101
International Journal of Pediatric Otorhinolaryngology, 24 (1992) 101-l 10 0 1992 Elsevier Science Publishers B.V. All rights reserved 0165-5876/92/$05.00
PEDOT
00809
Tympanometry and otoscopy prior to myringotomy: issues in diagnosis of otitis media Terese
Finitzo
‘, Sandy Friel-Patti
‘, Kathleen
Chinn ’ and Orval Brown ’
Dallas Cooperative Project on Early Hearing and Language with a Neuroscience Center, Methodist Medical Center Dallas, TX and h lJnic,ersity of Texas at Dallas, Callier Center for Communication Disorders. Dabs, TX and ’ Uniclersity of Texas, Southwestern Medical Center, Department of Otolaryngology, Dallas, TX (USA) (Revised
(Received 29 August 1991) version received 2 December 1991) (Accepted 5 December 199 1)
Key words: Validity; Sensitivity and specificity; Otitis media; Effusion
Myringotomy;
Otoscopy;
Tympanometry;
Abstract
Tympanometry and pneumatic otoscopy were compared to findings at myringotomy in 86 children (163 ears). Seventy percent of the ears (11.5) had effusion, as revealed by myringotomy. Sensitivity and specificity for tympanometry were 90% and 86%, respectively. Sensitivity and specificity for pneumatic otoscopy were 93% and 58%, respectively. A chi-square was performed to compare the sensitivity and specificity of tympanometry to otoscopy, revealing tympanometry significantly better at determining non-effusion states. Additionally, a combined otoscopy and tympanometry sensitivity and specificity were calculated for those otoscopy and tympanometry determinations in agreement, revealing both sensitivity and specificity above 90%. A Fisher’s exact probability test revealed no significant differences for the accuracy of tympanometry over otoscopy when the determinations of each were not in agreement. Implications of these results are discussed.
Introduction
Chronic otitis media with effusion COME) or persistent middle ear effusion is a condition seen in one-third of all children under 3 years of age [ll]. The Correspondence to: Kathleen Chinn, M.S., Callier Center for Communication Rd., Dallas, TX 75235, USA. Address reprint requests to: Orval Brown, M.D., Department of Otolaryngology, School. 5323 Harry Hines Blvd., Dallas, TX 75235, USA.
Disorders,
1966 lnwood
Southwestern
Medical
102
complications of persistent effusion, especially hearing loss and speech/language delay, lead many pediatricians to refer their young patients for myringotomy and ventilation tube placement. As myringotomy is an invasive procedure requiring anesthesia, it is important that physicians accurately identify and document chronic middle ear effusion prior to the recommendation of tympanostomy tubes. That is, sensitivity and specificity should be high. The two, common non-invasive methods used to determine middle ear status are tympanometry and pneumatic otoscopy, and neither achieves a sensitivity and specificity of 100%. Stool and Flaherty [lo] have suggested that for a physician/ otoscopist to be ‘validated’, sensitivity, the ratio of the number of true positives to the number of ears with effusion at myringotomy, must be 90%. Specificity, the ratio of the number of true negatives to the number of ears without effusion at myringotomy [4], must be 80%. To date, few training programs in either pediatrics or otolaryngology provide sufficient opportunity for a physician to acquire otoscopy skills to meet these standards. As part of an ongoing study to examine long-term sequelae from otitis media, we routinely obtain screening tympanograms in a child’s home or day care setting. Pneumatic otoscopy is not available in those settings. Based on the suggestions of Brostoff and Cantekin [3], a type B tympanogram is considered to be indicative of otitis media; whereas an A tympanogram is indicative of an absence of fluid. Intermediate tympanogram types are classified as effusion absent, except when otoscopic information is available to confirm the presence of effusion. A summary of our approach is shown in Table I [8]. In this study, we asked whether these decisions, based on a screening tympanogram, were in agreement with findings obtained at myringotomy. Tympanometry and pneumatic otoscopy, were performed and compared to findings obtained at myringotomy for 86 children in an effort to establish relations among immittance, otoscopy and myringotomy, obtained prior to anesthesia. Methods Subjects
The 86 subjects ranged in age from 6 months to 9 years, 1 month; 59 were males and 27 were females. The mean age for both females and males was 31 months. The subjects were seen at Children’s Medical Center in Dallas, Texas. All subjects were to undergo myringotomy for the placement of ventilation tubes as treatment for recurrent otitis media or persistent otitis media with effusion. Table II displays demographics of subjects and diagnosis at the time of surgery. Informed consent was obtained, according to the guidelines for the use of human subjects, from the Institutional Review Board at the University of Texas Southwestern Medical School. Procedures
First, each subject was seen by a certified audiologist who performed acoustic immittance using a Maico Screening Immittance Bridge (Model No. 610). Tym-
TABLE
I
Tympnnogram (Maico
classification
using automated
immittance
instrument
*
610)
Type
Description
A
Peak compensated near 0 daPa
immittance
> 0.2 mmhos
A’
Peak compensated immittance between - IO and - 99 daPa
> 0.2 mmhos
A_
Positive peak compensated immittance > 0.2 mmhos at + 50 daPa or greater
A,
Reduced between
C,
Peak compensated immittance between - 100 and - 150 daPa
C,
Peak compensated immittance between - 151 and -200 daPa
C,
Peak compensated immittance negative than - 200 daPa
B
Compensated immittance with no observable change in peak pressure from + 200 to - 400 daPa
mmhos
indicates
millimhos;
daPa,
peak immittance < 0.2 mmhos - 99 and + 49 daPa
more
dekapascals.
