Accumulation of factors influencing children's middle ear disease: risk factor modelling on a large population cohort

Data were analysed from a large national birth cohort to examine cumulative and interactive prediction from various risk factors for childhood middle ear disease, and to resolve conflicting evidence arising from small and incompletely controlled studies. The large sample size permitted appropriate covariate adjustment to give generality, and permit demographic breakdown of the risk factors.

A large multi-purpose longitudinal birth cohort study of all births in the UK in one week in 1970, with multiple questionnaire sweeps.

Over 13,000 children were entered into the original cohort. Data on over 12,000 children were available at the five year follow up.

For children at 5 years, parent reported data were available on health and social factors including data on two markers for middle ear disease: the occurrence of purulent (nonwax) ear discharge and suspected or confirmed hearing difficulty.

In those children who had ever had reported hearing difficulty (suspected or confirmed), after control for socioeconomic status, three of the classic factors (male sex, mother's smoking habits since birth, and attending day care) were significantly more frequent. In those who had ever had ear discharge reported, only mother's smoking habit since birth was significantly more frequent. However, it showed an orderly dose response relation. In addition, a derived general child health score was found to be significantly associated with both the middle ear disease markers. Control for this variable in the analysis of those having reported hearing difficulty reduced the effect size of mother's smoking habit, but it remained statistically significant. For reported ear discharge, even after control for the general health score and social index, mother's smoking habits and day care attendance were both significant predictors. Mother's (but not father's) smoking habits and day care attendance were found to be significant risk factors for middle ear disease. Breast feeding effects were weak and did not generally survive statistical control.

A child having all three risk factors (attends day care, a mother who smokes, and male sex) is 3.4 times more likely to have problems with hearing than a child who has none, based on cumulative risk. Further studies should focus on preventative risk modification and well specified intervention.