International Journal of Pediatric Otorhinolaryngology 75 (2011) 527–531

Contents lists available at ScienceDirect

International Journal of Pediatric Otorhinolaryngology

journal homepage: www.elsevier.com/locate/ijporl

Tympanic membrane perforation: Size, site and hearing evaluation

Issam Saliba *, Anthony Abela, Pierre Arcand
Sainte-Justine University Hospital Center (CHU SJ), 3175, Côte Sainte-Catherine, Department of Otorhinolaryngology, Montreal, QC, H3T 1C5, Canada

A R T I C L E I N F O A B S T R A C T

Article history: Objective: To assess different clinical scales of TM perforation size; to evaluate the effect of the size and
Received 9 December 2010 the site of a perforation on the hearing level and frequencies.
Received in revised form 10 January 2011 Methods: Prospective study. Observers had subjectively estimated the size in millimeter and in
Accepted 10 January 2011
percentage of a particular perforation; objectively computerized measures of TM perforations area were
Available online 22 January 2011
analyzed. Agreement between different measures was studied. Cases with postoperative intact TM and
an air-bone gap (ABG) of 10 dB or less were studied.
Keywords:
Results: Global mean preoperative ABG was 21.8  17 dB. Preoperative ABG was different between small-
Perforation
Tympanic membrane
large and small-total perforations (p = 0.001). Difference of the preoperative ABG was statistically significant
Size between perforations filling up the four quadrants and perforation limited to one quadrant in the postero-
Percentage inferior, antero-superior and antero-inferior site. No statistically difference between perforation sites was
Millimeter identified for each affected frequency. Difference is statistically significant (p = 0.001) between the 250 Hz
Classification and the other frequencies for the medium, large and total perforations. After myringoplasty bone conduction
improvement was statistically significant for the frequencies 500 (p = 0.04), 1000 (p = 0.04) and 2000 Hz
(p = 0.011). Agreement was large enough when TM perforation size was expressed in percentage and absent
when expressed in millimeter.
Conclusion: TM perforation can be clinically estimated quite precisely as a percentage of the TM area.
Conductive hearing loss is frequency dependent; with the greatest loss occurring at the lowest sound
frequencies. Hearing loss does not depend on the perforation’s location.
ß 2011 Elsevier Ireland Ltd. All rights reserved.

1. Introduction This estimated size is limited by inter-observer errors and by the

Tympanic membrane (TM) perforations are a common phe-
nomenon and usually the result of infection, ventilation tube
insertion or trauma [1,2]. The size and the site of TM perforations
are variable. Accurate evaluation of the TM perforation is an
important guide for informed management of this problem.
Larger TM perforation is thought to be unlikely to heal with
conservative management and probably require surgical interven-
tion. However, there is debate in the literature over whether the
size of TM perforation is predictive of surgical success [3–5]. In
addition, it has been established that the larger the perforation, the
greater the decibel loss in sound perception [6].
The site of the perforation is believed by some schools of
thought to have a significant effect on the magnitude of hearing
loss [2]. However, some workers believe that there is no significant
effect associated with the location of the TM perforation [7].
Commonly employed methods of quantitative analysis of TM
perforation include crude estimation of the perforation diameter in
millimeter or a percentage estimation of the TM perforation area.
* Corresponding author. Tel.: +1 514 345 4857; fax: +1 514 737 4822.
E-mail address: issam.saliba@umontreal.ca (I. Saliba).
TM perforation shape.
In regards to this divergent opinion, it is essential to standardize
an accurate measurement of TM perforation size in order to enable
objective management, to apply a unique worldwide scale to
compare results of different myringoplasty techniques and to
evaluate successful myringoplasty results of different TM perfora-
tion size.
In this study, we aim to assess different clinical scales of TM
perforation size, to evaluate the effect of the size and the site of TM
perforation on the hearing level and on the hearing frequencies.
2. Materials and methods
This prospective study was conducted between January 2008
and March 2010 at our tertiary care center. Ethical clearance for the
study was obtained from our review board ethical committee. All
patients were adults and responded to the following inclusion
criteria: (1) TM perforations present for at least 6 months, (2)
without evidence of active chronic otitis media, cholesteatoma or
retraction pocket formation. Excluded were those with purulent
discharge, suspected cholesteatoma and unidentified anterior rim
of the perforation. The investigation included 156 patients that had

