Clinical Focus

Grand Rounds

The Use of Tympanometry and Pneumatic Otoscopy for Predicting Middle Ear Disease
Paula K. Harris
Midwest Ear, Nose and Throat Clinic, Herrin, IL

Kathleen M. Hutchinson
Miami University, Oxford, OH

Joseph Moravec
Joseph J. Moravec, Inc., Facial Surgery Center, Cincinnati, OH

Purpose: Otitis media is the most common condition diagnosed by pediatricians and is estimated to affect approximately 70% of the pediatric population. The goal of this study was to evaluate the effectiveness of otoscopy and multifrequency tympanometry (MFT ) for diagnosis of otitis media in children. Method: Twenty-one children, age 1 to 10 years, who were seeking medical treatment for suspected middle ear disease were selected to participate. Data were collected prior to myringotomy to determine the sensitivity and specificity rates of the following otologic and audiologic measures: (a) pneumatic otoscopy, (b) conventional tympanometry, and (c) MFT. For this study, the gold standard, myringotomy, was used along with pneumatic otoscopy to determine the effectiveness, sensitivity, and specificity of conventional

226-Hz tympanometry, 678-Hz tympanometry, and 1000-Hz tympanometry to predict middle ear disease. Results: The diagnoses provided with pneumatic otoscopy and tympanometry were both similar, agreeing in diagnosis 80%100% of the time. The diagnoses from 678-Hz and 1000-Hz tympanometry were nearly equal and proved to detect abnormality at a higher rate. Conclusions: MFT is recommended on a routine basis with children having a history of otitis media, or else abnormal or notched 226-Hz tympanograms. Further research with a larger sample size will illuminate the possible predictive potential of MFT in otitis media. Key Words: multifrequency tympanometry, conventional tympanometry, pneumatic otoscopy

titis media is the most common condition diagnosed by pediatricians and is estimated to affect approximately 70% of the pediatric population (Parthasarathy, Bhat, & Malur, 2003). Gelfand (2001) describes otitis media as an inflammation of the middle ear that leads to a production of fluid. Presence of fluid can result in a slight to moderate hearing loss. The impairment of hearing resulting from otitis media occurs during the years most crucial for speech and language development. Research has shown that hearing impairment during the critical language development years can be associated with deficits affecting auditory processing, language development, and cognitive and

academic skills (Brown, 1994; Downs, 1995; Friel-Patti & Finitzo, 1990; Gravel & Wallace, 1992; S. Jerger, Jerger, Alford, & Abrams, 1983; Menjuk, 1992). Otitis media can take on several forms depending on the development and course of the disease, whether a bacterial infection is present, the nature of the fluid in the middle ear space, and the kinds of complications that occur (Gelfand, 2001). Otitis media is further defined by the developmental course of the disease. Northern and Downs (2002) describe two main types of otitis media in relation to disease development and severity. Acute otitis media (AOM ) occurs when microorganisms are present and has 3

American Journal of Audiology  Vol. 14  313  June 2005  A American Speech-Language-Hearing Association 1059-0889/05/1401-0003

a sudden onset accompanied by severe ear pain and fever. Chronic otitis media (COM ) is a more severe stage of infection and occurs concurrently with an infection in the middle ear cavity and the mastoid space; it can also be accompanied by a perforation of the tympanic membrane with discharge. Symptoms of otitis media can include severe ear pain and fever (Northern & Downs, 2002). Parents may not relate these symptoms to otitis media because symptoms can occur with other childhood illnesses. Signs of otitis media include crying, irritability, ear drainage, tugging of the ears, and lack of response to environmental sounds or peoples voices. It is also possible for a child with otitis media to be asymptomatic. For this reason, otitis media is sometimes referred to as a silent problem. Pneumatic otoscopy is used by many physicians to detect middle ear effusion. The purpose of pneumatic otoscopy is to determine the mobility and appearance of the tympanic membrane. Toner and Mains (1990) assessed the sensitivity and specificity of pneumatic otoscopy in predicting the presence of middle ear effusion by comparing outcomes to myringotomy results. Sensitivity refers to the proportion of people with the disease who have a positive test result. Specificity refers to the proportion of people without the disease who have a negative test result. The results indicated that pneumatic otoscopy had a sensitivity of 87% and specificity of 89%. When a test has a high sensitivity rating applied to the finding, a negative result rules out the disease. Finitzo, Friel-Patti, Chinn, and Brown (1992) and Vaughn-Jones and Mills (1992) also found similar percentages in sensitivity and specificity with pneumatic otoscopy. Several other studies involving general practitioners have shown that physical findings used to predict AOM, such as redness or tympanic membrane retraction, have poor sensitivity and specificity (Jensen & Lous, 1999; Johansen, Lildholdt, Damsbo, & Eriksen, 2000). Simple otoscopy has been shown to be successful in providing an accurate diagnosis when the signs of effusion are obvious, such as the presence of fluid-filled bubbles behind the tympanic membrane (Sassen, van Aarem, & Grote, 1994). Pneumatic otoscopy was used in this study because it has proved to be more successful and, at times, as accurate as tympanometry in predicting middle ear disease when performed by a skilled and experienced user with the proper equipment (Finitzo et al., 1992). Even when investigators found high sensitivity and specificity percentages, it was recommended that an audiologic follow-up with tympanometry be completed to confirm otoscopic results and define the middle ear condition (Gates, Avery, Hearne, & Holt, 1986; Sassen et al., 1994). Two types of tympanometry routinely used in clinical practice are conventional and multifrequency tympanometry (MFT ). Conventional tympanometry involves the use of one low-frequency probe tone (226 Hz) and the measurement of one component, generally the acoustic admittance vector. Palmu, Puhakka, Rahko, and Takala (1999) investigated the diagnostic value of conventional tympanometry. Fifty eight infants, ranging between the ages of 2 and 11 months, were evaluated with tympanometry and pneumatic otoscopy performed by a physician before a myringotomy. The results 4 American Journal of Audiology  Vol. 14  313  June 2005

