Is tympanocentesis safe and does it improve outcomes following episodes of acute otitis media? Undoubtedly, we need to improve diagnostic strategies for acute otitis media. Several strategies recommended by Pichichero in an article in this issue of American Family Physician¹ are valuable, such as assuring adequate illumination by replacing old otoscope bulbs, removing cerumen and adding a soft rubber sleeve (e.g., Welch Allyn SoftSpec) to aid in conducting pneumatic otoscopy. For many children, the examination reveals the presence of a bulging red or cloudy eardrum with impaired movement and, in 5 to 10 percent of these children, the examination yields a perforated tympanic membrane with draining pus.² In such examinations, the physician can be confident of the diagnosis and proceed with the preferred treatment strategy.³ However, this leaves a group of children in whom the diagnosis remains uncertain.

In our international study of acute otitis media,² family physicians were uncertain of their diagnosis in 42 percent of children less than one year of age, in 34 percent of children 13 to 30 months of age, and in 27 percent of children older than 30 months. Pichichero¹ recommends increasingly aggressive strategies to deal with these children: tympanometry or acoustic reflectometry and, if the physician is still uncertain of the diagnosis, tympanocentesis.

However, in considering these options, we must keep in mind a basic principle of medicine—first do no harm, a mandate that has been extended by the concept of Patient Oriented Evidence That Matters (POEMs).^4,5 The concept of POEMs suggests two important principles. First, treat the patient, not the chart (or the petri dish!)—be sure that the outcome of interest is important to the patient. Second, rather than opinion, seek evidence as a basis for evaluating alternatives before changing practice. In the child with a painful ear and possible acute otitis media, the outcomes of interest include short-term symptomatic improvement, avoidance of complications and prevention of recurrence.

In addition to tympanocentesis, the family physician has another important diagnostic strategy available: time! Simply maintaining contact with the family and treating the child for pain (with acetaminophen or ibuprofen) will lead to marked improvement in 65 to 80 percent of children within 24 to 48 hours and help clarify the diagnosis in others.^6,7 A watchful waiting approach is safe and will help avoid development of antibiotic resistance in those who improve on their own.^8,9 Pneumococcal infections are the infections least likely to remit without the use of antibiotics; therefore, if the new pneumococcal vaccine proves as useful as early data suggest, we may see a shift in the microbial spectrum with a concomitant increase in rates of early resolution of acute otitis media without the use of antibiotics.

I am not convinced that the data presented in Pichichero's article¹ indicate that routine tympanocentesis for infants and children in whom the diagnosis is uncertain would lead to better patient outcomes. First, as therapy, tympanocentesis provides no improvement either alone or in combination with an antibiotic in the treatment of acute otitis media.⁷ Second, I am concerned that even a low rate of adverse consequences—damage to the bony structure of the middle ear, laceration of the ear canal, persistent perforation, cholesteatoma—may outweigh the modest anticipated benefit or early symptom improvement from confirming the diagnosis and starting treatment (one to two days earlier in 10 to 20 percent of children).^6,7

The data on which Pichichero¹ bases his assessment of the safety of tympanocentesis were obtained from one physician who conducts 30 to 100 procedures a year, a rate unlikely to be generalized to most family physicians. The squirming, crying infant or toddler, in whom I most often have difficulty making a diagnosis with certainty, is also the one for whom I would be most concerned with performing tympanocentesis.

We also need to consider the consequences of a failed tympanocentesis attempt. It has been suggested that the next step would be an attempt following mild sedation.¹ In my practice setting, the time and effort required to obtain consent, educate the parent regarding care of the sedated child, administer and wait for the sedation to take effect, and help parents manage the occasional side effects of sedation all suggest that this would be a practice unsuitable for routine adoption, even if it were to lead to improved patient outcome. This leads to the last option: referral to an otolaryngologist. However, such referral usually requires a delay, and activates by default the watchful waiting approach, or, given the heightened parental concern invoked by the failed tympanocentesis, the presumptive use of an antibiotic.

We also have no evidence that the Improving Outcomes in Acute Otitis Media workshop (mentioned in the Pichichero article),¹ at which middle ear mannequins are used to teach the procedure, will result in the skills required for family physicians to perform tympanocentesis with the support available in most offices. Nor is there evidence that these skills learned at the procedures workshop would be maintained, given the few times a year most family physicians would use them.

Until more convincing data regarding the benefits and adverse effects of diagnostic tympanocentesis are available, it is prudent to use alternative diagnostic strategies, including those recommended by Pichichero.¹ Tympanocentesis is best reserved for the child with persistent acute otitis media unresponsive to multiple courses of antibiotics or for the child with a complex history, anatomic or immunologic abnormality, or other complication. In such situations, referral to an otolaryngologist might be the best option.

For the child with a painful ear or abnormal otoscopic findings, an uncertain diagnosis is best managed by treating symptoms. Seventy to 90 percent of painful ear episodes develop during a viral upper respiratory infection.¹⁰ A bulging or full tympanic membrane is a marker of an obstructed eustachian tube. Children whose tubes are not obstructed or intermittently obstructed from the upper respiratory infection often have normal position or retracted eardrums (see Table 2 in the Pichichero article¹ ). Depending on other findings, these patients might have acute otitis media 30 to 90 percent of the time,³ and are the patients in whom we have the most diagnostic uncertainty, even though they are also the ones whose symptoms might clear most quickly. Therefore, in children whose ear findings leave the physician uncertain, treatment of symptoms and reevaluation if symptoms persist are, in my opinion, the best options. An invasive procedure, even if usually benign, is unjustified when more than 80 percent of patients will improve rapidly without intervention,⁷ and the remainder are not placed at increased risk by delaying antibiotics.