Does Tympanometry Predict Antibiotic Usage in Acute Otitis Media?

INTRODUCTION Acute Otitis Media (AOM) is the most common childhood infection for which antibiotics are prescribed.{1-2} Several important factors have altered the prescription practices for AOM in the past decade. North America has seen an epidemic of microbial resistance, attributable in a large part to the over-prescription of antibiotics.{3-4} Evidence has shown that AOM is commonly over diagnosed and antibiotics are prescribed unnecessarily.{5-7} In addition a number of studies comparing antibiotic treatment to placebo, although proving superiority for antibiotic treatment, had very high rates of resolution with placebo. A meta analysis of 7 RCT's show a 73% success rate for Amoxicillin versus 60% for placebo.{8}

These results prompted interest in a watch and wait strategy for the treatment of AOM for children 6 months to 10 years of age. Three important studies were performed assessing the outcomes of delaying antibiotics for 48-72H, two in primary care{9-10} and one in a pediatric emergency department.{11} All three showed no difference in outcomes with 2-4 week follow-up. There was a significant reduction in antibiotic usage with the watch and wait approach, a reduction of about 70%. These studies led both the American Academy of Pediatrics and the Canadian Pediatric Society to recommend adopting a conservative non-antibiotic treatment strategy for AOM in the first 48-72H.{12} However in early 2011, 2 papers were published that suggested that antibiotic use in AOM was beneficial.{13, 14} What set both of these two papers apart from other papers was the definition of acute otitis media: in one study the children were enrolled by study clinicians who were otoscopists who had successfully completed an otoscopic validation program, in the second study middle ear fluid had to be present by means of pneumatic otoscopy. The question then arises, is the treatment of otitis media really a question that concerns the treatment, or is it a question of diagnosis? The published correspondence that these two papers generated, suggested that diagnosis of AOM is indeed a concern. A technique named tympanometry could prove to be useful.

Tympanometry is a well established technique for documentation of middle ear effusion. Tympanometry measures the compliance of the tympanic membrane over a range of pressures. A graphic curve is generated which can be compared to normals or examples of pathologic conditions. Primary care physicians have used this technique successfully among children with and without a diagnosis of AOM.{15} Two studies involving tympanometry and otitis media{16-17} have shown tympanograms consistent with clear middle ear effusion(Type B) in 50-60% of patients clinically diagnosed with AOM. This strongly suggests that AOM is being over diagnosed! Interestingly Spiro et al showed that rates of antibiotic prescription were not different in physicians who were aware of tympanogram results and those who were blinded. This indicates a lack of knowledge around the impact on outcome of the initial tympanogram.

Currently at the Stollery, (and as suggested by CPS and AMA) well children who are diagnosed with AOM are given a prescription for antibiotics, but told to wait 48 hours before filling them, in the expectation that many of these children will have a complete resolution of their symptoms. They are encouraged to take analgesics to keep them comfortable. However, the current state of knowledge indicates that antibiotics provide superior symptom control in the first 24-48H and provide superior resolution rates to placebo in clinically diagnosed AOM. The delayed antibiotic strategy has similar long term outcomes to immediate antibiotic treatment. It is not clear if patients with middle ear effusion documented by adjunctive tests, such as tympanometry represent a distinctly different group than those diagnosed by observation of the tympanic membrane only. Spiro's study indicates that physicians do not know and therefore are not changing their practice on the basis of tympanogram results.

This then brings us to the investigators' primary research question. Are those children aged 6 months to 16 years who attend the Stollery Emergency Department and are diagnosed with AOM (and sent home with conservative management) more likely to fill a prescription for antibiotics over the next 7 days if their tympanogram is type B rather than types A or C?

The investigators propose an observational study, within a watch and wait strategy, to determine if patients clinically diagnosed with AOM and having a type B tympanogram have higher rates of eventual antibiotic usage. Secondary outcome measures would include symptoms of pain, fever and impaired sleep than those without clear evidence of middle ear effusion.

If those patients with a B type tympanogram have a high rate of eventual antibiotic usage, delaying their treatment is not significantly reducing our overall antibiotic usage. Furthermore, the investigators could reduce the pain suffered by these patients in those first 48-72 hours. The corollary to this may be that those patients with other tympanogram patterns may have comparatively low need for antibiotics. This study would help to more accurately identify those patients likely to derive significant benefit from antibiotic treatment for AOM.

PROJECT DESIGN AND WORK PLAN

Design: The study is a non-interventional analysis of outcome. This is an observational cohort study.

Intervention and sampling method: All patients diagnosed with OM will be approached for study enrolment within the hours which an RA is available. Tympanometry will be conducted on every participant in both ears. The clinician will not be privy to the result of the tympanogram. The tympanograms will be assigned a B or A/C shape, upon review by the principle investigators. In cases of disagreement a third reviewer will review the tympanogram. Patients will be followed with a phone call in 3 days and again at 8 days. They will be asked to rate their symptoms (Acute Otitis Media-Severity of Symptom Scale){18}, in addition to specific scripted interview questions for the study (under development) including a determination of antibiotic usage for the infection. The investigators will also review the provincial pharmacy database to confirm if antibiotics prescriptions were filled.

Training: Research assistants will receive one-on-one training with the principal investigators on tympanography technique.

Recruitment: Updates regarding the study will occur at the monthly staff meetings, posters describing the study will also be placed throughout the ED. Research Assistants will be visible in the department and will remind staff of the study.

Primary Outcome: Are those children aged 6 months to 16 years who attend the Stollery Emergency Department and are diagnosed with AOM (and sent home with conservative management) more likely to fill a prescription for antibiotics over the next 7 days if their tympanogram is type B rather than types A or C? Secondary Outcomes: Does a type B tympanometry curve predict increases in (1)Reported Pain (2)Analgesic use (3)Fever (4)Sleep Disturbance (5)Oral intake versus type A/C tympanograms (6) Proportion of patients diagnosed with OM who have either type A or type A and C tympanograms.

Sample Size: A logistic regression will be performed using antibiotic use in 7 days as the dependent variable. Given 4 independent variables, the investigators need at least 40 (80 would be ideal) subjects who fill the antibiotic prescription. Enrolling 137 patients would give us an alpha and Beta each of 0.05, and would allow us to do Chi Squared test looking at a two by two table of Antibiotic (Yes or No) vs Tympanometry Graph (Type B vs other).

Analysis: Antibiotic use in the 7 days following enrolment will be analyzed by logistic regression. Independent variables will include age, Tympanogram type A/C, Viral upper respiratory symptoms for ≥ 5days prior to presentation to the ED, and pain requiring analgesia in the Emergency or at home within 6H of presentation. Quantitative and qualitative data will be described. Antibiotic drug data will be checked in NetCare to see if there is a discrepancy between verbal reports and NetCare.

Feasibility: Data from the investigators' emergency department indicates over 900 children per year present with AOM, and over the past 6 months of winter (Oct-Mar) there have been about 540. Based on previous results {9-11} the investigators predict an antibiotic usage rate of 30%. To collect 40 patients who fill a prescription for antibiotics the investigators will need 135 patients with otitis. If antibiotic usage rates are higher this number will decrease. Accounting for about a 15% drop out rate the investigators will need to enroll about 155 children. Conservatively assuming recruitment of 30% of children with AOM the investigators anticipate 24 weeks to reach the investigators' sample size.