panograms were classified as shown in Table I. Second, a pediatric otolaryngologist used a standard Welch Allyn pneumatic otoscope with halogen head and standard squeeze bulb to determine the presence or absence of effusion. The otolaryngologist performed the otoscopic examination without knowledge of the tympanometric
TABLE
II
Subject demographics
No. of Subjects Mean age (yr; mo) Age range
Males
Females
Total
59 2; 7
27 2: 7
86 2; 7
0; 6-9;
1
0; 11-l;
3
0; 6-9:
(yr: mo) Persistent Recurrent
OME a OME ’
9 49
a OME lasting three months or longer. ’ Multiple episodes of OM exceeding 6 in a year. * One chart was unavailable for review of ‘diagnosis
7 20
at time of surgery’.
16 69 *
1
104
screening results. Myringotomy was performed in the operating room using a standard operating microscope. Presence or absence of effusion and type of fluid were noted.
Results
Results will be subdivided and discussed in four sections: first, tympanometry to myringotomy; second, otoscopy to myringotomy; and third, the combined results of otoscopy and tympanometry to myringotomy. Finally, data will be presented to address the question of which of the procedures is more accurate overall when compared to myringotomy. The total number of ears examined with tympanometry, otoscopy, and myringotomy was 172. In nine cases, one ear could not be tested due to cerumen impaction or lack of patient compliance, bringing the number of ears to 163. Of the 163 ears, effusion was present in 115 ears (71%) at myringotomy. Validation of tympanometry to myringotomy
Table III is a display of the findings comparing tympanometry to myringotomy. Acoustic immittance measures identified 26 ears with type A or A’ tympanograms, indicating the absence of effusion. Agreement to findings at myringotomy occurred for 19 ears (73%). Sixty-eight ears, demonstrated type B tympanograms, suggesting the presence of effusion. Sixty-five ears or 96% of these cases were in agreement with myringotomy. There were 69 ears (42%) with ‘transition’, tympanograms (Table III). That is, not clear A or B types but type C,, C,, C,, A+ or A, status. Sixty-two percent of the ears with transition tympanograms revealed effusion when myringotomy was performed, without a clear pattern for C versus A subtypes. Effusion was present in 67% of the C tympanograms: 7 of 11 C,, 4 of 5 C, type, and 19 of 29 C,. Only 54% of the A subtypes manifested effusion (7 of 11 type A+ and 6 of 13 A,).
TABLE
III
Comparison of all tympanogram types with myringotomy
Typeof tympanogram B A, A’ G c2 c3 A+ 4
Totals
Ears with OME
Ears clear
Total ears
Ears with ome
In group OME
Ears clear
In group clear
(%)
(%o)
(o/o)
(o/o)
65 I 7 4 19 7 6
3 19 4 1 10 4 7
68 26 11 5 29 11 13
56.5 6 6 3.4 16.5 6 5.2
95.5 26.9 63.6 80 65.5 63.6 46.1
6.2 39.5 8.3 2 20.8 8.3 14.5
4.4 73 36.3 20 34.4 36.6 53.8
115
48
163
70.5
N/A
29.4
N/A
105 TABLE
IV
Tympanomrtry
sensitir~ity and specificity A, A’ und R typo only Myringotomy yes _____-
B tympanogram
.Myringotomy no COME absent) false-positives 3 ears cell I3
A or A tympanogram
true-negatives I9 ears cell 11 -_I
Formula:
sensitivity
A = k+c
X
100
6.5 -- := 0.902 x 100 = 90% 72
specificity
1) = n+B
X
100
I9 = 0.863 x 100 = 86% 22
Sensitivity is computed by the number of true positives divided by the number of true-positives plus false-negatives multiplied by 100. Using a 2 x 2 matrix the formula is stated as: A
. . .
sensitivtty = ~A + C x 100 Thus, as shown in Table IV for the A, A’ and B tympanograms, Specificity is computed as: D
specificity = -
D+B
sensitivity is 00%.