0165-5876/$ – see front matter ß 2011 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.ijporl.2011.01.012
[()TD$FIG]
528 I. Saliba et al. / International Journal of Pediatric Otorhinolaryngology 75 (2011) 527–531
Fig. 1. Percentage of tympanic membrane (TM) perforation area as measured by the Adobe acrobat 8 professional software. In this example: TM area = 9644.03 mm2 and the
perforation area (Pa) = 782.34 mm2. The percentage of the perforation area = 100  Pa/TM area = 8.1%. This perforation is classified in the grade I (small
perforation = perforation less than 25% of the TM surface).
myringoplasty for 172 TM repair. 16 cases had bilateral TM
perforation.
2.1. Outcome measures
Our study has two elements: first, measuring the TM
perforation size and second identifying the audiologic correlation
to the perforation criteria. To do so, ears were examined with an
otomicroscope and a photo-endoscopic image of each TM
perforation was taken immediately before the procedure. Photos
were taken by using a 4 mm diameter, 6 cm length and 0 degree
angulations’ endoscope (KarlStorz, Germany) connected to a Nikon
coolpix 4500 digital camera through a 590-70 connector (Karl-
Storz, Germany). In order to measure the inter-observer agreement
for TM perforation size all photos were examined independently
by 6 physicians (2 senior otolaryngology residents, 1 otology and
neurotology fellow and 3 senior otorhinolaryngologists of at least 5
years experience) all of whom were randomly selected from their
colleagues at our university program. Each physician found out the
following criteria: (1) the site of the perforation (ear drum was
virtually divided into 4 quadrants: postero-superior (Ps), postero-
inferior (Pi), antero-superior (As) and antero-inferior (Ai)), (2) the
size expressed in millimeter for the largest TM perforation
diameter; measures were done subjectively compared to the
mean diameter of a tympanic membrane which is 8 mm and (3)
the size of the TM perforation expressed in percentage compared
with the tympanic membrane surface (TMS); Once the observer
estimates the percentage of the perforation, he represents the case
in one of the four grades of TM perforation classification reported
in 2008 [8]. Grade I: small (S) for perforation less than 25% of TMS;
grade II: medium (M) for perforation between 25% and 50% of the
TMS; grade III: large (L) for perforation between 50% and 75% of the
TMS and grade IV: total (T) for perforations more than 75% of the
TMS.
To measure the intra-observer agreement, otorhinolaryngolo-
gists examined again, 10 days later, the 172 TM perforation series.
To eliminate bias, photos order of presentation was different from
the previous analysis. After all observers had subjectively
estimated the size of each particular perforation and in order to
measure objectively the TM perforations area, all photos were
analyzed by the Adobe acrobat 8 professional software which
compute the percentage of the perforation area compared with the
TM surface. An example of this calculation is represented in Fig. 1.
Postoperatively, the status of the ear drum and the hearing level
were the main criteria for measuring outcome. Hearing improve-
ment was assessed using the audiogram results obtained for at
least 4 months postoperatively. In order to identify ears where TM
perforation was the sole cause of conductive hearing loss we
studied cases with postoperative intact TM and an air-bone gap
(ABG) of 10 dB or less. For this group of patient, pre and
postoperative hearing tests were analyzed. The outcome measures
included the mean ABG value at the frequencies 0.5, 1, 2 and 4 kHz,
the difference between air conduction and bone conduction
threshold at each frequency between 0.25 and 4 kHz and the
bone conduction threshold.
2.2. Statistical analysis
The kappa statistics were used to measure the agreement
between the different physicians’ measures expressed as percent-
age.
A variance analysis with repeated measures and Chi square
tests was performed for hearing statistical analysis.
Correlation test and Student t-test for paired data were used to
determine the degree of agreement between two observers for
continuous data and Bland–Altman plot is used to analyze this
agreement. It is common to compute the limits of agreement
during Bland–Altman analysis. This is usually specified as
bias  1.96  STD (average difference  1.96 times standard devia-
tion of the difference).
A p-value of 0.05 or less was considered statistically significant
for all analysis.
3. Results
Postoperative ABG of 10 dB or less was achieved in126 patients;
only these 126 cases where considered for the remainder of the
study keeping the perforation as the sole cause of air-bone gap.
Their preoperative perforations were classified objectively by the
Adobe acrobat 8 professional software in one of the four grades: I
(small perforation; N = 40), II (medium perforation N = 43), III
(large perforation; N = 30) and IV (total perforation; N = 13). The
[()TD$FIG] [()TD$FIG]
I. Saliba et al. / International Journal of Pediatric Otorhinolaryngology 75 (2011) 527–531 529

