Learning Objectives: To better understand the relationship of squamous forms of otitis media to mucosal disease, based upon a systematic review of longitudinal epidemiological studies. To use these data to create a map of disease relationships.

Introduction: Clinical experience suggests that cholesteatoma often arises in individuals with a history of prior mucosal otitis media, or a history of tympanic retraction. I set out to exploit existing longitudinal studies to ascertain the relation of these disease entities, specifically to assess the relation of mucosal disease to tympanic retraction and cholesteatoma.

Methods: I searched the pubmed database using terms “tympanic retraction” OR “cholesteatoma” AND “epidemiology”. 708 articles were returned. Titles and subtracts were screened for relevance. Only longitudinal prospective or retrospective studies were included. Articles on syndromic or special populations were excluded. 24 articles were included for review, and the full text of these articles was evaluated to identify further references.

Results: Differing populations and definitions make meta-analysis inappropriate. There is no evidence of association of acute otitis media with risk of squamous disease. Presence and duration of chronic otitis media with effusion is associated with risk of subsequent squamous disease. Tympanic membrane retraction shows variable chronology, with many retractions resolving, and development of new retractions rare. Cholesteatoma remains a rare complication, and is predisposed to by TM retraction, but almost certainly also arises de novo, perhaps in those with subclinical disease. There is no evidence that grommet insertion reduces risk. These relationships can be constructed into a map of the inter-relation of disease, akin to the landscape map for mucosal otitis media I have previously constructed (Audiol Neurotol 2014;19:210–223).

Conclusions: Existing epidemiological studies can be used to construct a map of the relation of mucosal to squamous forms of otitis media, and so help to better understand epidemiological correlates, and to hypothesise pathophysiological relations.