showed that tympanometry was successful in determining middle ear disease in 94% of the cases with a sensitivity of 70%; the specificity was 98% with a positive predictive value of 93% and a negative predictive value of 94%. (The positive predictive value is how many of the test-positives truly have the disease. The negative predictive value is how many of the test-negatives in fact do not have the disease.) Watters, Jones, and Freeland (1996) also evaluated children using conventional tympanometry before myringotomy and found a high sensitivity of 91% for predicting middle ear effusion and a specificity of 79%. Both studies support the use of conventional tympanometry for ensuring the correct diagnosis of otitis media in children. Unfortunately, the diagnostic use of conventional Type As and Type C tympanograms has been limited because there is substantial overlap in the range of values recorded from normal and diseased ears (Shanks & Shelton, 1991). Negative tympanometric peak pressure does not adequately separate normal ears from ears with middle ear fluid (Nozza, Bluestone, Kardatzke, & Bachman, 1994). Tympanometric pressure varies widely in children with normal middle ear systems. Lildholdt (1980) found over 25% of children with normal ears having values of 250 daPa during some seasons. Similarly, the range of static compliance values found with various types of disorders overlaps the normal range. In the present investigation, MFT refers to the use of probe frequencies higher than 226 Hz. The higher frequency probe tone is usually 678 Hz or 1000 Hz (Fowler & Shanks, 2002). Research has shown that MFT is a useful clinical tool used to detect pathologies such as otosclerosis, ossicular chain disruption, tympanic membrane abnormalities, and otitis media with effusion (Ferekidis, Vlachou, Douniadakis, Apostolopoulos, & Adamopoulos, 1999). MFT typically records and graphs two separate tympanograms for susceptance (B) and conductance (G) instead of a single admittance (Y ) tympanogram (Gelfand, 2001). Vanhuyse, Creten, and Van Camp (1975) identified four normal types of B-G tympanogram types named for the amount of B and G peaks. Normal tympanograms are labeled 1B1G, 3B1G, 3B3G, and 5B3G (see Figure 1) on the basis of their shapes and configurations. They are named on the basis of the number of positive and negative peaks and must also meet a criterion for tympanogram width (see Table 1). Normal distance between the outermost peaks should be 75 daPa for 3B3G tympanograms and 100 daPa for 5B3G tympanograms to be considered within standard limits (Gelfand, 2001). Gelfand also reports that the width for the G tympanogram will tend to be narrower than for the B tympanogram. Figure 2 shows examples of abnormal MFT results with violations of criteria from Table 1. Vlachou, Ferekidis, Tsakanikos, Apostolopoulos, and Adamopoulos (1999) compared the results of conventional and MFT in 86 children affected by AOM. Tympanometry measures were taken the day of initial evaluation for middle ear disease and again 3 months later. Vlachou et al. categorized conventional tympanograms as normal with either Type A or C for determination of otitis media resolution. Clinical diagnosis was also based on

Figure 1. VanHuyse classification system. Based on Gelfand (2001).

Table 1. 678-Hz and 1000-Hz Vanhuyse tympanogram classification criteria. Normal types of tympanograms 1B1G 3B1G 3B3G 5B3G

Criteria for normal shape 1. The number of conductance peaks must not exceed three, and the number of susceptance peaks must not exceed five. 2. The distance between the outermost conductance maxima must not exceed the distance between the susceptance maxima. 3. The distance between the outermost maxima must not exceed 75 daPa for 3B3G tympanograms shapes and 100 daPa for 5B3G tympanogram shapes.

Note. Based on Gelfand (2001).

otalgia, fever, irritability, and otoscopy. The comparison revealed that many children who experienced recurrent otitis media or persistence of middle ear effusion during the follow-up appointment had normal compliance (Type A

or C ), indicating the resolution of middle ear effusion. MFT revealed abnormal or borderline tympanograms, indicating that the disease had not completely resolved. These findings suggested that MFT is able to detect subtle
Harris et al.: Use of Tympanometry and Pneumatic Otoscopy

Figure 2. Abnormal MFT examples for Table 1 criterion.

