Objective: To assess the clinical and laboratory features of acute otitis media (AOM) in infants younger than 2 months, to look for factors predicting bacterial otitis, and to evaluate the accuracy of AOM diagnosis among paediatricians.
Methods: The study population comprised a cohort of 277 hospitalised infants up to 61 days old that were treated for the first episode of AOM in a paediatric department. We reviewed their medical records and analysed the demographic, clinical and laboratory data, and the diagnosis made by both paediatricians and otolaryngologists.
Results: Presenting symptoms were mainly respiratory (70.0%) and fever (62.5%). The most common pathogens were Streptococcus pneumoniae and Haemophilus influenzae. Gram-negative bacilli grew in 10.5% of the infants. Multivariate analysis revealed that AOM in the second month of life was associated with male gender, concurrent bronchiolitis and diarrhea. Although high leukocyte count was associated with bacterial pathogen, more than 70% of the patients with positive culture had normal white blood cell counts. The paediatrician diagnosed only 45% of the patients subsequently diagnosed with AOM by an otolaryngologist.
Conclusions: The absence of predictors for bacterial infection in more than 70% of bacterial AOM suggests that empirical antibiotic treatment should be advised for the young infants with AOM even when afebrile and with normal laboratory profile. A low diagnostic rate of AOM by the paediatrician emphasizes the need for improvement in examination skills and instrumentation to allow a thorough ear evaluation in children of a very young age.
Statistics from Altmetric.com
Acute otitis media (AOM) is a common childhood disease and the most common cause for antibiotic prescriptions in children.1 2 Owing to the immaturity of the immune system and the eustachian tube anatomy in infancy, infants are prone to develop AOM,3 and indeed the incidence is highest in the first 2 years of life.4 Early onset AOM predicts later recurrent or chronic otitis media and impaired hearing.5 Although infectious diseases in early infancy are common and responsible for substantial mortality and morbidity, there are relatively few studies of ear infections at this very young age.6–11 Several trends in AOM characteristics have been recently reported including a change in its prevalence and in the profile of pathogenic bacteria.12–15 These changes have not been evaluated in infants under 2 months of age, who were excluded from most studies addressing AOM.2
The diagnosis of AOM requires signs and symptoms of middle ear inflammation, such as earache or ear pulling,2 which are irrelevant in very young infants where symptoms are nonspecific. In addition, it is difficult to see their eardrums due to the narrow external canal and the unique appearance of the normal tympanic membrane at this age.8 16 There are no data about the accuracy of AOM diagnosis in this age group, and the paediatrician often perceives that the diagnosis of AOM is based on guesswork rather than accurate clinical evidence.
Our aims in this study were to summarise the clinical and laboratory data in a large cohort of young infants under 2 months of age with AOM, to look for factors predicting bacterial otitis especially in the presence of diagnostic difficulties, and to evaluate the accuracy of the paediatrician’s diagnosis of AOM in this age group.
Bikur Cholim General Hospital, located in the centre of Jerusalem, is a close referral medical centre for heavily populated Arab and Jewish neighbourhoods. The paediatric department, affiliated to the Hebrew University Hadassah Medical School, has 36 in-patient beds. The study population comprised a cohort of all hospitalised infants aged up to 61 days treated for the first episode of AOM at the paediatric department during a period of 9 years (1994–2003). All patients were hospitalised with suspected infectious disease and were examined by a paediatrician and subsequently (within up to 12 h) by an otolaryngologist using an operating binocular microscope.
The diagnosis was based on clinical symptoms (fever, irritability) and signs of middle-ear inflammation (redness, opacification, fullness or bulging of tympanic membrane).2 The external ear canal was cleared of cerumen with a curette or by suction and then sterilised with alcohol. In patients with AOM, tympanocentesis and middle ear fluid (MEF) aspiration were performed. Aerobic cultures were processed in the hospital microbiology laboratory. In all febrile patients additional neonatal sepsis work-up was performed including blood, urine and cerebrospinal fluid cultures. Data were extracted retrospectively from the patients’ charts and included: age, sex, hospitalisation date, birth order, birth weight and gestational age, perinatal morbidity, feeding habits, previous antibiotic treatment, presenting symptoms and signs, laboratory data, clinical course and concomitant diseases. Fever was defined as rectal temperature >38°C. For the study analysis, we considered all Staphylococcal species to be contaminants since these are common in superficial skin cultures and thus difficult to distinguish between a real cause of infection and a contamination of the bacterial culture.
