Abstract
Acute otitis media (AOM) is over-diagnosed. Symptoms predicting AOM are neither sensitive nor specific. Undue reliance on tympanic membrane (TM) redness and failure to assess TM mobility with pneumatic otoscopy contribute to inaccurate assessment of physical findings. The features distinguishing AOM (for which antibiotics are appropriate) and otitis media with effusion(OME; for which antibiotics are infrequently indicated) are often not accurately appreciated. Tympanometry and acoustic reflectometry can be useful adjunctive tools to aid diagnosis and follow-up. Selective use of tympanocentesis may improve diagnostic accuracy because it validates or refutes the physicians' clinical impression.

Most family physicians are confident that they can make the diagnosis of acute otitis media (AOM) with relative ease in the pediatric patient. Yet, in one study, 75 children judged by family physicians to have AOM were sent to otolaryngologists for tympanocentesis. Only 11 (15%) of these children actually had AOM according to the specialists based on examinations with otomicroscopy, of whom 7 (9% of the total) actually had a middle ear pathogen isolated by tympanocentesis (1). Consider the consequences of over-diagnosis - a prescription of antibiotics almost always follows. If the patient did not actually have AOM then the patient is placed at risk for the adverse consequences of antibiotic therapy unnecessarily. If the patient is declared to not have responded satisfactorily, use of broader spectrum antibiotics is advocated at a follow-up examination. Repeated episodes of AOM and particularly repeated episodes of AOM which appear to have not responded to antibiotics often leads to referral for tympanostomy tube surgery, sometimes with concurrent adenoidectomy.

Over-diagnosis of AOM also has an impact in clinical research trials because it produces an over estimation of efficacy for all treatments and a blurring of any actual differences in efficacy between treatments. We lack standardized and stringent criteria for the clinical diagnosis of AOM and for the characterization of improvement or cure. In a review of the literature, Hayden found 19 different sets of criteria for making the diagnosis of AOM were used in 26 different clinical trials (2). Some proportion of children did not actually have AOM but either had otitis media with effusion (OME), or no disease at all. After enrollment in an antibiotic study, these children probably would be classified as cured or improved even though they did not have AOM in the first place.

The classic findings of AOM, such as fever and earache are sometimes absent even in cases confirmed by tympanocentesis. A bulging, red, immobile tympanic membrane is highly associated with AOM. However, many physicians rely on redness of the ear drum as their main diagnostic clue. Crying (and most children cry when their ears are examined), removal of cerumen with associated irritation of the auditory canal, and fever can cause redness of the ear drum in the absence of any middle ear infection whatsoever. Most of all, when a parent brings a child to the physician with a report of sleeplessness, irritability, rhinorrhea and perhaps fever, the temptation is great to see at least a little bit of redness or fluid behind the ear drum as justification for an antibiotic prescription (3).

Definition
Distinguishing episodes of otitis media as acute suppurative otitis media (AOM) versus otitis media with effusion (OME) is critical (4). The distinction between these entities should be clear. AOM is defined by the presence of symptoms of acute illness and a tympanic membrane (TM) under positive pressure (full or bulging). OME is defined by the absence of symptoms and signs of acute infection (other than reduced hearing) and a TM under negative pressure (retracted) or no pressure (neutral position) with the presence of fluid in the middle ear space. This distinction is essential for management decision making. Antibiotics are indicated for AOM but may be appropriately deferred for children with OME, in agreement with recommendations by the US Agency for Healthcare Policy and Research (4).

Interpreting Symptoms and Signs of AOM
Niemela et al (5) carried out a prospective survey collecting information on 354 consecutive children visiting a pediatrician, otolaryngologist or family physician because of any kind of acute respiratory illness to compare their symptoms with children who had AOM. The symptoms and signs observed at home were recorded by the parents before the visit and the findings upon physical examination were recorded later by the physician. AOM was diagnosed in 191 patients (54%). The most important symptoms increasing the likelihood of AOM were ear-related symptoms, such as earache, rubbing of the ear or report of the ear feeling blocked. However, only two-thirds of the children younger than two years of age with AOM had any ear-related symptoms. Fever, earache or crying/irritability were present in 90% of patients with AOM but also in 72% of patients without it. Many of the symptoms thought to be related to AOM, such as fever, cough, loss of appetite, diarrhea or vomiting were not more common in children with AOM than in children with other acute illnesses. Further, the duration of the symptoms was not markedly different in children with AOM in comparison to children with other diagnoses. Heikkinen and Ruuskanen (6) prospectively examined of 302 children younger than four years of age attending day care centers during episodes of upper respiratory infections. Forty percent of the children with AOM never complained or had symptoms of an earache, 31% had no fever, and half slept soundly. Kontiokari et al (7) prospectively examined 857 healthy day care children during upper respiratory infections. The parents were able to predict the presence of AOM in their own child with a sensitivity and specificity of 71% and 80%, respectively (positive predictive value 51%, negative predictive value 90%).