changes in the middle ear system following otitis media that conventional tympanometry was unable to detect, indicating the persistence of a pathology. Middle ear mechanics are changed in some children who have recovered from COM, and the alteration in mechanics may not be reflected in the conventional 226-Hz tympanogram (Hanks & Robinette, 1993). Consensus has not been reached among physicians with regard to diagnosis of otitis media. This divergence is demonstrated in different inclusion criteria used in clinical trials and by the wide range of sensitivity and specificity value performance (Stool et al., 1994). Research traditionally focuses on the presence or absence of fluid at myringotomy as the gold standard for determination of otitis media ( Marchant et al., 1986). Both tympanometry and otoscopy are widely used diagnostic procedures for identification of middle ear effusion in children and adults. Yet neither method is a gold standard for identification 6 American Journal of Audiology  Vol. 14  313  June 2005

of fluid. The diagnostic efficiency of tympanometry to establish the presence or absence of middle ear effusion in conjunction with the use of myringotomy findings has been well documented (Bluestone & Cantekin, 1975; Cantekin et al., 1980; Fria, Cantekin, & Probst, 1980; Palmu et al., 1999; Paradise, Smith, & Bluestone, 1976; Shurin, Pelton, & Finkelstein, 1977; Watters et al., 1996). Although these studies support myringotomy to be the gold standard, previous work also supports the clinical use of tympanometry without myringotomy for providing an objective measure of the presence or absence of fluid in the middle ear. Unlike myringotomy, tympanometry can be performed on all patients in the clinical and educational setting to determine the health of the middle ear (Marchant et al., 1986). Myringotomy is an invasive, expensive procedure that can only be justified in patients with specific pathologies, such as suspected prolonged middle ear effusion or recurrent otitis media.

In the present study, the sensitivity and specificity of pneumatic otoscopy, conventional tympanometry, and MFT were evaluated using myringotomy as the gold standard. True sensitivity and specificity values were established by a surgeon who was unaware of tympanometry test results. The purpose of this study was to clarify whether MFT results offer data that could be related to the diagnosis of otitis media, so as to contribute to an early prognosis and intervention.

Following the pneumatic otoscopy and tympanometry measurements, the physician separately performed the myringotomy and placement of the PE tubes. He reported the presence or absence of fluid in the middle ear and the type of fluid, if any, that was present. The presence of fluid was classified as absent, minimal fluid, or much fluid. The results of the pneumatic otoscopy and tympanometry were then compared with the surgical findings of the middle ear status and analyzed statistically.

Method
Participants
A total of 21 new patients being seen by an otolaryngologist participated in this study. The participants consisted of 13 boys and 8 girls, age 1 to 10 years, who were seeking medical treatment for suspected middle ear disease.

Statistical Analysis
Statistical analyses were performed using SPSS (Version 10.0). Comparisons were made between the otoscopic diagnosis, tympanometry, and myringotomy results. An alpha level of .05 was used, and contingency tables were produced to illustrate the comparisons between the testing procedures. Based on the contingency tables, chi-square tests of independence were performed for each of the variables involved in this study. From the comparisons of test procedures, the number of agreements and disagreements were computed and percentage of agreement was calculated. The ability to predict myringotomy results, by using measurements from pneumatic otoscopy and tympanometry as possible predictor variables, was analyzed using a logistic regression model in which presence of fluid was the dependent variable and each of the test procedure predictors were covariates. Appropriate p values, point estimates, and confidence intervals for odds ratio were determined.

Procedure
The otologic and audiologic measurements used in this project consisted of pneumatic otoscopy, conventional tympanometry, and MFT. All were performed on each participant bilaterally before a myringotomy and placement of pressure equalization (PE ) tubes. The physician examined the mobility and appearance of the tympanic membrane with a pneumatic otoscope for each participant. The physician then classified mobility as normal mobility, decreased mobility, or no mobility, and appearance was classified as normal, thickened or inflamed, retracted, or perforated. Both the conventional and MFT measurements were obtained using a Grason-Stadler-33 Middle Ear Analyzer. Three separate tympanometry measurements were obtained on each ear of each patient by using a low-frequency 226-Hz probe tone and two high-frequency (678 Hz and 1000 Hz) probe tones. The Jerger classification system (J. Jerger, 1970) was used to identify the conventional low-tone tympanograms as either Type A, As, B, or C. Table 2 presents the criteria system for classifying the different 226-Hz tympanogram types; Figure 3 illustrates the configurations. The Vanhuyse classification system (Vanhuyse et al., 1975) was used to categorize the shape of the high-frequency tympanograms. The criteria system used for classifying the high-frequency 678-Hz and 1000-Hz tympanograms is presented in Table 1 and in Figure 1.

Results
A total of 21 participants were examined using pneumatic otoscopy to determine mobility and overall appearance of the tympanic membrane. Tympanograms and myringotomy were successfully completed and recorded for statistical analysis on 35 of the 42 ears. Five examinations were excluded due to the presence of patent PE tubes and occluding cerumen. Two false-negative examinations were excluded due to the suggested effects of nitrous oxide on the middle ear during anesthesia. The child in this case resisted preoperative tympanometric examination, and tympanometric findings were obtained on each ear under general anesthesia. The operating surgeon found mucoid

Table 2. 226-Hz tympanogram classification criteria. Type A As B C Criteria Static compliance: 0.351.25 ml Peak pressure: 100 to +100 daPa Static compliance: <0.35 Peak pressure: 100 to +100 daPa Flat curve, no distinct peak Static compliance: 0.351.25 ml Peak pressure: > 100 Clinical interpretation in the study Normal middle ear Decreased static compliance, no fluid Fluid in middle ear Negative pressure, possible fluid in the middle ear

Note. Based on Jerger (1970) and Silman, Silverman, & Arick (1992).