Statistical significance for the differences in continuous variables was tested by the non-parametric Mann–Whitney U test. Comparison of categorical variables was performed by the χ2-square test or the Fisher exact test for small samples. A p value of <0.05 was considered significant. For the comparison of patients with positive and negative MEF cultures, we included only patients in whom tympanocenthesis was performed on their admission day. To evaluate factors/co-morbidities to disease in older age (30 to 60 days after birth) and to evaluate risk factors for bacterial pathogen, we performed multivariate analysis by a logistic regression model, which included all factors found by univariate analysis to have p value below 0.2. In the age analysis, we have also included the specific bacterial pathogens as well. SPSS software (release 12.0.2) was used for statistical analysis. The study was approved by the institutional ethics review board of Bikur Cholim Hospital, Jerusalem.
A total of 277 infants aged less than 61 days, (median 40 range 5 to 61) were enrolled during the 9-year study period. Male:female ratio was 1.45. The patient’s characteristics are summarised in table 1. Almost 66% of the patients were admitted in the winter months between November and March, peaking with 21.2% of infants in January.
Presenting symptoms and signs
The most common presenting symptoms were respiratory complaints (70.0%) and fever (62.4%), followed by decreased appetite, restlessness, vomiting, and lethargy (table 2). Most patients were referred and hospitalised for neonatal sepsis work-up. In 29 of 173 febrile patients (16.8%), the fever lasted for more than 1 day. Interestingly, in 41 of 104 (39.4%) infants without a history of fever, a temperature higher than 38.0°C or higher was documented on admission. Thus, a total of 214 infants (77.3%) of the study group presented with fever. Within 24 h of admission, 184 patients (66.4%) had a temperature higher than 38.0°C, of them 110 had a temperature >38.5°C. White blood cell (WBC) counts were below 5000 cells/μl in five patients, and above 15 000 cells/μl in 58 (20.9%). Elevated C-reactive protein (CRP) (>5 mg/ml) or erythrocyte sedimentation rate (ESR) (>30 mm Hg/h) were present in 52 of 85 (61.2%) and 81 of 129 (62.8%) tested, respectively.
Tympanocentesis and bacterial pathogens in MEF
A tympanocentesis was performed in 259 (93.5%) patients. In the remaining patients, the parents refused the procedure. In 169 infants, the tympanocentesis was performed prior to initiation of antibiotic treatment, and in 54, after up to 5 days of antibiotic treatment in hospital. The drained MEF was described by the otolaryngologist as purulent in 231 patients and mucoid in the rest.
A bacterial pathogen (table 3) was found in aspirated MEF of 119 patients (46.3% of total 259 tympanocenteses). The growth rate, as expected, was significantly lower in patients who had received antibiotics in hospital prior to tympanocentesis (20.4% versus 52.7%, p = 0.0001). The characteristics of infants with and without bacterial pathogens in MEF culture are shown in tables 1 and 2. The most common pathogens were Streptococcus pneumoniae and Haemophilus influenzae (table 3). In five patients, both these bacteria were isolated. S pyogenes grew in nine (3.5%) patients. Enteric Gram-negative bacteria were found in 27 (10.5%) patients, the most common were Klebsiella pneumoniae and Escherichia Coli. In the remaining patients, the effusion was either sterile or grew Staphylococcal species. Interestingly, patients with S pyogenes were younger (28 (SD 9) days) than patients with other pathogens (39 (SD 14) days, p = 0.011). No such trend was observed with the other bacteria. It should be noted that patients who had positive cultures had higher rate of leukocytosis (28.7% versis 13.4% p = 0.008). By multivariate analysis, only WBC counts above 15 000/µl were associated with isolation of bacterial pathogen (adjusted odds ratio 2.6 (CI 95% 1.27 to 5.3), p = 0.009). No other significant association was found in this cohort between either positive MEF culture or specific pathogen and the demographic characteristics, clinical presentation and laboratory data (table 2).