Fig. 2. Results of preoperative mean air-bone gap in decibel (dB) presented by the
sites of tympanic membrane (TM) perforation. No postero-anterior perforation was
noted in the superior part of the TM. TM was virtually divided into 4 quadrants by
one line along the malleus handle and another line at right angles crossing the
umbo. Ps: postero-superior; As: antero-superior; Pi: postero-inferior; Ai: antero-
inferior.
Fig. 3. Preoperative air-bone gap by frequency for the four tympanic membrane
perforations grade. Grade I = small for perforation less than 25% of TMS; grade
II = medium for perforation between 25% and 50% of the TMS; grade III = large for
perforation between 50% and 75% of the TMS and grade IV = total for perforations
more than 75% of the TMS.

preoperative mean ABG was 13.75  8.6 dB, 21.31  10.4 dB,
28.1  27 dB and 34.40  11 dB for the grade I, II, III and IV,
respectively. The global mean preoperative ABG was 21.8  17 dB.
Preoperative ABG was not different between the grade I–II, II–III, II–IV
and III–IV (p > 0.05). However p = 0.001 for the ABG difference
between grades I–III and I–IV.
3.1. Site and size of TM perforation
Sites of perforations and mean level of their preoperative ABG
are represented in Fig. 2. Most perforations were found in the
inferior part of the tympanic membrane. Posterior, anterior and
superior parts of the TM were equally affected (p > 0.05). No
postero-anterior perforation developed in the superior part of the
ear drum. Difference of the preoperative ABG was statistically
significant when comparing perforations occupying simultaneous-
ly the four quadrants (Ps–Pi–As–Ai; 34.4  10.2 dB) and perforation
limited to one quadrant in the Pi (p = 0.007; 14.3  7.6 dB), As
(p = 0.017; 13  7.5 dB) and Ai (p = 0.006; 15.2  11.2 dB) quadrant.
However, ABG was not statistically different for perforations in the Ps,
Pi, As and Ai quadrants neither for the remaining combination not
aforementioned. Even though preoperative ABG increases when two,
three or four quadrants were affected by the perforation, the
difference was not statistically significant (Table 1).
250 and 500 Hz frequencies are more affected by the
perforations than frequencies of 1 kHz or above (p < 0.05).
However no statistically difference between perforation sites
was identified for each affected frequency (p = 0.138). While for
the small perforation (grade I) no conductive loss difference was
found between frequencies (p = 0.072), difference is
statistically significant (p = 0.001) between the 250 Hz and
the other frequencies for the grade II, III and IV perforations’
(Fig. 3).
3.2. Bone conduction
After myringoplasty bone conduction (BC) improvement was
statistically significant for the frequencies 500 (p = 0.04), 1000
(p = 0.04) and 2000 Hz (p = 0.011). However this improvement was
not clinically significant. Postoperative and preoperative mean
differences were 3.6, 2.7 and 2.4 dB for the frequencies 500, 1000
and 2000 Hz, respectively. Preoperative BC threshold was no
different by comparing the sites neither by comparing the size of
TM perforation.
3.3. Size expressed in percentage
The inter-observer agreement between the 6 physicians was
substantial and almost perfect. Kappa measures vary from 0.65 to
0.83.
The intra-observer agreement was almost perfect where kappa
measures were above 0.81.

Table 1
Preoperative air-bone gap in decibel by site of tympanic membrane perforation. N: number of perforation; SD: standard deviation; SE: standard error; Ps: postero-superior;
As: antero-superior; Pi: postero-inferior; Ai: antero-inferior.