Harris et al.: Use of Tympanometry and Pneumatic Otoscopy

Figure 3. Jerger classification system for low-frequency tympanograms. Based on Jerger (1970).

effusion in each ear. Research suggests the use of nitrous oxide during the anesthesia process causes effusion to be displaced into the eustachian tube, increases middle ear pressure, and disrupts tympanometry outcomes (Rees & Freeland, 1992). Gates and Cooper (1980) also found rising middle ear pressure during gas anesthesia but concluded that it was due to assisted mask ventilation. Analysis of pneumatic otoscopy, tympanometry, and myringotomy was restricted to the 35 successfully examined ears.

Association Between Myringotomy and Tympanometry

The results of statistical analyses between myringotomy and conventional low-tone (226 Hz) tympanograms are represented in Table 3. The table shows the predictive values in relation to low-tone 226-Hz tympanograms and middle ear fluid. Data (c2 = 12.182, p = .007) indicate that Type A tympanograms accurately predicted the absence of fluid in 70% (7 of 10) of the cases and that a Type B tympanogram accurately predicted the presence of fluid in 8 American Journal of Audiology  Vol. 14  313  June 2005

100% (12 of 12) of the cases. The results for Type As and C tympanograms revealed approximately a 50% predictive value for determining middle ear fluid. In terms of detecting effusion, 226-Hz tympanometry had a sensitivity of 80% and a specificity of 100%. A logistic regression analysis was completed using Type A as the normal category and Type As, Type B, and Type C collapsed together as the abnormal category. The results revealed that there was not enough evidence to accurately predict the presence of fluid ( p = .08); the point estimate for the odds ratio was 8.7 with a 95% confidence interval (0.62, 540.2). Table 4 shows the agreement between myringotomy results and 678-Hz tympanometry (c2 = 11.26, p = .002). The results revealed an 87.5% (7 of 8) accuracy rating for 678-Hz tympanograms to predict the absence of middle ear fluid and a 77.8% (21 of 27) accuracy in predicting the presence of fluid in the middle ear. The results for 678-Hz tympanometry revealed a sensitivity of 95% to detect middle ear effusion and a 54% specificity. The results of a logistic regression analysis revealed that the ability of 678-Hz tympanometry to predict the presence

Table 3. Predictive values of 226-Hz tympanogram results based on myringotomy results. Myringotomy results Absence of fluid 226-Hz tympanogram results Normal (absence of fluid) Abnormal (presence of fluid) Type Type Type Type A As B C n 7 1 0 5 % 70 50 0 45.5 Presence of fluid n 3 1 12 6 % 30 50 100 54.5 n 10 2 12 11 Total number % 100 100 100 100

Note. c2 = 12.182; p = .007; sensitivity = 80%; specificity = 100%.

of fluid was significant ( p = .009). The point estimate for the odds ratio is 10.96 with 95% confidence interval (1.16, 8), which indicates a positive association between the presence of fluid and abnormal 678-Hz tympanometry. Table 5 shows the relationship between myringotomy and tympanometry using the 1000-Hz probe tone. These results revealed a 100% (7 of 7) accuracy for 1000-Hz tympanometry to predict the absence of middle ear fluid and a 78.6% (22 of 28) accuracy in predicting the presence of fluid in the middle ear. The results for 1000-Hz tympanometry revealed a sensitivity of 100% to detect middle ear effusion and a 54% specificity. The results of a logistic regression analysis revealed that the ability of 1000-Hz tympanometry to predict the presence of fluid was significant ( p = .009). The point estimate for the odds ratio is 10.96 with 95% confidence interval (1.16, 8), which indicates a positive association between the presence of fluid and 1000-Hz tympanometry.

relationship between pneumatic otoscopy results and the health status of the middle ear as revealed by myringotomy. The results of a logistic regression analysis for pneumatic otoscopy revealed that the procedure was significant in predicting the presence of fluid ( p = .0005). The point estimate for the odds ratio is 27.5 with a confidence interval of 95% (3.015, 8), which suggests that the odds of finding fluid given an abnormal otoscopy are greater than finding fluid given a normal result. These results indicate a positive association between fluid presence and abnormal otoscopy.