x 100
Thus, as shown in Table IV for the A, A’ and B tympanograms, specificity is 86%. If one adds the transition type tympanograms into the analysis and considers them negative as suggested by Brostoff and Cantekin 131 then sensitivity is 57% and specificity is 93% (Table V). Validation of otoscopy to myringotom)
Otoscopy and pneumatic otoscopy on the same 163 ears as reported above identified the presence of effusion in 127 of the 163 (77.9%) ears. Criteria for determination of effusion consisted of a combination of three tympanic membrane characteristics: (al condition, (b) color, and cc) mobility. A normal luster, translu-
TABLE
V
Tympanometry sensitivity and specificity for all types Myringotomy yes (OME present)
Myringotomy no COME absent)
B tympanograms
true-positives 65 ears cell A
‘false-positives 3 ears cell B
AANDC tympanograms
false-negatives 50 ears cell C
true-negatives 45 ears cell D
Formula:
sensitivity
g
specificity
A = ~ A+C
x 100
= 0.565 x 100 = 57% D = ~ D+B
x100
cent, silver colored tympanic membrane that moved freely was determined to be effusion absent. Retracted eardrums that had rebound mobility only were considered to have only negative pressure unless that retraction was superior and extreme, then adhesive OM was suspected. A red tympanic membrane with accompanying fever and pain was considered to be characteristic of acute otitis media. Middle ear effusion was determined by eardrums which were opaque, had obscured landmarks or retraction, with immobility to pneumatic otoscopy. Other criteria included the presence of air fluid levels or bubbles in the middle ear space. Each effusion determined ear was classified as purulent, mucoid or serous fluid. Purulent effusion is characterized by a reddened tympanic membrane with yellowish fluid behind the eardrum. Mucoid effusion is characterized by a dull tympanic membrane. Serous effusion is clear to yellowish-clear fluid and can be accompanied by freely moveable bubbles in the middle ear space. Table VI is a display of the findings comparing otoscopic examination to myringotomy. Sensitivity when compared to myringotomy was 93% and specificity was 58%: Combined otoscopy and myringotomy
Sensitivity and specificity were calculated for tympanometry (76 Type A, A’ and B tympanograms only) and otoscopy combined compared to myringotomy (Table VII). This combination resulted in a sensitivity of 98.3% and a specificity of 92.8%. Keep in mind that these values reflect situations in which the otoscopist and the tympanogram were in clear agreement. Sensitivity was calculated using tympanom-
107 TABLE
VI
Otoscopy srnsilicify and specificity Myringotomy ye.s COME present)
Myringotomy no COME absent)
0toscopy yes OME
true-positives 107 ears cell A
false-positives 20 ears cell B
0t0scopy no OME
false-negatives 8 ears cell C
true-negatives 28 ears cell D
Formula:
sensitivity
A = ___ Xl00 .4 + c
107 ~ = 0.930 x IO0 = 93% II5
specificity
D = DiB
x IO0
G = 0.583 x 100 = 58%
TABLE
VII
Combined sensitir’ity and specificity Myringotomy yes (OME present)
Myringotomy no (OME absent)
Otoscopy yes Tympanogram B
true-positives 61 ears cell A
false-positives I ear cell B
Otoscopy no Tympanogram A, A’
false-negatives 1 ear cell C
true-negatives 13 ears cell D
Formula:
sensitivity
;
specificity
A = __ A+C
x100
= 0.983 x 100 = 98.3% D = __ x 100 I4 II + B
13 G = 0.928 x 100 = 92.8%
108 TABLE
VIII
Fisher exact probability test Effusion absent
Effusion present
otoscopy correct
6
4
Tympanometry correct
6
2
etry and otoscopy indicating effusion present. Specificity was calculated tympanometry and otoscopy indicating absence of effusion.
using
Accuracy of tympanometry and otoscopy as compared to myringotomy
A chi-square analysis for two independent samples was performed in order to ascertain whether the sensitivity and/or specificity of tympanometry to myringotomy was significantly better than that of otoscopy to myringotomy. The chi-square analysis did not reveal a significant difference in tympanometry versus otoscopy in sensitivity; but revealed a significant difference in the success of tympanometry versus otoscopy in determining the absence of effusion when one limits the analysis to A, A’ and B tympanogram types. A Fisher’s exact probability test [9] was undertaken to compare the accuracy of tympanometry to myringotomy and otoscopy to myringotomy. Cases where the two procedures disagreed are included in this analysis, in order to determine which procedure was correct more often when there was disagreement. Results presented in the two-by-two matrix (Table VIII>, indicate no significant difference between the two tests. When determinations differed, tympanometry was correct lo/18 times or 56%, while otoscopy was correct 8/18 times or 44% (P > 0.05).
Discussion
The results of this study comparing the sensitivity and specificity of pneumatic otoscopy and a screening immittance instrument to myringotomy have implications for diagnosis of otitis media in pediatric practices. First, the sensitivity of tympanometry, that is the ability of the type B tympanogram to identify the presence of effusion, is high at 90% when ‘transition’ tympanograms are eliminated from the analysis. Even specificity of the screening bridge used here is to be considered good at 86%. For the otoscopic examination, sensitivity was high (93%); however, specificity was only 58%. The data do suggest that when the otoscopist and tympanometric findings are in agreement, sensitivity and specificity both exceed 90%. But what of the situation where there is not agreement between measures? These data were examined three ways. Sensitivity and specificity for both measures were compared as well as the outcome when the two measures disagreed.