In a study of 85 infants and children, Schwartz et al (8) made a diagnosis of AOM based on pneumatic otoscopy findings of diminished mobility of the TM, bulging contour and color. The diagnosis was validated by tympanocentesis and positive culture of middle ear fluid. A poorly mobile, bulging, yellow, and opacified TM were the most predictive of finding AOM; unfortunately, they were all seen in only 19% of the studied patients. The color of the TM did not correlate at all with pathogen isolation from the middle ear exudate. Seventy-seven percent of children with AOM had no fever and 28% had no ear pain or irritable behavior.

Karma et al (9) used a multi-variable analysis of TM findings to find the strongest associations with the presence of AOM. Their report was tabulated by Pelton (10). Combining the TM position (normal, retracted or bulging), color (normal, red or opaque = "cloudy") and mobility (normal or impaired = reduced) could result in 99% predictive value (Table 1) when compared with the findings at tympanocentesis. The presence of a normal or retracted position of the TM with slightly impaired or normal mobility was unlikely to be associated with AOM unless the fluid behind the membrane was cloudy. A slightly red TM with normal TM position and mobility had a predictive value of 7% for AOM. A distinctly red TM in normal position and normal mobility had a 15% predictive value.

In 1992, Kaleida and Stool (11) described an otoscope validation program to assess accuracy regarding the diagnosis of middle ear effusion. Children were examined by the investigators concurrently with the physician in training. The children had a history of recurrent or chronic otitis media or both and had been scheduled for elective myringotomy (with or without tympanostomy tube placement). The children underwent general anesthesia after the examinations and a tympanocentesis was performed. The otoscopists examination was therefore validated by the middle ear findings. Using this approach, the investigators found that physicians became more accurate in diagnosing the presence than the absence of middle ear effusion. The mean sensitivity and mean specificity for the group as a whole were 87% and 74%, respectively. The chief benefit of the program was an improvement in diagnostic skills; the chief limitation was the large amount of time required.

Finitzo et al (12) compared tympanometry and pneumatic otoscopy with findings of myringotomy in 86 children (163 ears). Seventy percent of the ears had an effusion as revealed by myringotomy. Abnormal tympanograms consistent with middle ear effusion were flat. The sensitivity and specificity for tympanometry were 90% and 86%, respectively; for pneumatic otoscopy, it was 93% and 58%, respectively. In 40% of the examined children, family physicians over-diagnosed the presence of middle ear effusion. Barnett et al (13) compared acoustic reflectometry and other diagnostic techniques for detection of middle ear effusion in children with middle ear disease in their study of 299 ears. When the spectral gradient angle of the acoustic reflectometry instrument was < 49, 88% of children were found to have middle ear effusion with tympanocentesis. In contrast, when the spectral gradient angle was > 95, 17% of children had middle ear effusion. Sensitivity, specificity, positive predictive value and negative predictive value of acoustic reflectometry and tympanometry compared favorably in predicting the presence or absence of middle ear effusion at surgery. Lampe and Schwartz (14) summarized the value of acoustic reflectometry in children with AOM. Six studies evaluating 12,333 ears were described. The gold standard of tympanocentesis was available for comparison in several studies whereas in others, pneumatic otoscopy findings represented the gold standard. Acoustic reflectometer sensitivity varied from 54% to 94% and specificity from 59% to 83%.

