Discussion:

The rate of 93% of TM closure at one year reported in this series corresponds to other large paediatric studies, ranging from 74 to 97% (1,5–7,9–12,15–18). However, our 81% TM satisfaction rate at one year, which includes complications linked to Eustachian tube malfunction, seems to us more appropriate and clinically relevant in paediatric patients. Furthermore, when associating satisfactory anatomical and audiometric result, success rate falls to 66% at one year. It has been shown in previous studies that OME, re-perforation and tympanic retractions are commonplace in the closed TM of the child (7,9,11,13,18). We believe that these complications directly linked to the correct surgical TM closure, should be taken into account to determine if the procedure was a success. As these complications sometimes occur between one and three years after surgery, we believe that the minimum duration of follow-up should be of three year, for the surgery to be considered successful (9,10).
At 3 years, the rate of reperforation was very low (one case out of 66, 2%) but the rate of myringitis was surprisingly high (9%) and could be a consequence of the use of cartilage to close the TM. Although no statistical link has been established, it has been described previously in other cartilage tympanoplasty studies (14,15). No specific clinical characteristic was identified in children with myringitis. The rate of myringitis remained important over time (5% at one year and 9% at three years). However, this may also be explained by the smaller number of patients at the three-year follow-up (66 vs 139) and that patients with post-operative complications are more likely than asymptomatic patients to return for long-term follow-up after myringoplasty.
The univariate study showed that the size and location of the TM perforation were not risk factors (the success rate for anterior and larger perforations was only of 76% but was not significantly lower) and found significantly better results when the procedure was performed at eight-years old or older. In the literature, the role of size and location of the perforation has been reported as a factor influencing the results, large and anterior locations being associated with poorer outcomes (8,9,17). The role of age is controversial, as it has not been found to be a significant factor for success in some studies (5–8), while others also consider younger age being a risk factor, setting the cut-off at eight (9), nine (11,12) or ten years of age (10). Another study found that children operated on before the age of eight had a significantly higher rate of re-perforation (19).
Concerning other factors, as in our study, craniofacial comorbidities were not associated with poorer outcomes in three studies (7,14,19), including patients with velopalatal clefts. Too few patients with such comorbidities were included in our cohort to properly address their influence on the rate of successful tympanic membrane perforation closure. Perioperative tympanic cavity inflammation has previously been identified as a risk factor (5) however as in our study, a meta-analysis found statistically non-significant decreased success rate (8). Chronic otitis in the contralateral ear has been associated with poorer outcomes in some studies (5,8,20) but, has also as in our series not been associated with statistically significant results (17). This discrepancy might result from the fact that following a previous publication (5), ipsilateral inflammatory mucosa and contralateral OME have been considered relative contraindications of tympanic closure in our team. Consequently, only very few patients with these two factors were included in the present series, biasing the statistical analysis.
Two main limitations can be identified in this study. First the lost-to-follow-up rate which could induce selection bias influencing our success rate, as those patients might have stopped their follow-up due to the absence of symptoms or sought a follow-up elsewhere if complications occurred. Second is the high proportion of children with a history of craniofacial malformations (13%), and especially velopalatal clefts (8%), in our cohort. One of the reasons is that our paediatric hospital concentrates a large number of national referral centres for malformative diseases. Those children are more likely to develop ear ventilation problems, OME and retraction after tympanoplasty (14). Additionally, the high percentage of craniofacial malformations in our series and the fact that most children with such anomalies benefit from one consultation per semester or per year throughout their childhood and adolescence in our center, might explain the longer median follow-up duration in our study as compared to most previous publications, thus inducing a follow-up bias
Thus, a minimum of eight years of age at the time of surgery seems to be preferable and long-term follow-up over a few years would be advisable complications may occur in the long-run.