Recall that the chi-square analysis did not reveal a significant difference between otoscopy and tympanometry for sensitivity - but findings do suggest that in this study, the accuracy of tympanometry exceeded otoscopy in determining the absence of effusion. Thus, while there was no significant overall difference when the two measures were in disagreement, the data from this study suggest that tympanometry is more accurate in determining specificity than otoscopy. While the otoscopist did not achieve the specificity standard suggested by Stool and Flaherty [lOI, certification was not the purpose of this study. It is probable that the values obtained here reflect those found in many clinical practices. Bluestone and Cantekin report that the two pediatricians who participated in their study prior to pneumatic otoscopy training obtained sensitivity ratings of 96% and 68%~ and specificity ratings of 68% and 64% [2]. The utilization of a screening immittancc instrument may facilitate increased accuracy in the diagnosis of otitis media. especially in the presence of B tympanograms. A limitation of this study perhaps is that we have not treated the two ears of the individual children as paired organs. Le et al. [5] suggest factors that influence effusion (i.e., eustachian tube dysfunction) are usually present for both ears. Rockette and Casselbrant [7] raise the same point and suggest that a child not be considered effusion positive unless both ears are determined to have effusion. If we had followed this suggestion and considered the ears as paired organs. thus counting a child effusion positive only when both ears were determined to have effusion, then our sensitivity would have remained good while our specificity would have improved greatly, The procedure is valid and appropriate for incidence and epidemiology studies of OME in that bilateral effusion is more common than unilateral, however, we believe for the purpose of this study, as with other validation studies [1,6,7], it is appropriate that each ear be considered as an individual datum. Another limitation of our study may be our decision not to include the ‘transition’ tympanograms in the initial sensitivity and specificity analysis. These transition tympanograms also present a conundrum for the practitioner. In actual clinical practice, these tympanogram types are common, and our data suggest that presence and absence of fluid occurs with nearly equal frequency with these transitional tympanograms. Data need to be interpreted in light of the small numbers for some types. Additional study is needed to determine if C tyrnpanograms have a higher incidence of associated effusion than some of the A subtypes. However, the suggestions of Brostoff and Cantekin [3] can bc followed at this time. In the presence of these ‘transition’ tympanograms, the findings on otoscopy can be relied on to determine the diagnosis of OME. Clearly, this advice is strong with an otoscopic determination that fluid is present. But the decision that fluid is absent may be little better than chance without a validated otoscopist whose specificity reaches or exceeds the recommended 80%. Such results argue for the need to include validation procedures in both pediatric and otolaryngologic training programs and perhaps even to encourage physicians in clinical practice to consider validation as part of continuing education.
110
Acknowledgements
We acknowledge the contribution of Catherine Nolan Balay, M.S., and Karen Clinton Brown, M.S., for assistance with data collection. Some subjects in this study were patients at Children’s Medical Center at Dallas, Texas, while others were taken from the Dallas Cooperative Project on Early Hearing and Language Development. This study was funded in part by NIH NINCDS grant No. 1 R01 19675 and the American Hearing Research Foundation, Chicago, Illinois.
References 1 Babonis, T.R., Weir, M.R. and Kelly, P.C., Impedance tympanometry and acoustic reflectometry at myringotomy, Pediatrics, 87 (1991) 475-480. 2 Bluestone, C.D. and Cantekin, E.I., Design factors in the characterization and identification of otitis media and certain related conditions, Ann. Otol. Rhinol. Laryngol., 88, Suppl. (1979) 13-28. 3 Brostoff, L. and Cantekin, E.I., Observer invariant diagnosis of middle ear effusions. In: D. Lim, C.O. Bluestone, J.O. Klein, J.O. Nelson (Eds.), Recent Advances in Otitis Media with Effusion: Proceedings of the Fourth International Symposium, B.C. Decker, Philadelphia, 1988. 4 Cantekin, E.I., Bluestone, CD., Fria, T.J., Stool, S.E., Berry, Q.C. and Sabo, D.L., Identification of otitis media with effusion in children, Ann. Otol. Rhino]. Laryngol., 89, suppl. 68 (1980) 190-195. 5 Le, C.T., Freeman, D.W. and Fireman, B.H., Evaluation of ventilating tubes and myringotomy in the treatment of recurrent or persistent otitis media, Pediatr. Infect. Dis. J., 10 (1991) 2-11. 6 Moller, H. and Tos, M., Point and period prevalence of otitis media with effusion evaluated by daily tympanometry, J. Laryngol. Otol., 104 (1990) 937-941. 7 Rockette, H.E. and Casselbrant, M., Methodologic issues in screening for otitis media. In D. Lim, C. Bluestone, J.O. Klein and J. Nelson (Eds.), B.C. Decker, Philadelphia, 1988, pp. 42-44. 8 Roland, P.S., Finitzo, T., Friel-Patti, S., Brown, K.C., Stephens, K.T., Brown, 0. and Coleman, J.M., Otitis media: incidence, duration and hearing status, Arch. Otolaryngol. Head Neck Surg., 115 (1989) 1049-1053. 9 Siegel, S., The case of two independent samples. In S. Siegel (ed.), Non-parametric Statistics: For The Behavioral Sciences, Chapter 6, McGraw-Hill Book Company, New York, (1956) 96-104. 10 Stool, S.E. and Flaherty, M.R., Validation of diagnosis of otitis media with effusion, Ann. Otol. Rhinol. Laryngol., 92, Suppl. 107 (1983) 5-6. 11 Teele, D.W., Klein, J.O. and Rosner, B.A., Otitis media with effusion during the first three years of life and development of speech and language, Pediatrics, 74 (1984) 282-287.