N Mean SD SE Minimum Maximum

Ps 5 17.3 4.79 2.14 11.25 22.50

As 10 13 7.61 2.40 0.00 22.50
Pi 16 14.4 7.58 1.89 0.00 30.00
Ai 19 15.2 11.15 2.56 0.00 36.25
Ps–Pi 20 21 11.87 2.65 0.00 47.50
As–Ai 6 32.1 9.53 3.89 21.25 45.00
Pi–Ai 15 28.6 14.55 3.75 7.5 60.00
Ps–Pi–Ai 14 17.1 36.54 9.76 10.7 38.33
Ps–Pi–As–Ai 21 34.4 10.19 2.22 17.5 56.25

Total 126 21.7 17.16 1.52 0.00 60

530 I. Saliba et al. / International Journal of Pediatric Otorhinolaryngology 75 (2011) 527–531

Table 2
Size expressed in percentage: kappa measures of inter-observers and intra-observers measures (S1–S1i, S2–S2i and S3–S3i). R5a and R5b represent the two senior
otolaryngology residents, F1 represents the otology and neurotology fellow, S1, S2 and S3 represent the 3 senior otorhinolaryngologists of at least 5 years experience and S1i,
S2i and S3i represent the same 3 senior otorhinolaryngologists 10 days after the first perforation size measure.

Observer Kappa of subjective measures Kappa of objective measures

R5b F1 S1 S2 S3 S1i S2i S3i

R5a 0.737 0.754 0.714 0.713 0.729 0.659 0.667 0.652 0.620
R5b 0.763 0.740 0.822 0.771 0.710 0.703 0.706 0.661
F1 0.726 0.807 0.774 0.734 0.731 0.728 0.675
S1 0.800 0.800 0.779 0.764 0.755 0.703
S2 0.832 0.823 0.828 0.820 0.768
S3 0.818 0.809 0.803 0.733
S1i 0.818 0.820 0.921
S2i 0.801 0.928
S3i 0.930

Comparison of the subjective measures to the objective
computerized measures showed a kappa measure oscillating from
0.62 to 0.93. Each group of specialist significantly estimated the
sizes of the perforations and was significantly accurate and similar
to the objective software measures (Table 2).
the TM can result in a conductive hearing loss that ranges from
negligible to 50 dB. Until recently, the effects of TM perforations on
middle-ear sound transmission were not well characterized.
In a normal ear, the sound pressure difference across the TM
provides the primary drive to motion of the TM and ossicles [9].