Association Between Conventional Tympanometry and MFT

Research findings have suggested that MFT using highfrequency probe tones is able to detect subtle changes in the middle ear system following otitis media that conventional tympanometry was unable to detect, indicating the persistence of pathology (Vlachou et al., 1999). Although the participants in this study were being treated for the presence of otitis media, and not follow-up after treatment, a statistical analysis was performed to determine any correlations present between the different tympanometry types used in the study (226 Hz, 678 Hz, and 1000 Hz). The results are represented in Tables 7 and 8. The results shown in Table 7 revealed that when a Type A (normal ) tympanogram was recorded with 226-Hz tympanometry, a normal morphology was recorded 80% (8 of 10) of the time with 678-Hz tympanometry (c2 = 25.92, p = .000). When a Type As, B, or C tympanogram was recorded with 226-Hz tympanometry, an abnormal morphology was recorded with 100% (25 of 25) accuracy using 678-Hz tympanometry. This analysis revealed a strong

Association Between Pneumatic Otoscopy and Myringotomy

Pneumatic otoscopy measures were used to predict tympanic membrane mobility compared with myringotomy results. The results revealed that when the tympanic membrane was categorized as having normal mobility there was an 81.8% (9 of 11) accuracy in predicting the absence of fluid in the middle ear (see Table 6). When the tympanic membrane was considered to have decreased mobility there was a 69.2% (9 of 13) accuracy in predicting the presence of fluid in the middle ear (c2 = 14.8, p = .000). When the tympanic membrane was considered to have an absence of mobility, the accuracy in predicting fluid in the middle ear was 100% (11 of 11). There appears to be a strong

Table 4. Predictive values for 678-Hz tympanometry results based on myringotomy results. Myringotomy results Absence of fluid 678-Hz tympanogram results Normal (absence of fluid) Abnormal (presence of fluid) n 7 6 % 87.5 22.2 n 1 21 Presence of fluid % 12.5 77.8 n 8 27 Total % 100 100

Note. c2 = 11.264; p = .002; sensitivity = 95%; specificity = 54%. Harris et al.: Use of Tympanometry and Pneumatic Otoscopy

Table 5. Predictive values for 1000-Hz tympanometry results based on myringotomy results. Myringotomy results Absence of fluid 1000-Hz tympanogram results Normal (absence of fluid) Abnormal (presence of fluid) n 7 6 % 100 21.4 n 0 22 Presence of fluid % 0 78.6 n 7 28 Total % 100 100

Note. c2 =14.808; p = .000; sensitivity = 100%; specificity = 54%.

Table 6. Predictive values for pneumatic otoscopy results based on myringotomy results. Pneumatic otoscopy results Absence of fluid Mobility of tympanic membrane results Mobile Decreased mobility No mobility n 9 4 0 % 81.8 30.8 0 n 2 9 11 Presence of fluid % 18.2 69.2 100 n 11 13 11 Total % 100 100 100

Note. c2 = 16.130; p = .000; sensitivity = 95%; specificity = 54%.

Table 7. Predictive values of 226-Hz tympanogram results based on 678-Hz tympanometry results. Normal 678-Hz tympanogram results 226-Hz tympanogram results Type Type Type Type A As B C n 8 0 0 0 % 80 0 0 0 Abnormal 678-Hz tympanogram results n 2 2 12 11 % 20 100 100 100 n 10 2 12 11 Total % 100 100 100 100

Note. c2 = 25.926; p = .000.

Table 8. Predictive values of 226-Hz tympanogram results based on 678-Hz tympanometry results. Normal 1000-Hz tympanogram 226-Hz tympanograms Type Type Type Type A As B C n 7 0 0 0 % 70 0 0 0 Abnormal 1000-Hz tympanogram n 3 2 12 11 % 30 100 100 100 n 10 2 12 11 Total % 100 100 100 100

Note. c2 = 21.875; p = .000.

agreement between 226-Hz tympanometry and 678-Hz tympanometry in predicting the presence or absence of fluid in the middle ear. The results shown in Table 8 revealed that when a Type A (normal ) tympanogram was recorded with 226-Hz tympanometry, a normal morphology was recorded 70% (7 of 10) of the time with 1000-Hz tympanometry (c2 = 21.87, p = .000). When a Type As, B, or C tympanogram 10 American Journal of Audiology  Vol. 14  313  June 2005

was recorded with 226-Hz tympanometry, an abnormal morphology was recorded with 100% (25 of 25) accuracy with 1000-Hz tympanometry. This analysis also revealed a strong agreement between 226-Hz tympanometry and 1000-Hz tympanometry in predicting the presence or absence of fluid in the middle ear. Table 9 illustrates that the predictive diagnoses between 678-Hz and 1000-Hz tympanometry were very similar (c2 = 29.53, p = .000).

Table 9. Predictive values of 1000-Hz tympanometry results based on 678-Hz tympanometry results. Normal 678-Hz tympanogram results 1000-Hz tympanogram results Normal 1000-Hz tympanogram Abnormal 1000-Hz tympanogram Note. c2 = 29.531; p = .000. n 7 1 % 100 3.6 Abnormal 678-Hz tympanogram results n 0 27 % 0 96.4 n 7 28 Total % 100 100

In summary, for the cases identified as abnormal by low-frequency tympanograms, the low-frequency results and the MFT results were in 100% agreement. However, low-frequency tympanometry identified some cases as normal that were classified as abnormal by MFT.