How can we improve diagnostic accuracy in AOM?
Adequate illumination is critical. Barriga et al (15) examined otoscopes located at the private offices of 96 physicians. The light output of each unit was measured and then remeasured with a new bulb, and when possible, a new battery was placed in the unit. Approximately 1/3 of the otoscopes had a sub-optimal output of light. Replacement of the bulb restored adequate illumination for 80% of these otoscopes. One-third of the physicians reported changing otoscope bulbs less often than every two years (as is recommended) and almost half of the rechargeable batteries inspected were outdated.

The TM cannot be adequately seen while partially occluded by cerumen. (8, 11, 16) Some type of cerumen removal curette is essential and should be regularly used. If the wax is dry or deep in the auditory canal then cerumenolytics and/or warm water irrigation may be necessary. Parents should be taught how to restrain an uncooperative child for wax removal and proper examination of the TM. Disposable specula are too short to allow optimal TM visualization; the widest caliber,
re-usable specula should be used to maximize illumination of the TM.

Pneumatic otoscopy has been advocated as an important adjunct to assist in diagnostic accuracy of AOM (Table 1) (3, 8-11, 16-19) yet most physicians find it inconvenient or remain unconvinced. The ear speculum must create an air seal against the external auditory canal to successfully cause movement of the TM; a rubber sleeve over the speculum may reduce the discomfort of the process. Four characteristics of the TM should be evaluated and described in every examination (position, mobility, color, degree of translucency). The normal TM is in the neutral position (neither retracted nor bulging), pearly gray, translucent and responds briskly to positive and negative pressure, indicating an air filled space. The abnormal TM may be retracted or bulging, and immobile or poorly mobile to positive and/or negative air pressure. The color of the eardrum is of minor importance although patients with AOM more often have a red TM. The key differentiating features of AOM compared to OME on physical exam relate to TM position. In AOM the TM almost always is bulging and in OME it is usually retracted or, occasionally, it is in neutral position. The TM is thickened in both AOM and OME, thereby reducing visibility through it. Sometimes a yellow or grayish middle ear effusion can be seen behind the TM in either condition.

Tympanometry and acoustic reflectometry each have attributes which make them of value in providing information about the possible presence of a middle ear effusion (13, 14, 20). The sensitivity, specificity, positive predictive value and negative predictive value of the two instruments has been assessed in comparison with pneumatic otoscopy, audiometry and tympanocentesis findings (13). A comparison of the two instruments in office practice has also been made (21); neither is a perfect instrument and both have some limitations. Both instruments have portable models which allow them to be carried from one examination room to another. Acoustic reflectometry has the advantage of not requiring a seal within the canal which improves it's usefulness in the crying child because a reading can be obtained when a child stops crying to take a breath. Tympanometry provides additional information about actual pressures within the middle ear space (20). Both instruments can be connected to printers, thereby providing a permanent record of the readings for comparison at subsequent examinations and for documentation to health insurers.

Family physicians who see young children with AOM may be called upon to perform tympanocentesis in selected patients in their office as recommended by the CDC (22). The gold standard in clinical trials for the diagnosis of AOM has always been tympanocentesis determination for the presence of middle ear fluid with subsequent culture for identification of causative pathogens (17-19, 22, 23). Selective use of tympanocentesis may improve diagnostic accuracy because it validates or refutes the physicians' impression after visual examination. Few experiences are more self-educating than to diagnose AOM and then find no fluid with tympanocentesis. Why are so few family physicians utilizing this procedure in their office practices? Some fear complications and yet, after proper training, the procedure has a very satisfactory safety record (Table 2). Certainly proper restraint of the patient and excellent visualization of the TM are essential; mild sedation may also be helpful in some cases (24) (Table 3). Table 4 outlines when tympanocentesis should be performed and it's benefits. It would appear we need post-graduate training through continuing medical education courses to allow more physicians to learn and practice the use of tympanocentesis (22, 23) and this is now available (Outcomes Management Educational Workshops: 1-877-EAR-OMEW or www.omew.com)

Concluding Remarks
The average physician examines the ears of more than 30 patients each working day (60 ears), 5 days per week for about 40 weeks per year for a total of more than 14,000 ears per year. Nevertheless, published literature suggests that AOM is very frequently over-diagnosed. We need to make an effort to improve the accuracy of this diagnosis. The Centers for Disease Control has initiated a campaign to promote the judicious use of antibiotics, especially by differentiation of AOM from OME (Table 5)(25).

References