Variability of Eustachian Tube Function: Comparison of Ears With Retraction Disease and Normal Middle Ears


  • Supported by grants from the Joint Committee North Medical Care Region, the County Council of Norrbotten, Umeå University, the Swedish Medical Research Council (#K99-73X-06578-17B), and the Foundation of Acta Oto-Laryngologica.


Objective To explore the short-term and long-term variability of tubal opening and closing in ears with advanced retractions and in healthy ears.

Study Design/Methods Twenty ears with retraction type middle ear disease (R-MED) and 20 normal ears underwent direct recording of the middle ear pressure during repeated forced openings, equalization of +100 daPa and −100 daPa by swallowing, Valsalva inflation, and forceful sniffing. Tests were performed twice (separated by 30 min) on each of 2 days separated by 3 to 4 months.

Results There was considerable intraindividual variability of the forced opening pressure and the closing pressure in both groups, within as well as between sessions and test days. Although the variability was 1.5 to 2 times higher in ears with retraction than in the normal group, mean Po and Pc did not differ between the groups. Compared with normal ears, ears with retraction changed more frequently from a positive to negative test response, or vice versa, when re-tested after 30 minutes. Rates of positive response in the equalization and Valsalva tests were significantly lower in diseased ears compared with normal ears.

Conclusions Eustachian tube opening and closing functions vary more in ears with retraction disease than in normal ears, which is consistent with the variable clinical course of R-MED and implies that single tubal function tests have little prognostic value on the individual level.


Tubal malfunction is related to different pathological conditions such as retractions of the tympanic membrane and secretory otitis media (SOM). A connection between negative middle ear pressure (Pme) and retraction of the tympanic membrane was early recognized. A causal relationship between tubal obstruction, development of negative Pme with subsequent effusion of fluid in the middle ear, and retraction of the tympanic membrane was postulated—the hydrops-ex-vacuo theory. This theory has been questioned for various reasons. First, anatomical tubal obstruction is rare. Second, the only demonstrated physiological way to create high negative Pme is active evacuation by sniffing, a mechanism that presupposes tubal closing failure. 1 Third, the hydrops-ex-vacuo theory was based on the assumption that gas is continuously absorbed by the middle ear mucosa, but several recent studies have shown that transmucosal gas transport is bidirectional. 2,3 Also, Pme can be maintained at ambient pressure or even be positive without tubal openings in healthy ears. 4,5 These findings support the view that gas exchange over the mucosa is the basic mechanism for regulation of the Pme, whereas the eustachian tube serves as a release valve for excessive positive Pme and can also open by muscular forces to equalize excessive negative Pme.

It is likely that the tubal capacity to open becomes more important in pathological conditions with impaired transmucosal gas exchange. Impaired tubal opening function has been focused as a main problem in retraction pockets, atelectasis, adhesive ears, and in retraction type cholesteatoma—conditions for which the common term retraction type middle ear disease (R-MED) was suggested. 6 Thus tubal opening function has been extensively investigated, whereas tubal closing function has been less studied. However, Magnuson 1 proposed closing failure as a pathogenic factor for the development of tympanic membrane retractions, because habitual sniffing can cause high negative Pme.

Although both the clinical course and experimental findings in R-MED suggest a varying and unstable function, there are few systematic studies on the variability and dynamics of the tubal function in pathological conditions. 7 In studies in children with SOM, pronounced intraindividual variability was reported (Unpublished data). 8–10 In clinical and experimental situations, conclusions are often based on results of a single tubal function test. To better evaluate such results and to find out whether it is possible to establish a normal reference, it is of interest to explore the variability in different pathological conditions as well as in healthy ears.

The aim of the present study was to investigate the variability over time for responses in different tubal function tests in ears with R-MED and in normal ears. Variability was explored within and between tests, for individuals and for groups.



The study comprised 20 ears with R-MED of 15 subjects and 20 healthy middle ears of 18 subjects. Subjects with diseased ears were patients at the ear surgery unit of the ENT department of the County Hospital of Boden during 1997 and 1998. Their mean age was 42 years (range, 12–75 y). The healthy group, of which the mean age was 45 years (range, 28–56 y), comprised 16 healthy volunteers from the staff of the ENT department and two subjects who had bilateral tympanostomy tubes inserted before hyperbaric oxygen treatment (for diabetic lesions and localized mandibular osteomyelitis, respectively). Common inclusion criteria for both groups were oral and written informed consent to participate in the study and absence of obvious signs of inflammatory disease affecting the upper airways such as a common cold, sinusitis, or allergy. Criteria for inclusion in the retraction group were dry ear and irreversible retraction pocket or atelectasis of the tympanic membrane with adhesion to medial structures or bone erosion, or retraction type cholesteatoma. Ears with perforations were included when the perforation was obviously secondary to a retraction. The specific criterion for inclusion in the normal group was absence of persistent or recurrent middle ear disease. Distributions of age, sex, side, pathological findings, and previous history are summarized in Tables I and II.