International Journal of Pediatric Otorhinolaryngology, 24 (1992) 101-l 10 0 1992 Elsevier Science Publishers B.V. All rights reserved 0165-5876/92/$05.00
PEDOT
00809
Tympanometry and otoscopy prior to myringotomy: issues in diagnosis of otitis media Terese
Finitzo
‘, Sandy Friel-Patti
‘, Kathleen
Chinn ’ and Orval Brown ’
Dallas Cooperative Project on Early Hearing and Language with a Neuroscience Center, Methodist Medical Center Dallas, TX and h lJnic,ersity of Texas at Dallas, Callier Center for Communication Disorders. Dabs, TX and ’ Uniclersity of Texas, Southwestern Medical Center, Department of Otolaryngology, Dallas, TX (USA) (Revised
(Received 29 August 1991) version received 2 December 1991) (Accepted 5 December 199 1)
Key words: Validity; Sensitivity and specificity; Otitis media; Effusion
Myringotomy;
Otoscopy;
Tympanometry;
Abstract
Tympanometry and pneumatic otoscopy were compared to findings at myringotomy in 86 children (163 ears). Seventy percent of the ears (11.5) had effusion, as revealed by myringotomy. Sensitivity and specificity for tympanometry were 90% and 86%, respectively. Sensitivity and specificity for pneumatic otoscopy were 93% and 58%, respectively. A chi-square was performed to compare the sensitivity and specificity of tympanometry to otoscopy, revealing tympanometry significantly better at determining non-effusion states. Additionally, a combined otoscopy and tympanometry sensitivity and specificity were calculated for those otoscopy and tympanometry determinations in agreement, revealing both sensitivity and specificity above 90%. A Fisher’s exact probability test revealed no significant differences for the accuracy of tympanometry over otoscopy when the determinations of each were not in agreement. Implications of these results are discussed.
Introduction
Chronic otitis media with effusion COME) or persistent middle ear effusion is a condition seen in one-third of all children under 3 years of age [ll]. The Correspondence to: Kathleen Chinn, M.S., Callier Center for Communication Rd., Dallas, TX 75235, USA. Address reprint requests to: Orval Brown, M.D., Department of Otolaryngology, School. 5323 Harry Hines Blvd., Dallas, TX 75235, USA.
Disorders,
1966 lnwood
Southwestern
Medical
102
complications of persistent effusion, especially hearing loss and speech/language delay, lead many pediatricians to refer their young patients for myringotomy and ventilation tube placement. As myringotomy is an invasive procedure requiring anesthesia, it is important that physicians accurately identify and document chronic middle ear effusion prior to the recommendation of tympanostomy tubes. That is, sensitivity and specificity should be high. The two, common non-invasive methods used to determine middle ear status are tympanometry and pneumatic otoscopy, and neither achieves a sensitivity and specificity of 100%. Stool and Flaherty [lo] have suggested that for a physician/ otoscopist to be ‘validated’, sensitivity, the ratio of the number of true positives to the number of ears with effusion at myringotomy, must be 90%. Specificity, the ratio of the number of true negatives to the number of ears without effusion at myringotomy [4], must be 80%. To date, few training programs in either pediatrics or otolaryngology provide sufficient opportunity for a physician to acquire otoscopy skills to meet these standards. As part of an ongoing study to examine long-term sequelae from otitis media, we routinely obtain screening tympanograms in a child’s home or day care setting. Pneumatic otoscopy is not available in those settings. Based on the suggestions of Brostoff and Cantekin [3], a type B tympanogram is considered to be indicative of otitis media; whereas an A tympanogram is indicative of an absence of fluid. Intermediate tympanogram types are classified as effusion absent, except when otoscopic information is available to confirm the presence of effusion. A summary of our approach is shown in Table I [8]. In this study, we asked whether these decisions, based on a screening tympanogram, were in agreement with findings obtained at myringotomy. Tympanometry and pneumatic otoscopy, were performed and compared to findings obtained at myringotomy for 86 children in an effort to establish relations among immittance, otoscopy and myringotomy, obtained prior to anesthesia. Methods Subjects
The 86 subjects ranged in age from 6 months to 9 years, 1 month; 59 were males and 27 were females. The mean age for both females and males was 31 months. The subjects were seen at Children’s Medical Center in Dallas, Texas. All subjects were to undergo myringotomy for the placement of ventilation tubes as treatment for recurrent otitis media or persistent otitis media with effusion. Table II displays demographics of subjects and diagnosis at the time of surgery. Informed consent was obtained, according to the guidelines for the use of human subjects, from the Institutional Review Board at the University of Texas Southwestern Medical School. Procedures
First, each subject was seen by a certified audiologist who performed acoustic immittance using a Maico Screening Immittance Bridge (Model No. 610). Tym-
TABLE
I
Tympnnogram (Maico
classification
using automated
immittance
instrument
*
610)
Type
Description
A
Peak compensated near 0 daPa
immittance
> 0.2 mmhos
A’
Peak compensated immittance between - IO and - 99 daPa
> 0.2 mmhos
A_
Positive peak compensated immittance > 0.2 mmhos at + 50 daPa or greater
A,
Reduced between
C,
Peak compensated immittance between - 100 and - 150 daPa
C,
Peak compensated immittance between - 151 and -200 daPa
C,
Peak compensated immittance negative than - 200 daPa
B
Compensated immittance with no observable change in peak pressure from + 200 to - 400 daPa
mmhos
indicates
millimhos;
daPa,
peak immittance < 0.2 mmhos - 99 and + 49 daPa
more
dekapascals.