AOM in the first and the second month of life
Eighty-nine infants were younger than 31 days, and 188 between 31 and 61 days of age (table 4). The demographic characteristics (perinatal morbidity, prematurity rate, nutrition) were similar in both groups, except for the male predominance (p = 0.027) and substantial proportion of birth weight below 2.5 kg in the older infants (p = 0.044), also younger infants had more siblings (p = 0.045). As expected, antibiotic treatment prior to admission was more common in the older infants (13.8% versus 6.7%, p = 0.05). In this older subgroup, respiratory symptoms and concomitant bronchiolitis were more common. Lumbar puncture was performed more frequently in the younger infants (p<0.001). The prevalence of the various bacterial pathogens, including the enteric Gram-negative bacteria, was similar in both groups except for S pyogenes, which was more prevalent in the younger infants (p = 0.033). Multivariate analysis comparing the older age group with neonates younger than 1 month revealed that AOM in the older infants was associated with male gender (AOR 2.05, CI 95% 1.19 to 3.53), bronchiolitis (AOR 2.62, CI 95% 1.50 to4.56) and diarrhea (AOR 3.6, CI 95% 1.27 to 10.21), (model sensitivity is 92.2 and specificity 71.2).
Diagnosis of otitis by the paediatrician
AOM (unilateral or bilateral) was diagnosed by the paediatrician in only 126 of the 277 (45.5%) patients who were subsequently diagnosed with AOM by the otolaryngologist. In 132 patients, the paediatrician either reported technical difficulties (narrow canal or obstructing cerumen, 93 patients, 33.6%), or incorrectly described normal tympanic membranes (39 patients, 14.1%). In 142 infants, bilateral AOM was diagnosed by the otolaryngologist. The paediatrician found acute otitis (bilateral in 52, unilateral 8) in only 60 of these infants (42.2%), and failed to diagnose in the remaining 82 patients (57.8%). In 19 cases (6.9%) the paediatrician’s ear examination was not documented.
Concomitant infections were diagnosed in 151 patients (54.5%). Bronchiolitis was by far the most common (127), followed by conjunctivitis (29), urinary tract infection (UTI; 13), aseptic meningitis (9), upper respiratory tract infection and pneumonia (7 each), acute gastroenteritis (6), bacteraemia (2- S pneumoniae and E coli). The same pathogen grew in MEF and another fluid in three infants (pneumococcal bacteremia and UTI owing to E coli and Klebsiella). Certain MEF pathogens were associated with additional infections: Haemophilus influenzae was associated with the presence of conjunctivitis (11/43 p<0.001), Pseudomonas aeruginosa with the presence of acute gastroenteritis (2/6 p<0.001).
According to the recent literature the predominant pathogens of AOM in children are S. pneumoniae (up to 40%), non typable H. influenzae (25–30%) and Moraxella catarrhalis (10–20%). Gram-negative bacteria are very rare.14 17 Most of the current knowledge concerning the pathogens of ear infection in very young infants is based on studies performed between 1970 and 1989.5 7 8 10 11 Our observation of significant rate of enteric Gram-negative bacteria (10.5% of the patients) is in concordance with the reported rate of 7.1–18.9%.5 7 8 10 11 This observation is not surprising, as these bacteria are frequent pathogens in neonatal infection. This emphasizes the importance of early tympanocentesis in young infants with AOM especially owing to the drug resistance increase reported for these pathogens in the community.18–22 In another study from Israel, Turner et al, who found 8% enterobacteria in MEF of 137 infants younger than 2 months, doubted the birth canal origin of these organisms.9 Indeed, our finding of equal incidence of enteric bacteria in up to and above 1 month of age, supports this doubt, as a higher incidence in the first month would have been expected if related to the bacterial flora of the birth canal.
As in previous reports, we have also observed both the predominance of S pneumoniae and H influenzae, and the rarity of Moraxella as causative pathogens of AOM in the first 2 months of life.6–11 It is of note that pneumococcal vaccine has not been included in routine immunisation programme in Israel. The rate of S. pyogenes in our patients (3.5%) was similar to that reported in older children23 24 and in young infants.6 9 This pathogen was more common among infants aged less than 1 month. That is in contrast to the notion that in older children the proportion of AOM caused by S pyogenes increases with age.24
The recent clinical practice guideline on the management of children aged from 2 months through 12 years with AOM recommends antibiotic treatment to all infants younger than 6 months and optional withholding of antibiotics in children older than 2 years of age.2 However, there are no specific guidelines for infants younger than 2 months, in whom factors predicting bacterial growth in MEF have not been addressed. In our study of 277 infants aged less than 2 months, only a high WBC count was found to be associated with bacterial growth in MEF. However, this was found in only 28.7% of patients with positive cultures leaving the rest without specific predictors. In the absence of such predictors it would seem appropriate that all infants aged under 2 months with AOM, including afebrile infants and those with normal laboratory profile, should receive antibiotic treatment, as recommended by the current clinical guidelines for young infants.