Table Table 1.. Distribution of Hearing Level, Side, and Sex.
original image

*Pure-tone average (mean of 0.5–1.0–2.0 kHz).

Table Table 2.. History and Pathological Findings in Ears With Retraction and in Normal Ears.
original image

*Tympanoplasty and mastoid surgery.

†Refers to ears of subjects who are habitual sniffers.


Pressures in the middle ear and in the nasopharynx were recorded directly and simultaneously during the performance of the forced opening, equalization, Valsalva, and sniff tests. The test procedure, here termed session, was performed twice with 30 minutes intervening on each of 2 days. The mean interval and SD between test days were 3.4 ± 1.3 months for the retraction group and 3.7 ± 1.3 months for the normal group. The times of entry in the study showed no seasonal predominance. Ears without perforation in the retraction had ventilation tubes throughout the study. In normal ears a myringotomy sized 0.5 to 1 mm was made each test day. For topical anesthesia, 90% phenol solution was used.

For the second test day, two subjects in the retraction group did not return. In the normal group, there were four dropouts, of whom one moved from the area, one was disabled by disease, and two experienced transient pulsatile tinnitus or discomfort during healing. All myringotomies healed completely without sequels within 3 weeks.


The equipment, which is described in detail elsewhere, 10 included a headpiece with three pressure transducers, a mechanical manometer for calibration, an air pump, a data acquisition system, and a computer. Two pressure transducers were connected to each ear canal via catheters and rubber probes used for tympanometry and a third was connected to the nose via a short catheter with a plastic foam cuff lodged in one nostril. The air pump was connected to the two ear pressure transducers via catheters and stopcocks. The data acquisition system included a work station with amplifiers and analogue-to-digital converters, a software package, and the computer.

Test Procedure

Forced opening test.

To evaluate the resistance to tubal opening and the forces that act to close an open tube, Pme was continuously increased by the air pump at a constant flow rate of 8 mL/min until the tube was forced open, as indicated by a sudden pressure decrease. At this point the air pump was switched off manually. Pme at the point of deflection was termed forced opening pressure (Po). The pressure decrease was continuously monitored until a residual positive pressure, called the closing pressure (Pc), was established, or until the pressure equaled the ambient pressure (the zero level). In each test session, at least six forced openings were performed with an interval of at least 3 seconds.

Pressure equalization.

To evaluate the ability to open the tube by muscular force, +100 daPa and −100 daPa were applied in the middle ear by the air pump. If equalization did not occur on the first attempt, the subject was asked to swallow at least 5 small sips of water. Equalization was considered positive if the induced test pressure changed by at least 10 daPa in the direction of the ambient pressure.

Valsalva test.

To evaluate the ability to inflate the middle ear actively, five Valsalva maneuvers were performed. The nose was pinched by the investigator to assure a correct position of the nasal cuff. If Pme increased at least once, the test was considered positive.

Sniff test.

To evaluate the tubal capacity to resist a high negative nasopharyngeal pressure, five vigorous sniffs were performed. If Pme decreased at least once, the test was considered positive.

Pme was instrumentally equalized to ambient pressure between the different tests. Tests for leakage were performed by a manometer before, after, and when needed, during the experiment. The sampling interval was 0.05 seconds, allowing for recording of rapid pressure changes.

Statistical Analysis

In each group variability was analyzed within sessions, between sessions, and between days, on the individual as well as on the group level. On the individual level, variability within a session referred to the variation of Po and Pc at repeated forced openings within a test session.

The mean of the third through sixth forced openings in each session was chosen to represent the result of the test. These measurements were considered the most representative with regard to the study question because, on the group level, the very first forced opening pressure (though not the first closing pressure) was significantly higher than those that followed, which showed a slight linear decrease (r = 0.10, Pearson's regression coefficient). This is a parallel to the Naunton-Galluser phenomenon, which refers to the improvement of tubal opening function with repeated swallowing. 11

For analysis of the variability of Po and Pc, univariate and multivariate ANOVA was used. The intraindividual variability of Po and Pc within and between sessions and days was estimated in terms of SD, by means of variance components estimation for each session separately, and by multilevel analysis of all data. The magnitude of individual changes in Po and Pc between sessions and test days was also calculated. As to the interindividual variability, differences in Po and Pc with regard to outcomes of other tests, and to various host factors were analyzed by repeated measures ANOVA, because repetitious values within each test were dependent within each ear. After the analysis based on all four sessions together, separate analyses were performed for each session.