3.4. Size expressed in millimeter 4.1. Size, site and hearing loss

Difference between measures expressed in millimeter was
statistically and highly significant (p < 0.001) for all the 6
physicians. The average difference between two measures varies
between 0.2  1.05 mm and 1.09  1.03 mm. This disagreement is
evident on the Bland–Altman plot (Fig. 4). For the same observer at 10
days interval, average difference between measures was
0.13  0.69 mm and statistically significant (p = 0.015). Disagree-
ment between observers for perforation size expressed in millimeter
is obvious for the small and for the large perforations (Fig. 4).
4. Discussion
The tympanic membrane serves as a key component of the
[()TD$FIG]tympano-ossicular system for sound transmission. Perforations of
Fig. 4. Example of Bland–Altman plot showing the correlation between observers F1
and S1 for the perforation size expressed in millimeter. It represents the 126
perforations. Multiple measures had the same mean size of tympanic membrane
perforation. We notice that the average difference is 1.09 mm. Disagreement
between observers is obvious for perforations above and below 4 mm.
Tympanic membrane protects the middle ear cleft from
infection and shields the round window from direct sound waves.
This shield is necessary to create a phase differential so that the
sound wave does not impact on the oval and round window
simultaneous [10]. For instance, posterior perforations are
believed to be worse than the anterior ones because of the direct
exposure of the round window to sound waves. Perforations at or
near the site of tympanic membrane attachment to manubrium
have more severe effects than those of comparable size at different
sites [11].
Our statistical analysis of the TM perforation site in patients
with conductive hearing loss showed no correlation with the
magnitude of hearing losses recorded unless the postero-superior
quadrant is affected. The potential effect of loss of the round
window baffle on hearing thresholds was not, however, demon-
strated by our findings. Hearing loss was not greater with
perforations affecting the postero-inferior quadrant (over the
round window area) compared with those sited antero-inferiorly.
No difference was found between postero-superior quadrant
perforation alone and a larger perforation affecting the four
quadrants. This is in agreement with the negative effect of a direct
sound wave simultaneously on both inner ear windows. It appears
as though exposure of the ossicular chain (postero-superior) was
more of a factor. Similar result was reported by Mehta et al. [12].
Sound transmission with small perforations doesn’t depend on
perforation site. Because the wavelengths of sound at audiometric
frequencies of 4 kHz or less are larger than the middle-ear
dimensions, the phase-cancellation effect at the round window
should theoretically be similar for anterior and posterior perfora-
tions [9].
Preoperative ABG showed no difference for small perforation
affecting only Ps, Pi, As and Ai quadrants. The only statistically
difference of ABG is between small (grade I) and large (grade III) or
total (grade IV) perforations.
The tympanic membrane vibrates at the same frequency as the
incoming sound, and in turn, causes the ossicular chain to vibrate
at that same frequency. Perforation-induced losses are greatest at
the lowest frequencies and decrease as frequency increases. The
size of the perforation was also an important determinant of the
hearing loss. Larger perforations result in larger hearing losses an
effect that was present at all audiometric frequencies by a
reduction in driving pressure across the TM. This driving pressure
is minimal when perforation is small; therefore small perforations,
 I. Saliba et al. / International Journal of Pediatric Otorhinolaryngology 75 (2011) 527–531
in opposition to medium, large and total perforations, do not affect
significantly nor the low neither the high frequency.
Isolated postero-superior and isolated antero-superior perfora-
tions were noted. No postero-anterior perforation was identified in
the superior part of the TM. It seems that malleus handle prevent
the extension of the perforation. However, total perforation could
be the result of an upper expansion from the inferior quadrants.
4.2. Bone conduction
After myringoplasty, closure of air-bone gap is well-known, but
improvement in bone conduction threshold (BCT) is not commonly
observed. Huizing demonstrated that in addition to otosclerosis,
treatment of the middle ear disorder may lead to BCT improvement
[13]. We report a BC statistical improvement at 500, 1000 and
2000 Hz after myringoplasty. BCT is not always a true representa-
tion of the sensorineural hearing capacity of the inner ear. When
sound is applied to the skull, its transmission is not solely via the
bone to the cochlea. There are also some anatomic structures like
external auditory meatus and ossicular chain contribution to bone
conduction. In individuals with conductive impairment, ossicular
chain transmission is not so prominent, and the bone conduction
threshold is not a measure of the same thing as in normal
individuals [14].
4.3. Size measurement of TM perforation