panograms were classified as shown in Table I. Second, a pediatric otolaryngologist used a standard Welch Allyn pneumatic otoscope with halogen head and standard squeeze bulb to determine the presence or absence of effusion. The otolaryngologist performed the otoscopic examination without knowledge of the tympanometric
TABLE
II
Subject demographics
No. of Subjects Mean age (yr; mo) Age range
Males
Females
Total
59 2; 7
27 2: 7
86 2; 7
0; 6-9;
1
0; 11-l;
3
0; 6-9:
(yr: mo) Persistent Recurrent
OME a OME ’
9 49
a OME lasting three months or longer. ’ Multiple episodes of OM exceeding 6 in a year. * One chart was unavailable for review of ‘diagnosis
7 20
at time of surgery’.
16 69 *
1
104
screening results. Myringotomy was performed in the operating room using a standard operating microscope. Presence or absence of effusion and type of fluid were noted.
Results
Results will be subdivided and discussed in four sections: first, tympanometry to myringotomy; second, otoscopy to myringotomy; and third, the combined results of otoscopy and tympanometry to myringotomy. Finally, data will be presented to address the question of which of the procedures is more accurate overall when compared to myringotomy. The total number of ears examined with tympanometry, otoscopy, and myringotomy was 172. In nine cases, one ear could not be tested due to cerumen impaction or lack of patient compliance, bringing the number of ears to 163. Of the 163 ears, effusion was present in 115 ears (71%) at myringotomy. Validation of tympanometry to myringotomy
Table III is a display of the findings comparing tympanometry to myringotomy. Acoustic immittance measures identified 26 ears with type A or A’ tympanograms, indicating the absence of effusion. Agreement to findings at myringotomy occurred for 19 ears (73%). Sixty-eight ears, demonstrated type B tympanograms, suggesting the presence of effusion. Sixty-five ears or 96% of these cases were in agreement with myringotomy. There were 69 ears (42%) with ‘transition’, tympanograms (Table III). That is, not clear A or B types but type C,, C,, C,, A+ or A, status. Sixty-two percent of the ears with transition tympanograms revealed effusion when myringotomy was performed, without a clear pattern for C versus A subtypes. Effusion was present in 67% of the C tympanograms: 7 of 11 C,, 4 of 5 C, type, and 19 of 29 C,. Only 54% of the A subtypes manifested effusion (7 of 11 type A+ and 6 of 13 A,).
TABLE
III
Comparison of all tympanogram types with myringotomy
Typeof tympanogram B A, A’ G c2 c3 A+ 4
Totals
Ears with OME
Ears clear
Total ears
Ears with ome
In group OME
Ears clear
In group clear
(%)
(%o)
(o/o)
(o/o)
65 I 7 4 19 7 6
3 19 4 1 10 4 7
68 26 11 5 29 11 13
56.5 6 6 3.4 16.5 6 5.2
95.5 26.9 63.6 80 65.5 63.6 46.1
6.2 39.5 8.3 2 20.8 8.3 14.5
4.4 73 36.3 20 34.4 36.6 53.8
115
48
163
70.5
N/A
29.4
N/A
105 TABLE
IV
Tympanomrtry
sensitir~ity and specificity A, A’ und R typo only Myringotomy yes _____-
B tympanogram
.Myringotomy no COME absent) false-positives 3 ears cell I3
A or A tympanogram
true-negatives I9 ears cell 11 -_I
Formula:
sensitivity
A = k+c
X
100
6.5 -- := 0.902 x 100 = 90% 72
specificity
1) = n+B
X
100
I9 = 0.863 x 100 = 86% 22
Sensitivity is computed by the number of true positives divided by the number of true-positives plus false-negatives multiplied by 100. Using a 2 x 2 matrix the formula is stated as: A
. . .
sensitivtty = ~A + C x 100 Thus, as shown in Table IV for the A, A’ and B tympanograms, Specificity is computed as: D
specificity = -
D+B
sensitivity is 00%.