Outcomes of the equalization, Valsalva, and sniff tests were analyzed for frequencies of positive response. As a measure of the intraindividual variability, the rate of responses that changed from positive to negative, or vice versa, within a test day was calculated for each test. The χ2 test or Fisher's exact test was used for relationships between host factors and outcomes of equalization, Valsalva, and sniff tests. Separate analyses were performed for each session. Relationships were considered statistically significant when P < .05.


Forced Opening Test

Mean Po and Pc did not differ between the groups (Fig. 1). Also, no difference between test sessions or test days was detected (Table III). Further analysis of the retraction group showed that Po and Pc were more than 30% higher in the four ears with previous major middle ear and mastoid surgery compared with those without major surgery, in the eight ears with perforation compared with those without perforation, and in the eight ears with cholesteatoma compared with those without cholesteatoma. Although differences were statistically significant, the subgroups were too small for general conclusions.

Figure Fig. 1..

Box plot showing the distribution of Po (shaded boxes) and Pc (open boxes) in ears with retraction and in normal ears. Boxes represent the middle 50% of all values. The horizontal line indicates the median. Bars represent the upper and lower 25%, respectively. There were no statistically significant differences in mean Po or mean Pc between the groups.

Table Table 3.. Forced Opening (Po) and Closing Pressure (Pc) in daPa at Different Test Days and Sessions.
original image

There were no statistically significant differences between test sessions and test days within each group, or between groups within the sessions.

SD = standard deviation.

The Po and Pc were stable on the group level but showed a considerable intraindividual variability within as well as between sessions and test days, which was approximately 1.5 to 2 times higher in diseased ears than in normal ears (Table IV). Within individual ears, differences in Po and Pc between sessions and days were more widely distributed in the retraction group than in the normal group (Fig. 2).

Table Table 4.. Intra-individual Variability Expressed as Relative Standard Deviation (SD/Mean Value) for Po and Pc Within Sessions, Between Sessions, and Between Test Days*
original image

*Analyzed by multilevel analysis.

†Refers to the variability of the third through sixth opening within each session.

Figure Fig. 2..

Box-plots illustrating the distribution of individual differences in Po (shaded boxes) and Pc (open boxes) between sessions and between days. Outliers and extreme values are excluded. As shown by larger boxes and longer bars, ears with retraction varied more than normal ears.

Four tubes in three subjects in the retraction group were patulous in both sessions for the same test day, and one tube in the normal group was patulous throughout the study. At the opposite extreme, three tubes in three subjects in the retraction group could not be forced opened by the maximally applied Pme (1000 daPa) in one or two sessions, and no blocked tubes were found in the normal group. None of the tubes in the retraction group remained patulous or blocked throughout all sessions.

Equalization, Valsalva, and Sniff Tests

Frequencies of positive response were higher in normal compared with diseased ears (Fig. 3). Because some ears changed from a negative to positive response from one session to the next, or vice versa, rates of positive response per test day were higher than for the single sessions. Except for the sniff test, rates of unstable response were generally higher in the retraction group (Table V), indicating a higher variability than in normal ears. Rates of positive and unstable response did not change significantly between sessions or test days in any of the groups.

Figure Fig. 3..

Percentage of positive response in the equalization, Valsalva, and sniff tests. In diseased and normal ears, the rates of positive response for equalization of +100 daPa were 55% and 93%, respectively; for equalization of −100 daPa, 26% and 89%; for the Valsalva test, 48% and 90%; and for the sniff test, 26% and 42%. Differences between groups were statistically significant for all except the sniff test.

Table Table 5.. Variability of Responses in Equalization, Valsalva, and Sniff Tests in Ears With Retraction and in Normal Ears.
original image

*n = Number of pairwise comparisons between sessions on the same day.

Numbers represent the percentage of responses that changed from a positive to negative response, or vice versa, between two sessions the same day.

Relations Between Test Outcomes

The Po and Pc correlated well both in the retraction group and in the normal group (r = 0.79 and 0.88, respectively, Pearson's correlation coefficient, P < .001). Po and Pc were not closely related to responses in other tubal function tests, except that for both groups ears with a positive response in the sniff test had lower values for Po and Pc compared with ears with a negative sniff test. Furthermore, healthy ears with a positive Valsalva test had lower Po and Pc values than healthy ears with a negative test.

