Bulging Tympanic Membrane — Acute Otitis Media

Test Characteristics

Metric	Value
False-negative rate	47% (sensitivity 53%)
False-positive rate	3% (specificity 97%)
Bayes factor (positive test)	20×
Bayes factor (negative test)	1/2×
Base rate	50% of children with ear pain and fever

Interpreting Results

Scenario	Prior	+ Result	− Result
Child with febrile URI, no ear-specific symptoms	10%	20 × 10% ≥ 100%10%× 20100%	1/2 × 10% = 5%10%÷ 25%
Child with ear pain and fever	50%	20 × 50% ≥ 100%50%× 20100%	1/2 × 50% = 25%50%÷ 225%

20 × 10% ≥ 100%10%× 20100%: exact posterior is 69%. 20 × 50% ≥ 100%50%× 20100%: exact posterior is 95%. Negative-test exact posterior at 50% prior is 33% (simple approximation gives 25%). Bulging is highly specific for middle ear effusion in acutely symptomatic children — when present, it strongly indicates AOM. The negative Bayes factor is weak (1/2×) because nearly half of true AOM cases lack frank bulging on examination; cloudiness and distinctly impaired mobility were each more sensitive than bulging in the Karma data. Distinguish AOM (bulging, opaque, often hyperaemic) from OME (retracted or neutral, air-fluid level, no bulging). Crying alone can redden a TM and produce a false impression of inflammation — judge color in a calm child.

• + result: at a coin-flip prior (child with ear pain and fever, 50%), risk pushes to 95% — strong confirmation
• − result: drops to 25% (exact 33%) — bulging misses about half of AOM, so its absence doesn't rule it out

Sources:

• Karma PH, Penttilä MA, Sipilä MM, Kataja MJ. Otoscopic diagnosis of middle ear effusion in acute and non-acute otitis media. I. The value of different otoscopic findings. Int J Pediatr Otorhinolaryngol. 1989;17(1):37–49. Pooled sens 53% and spec 97% computed from Tables II and V (acute visits, bulging) vs myringotomy across 5462 visits.
• Rothman R, Owens T, Simel DL. Does this child have acute otitis media? JAMA. 2003;290(12):1633–1640. Re-analysis of Karma data with verification-bias adjustment.

---