It is important first to check for the inter-observer agreement as
well as the intra-observer agreement. Kappa statistics were
created for the measurement of observer’s agreement for
categorical data. Unlike the simple percent agreement, kappa
statistics offers the advantage to compare the agreement against
that which might be expected by chance. A kappa close to 1
represents a perfect agreement while a one close to 0 represents a
poor agreement. A kappa higher than 0.4 symbolize a moderate
agreement while kappa higher than 0.6 could be considerate a
substantial agreement. Thus kappa of the TM perforation
expressed in percentage is high and has led to the same conclusion.
While agreement for TM perforation size expressed in percentage
was large enough for the inter-observer, the intra-observer
measurement and for the objective software measurement,
disagreement between observers for perforations expressed in
millimeter is obvious and independent of the TM perforation size.
Error is significant for small as well as for larger perforation. To
compare objectively different myringoplasty techniques and their
results, it is preferential to characterize clinically a TM perforation
by describing three criteria: (1) the size of the perforation; we
suggest to avoid the Millimetric representation and to use the
percentage based classification described in 2008; (2) the site of
the perforation and (3) the type of the perforation (central or
marginal) for all the risks that marginal perforation can lead to.
Our findings do conflict with the study by Ibekwe et al. They
conclude that visual estimation in approximating size as a
531
percentage of the total area of the TM perforation is imprecise
[15]. Accurate clinical determination of the size of a perforation is
essential. Although the use of quadrants to assess size in
percentage is just an estimate, it provides a visual template to
assist in the estimation of small or larger perforation with a good
agreement between observers. Upon our result, there is no need for
a complicated, expensive computer-based video-otoscopy system
that precisely calculates the perforation size relative to the TM
area.
5. Conclusion
Tympanic membrane perforation can be clinically estimated
quite precisely as a percentage of the TM area. Accurate estimation
of the perforation size is helpful in the clinical management and in
the comparison of different myringoplasty techniques and their
results.
This prospective clinical study of TM perforations showed that
conductive hearing loss is frequency dependent, with the greatest
loss occurring at the lowest sound frequencies, the hearing loss
increases as the size of the perforation increases and the hearing
loss does not depend on the perforation’s location.
References
[1] T.S. Ibekwe, G.T. Ijaduola, O.G. Nwaorgu, Tympanic membrane perforation among
adults in West Africa, Otol. Neurotol. 28 (2007) 348–352.
[2] S.E. Voss, J.J. Rosowski, S.N. Merchant, W.T. Peake, Non-ossicular signal transmis-
sion in human middle ears: experimental assessment of the ‘‘acoustic route’’ with
perforated tympanic membranes, J. Acoust. Soc. Am. 122 (2007) 2135–2153.
[3] B. Kotecha, S. Fowler, J. Topham, Myringoplasty: a prospective audit study, Clin.
Otolaryngol. 24 (1999) 126–129.
[4] N.A. Bhat, R. De, Retrospective analysis of surgical outcome, symptom changes,
and hearing improvement following myringoplasty, J. Otolaryngol. 29 (2000)
229–232.
[5] L. Pignataro, L.G.D. Berta, P. Capaccio, A. Zaghis, Myringoplasty in children:
anatomical and functional results, J. Laryngol. Otol. 115 (2001) 369–373.
[6] J.H. Black, S.A. Hickey, P.J. Wormald, An analysis of the results of myringoplasty in
children, Int. J. Pediatr. Otorhinolaryngol. 31 (1995) 95–100.
[7] N.J. Roland, R.D.R. McRae, A.W. McCombe, From chronic suppurative otitis media,
in: Key Topics in Otolaryngology and Head and Neck Surgery, 2nd edition, Bios
Scientific Publishers, Wales, 2001, pp. 38–41.
[8] I. Saliba, Hyaluronic acid fat graft myringoplasty: how we do it, Clin. Otolaryngol.
33 (December (6)) (2008) 610–614.
[9] S.E. Voss, J.J. Rosowski, S.N. Merchant, W.T. Peake, Middle-ear function with
tympanic-membrane perforations. I. Measurements and mechanisms, J. Acoust.
Soc. Am. 110 (2001) 1432–1444.
[10] F.O. Ogisi, P. Adobamen, Type 1 Tympanoplasty in Benin: a 10-year review, Niger.
Postgrad. Med. J. 11 (2004) 84–87.
[11] S.W. Ahmad, G.V. Ramani, Hearing loss in tympanic membrane perforations, J.
Laryngol. Otol. 93 (1979) 1091–1098.
[12] R.P. Mehta, J.J. Rosowski, S.E. Voss, E. O’Neil, S.N. Merchant, Determinants of
hearing loss in perforations of the tympanic membrane, Otol. Neurotol. 27
(February (2)) (2006) 136–143.
[13] E.H. Huizing, Bone conduction loss due to middle ear pathology: pseudopercep-
tive deafness, Int. Audiol. 3 (1964) 89–98.
[14] M.M. Paparella, T. Morizono, C.T. Le, et al., Sensorineural hearing loss in otitis
media, Ann. Otol. Rhinol. Laryngol. 93 (1984) 623–629.
[15] T.S. Ibekwe, O.G. Nwaorgu, A.A. Adeosun, D.D. Kokong, H.O. Lawal, P.O. Okundia,
P.A. Onakoya, Assessments of the size of tympanic membrane perforations: a
comparison of clinical estimations with video-otoscopic calculations, Ear Nose
Throat J. 87 (October (10)) (2008) 567–569.