Outcome Measures
Two outcome measures were studied: the presence of bilateral OME in those children who still had it at the 3-month follow-up examination, and unilateral or bilateral OME in all children with OME, whether unilateral or bilateral, at the follow-up examination.
Definitions
OME was defined as the presence of fluid in the middle ear cavity behind an intact tympanic membrane without signs or symptoms of acute infection. The presence of middle ear effusion was defined by the C2-curve and the B-curve because of their high positive predictive values (88% and 54%, respectively).28,29 Recurrent URTI was defined as 6 or more episodes of a common cold or nasal catarrh in the past 12 months, each episode not exceeding 4 weeks. Recurrent AOM was defined as 3 or more episodes in the past 12 months. A child failed the Ewing test if he or she did not respond to the sound stimuli in any of the tests. Bottle feeding meant that no breastfeeding had been used at all.
Statistical Analysis
Bivariate analyses were carried out to test the association between determinants and outcome assessed at the 3-month follow-up visit. The strengths of the associations were expressed as odds ratios with 95% confidence interval limits.30 Associations were also tested by chi-square analysis. All factors showing a P value equal to or less than 0.2, and all factors that were clinically relevant were subjected to logistic regression analysis.31 The goal of the regression analysis was to find a set of determinants for persistent OME. From those factors that were significant we derived a prediction score by adding the betas (eg, prediction score=bfactor1+bfactor2+bfactor3).31 The discriminating ability of these predictions was assessed by means of receiver operating characteristic curve analysis. The size of the area under the curve (AUC) is a measure for the discriminating capability, the ability to separate patients into groups classified according to the determinants.32 A test that does not discriminate will have an AUC of 0.5; a perfect test will have an AUC of 1.0.
Results
A total of 593 children were suspected of having bilateral OME at the initial visit, but the tympanograms were normal for 82 (14%), and no classifiable tympanogram could be obtained for 11 (2%). Sixteen of the remaining 500 children were excluded because they had been given an antibiotic in the preceding 4 weeks, and the parents of 51 children refused consent. Thus a total of 433 children (73%) were included in our study. Eighty children (18%) had more than 3 entry criteria, and 231 (53%) had more than 2. The most common combinations of complaints were subjective hearing loss with recurrent URTI (19%) and subjective hearing loss with mouth breathing and snoring (14%).
Twelve children failed to attend the 3-month follow-up visit, and in 24 no classifiable tympanogram could be obtained ([Figure]). We excluded these 36 children from the 3-month follow-up as they did not differ significantly from the evaluation group in respect to sex, age, month of entry, presence of URTI at the initial visit, or complaints at entry. Of the remaining 397 children (92%), 223 (56%) still had C2-typanogram and B-tympanogram in both ears, 81 (20%) in one ear, and 93 (24%) had A-tympanogram or C1-tympanogram. Patient characteristics at the initial visit are shown in [Table 1] (Please see pages 609 -611 for all tables.)
Persistent Bilateral OME
[Table 2] shows the results of the bivariate analysis of determinants for persistent bilateral OME. Entry into the study between June and November showed a weak association (P=.05). The presence of AOM or URTI during the examination showed a strong association with bilateral OME (P=.005 and P <.001, respectively).
[Table 3] shows the results of the logistic regression analysis. Only no adenoidectomy showed any influence on the duration of OME (P <.05). The presence of AOM or URTI at the follow-up visit was associated with bilateral OME (P <.05 and P <.001, respectively). Because of the influence of URTI on OME, logistic regression analysis was carried out both with and without URTI at the follow-up visit. When URTI was present no determinants were found. When it was absent, the following 3 determinants were prognostic of bilateral OME: (1) no adenoidectomy (P <.05; adjusted odds ratio [OR]=2.75; 95% confidence interval [CI], 1.22-6.20); (2) an episode of AOM in the first year of life (P <.05; adjusted OR=1.59; 95% CI, 1.03-2.46); and (3) entry into the study between June and November (P <.05; adjusted OR=2.07; 95% CI, 1.11-3.86). No good set of discrimant factors was found when adding together the b coefficients for an episode of URTI and AOM on follow-up, no adenoidectomy, and an episode of AOM in the first year of life (AUC=0.61; standard error [SE]=0.029).