Discussion
This study demonstrated good agreement among pneumatic otoscopy, conventional tympanometry, MFT, and myringotomy results. In general, an abnormal lowfrequency tympanogram would also be detected with MFT. However, a normal low-frequency tympanogram in some ears might result in an abnormal MFT. We found pneumatic otoscopy to be effective for predicting the presence of fluid in the middle ear. The results revealed an 81.8% (9 of 11) accuracy rating to predict the absence of fluid in the middle ear and a 100% (11 of 11) accuracy rating to predict the presence of fluid in the middle ear. The sensitivity (84.6%) and the specificity (100%) were also found to be consistent with the findings of previous research studies that documented the sensitivity of pneumatic otoscopy to be between 87% and 93% and the specificity to be between 58% and 91.4% (Finitzo et al., 1992; Toner & Mains, 1990; Vaughn-Jones & Mills, 1992). Unlike other studies of diagnostic indices, this analysis established the presence or absence of fluid by myringotomy. The results of the present study demonstrate that pneumatic otoscopy is a good predictor of fluid presence, consistent with previous research supporting the usefulness of pneumatic otoscopy as a diagnostic tool. The use of 226-Hz tympanometry was also found to be a successful predictor of presence or absence of fluid in the middle ear. When 226-Hz Type A and Type B tympanograms were analyzed independently of Type As and Type C tympanograms, the results revealed a sensitivity rating of 80% and a specificity rating of 100% for Type A and Type B tympanograms to predict middle ear fluid. These findings are found to be comparable with prior research studies illustrating the sensitivity of Type A and Type B tympanograms to be between 80.2% and 90%, and the specificity of Type A and Type B tympanograms to be 71% to 98.8% (Finitzo et al., 1992; Mills, 1987; Toner & Mains, 1990; Vaughn-Jones & Mills, 1992). The results of the 226-Hz tympanometry were highly consistent with the results of the 678- and 1000-Hz tympanometry for determining fluid presence. However, low-frequency tympanometry identified 3 of 10 cases as

normal that were identified as abnormal by 678-Hz results; these same 3 normal cases were identified as abnormal by 1000-Hz results. It is possible that recordings of abnormal MFT results after COM indicate a delayed recovery of the middle ear system or some permanent changes not demonstrated by low probe tone tympanometry. MFT was found to have a high predictive value for determining the absence of fluid in the middle ear. The results show that MFT was successful in determining normal ears 87.5% of the time with 678 Hz and in 100% of the cases with 1000 Hz. The lower positive predictive value for determination of fluid presence of 77.8% for 678-Hz tympanometry and 78.6% for 1000-Hz tympanometry could be due to the fact that not all abnormal 678-Hz and 1000-Hz tympanogram configurations indicate fluid in the middle ear. Other middle ear pathologies, such as scarred or flaccid eardrums, can be associated with the abnormal morphology. MFT recordings could indicate persistence of changes in the mass and stiffness balance of the middle ear not demonstrated by conventional tympanometry. Therefore it is not surprising that we documented a sensitivity of 100% and a specificity rate of 54% for 1000-Hz tympanometry, and a sensitivity of 95% and a specificity of 54% for 678-Hz tympanometry for determination of fluid presence. Comparison of sensitivity and specificity rates of our results with past research findings is difficult. No data is available for evaluating the sensitivity and specificity of MFT using the same recording methods (speed of recording, recording direction, instrument sensitivity and pressure for ear canal estimation), participant population, and diagnostic indices compared with myringotomy outcomes. The success of a diagnostic protocol for middle ear effusion is affected by the validity of the eventual diagnosis. In our study, a conclusive diagnosis of middle ear fluid depended on whether effusion was actually found surgically, by myringotomy, a true gold standard for effusion. All of our test results were judged by whether they were validated by myringotomy outcomes within the same participant group. Pneumatic otoscopy was found to be a significant predictor for fluid presence. Otoscopic errors are minimized with pneumatic versus simple otoscopy (Toner & Mains, 1990). Likewise, strengths of conventional tympanometry versus MFT protocols were examined in conjunction with myringotomy results. Cases identified as abnormal by conventional tympanometry were also detected by MFT findings. However, low-frequency tympanometry identified some cases as normal that were classified abnormal by MFT. Conversely, MFT results better identified normal
Harris et al.: Use of Tympanometry and Pneumatic Otoscopy

11

middle ear function. When jointly testing MFT procedures, it was found that 678 Hz and 1000 Hz produced similar predictions of fluid presence. The logistic regression model comparing 678-Hz and 1000-Hz tympanometry indicates that the two procedures are very consistent with each other (see Table 9). The strong relationship between the results of the two high-frequency probes suggests that the use of more than one high-frequency probe tone would be redundant in a clinical setting. Research data collected for high-frequency tympanometry demonstrates a valid and reliable immittance test protocol for infants (Purdy & Williams, 2002). Tympanometric shapes seen in infants are different from those seen in adults. If, for instance, infants under the age of 1 year were included in our study, there may have been more differences between the results obtained from the MFT and low-frequency 226-Hz tympanometry. Several reports conclude that 226-Hz tympanogram shapes seen in infants are different than those seen in older children and adults (Keefe & Levi, 1996; Meyer, Jardine, & Deverson, 1997). Future work should evaluate infants under the age of 1 year using a protocol similar to that used in the present study. Future studies with a larger sample size would be beneficial. A larger sample size could allow the implementation of generalized estimating equations, which would permit data on both ears to be incorporated into the study as well as eliminate the need to collapse the data into two categories (normal and abnormal ). A larger sample size would also increase the statistical power of the results. Further research including greater numbers of effusion and noneffusion ears and follow-up myringotomy to identify middle ear fluid is required to validate the proposed protocol. The findings of the current study showed that tympanometry using probe tones of 226, 678, and 1000 Hz are sensitive audiologic measurements for detecting middle ear effusion, and that the results of the higher probe tones produced results that were highly consistent. We recommend a protocol incorporating both low- and high-frequency probe tone tympanometry. Our study showed that the results for Type As and C tympanograms revealed approximately a 50% predictive value for determining middle ear fluid; these tympanograms can be associated with middle ear fluid but are also associated with normal middle ear function. Some children with normal low-frequency tympanograms were found to have abnormal MFT results, perhaps as a result of monomeric eardrums or permanent mechanical disturbances of the middle ear. Based on these results and myringotomy outcomes, clinicians should use MFT with selected children having a history of otitis media, an airbone gap with a Type A tympanogram, or an abnormal 226-Hz tympanogram; MFT may further define Type A, As, and Type B designations when small or rounded peaks occur. A flat 226-Hz tympanogram is strong evidence for otitis media with effusion, but some high-impedance pathologies may be more evident using MFT. In various stages of COM, impedance may be altered by increased mass due to thick effusion in contact with the ossicles, or decreased compliance due to reduction of air volume of the mastoid and middle ear space. If stiffening effects seen 12 American Journal of Audiology  Vol. 14  313  June 2005