Many infants were hospitalised during the winter months. AOM is known to be a seasonal disease in older children, occurring predominantly in winter.17 This pattern is consistent with the appearance of viruses such as respiratory syncytial and influenza viruses. Indeed, 70% of our patients with AOM had respiratory symptoms and 45.8% were diagnosed with bronchiolitis. The association of acute bronchiolitis with AOM is known,25 26 and in a previous study on bronchiolitis, we found concomitant AOM in 53% of the patients.27 In these patients with bronchiolitis no difference in the clinical and laboratory findings between children with and without AOM was demonstrated,27 in concordance with our present findings in infants aged under 2 months. We also found that concomitant conjunctivitis was associated with H. influenzae growth in MEF. This association has been termed “otitis-conjunctivitis syndrome” in older children.28 Thus, the presence of concomitant conjunctivitis in this age group should alert the paediatrician of the possible H influenzae infection.
In our study the paediatricians failed to accurately diagnose AOM in more than half of the patients. This observation emphasizes the diagnostic difficulties of AOM, which is often missed by paediatricians in very young infants. Several studies have shown that even in older children physicians are uncertain of their diagnosis of AOM in as many as 40%.29–31 In recent studies, which have mainly focused on patients with otitis media with effusion, the diagnostic skills of both paediatricians and paediatric residents were evaluated using video-recorded otoendoscopic examinations of tympanic membrane, representing otitis and normal tympanic membranes.30 31 This mode of examination bypasses the difficulties caused by a narrow or obstructed ear canal, nevertheless correct diagnosis was obtained in only 36–54% patients.
To the best of our knowledge the present study is the first to evaluate the paediatricians’ clinical diagnostic accuracy of AOM in very young infants. The symptoms of otitis in this very young age are nonspecific, since symptoms such as earache or ear pulling are not relevant in this age group. Furthermore, it is often difficult to see the tympanic membrane due to the narrow external auditory canal, which is often occluded by cerumen. In addition, the normal tympanic membrane of the very young infant is unique in its thickness, opacity and the often absence of transparency.16 It lies in an extremely oblique position, making it difficult to distinguish from the canal wall.8 The microscopic otoscope increases the diagnostic accuracy by improved visualization of the tympanic membrane. Pneumatic otoscopy was not used in our study, and we doubt that it may further improve the diagnostic accuracy.
Although the majority of pathogens are similar to those reported in older children, a significant rate of Gram-negative bacteria may justify the practice of tympanocentesis in very young hospitalised infants and should be taken into consideration when choosing antibiotic treatment. The absence of specific predictors for bacterial infection in the majority of neonates suggests that the initiation of empiric antibiotic treatment in young infants with AOM may be appropriate even when afebrile and with normal laboratory profile. Finally, the low diagnostic rate of AOM by the paediatricians emphasizes the need for improvement of their examination skills and instrumentation for a thorough ear evaluation.
What is already known on this topic
The incidence of AOM is highest in the first 2 years of life.
There are a few data on clinical and laboratory features of AOM in infants under 2 months of age, who were excluded from most studies addressing AOM.
There are no data about the accuracy of AOM diagnosis in this age group and the paediatrician often perceives that the diagnosis of AOM is based on guesswork rather than accurate clinical evidence.
What this study adds
Presenting symptoms of very young infants with AOM were mainly respiratory and fever.
The most common pathogens were Streptococcus pneumoniae and Haemophilus influenzae. Gram-negative bacilli grew in 10.5% of the infants.
Bacterial pathogen growth in the middle ear fluid was not associated with any clinical or laboratory feature except for elevated WBC count.
Paediatricians diagnosed AOM in less than half of the patients subsequently diagnosed with AOM by an otolaryngologist.
Competing interests: None.
Ethics approval: Due to the retrospective character of the study, it was approved by the institutional ethics review board of Bikur Cholim Hospital in Jerusalem without need for signed patient consent.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.