Analyses for interrelations between responses in the equalization, Valsalva, and sniff tests were inconsequent in the normal group owing to very few negative responses; thus results concern the retraction group only. Ears that failed to equalize +100 daPa also failed to equalize −100 daPa. The ability to equalize +100 daPa was not related to the response in the Valsalva test, whereas ears that equalized +100 daPa were more often positive in the sniff test than ears that did not equalize +100 daPa (40% vs. 9%). Ears that equalized −100 daPa were more often Valsalva-positive than ears that did not (67% vs. 40%). These ears were also more often sniff-positive than those that did not equalize −100 daPa (43% vs. 20%). Valsalva-positive ears were more often sniff-positive than Valsalva-negative ears (40% vs. 15%), and sniff-positive ears were more often Valsalva-positive than sniff-negative ears (70% vs. 39%). These proportions represent means of the four sessions. At separate analyses for each session, differences were consistent although not statistically significant, owing to the small sample size.

Two kinds of valve mechanisms were found. For the first, 4 ears in 4 tests in the retraction group and 1 ear in 1 test in the normal group failed to eliminate a sniff-induced negative Pme, as indicated by a positive sniff test combined with inability to equalize −100 daPa and to perform Valsalva inflation. For the second, 8 ears in 12 tests in the retraction group and 2 ears in 4 tests in the normal group failed to release a positive residual Pme after Valsalva inflation, as indicated by a positive Valsalva test combined with inability to equalize +100 daPa and a negative sniff test.

Relations Between Test Outcomes and Host Factors

In the normal group, Po and Pc were 102 and 47 daPa higher, respectively, in female subjects than in male subjects. The difference was close to significant in the multilevel analysis. No statistically significant relationships with Po and Pc were found for side, hearing level, dysacusis, autophony, or recurrent otitis media in childhood. As to responses in the equalization, Valsalva, and sniff tests, the following trends were consistent throughout all sessions: male subjects were more often sniff-positive than female subjects in both groups. In the retraction group, equalization of +100 daPa and −100 daPa and Valsalva inflation were less often successful in ears in which dysacusis occurred, in ears of subjects with a sniffing habit, and in ears that had had recurrent otitis media in childhood.


There were two main findings in this study. First, the intraindividual variability of test responses was considerable and more pronounced in diseased ears than in normal ears for all tubal function tests. Second, although ears with retraction showed reduced capacity to equalize positive and negative Pme and to perform Valsalva inflation as compared with normal ears, no differences in mean values of Po or Pc were found between the groups.

Forced Opening Test

The Po and Pc, reflecting the tubal resistance to passive opening and tubal closing forces, respectively, varied considerably in the short-term as well as the long-term perspective, particularly in diseased ears. Previously, we reported a pronounced variability in children with SOM, albeit lower than in the diseased group in this study. 11 This variability is consistent with the fluctuating clinical course that is often seen in R-MED, and with the daily fluctuations that have been described in ears with atelectasis and retractions. 12,13 The pronounced variability and overlapping of results between healthy and diseased ears emphasize the low prognostic value of the forced opening test in ears with R-MED.

Variability of Po and Pc differed between diseased ears and normal ears, whereas the mean Po and mean Pc did not. In diseased ears, higher values of Po and Pc would be expected if tubal obstruction was associated with R-MED. We found blocked eustachian tubes only in three diseased ears, but interestingly, the tube became patulous in three diseased ears and one normal ear. None of the tubes in diseased ears remained blocked or patulous throughout the study. Lower values of Po and Pc were reported in ears with R-MED compared with the contralateral healthy ears in one study 14 and in comparison with normal ears in another study. 15 These findings demonstrate weaker closing forces rather than obstruction.

With a constant airflow from the pump, the mean rate of the pressure increase was 32 daPa/s in normal ears and 71 daPa/s in diseased ears, reflecting the reduced volume of the middle ear and mastoid in diseased ears. This implies a measuring bias because the value of Po increases with the rate of pressure increase. 16 In this study, the adjusted mean Po would be approximately 30 daPa lower in diseased than in normal ears, but the difference between the groups would not reach statistical significance.

Equalization, Valsalva, and Sniff Tests

Equalization tests reflect the capacity to open the tube by muscular force, while tubal openings at Valsalva inflation and sniffing are considered dependent on the pressure gradient between the nasopharynx and the middle ear. However, in both groups some subjects claimed that they facilitated air passage intentionally at inflation as well as sniffing by a straining maneuver in the pharynx and palate.

