Results:

In this study, 139 ears (127 patients) were included, 57 males and 70 females. Comorbidities were as follows: 16 patients (13%) had craniofacial malformations (including ten with a velopalatal cleft), 25 patients were diagnosed with allergic rhinitis (20%).
Mean age at surgery was 9.6 years ± 2.6 (Figure 2 ). Perforation was in the right ear in 74 cases (53% of ears) and in the left ear in 65 cases (47%). No clinical symptom was present in 49 cases (35%). Preoperative auditory evaluation is reported in Table 1 . The size of the perforation was available for 109 ears (78%): pinpoint size one ear (1%), up to a quarter of the eardrum’s surface in 34 ears (31%), up to half of the eardrum in 38 ears (35%) and more than half of the surface to total in 36 (33%). Location of the perforation was available in 137 ears (99%): posterior in 24 cases (18%), inferior in 28 cases (20%), anterior in 47 cases (34%), subtotal or central in 36 cases (26%) and multiple in two (2%).
All procedures were performed with an endaural approach and an underlay tragal cartilage graft. Procedures were performed by a senior paediatric otologist in 57 cases (47%) and by a junior otolaryngologist (10 years’ experience or less) in 65 cases (53%). There was a preoperative external auditory canal stenosis in 17 cases (12%). During the procedure, the atrial mucosa was inflammatory (abnormal mucosa or otorrhoea) in 45 ears (33%) and glue in the middle ear was found in nine ears (7%).
At one-year anatomical assessment follow-up, 129 (93%) tympanic membranes were closed. Satisfactory anatomical result (closure without OME, retraction, cholesteatoma or myringitis) was observed in 112 ears (81%). The unsatisfactory results are detailed in Table 2 . Notably, two children had developed a severe retraction pocket with cholesteatoma and nine an evolutive retraction pocket at one-year follow-up. Of those patients, 3/11 had cleft palate, 2/11 had contralateral tympanoplasty and mean age at surgery was 9.2 years.
Audiometry data one year after surgery were available for 127 ears (91%) and reported in Table 1 . Audiometric results were considered good (residual air conduction ≤ 20 dB) in 102 ears (80%). In five ears (4%) the bone conduction threshold was increased by more than 10 dB. Both anatomical and audiometric success (satisfactory anatomical and audiometric results) was obtained in 84 ears (66%).
The otoscopic data at three-years post-surgery were available for 66 ears (47% of 139 ears). Among the 52 ears which were satisfactory at one year, six (12%) became unsatisfactory at three years (including one reperforation, two OME and three cases of myringitis). Of the 14 ears which were unsatisfactory at one-year follow-up, 10 became satisfactory and four remained unsatisfactory (including one cholesteatoma, one perforation with myringitis and two cases of myringitis). Overall, at three years follow-up, the tympanic membrane was closed in 64/66 ears (97%) and the result was satisfactory in 56/66 ears (85%). Myringitis was the most frequent cause of poor result (six ears, 9%), detail in Table 2 . The median air conduction threshold was 15dB [12.5; 22.5] with 26/32 (81%) of ears considered as having a good functional result (auditory data available for 32 ears, detail in Table 1 ). Overall success at three years (both satisfactory anatomical and audiometric results) was obtained in 24/32 ears (75%). Air conduction audiometry was significantly better at one-year and three-years follow-up versus pre-operative values (difference at 1 year -6dB [-10;0] p<0.001; difference at 3 years -2dB [-13;1] p = 0.034).
The univariate analysis of risk factors of non-satisfactory TM at one-year follow-up identified only one significant risk factor. Patients operated on under the age of eight years had 31% of non-satisfactory results at one-year follow-up (versus 14% at the age of eight or over, p=0.024). The other risk factors (size, anterior or subtotal perforation, various associated pathologies, other age groups) were not associated with higher risk of non-satisfactory TM (Table 3 ). When analysing tympanic perforation locations separately (posterior, inferior/central, anterior, subtotal or multiple locations), no statistical difference in outcome at one year was found (p = 0.20).