x 100
Thus, as shown in Table IV for the A, A’ and B tympanograms, specificity is 86%. If one adds the transition type tympanograms into the analysis and considers them negative as suggested by Brostoff and Cantekin 131 then sensitivity is 57% and specificity is 93% (Table V). Validation of otoscopy to myringotom)
Otoscopy and pneumatic otoscopy on the same 163 ears as reported above identified the presence of effusion in 127 of the 163 (77.9%) ears. Criteria for determination of effusion consisted of a combination of three tympanic membrane characteristics: (al condition, (b) color, and cc) mobility. A normal luster, translu-
TABLE
V
Tympanometry sensitivity and specificity for all types Myringotomy yes (OME present)
Myringotomy no COME absent)
B tympanograms
true-positives 65 ears cell A
‘false-positives 3 ears cell B
AANDC tympanograms
false-negatives 50 ears cell C
true-negatives 45 ears cell D
Formula:
sensitivity
g
specificity
A = ~ A+C
x 100
= 0.565 x 100 = 57% D = ~ D+B
x100
cent, silver colored tympanic membrane that moved freely was determined to be effusion absent. Retracted eardrums that had rebound mobility only were considered to have only negative pressure unless that retraction was superior and extreme, then adhesive OM was suspected. A red tympanic membrane with accompanying fever and pain was considered to be characteristic of acute otitis media. Middle ear effusion was determined by eardrums which were opaque, had obscured landmarks or retraction, with immobility to pneumatic otoscopy. Other criteria included the presence of air fluid levels or bubbles in the middle ear space. Each effusion determined ear was classified as purulent, mucoid or serous fluid. Purulent effusion is characterized by a reddened tympanic membrane with yellowish fluid behind the eardrum. Mucoid effusion is characterized by a dull tympanic membrane. Serous effusion is clear to yellowish-clear fluid and can be accompanied by freely moveable bubbles in the middle ear space. Table VI is a display of the findings comparing otoscopic examination to myringotomy. Sensitivity when compared to myringotomy was 93% and specificity was 58%: Combined otoscopy and myringotomy
Sensitivity and specificity were calculated for tympanometry (76 Type A, A’ and B tympanograms only) and otoscopy combined compared to myringotomy (Table VII). This combination resulted in a sensitivity of 98.3% and a specificity of 92.8%. Keep in mind that these values reflect situations in which the otoscopist and the tympanogram were in clear agreement. Sensitivity was calculated using tympanom-
107 TABLE
VI
Otoscopy srnsilicify and specificity Myringotomy ye.s COME present)
Myringotomy no COME absent)
0toscopy yes OME
true-positives 107 ears cell A
false-positives 20 ears cell B
0t0scopy no OME
false-negatives 8 ears cell C
true-negatives 28 ears cell D
Formula:
sensitivity
A = ___ Xl00 .4 + c
107 ~ = 0.930 x IO0 = 93% II5
specificity
D = DiB
x IO0
G = 0.583 x 100 = 58%
TABLE
VII
Combined sensitir’ity and specificity Myringotomy yes (OME present)
Myringotomy no (OME absent)
Otoscopy yes Tympanogram B
true-positives 61 ears cell A
false-positives I ear cell B
Otoscopy no Tympanogram A, A’
false-negatives 1 ear cell C
true-negatives 13 ears cell D
Formula:
sensitivity
;
specificity
A = __ A+C
x100
= 0.983 x 100 = 98.3% D = __ x 100 I4 II + B
13 G = 0.928 x 100 = 92.8%
108 TABLE
VIII
Fisher exact probability test Effusion absent
Effusion present
otoscopy correct
6
4
Tympanometry correct
6
2
etry and otoscopy indicating effusion present. Specificity was calculated tympanometry and otoscopy indicating absence of effusion.
using
Accuracy of tympanometry and otoscopy as compared to myringotomy
A chi-square analysis for two independent samples was performed in order to ascertain whether the sensitivity and/or specificity of tympanometry to myringotomy was significantly better than that of otoscopy to myringotomy. The chi-square analysis did not reveal a significant difference in tympanometry versus otoscopy in sensitivity; but revealed a significant difference in the success of tympanometry versus otoscopy in determining the absence of effusion when one limits the analysis to A, A’ and B tympanogram types. A Fisher’s exact probability test [9] was undertaken to compare the accuracy of tympanometry to myringotomy and otoscopy to myringotomy. Cases where the two procedures disagreed are included in this analysis, in order to determine which procedure was correct more often when there was disagreement. Results presented in the two-by-two matrix (Table VIII>, indicate no significant difference between the two tests. When determinations differed, tympanometry was correct lo/18 times or 56%, while otoscopy was correct 8/18 times or 44% (P > 0.05).
Discussion
The results of this study comparing the sensitivity and specificity of pneumatic otoscopy and a screening immittance instrument to myringotomy have implications for diagnosis of otitis media in pediatric practices. First, the sensitivity of tympanometry, that is the ability of the type B tympanogram to identify the presence of effusion, is high at 90% when ‘transition’ tympanograms are eliminated from the analysis. Even specificity of the screening bridge used here is to be considered good at 86%. For the otoscopic examination, sensitivity was high (93%); however, specificity was only 58%. The data do suggest that when the otoscopist and tympanometric findings are in agreement, sensitivity and specificity both exceed 90%. But what of the situation where there is not agreement between measures? These data were examined three ways. Sensitivity and specificity for both measures were compared as well as the outcome when the two measures disagreed.