Responses varied more from test to retest in the diseased group than in normal ears. In contrast to Po and Pc, responses in the equalization and Valsalva tests differed significantly between diseased and normal ears. The proportions of positive response in each group are consistent with previous reports on tubal function in R-MED 1,14,17 and in normal ears. 18–20 Although reduced compared with normal ears, the rates of successful equalization of +100 daPa (55%), −100 daPa (26%), and Valsalva inflation (48%) in ears with retraction, and the extent of test-retest variability in this group are consistent with the finding that Pme is not constantly negative but fluctuates over time in atelectatic middle ears. 12,13

The sniff test was positive in 26% for the diseased group and in 44% for the normal group. Closing failure per se thus appears to be a common phenomenon. When combined with sniffing, negative Pme may occur. This should be harmless at normal equalization capacity. However, at locking of the tube and inability to equalize or inflate the ear, the persistent high negative Pm may promote the development of R-MED as proposed by Magnuson. 1 In this study, such a valve mechanism occurred in 20% and 5% of all positive sniff tests in diseased and normal ears, respectively. An inherent problem with a more pronounced closing failure is that the symptom, autophony, may trigger repetitious sniffing to lock the tube, 1,21 but at the cost of a high negative Pme and subsequent structural changes of the tympanic membrane.

Methodological Considerations

The primary issue of this study was to study the individual variability of test outcomes rather than the means and frequencies on the group level. Although none of these commonly used tests completely simulate physiological conditions, this should not affect the determination of the variability of responses. The significantly higher first Po represents a known source of variability within sessions. 16 Having confirmed this, we decided to use the third through sixth Po and Pc, because they showed little linear decrease and thus should provide a more reliable measure when comparing results between different sessions and test days. However, although quite stable on the group level, the third through the sixth Po and Pc varied considerably within individual ears in the retraction group.

Comparisons between different studies are difficult because Po and Pc are affected by various factors such as the rate of pressure increase, the body position, drugs, and physical and mental strain. For comparative purposes, the second, third, fourth, or fifth forced openings should provide a more reliable measure of the tubal resistance than the very first Po, which is significantly higher, possibly because of mucus in the tube that is cleared at the initial opening. However, the very first tubal opening in a tubal function test probably better reflects the physiological situation. The choice of which value to use depends on the issue studied and should be clearly defined.

Because the performance of the sniff-test is not standardized, rates of positive response differ between studies. For example, in ears with retraction-type cholesteatoma, Falk and Magnuson 17 reported 56% at up to 20 attempts to evacuate the ear, compared with the rate of 26% in this study, in which five attempts were employed. Likewise, rates of positive response in equalization and Valsalva tests differ depending on the performance and on criteria for positive response. Furthermore, the order in which tests are performed affects the outcome; tubal opening capacity improves with repeated openings, 11 whereas negative pressures and locking may impede opening. Thus standardization of tests is a precondition for comparisons between studies.

Results in tubal function tests in diseased ears are interpreted in relation to a normal reference. For ethical reasons, direct pressure recording in normal ears is only possible in small study populations, because a myringotomy is needed. The few previous studies in “normal” ears involve traumatic tympanic membrane perforations or contralateral healthy ears in patients with unilateral retraction disease. Such ears may not represent a completely normal population. It is thus of interest to relate findings with direct technique in different supposedly normal study populations, including that in the present study, to findings with indirect technique in larger populations to determine the consistency between these techniques. Studies on ears with intact tympanic membranes were performed in a pressure chamber by Bylander et al. 19 At comparable rates of pressure increase, our mean Po and mean Pc of the third through sixth forced opening in the normal group were 24% and 26% lower, respectively. Although not a part of our primary analysis, values based on the very first opening were 10% and 24% lower, respectively.


In individual ears with retraction, the tubal opening and closing function varied considerably over time, whereas variability was less pronounced in normal ears. Whereas the capacity to equalize positive and negative pressures and to inflate the ear by Valsalva's maneuver was significantly reduced in diseased ears compared with normal ears, there were no significant differences with regard to the passive opening and closing pressures in the forced opening test. In view of the extent of variability in diseased ears and the overlapping of results between diseased and normal ears in this study, single tests of the tubal opening and closing function cannot be used as prognostic tools on the individual level. When tubal function tests are used for other purposes, for example comparative studies on the group level, the findings in this study can be used as a reference when designing and evaluating such studies. However, lack of standardization of methods and procedures makes comparisons between different studies unreliable.


Approval was granted by the Ethical Committee of Umeå University Hospital.