on a tympanogram ( Type As) are accompanied by a lowfrequency conductive component, there is additional support to predict a stiffness-producing middle ear disease. However, if an increase in the mass component on the MFT is accompanied by a high-frequency conductive loss, there is additional support for a mass-producing middle ear pathology. We conclude that MFT and conventional tympanometry provide improved diagnostic methods for otitis media in children. Conclusions are based on agreement between tympanometry, otoscopy, and myringotomy measures. MFT seems to provide more detailed information about the mechanical and acoustic status of the middle ear than 226-Hz admittance tympanometry. Further research in this area will illuminate the possible predictive potential of MFT in otitis media.

Acknowledgments
Thanks to Bryan R. Bell and Jeremy L. Craft for statistics consultation.

References
Bluestone, C. D., & Cantekin, E. I. (1975). Design factors in characterization and identification of otitis media and certain related conditions. Annals of Otology, Rhinology, and Laryngology Supplement, 88(5), 1328. Brown, D. P. (1994). Speech recognition in recurrent otitis media: Results in a set of identical twins. Journal of American Academy of Audiology, 5, 16. Cantekin, E. I., Bluestone, C. D., Fria, T. J., Stool, S. E., Berry, O. C., & Sabo, D. L. (1980). Identification of otitis media with effusion in children. Annals of Otology, Rhinology, and Laryngology, 89, 190194. Downs, M. P. (1995). Contribution of mild hearing loss to auditory language learning problems. In R. Roeser (Ed.), Auditory disorders in school children (3rd ed., pp. 188200) New York: Thieme. Ferekidis, E., Vlachou, S., Douniadakis, D., Apostolopoulos, N., & Adamopoulos, G. (1999). Multi-frequency tympanometry in children with acute otitis media. Otolaryngology Head and Neck Surgery, 121(6), 797801. Finitzo, T., Friel-Patti, S., Chinn, K., & Brown, O. (1992). Tympanometry and otoscopy prior to myringotomy: Issues in diagnosis of otitis media. International Journal of Pediatric Otorhinolaryngology, 24, 101110. Fowler, C. G., & Shanks, J. E. (2002). Tympanometry. In J. Katz, R. F. Burkard, & L. Medwetsky (Eds.), Handbook of clinical audiology (5th ed., pp. 175204) Philadelphia: Lippincott Williams & Wilkins. Fria, T. J., Cantekin, E. I., & Probst, G. (1980). Validation of an automatic otoadmittance middle ear analyzer. Annals of Otology, Rhinology, and Laryngology, 89, 253256. Friel-Patti, S., & Finitzo, T. (1990). Language learning in a prospective study of otitis media with effusion in the first two years of life. Journal of Speech and Hearing, 33, 188194. Gates, G. A., Avery, C., Hearne, E. M., & Holt, G. R. (1986). Predictive value of tympanometry in middle ear effusion. Annals of Otology, Rhinology, and Laryngology, 95, 4650. Gates, G. A., & Cooper, J. C. (1980). Effect of anesthetic gases on middle ear pressure and presence of effusion. Annals of Otology, Rhinology, and Laryngology Supplement, 89, 6264.