Recall that the chi-square analysis did not reveal a significant difference between otoscopy and tympanometry for sensitivity - but findings do suggest that in this study, the accuracy of tympanometry exceeded otoscopy in determining the absence of effusion. Thus, while there was no significant overall difference when the two measures were in disagreement, the data from this study suggest that tympanometry is more accurate in determining specificity than otoscopy. While the otoscopist did not achieve the specificity standard suggested by Stool and Flaherty [lOI, certification was not the purpose of this study. It is probable that the values obtained here reflect those found in many clinical practices. Bluestone and Cantekin report that the two pediatricians who participated in their study prior to pneumatic otoscopy training obtained sensitivity ratings of 96% and 68%~ and specificity ratings of 68% and 64% [2]. The utilization of a screening immittancc instrument may facilitate increased accuracy in the diagnosis of otitis media. especially in the presence of B tympanograms. A limitation of this study perhaps is that we have not treated the two ears of the individual children as paired organs. Le et al. [5] suggest factors that influence effusion (i.e., eustachian tube dysfunction) are usually present for both ears. Rockette and Casselbrant [7] raise the same point and suggest that a child not be considered effusion positive unless both ears are determined to have effusion. If we had followed this suggestion and considered the ears as paired organs. thus counting a child effusion positive only when both ears were determined to have effusion, then our sensitivity would have remained good while our specificity would have improved greatly, The procedure is valid and appropriate for incidence and epidemiology studies of OME in that bilateral effusion is more common than unilateral, however, we believe for the purpose of this study, as with other validation studies [1,6,7], it is appropriate that each ear be considered as an individual datum. Another limitation of our study may be our decision not to include the ‘transition’ tympanograms in the initial sensitivity and specificity analysis. These transition tympanograms also present a conundrum for the practitioner. In actual clinical practice, these tympanogram types are common, and our data suggest that presence and absence of fluid occurs with nearly equal frequency with these transitional tympanograms. Data need to be interpreted in light of the small numbers for some types. Additional study is needed to determine if C tyrnpanograms have a higher incidence of associated effusion than some of the A subtypes. However, the suggestions of Brostoff and Cantekin [3] can bc followed at this time. In the presence of these ‘transition’ tympanograms, the findings on otoscopy can be relied on to determine the diagnosis of OME. Clearly, this advice is strong with an otoscopic determination that fluid is present. But the decision that fluid is absent may be little better than chance without a validated otoscopist whose specificity reaches or exceeds the recommended 80%. Such results argue for the need to include validation procedures in both pediatric and otolaryngologic training programs and perhaps even to encourage physicians in clinical practice to consider validation as part of continuing education.
110
Acknowledgements
We acknowledge the contribution of Catherine Nolan Balay, M.S., and Karen Clinton Brown, M.S., for assistance with data collection. Some subjects in this study were patients at Children’s Medical Center at Dallas, Texas, while others were taken from the Dallas Cooperative Project on Early Hearing and Language Development. This study was funded in part by NIH NINCDS grant No. 1 R01 19675 and the American Hearing Research Foundation, Chicago, Illinois.
References 1 Babonis, T.R., Weir, M.R. and Kelly, P.C., Impedance tympanometry and acoustic reflectometry at myringotomy, Pediatrics, 87 (1991) 475-480. 2 Bluestone, C.D. and Cantekin, E.I., Design factors in the characterization and identification of otitis media and certain related conditions, Ann. Otol. Rhinol. Laryngol., 88, Suppl. (1979) 13-28. 3 Brostoff, L. and Cantekin, E.I., Observer invariant diagnosis of middle ear effusions. In: D. Lim, C.O. Bluestone, J.O. Klein, J.O. Nelson (Eds.), Recent Advances in Otitis Media with Effusion: Proceedings of the Fourth International Symposium, B.C. Decker, Philadelphia, 1988. 4 Cantekin, E.I., Bluestone, CD., Fria, T.J., Stool, S.E., Berry, Q.C. and Sabo, D.L., Identification of otitis media with effusion in children, Ann. Otol. Rhino]. Laryngol., 89, suppl. 68 (1980) 190-195. 5 Le, C.T., Freeman, D.W. and Fireman, B.H., Evaluation of ventilating tubes and myringotomy in the treatment of recurrent or persistent otitis media, Pediatr. Infect. Dis. J., 10 (1991) 2-11. 6 Moller, H. and Tos, M., Point and period prevalence of otitis media with effusion evaluated by daily tympanometry, J. Laryngol. Otol., 104 (1990) 937-941. 7 Rockette, H.E. and Casselbrant, M., Methodologic issues in screening for otitis media. In D. Lim, C. Bluestone, J.O. Klein and J. Nelson (Eds.), B.C. Decker, Philadelphia, 1988, pp. 42-44. 8 Roland, P.S., Finitzo, T., Friel-Patti, S., Brown, K.C., Stephens, K.T., Brown, 0. and Coleman, J.M., Otitis media: incidence, duration and hearing status, Arch. Otolaryngol. Head Neck Surg., 115 (1989) 1049-1053. 9 Siegel, S., The case of two independent samples. In S. Siegel (ed.), Non-parametric Statistics: For The Behavioral Sciences, Chapter 6, McGraw-Hill Book Company, New York, (1956) 96-104. 10 Stool, S.E. and Flaherty, M.R., Validation of diagnosis of otitis media with effusion, Ann. Otol. Rhinol. Laryngol., 92, Suppl. 107 (1983) 5-6. 11 Teele, D.W., Klein, J.O. and Rosner, B.A., Otitis media with effusion during the first three years of life and development of speech and language, Pediatrics, 74 (1984) 282-287.
