Persistent Unilateral or Bilateral OME
[Table 4] shows the results of the bivariate analysis of determinants for persistent unilateral or bilateral OME. Absence of AOM at initial visit, attending daycare and less than 3 rooms in the house showed a weak association. The presence of URTI during examination was again associated with the presence of middle ear effusion.
Logistic regression analysis ([Table 5]) showed a significant association with the absence of AOM at the initial visit (P <.05) and with URTI at the follow-up visit (P=.001). The analysis was repeated again with and without URTI at the follow-up visit. In the presence of URTI no determinants were found; in the absence of URTI an episode of AOM in the first year of life showed a significant association (P <.05; adjusted OR=1.69; 95% CI, 1.15-2.49). Again, no good set of discrimant factors was found by adding the b coefficients of AOM at initial visit, history of AOM in the first year of life, daycare, and URTI at the follow-up visit (AUC=0.65; SE=0.03).
Discussion
Our entry criteria detected a high percentage (84%) of patients with bilateral OME. The presence of more than 1 criterion in the same child had no effect on the duration of OME. Our data show a strong association between the presence of AOM or URTI during examination and the finding of middle ear effusion. In the absence of URTI, the following were associated with persistent bilateral OME: no adenoidectomy, an episode of AOM in the first year of life, and entry into the study between June and November. An episode of AOM in the first year of life was associated with persistent unilateral or bilateral OME. It was not possible to construct a discriminant model prognostic of persistent bilateral or unilateral or bilateral OME as the AUCs of 0.61 and 0.65, respectively, were rather low.
A possible weak point in this study is that our results are based on tympanometry outcome after a 3-month interval. The 36 children excluded from the 3-month follow-up did not differ significantly from those included, so their exclusion probably had no influence on the results.
Some of the determinants identified in this study are closely related to URTI. For instance we found a weak association between attending daycare and persistent unilateral or bilateral OME in bivariate analysis, but this association disappeared in the logistic regression analysis. This could be explained by the close relationship between attending daycare and URTI-related factors in the occurrence of OME. This close association between OME and URTI has been reported before.33 The finding of this relationship suggests that occurrence of middle ear effusion during an URTI episode is a physiologic phenomenon, probably caused by eustachian tube obstruction and needing no further treatment. If this is true, children with URTI-related OME need only close follow up and no active treatment.
The factors of no adenoidectomy, early episodes of AOM, and the prevalence of bilateral OME during the months June to November were found in the absence of URTI and suggest a different, more pathophysiologic relationship with OME than when OME is found with URTI. As adenoid size seems not to be correlated with the pathogenesis of OME,34 the association with no adenoidectomy supports the viewpoint that the adenoids are an important factor in OME by providing an infectious focus in the nasopharynx or by their influence on the immunological status of the nasopharynx.34-36
The association between AOM early in life and persistence of OME has been reported previously.37 Children with early AOM episodes are frequently known as “otitis-prone.”38 They also have more recurrent AOM episodes and more episodes of URTI than average.
The exact relationship between AOM and OME is not known. AOM may precede OME, but it is also more frequent in patients with OME.39 Several studies have reported an association between OME and winter or early spring. This is probably the result of the increased incidence of URTI during this season. The period between June and November (summer and early autumn) is less likely to be associated with frequent episodes of URTI. Contrary to the findings of other studies, we found no association between age younger than 3 years and persistent OME.40
In our study we found that factors associated with persistent OME differ from those that increase the risk for occurrence of OME (attending daycare, bottle feeding, passive smoking).* This is an important finding. First, the absence or presence of the determinants for persistent OME may affect the choice of treatment. A more active attitude is needed if there is a higher chance for sequelae, such as language delay caused by conductive hearing loss, atelectasis of the tympanic membrane, and cholesteatoma, as is the case in persistent OME. In the absence of determinants for persistent OME a more watchful policy may be followed, especially if the child has no severe hearing loss, does not suffer from recurrent AOM episodes, and has no tympanic membrane abnormalities. This would lead to a reduction of overtreatment and to cost reduction. In the presence of determinants for persistent OME, children should be monitored more closely. When sequelae such as language delay or recurrent episodes with AOM are present, active treatment with antibiotics, tympanostomy tub insertion,5,9 or adenoidectomy34-36 is needed. We also feel that future research should be focused on the pathophysiological group of otitis-prone children and not on all children with OME. If we are capable in differentiating between those children who need further treatment from those children in whom OME is a physiological reaction on a URTI, overtreatment and costs may be greatly reduced.