Gelfand, S. A. (2001). Essentials of audiology (2nd ed.). New York: Thieme. Gravel, J. S., & Wallace, I. F. (1992). Listening and language at 4 years of age: Effects of early otitis media. Journal of Speech and Hearing, 35, 588595. Hanks, W. D., & Robinette, R. (1993). Cost justifications for multifrequency tympanometry. American Journal of Audiology, 2(3), 78. Jensen, P. M., & Lous, J. (1999). Criteria, performance and diagnostic problems in diagnosing acute otitis media. Family Practice, 16, 262268. Jerger, J. (1970). Clinical experience with impedance audiometry. Archives of Otolaryngology, 92(4), 311324. Jerger, S., Jerger, J., Alford, B. R., & Abrams, S. (1983). Development of speech intelligibility in children with recurrent otitis media. Ear and Hearing, 4(3), 138145. Johansen, E. C., Lildholdt, T., Damsbo, N., & Eriksen, E. W. (2000). Tympanometry for diagnosis and treatment of otitis media in general practice. Family Practice, 17, 317322. Keefe, D., & Levi, E. (1996). Maturation of the middle and external ears: Acoustic power-based responses and reflectance tympanometry. Ear and Hearing, 17, 361373. Lildholdt, T. (1980). Negative middle ear pressure. Variations by season and sex. Annals of Otology, Rhinology, and Laryngology, 89(Suppl. 68), 6770. Marchant, C. D., McMillan, P. M., Shurin, P. A., Johnson, C. E., Turczyk, V. A., Feinstein, J. C., & Panek, D. M. (1986). Objective diagnosis of otitis media in early infancy by tympanometry and ipsilateral acoustic reflex thresholds. Journal of Pediatrics, 109, 590595. Menjuk, P. (1992). Relationship of otitis media to speech and language development. In J. Katz, N. Stecker, & D. Henderson (Eds.), Central auditory processing: A transdisciplinary approach (pp. 187198). St. Louis, MO: Mosby. Meyer, S., Jardine, C., & Deverson, W. (1997). Developmental changes in tympanometry. British Journal of Audiology, 31, 189195. Mills, R. P. (1987). Persistent middle ear effusions in children with recurrent acute otitis media. Clinical Otolaryngology, 12, 97101. Northern, J. L., & Downs, M. P. (2002). Hearing in children (5th ed.). Philadelphia: Lippincott Williams & Wilkins. Nozza, R. J., Bluestone, C. D., Kardatzke, D., & Bachman, R. (1994). Identification of middle ear effusion using aural acoustic admittance and otoscopy. Ear and Hearing, 15, 310323. Palmu, A., Puhakka, H., Rahko, T., & Takala, A. (1999). Diagnostic value of tympanometry in infants in clinical practice. International Journal of Pediatric Otorhinolaryngology, 49, 207213. Paradise, J. L., Smith, C. G., & Bluestone, C. D. (1976). Tympanometric detection of middle-ear effusion in infants and young children. Pediatrics, 58(2), 198210. Parthasarathy, T. F., Bhat, V., & Malur, G. J. (2003). Otitis media: A primer on diagnosis and management. The Hearing Journal, 56(6), 4248.

Purdy, S. C., & Williams, M. J. (2002). High frequency tympanometry: A valid and reliable immittance test protocol for young infants. New Zealand Audiological Society Bulletin, 10(3), 924. Rees, G. L., & Freeland, A. P. (1992). The effects of anesthesia on tympanograms of children undergoing grommet insertion. Clinical Otolaryngology, 17, 200202. Sassen, M. L., van Aarem, A., & Grote, J. J. (1994). Validity of tympanometry in the diagnosis of middle ear effusion. Clinical Otolaryngology, 19, 185189. Shanks, J., & Shelton, C. (1991). Basic principles and clinical applications of tympanometry. Otolaryngologic Clinic of North America, 24(2), 299328. Shurin, P. A., Pelton, S. I., & Finkelstein, J. (1977). Tympanometry in the diagnosis of middle-ear effusion. New England Journal of Medicine, 296(8), 412417. Silman, S., Silverman, C. A., & Arick, D. S. (1992). Acousticimmittance screening for detection of middle-ear effusion. Journal of the American Academy of Audiology, 3, 262268. Stool, S. E., Berg, A. D., Berman, S., Carney, C. J., Cooley, J. R., Culpepper, L., et al. (1994). Managing otitis media with effusion in young children: Quick reference guide for clinicians (AHCPR Publication No. 94-0623). Rockville, MD: U.S. Department of Health and Human Services. Toner, J. G., & Mains, B. (1990). Pneumatic otoscopy and tympanometry in the detection of middle ear effusion. Clinical Otolaryngology, 15, 121123. Vaughn-Jones, R., & Mills, R. P. (1992). The Welch Allyn audioscope and microtymp: Their accuracy and that of pneumatic otoscopy, tympanometry, and pure tone audiometry as predictors of otitis media with effusion. Journal of Laryngology and Otology, 106, 600602. Vanhuyse, V. J., Creten, W. L., & Van Camp, K. J. (1975). On the W-notching of tympanograms. Scandinavian Audiology, 4, 4550. Vlachou, S., Ferekidis, F., Tsakanikos, M., Apostolopoulos, N., & Adamopoulos, G. (1999). Prognostic value of multiplefrequency tympanometry in acute otitis media. Journal for Oto-Rhino-Laryngology, 61(4), 195200. Watters, G. W., Jones, J. E., & Freeland, A. P. (1996). The predictive value of tympanometry in the diagnosis of middle ear effusion. Clinical Otolaryngology, 22, 343345. Received August 15, 2004 Revision received January 22, 2005 Accepted April 28, 2005 DOI: 10.1044/1059-0889(2005/002) Contact author: Kathleen M. Hutchinson, 2 Bachelor Hall, Miami University, Oxford, OH 45056. E-mail: hutchik@muohio.edu

Harris et al.: Use of Tympanometry and Pneumatic Otoscopy

13