics