Ann Otol Rhinal Laryngol99: 1990
CERAVITAL VERSUS PLASTIPORE IN TYMPANOPLASTY: A RANDOMIZED PROSPECTIVE TRIAL CHARLES
A.
MANGHAM, MD
ROGER C. LINDEMAN, MD SEATTLE, WASHINGTON
T~e hearing results a~d extrusion rates. for ossicular chain reconstruction using a new material called Ceravital, a bioactive glass ceramic, were compared with those for Plastipore, a porous polyethylene that is our standard for synthetic materials. Our null hypothesis was that the results of the two materials were similar. A Ceravital or a Plastipore prosthesis was randomly assigned to 112 consecutive patients. We had 6-month follow-up on 100 patients and 3-year follow-up on 80 patients. The average air-bone gap at 4,000 Hz was 6 dB less w~th Plasti~ore than with Ceravital (p = .036). Patients were .twice ~ likely t~ have an air-bone gap of 15 dB or lesswith Plastipore than with Ceravital (40% versus 21%; p= .061). Two of the 38 patients with Ceravital prostheses had a late hearing-result failure due to resorption of the prosthesis material. The materials had similar extrusion rates. We rejected our null hypothesis. The Plastipore group had better hearing results than did the Ceravital group.
KEY WORDS - ceramics, Ceravital, ossicular chain reconstruction, Plastipore, polyethylene, prosthesis, tympanoplasty.
INTRODUCTION
suIts for Ceravital and Plastipore partial prostheses, but that the results for Ceravital total prostheses were "unquestionably better" than the results for Plastipore prostheses. We did a cross-tabulation of their hearing results by material (Ceravital or Plastipore) controlled for prosthesis (partial or total) that showed an insignificant difference between Ceravital and Plastipore total prostheses (Gersdorff et al,? Table 7; degrees of freedom = 3, X2 =4.07, P = .255). A more accurate conclusion is that should their results with total prostheses remain constant with additional subjects, then the differences between prostheses would be significant.
Over the years, the standard for ossicular chain reconstruction has been the sculpted ossicle (for an example, see the article by Wehrs'). This is still the choice when the patient's own ossicle can be used; however, frequently the patient's disease process makes autologous ossicles unsuitable. The requirement to eliminate the risk of transmitted diseasessuch as acquired immunodeficiency syndrome makes homograft ossicles difficult to acquire in our practice. For this reason, like many other surgeons, we have turned to alloplastic materials for ossicular reconstruction. Plastipore, the trade name for porous polyethylene introduced in the 1970s, is our standard for alloplastic implant materials against which the performance of newer alloplastic materials may be compared. Plastipore fixes to middle ear structures by tissue growth into the pores of the material. Some of this ingrowth is an inflammatory response that may be responsible for some hearing result failures and prosthesis extrusion." Ceravital, the trade name for a specific glass ceramic, fixes to the middle ear by a chemical reaction between the glass ceramic and bone. The prospect of greater biologic compatibility leading to better hearing results and fewer prosthesis extrusions with Ceravital encouraged us to consider reevaluating our use of Plastipore. The purpose of this randomized prospective study is to compare the hearing results and extrusion rates of Plastipore with the hearing results and extrusion rates of Ceravital. Our null hypothesis is that no difference exists between materials.
METHODS
Subjects. The Research Advisory Committee of the Virginia Mason Research Center reviewed the research plan and the Institutional Review Board approved the written informed consent before patients were enrolled into the study. On the basis of study design considerations discussed below, we planned on having 120 surgical patients: 80 with a partial prosthesis, needed when the incus is absent and the stapes superstructure is present, and 40 with a total prosthesis, needed when both the incus and stapes superstructure are absent (Table 1). Beginning in April 1984, patients were accrued from our surgical practices. A third surgeon was added in TABLE 1. ENROLLMENT IN STUDY Material
Plastipore
When our study began in 1983, there were no published reports comparing results with Plastipore and Ceravital. In 1986, Gersdorff et al" reported that there was no significant difference between re-
Ceravital Total
Prosthesis Planned
Partial Total Partial Total
40 20 40 20 120
Actual
40 16 40 16 112
6-Month 3-Year Follow- Followup up
38 13 35 14 100
31 11 26 12 80
From the Virginia Mason Research Center, Seattle, Washington. Supported by grants from Xomed, Inc, Jacksonville, Florida, and the Virginia Mason Research Center, Seattle, Washington. Presented at the meeting of the American Otological Society, Inc. San Francisco, California, April 2-3, 1989.
112
Mangham & Lindeman, Ceravital Versus Plastipore
the second year of patient accrual. The partialprosthesis group was filled after 15 months. We terminated enrollment with 112 patients after the total prosthesis group was still unfilled after 24 months. The six patients enrolled by the third surgeon were eliminated because the surgeon requested to withdraw from the study. Six of the remaining 106 patients were eliminated because they failed to return for follow-up. We had 6-month follow-up on the remaining 100 patients and 3-year follow-up on 80 patients. Excluding patients who had revision operations, the average follow-up for the Ceravital group was 36.7 months, and it was 36.5 months for the Plastipore group; thus, the variable of duration of follow-up was equalized. Most of this report involves the 69 patients who had 80 ossicular prosthesis operations and both 6-month and 3-year follow-up. Four of the 69 had ossicular prostheses in both ears. Seven of the 69 had an ossicular prosthesis replaced at a revision operation during the enrollment period. The 69 patients included 32 males who were an average of 45.5 years of age and 37 females who were an average of 51.1 years of age. There were only two patients younger than 16 years of age; therefore, there should be little bias due to differences between results in children and results in adults. Study Design. In a sample-size analysis we assumed a worst-case scenario, ie, that preoperative and postoperative hearing results had a correlation coefficient no greater than 0.1. In order to detect a true difference of 10 dB between groups 80 % of the time with a risk of falsely detecting this difference no more than 5 % of the time, we found that we needed 37 subjects in each group. To detect a 5-dB difference, we would need 151 in each group. Although detecting a 5-dB difference was desirable, it Was impractical because we anticipated that we would enroll subjects at a rate of 75 per year and we wanted to finish enrollment in 18 months. In a post hoc analysis, the measured correlation between the preoperative pure-tone average air-bone gap (PTA AB gap) and the 3-year postoperative PTA AB gap Was 0.20; hence, our actual sample size requirement was 34 in each group to detect a lO-dB difference between groups.
Patients with partial prostheses have a greater chance of meeting a criterion for success than patients with total prostheses." To equally distribute the prosthesis type, we stratified the subjects into those with a partial prosthesis and those with a total prosthesis prior to randomizing between Ceravital and Plastipore. We assumed that other variables that affect hearing results such as stage of disease, and whether the operation was one-stage, twostage, or a revision, would be equally distributed between the two materials by the randomization process.s
113
Procedure. The surgeon told the circulating nurse whether he would need a partial or total prosthesis. The nurse then opened the appropriate envelope, labeled partial or total, and told the surgeon whether Ceravital or Plastipore was written on a card inside. The surgeon then used the material directed by the contents of the envelope and used the surgical technique appropriate for the prosthesis material. In general, Ceravital prostheses had bone pate placed on the lateral surface of the prosthesis platform, and Plastipore prostheses had cartilage - some homograft, some autograft - interposed between the platform of the prosthesis and the undersurface of the eardrum. After the operation, the surgeon filled out a data collection sheet. Data from these sheets were then entered into a data base management package (dBase III, Ashton-Tate, Culver City, Calif) run on a microcomputer (Compaq Deskpro, Compaq Computer Corporation, Houston, Tex) and analyzed with a statistical package (ySTAT, Ming Telecomputing, Inc, Lincoln Center, Mass).
The hearing results were reported as pure-tone averages (PTAs) of the difference between bone conduction thresholds and air conduction thresholds for pure tones of 500,1,000, and 2,000 Hz. We used bone conduction and air conduction threshold data from the same audiogram, rather than using the best preoperative bone conduction score compared with the postoperative air conduction score as suggested by Sheehy" (called residual conduction deficit in Gersdorff et al"), Average bone conduction scores improved after operation by 2.3 dB for the 6-month data and 1.0 dB for the 3-year data. In other words, Sheehy's method would have resulted in slightly better hearing results, because Sheehy's method would have us using slightly worse bone conduction scores to subtract from our postoperative air conduction scores. We considered the possibility that there was a correlation between a patient's air conduction threshold and the air-bone gap after operation. This might occur, for example, if patients with more severe disease had worse air conduction thresholds and their severity of disease limited the benefits of operation and left them with large air-bone gaps. We found no correlation between air conduction threshold and air-bone gap. We decided not to report word recognition scores. We found that having eight different audiologists giving live-voice tests at intensity levels that differed among preoperative tests and postoperative tests introduced variability for which we could not develop controls. Some surgeons differentiate between a protrusion of the prosthesis, in which a part of the prosthesis can be seen through a defect in the eardrum, and an extrusion, in which the whole platform of the prosthesis can be seen through a hole in the drum. The rationale is that protrusions may have spontaneous
Mangham & Lindeman, Ceravital Versus Plastipore
114
TABLE 2. HEARING RESULTS AT 3 YEARS EXPRESSED AS PERCENTAGE OF PATIENTS Postop Air-Bone Gap (dB)
Partial Ceravital, Sl (n =20)
Partial Ceravital, S2 (n=6)
Partial Ceravital, Subtotal (n =26)
Total Ceravital, Sl (n=5)
Total Ceravital, S2 (n= 7)
Total Ceravital, Subtotal (n=12)
:55 :510 :515 :520 :525 :530 :550
0 10 25 65 65 85 100
17 17 17 17 33 50 100
4 12 23 54 58 77 100
0 0 20 20 20 40 100
0 0 14 29 29 86 100
0 0 17 25 25 67 100
3 8 21* 45 47 74 100
Postop Air-Bone Gap (dB)
Partial Plastipore, Sl (n =22)
Partial Plastipore, S2 (n=9)
Partial Plastipore, Subtotal (n = 31)
Total Plastlpore, Sl (n=5)
Total Plastipore, S2 (n=6)
Total Plastipore, Subtotal (n =11)
Plastipore, Total (n =42)
0 0 33 33 50 67 100
0 9 27 27 45 73 100
2 19 40* 50 64 79 100
5 0 3 :55 23 27 11 :510 45 50 33 :515 58 64 :520 44 82 44 71 :525 91 81 :530 56 100 100 :550 100 51 - surgeon 1, 52 - surgeon 2. "Difference betweenCeravital and Plastipore for s 15 dB, x' = 3.50, p = .061.
eardrum healing, whereas extrusions do not heal. We labeled both protrusions and extrusions as "extrusions" if there was any prosthesis showing at any time during the 3-year follow-up. This resulted in an overestimate of extrusion rate by about 20 %, since some of the protrusions have since healed. To determine if differences in technique could account for the difference in surgeons' results, we had to identify the surgeons. In order to make the reader unaware of which surgeon used which technique, we have randomly assigned one of the surgeons the code name "surgeon 1," and the other the name "surgeon 2."
RESULTS Differences Due to Material and Type oj Prosthesis. The 3-year hearing results are shown in Table 2.
The first column shows seven outcomes for differences between air conduction and bone conduction PTA scores ranging from 5 dB or less to 50 dB or less. The second column shows the results for surgeon 1 using partial prostheses. In column 2, data rows 1 through 7 show the results with use of partial Ceravital prostheses, and data rows 8 through 14 show the results of partial Plastipore prostheses. The third column shows the results for surgeon 2 using partial prostheses. The fourth column has the average hearing results with partial prostheses for surgeon 1 and surgeon 2 obtained by averaging the results in columns 2 and 3. The fifth column shows the results for surgeon I using total Ceravital prostheses and total Plastipore prostheses. The sixth column shows the results for surgeon 2 using total prostheses. The seventh column has the average hearing results with total prostheses for surgeon 1 and surgeon 2, obtained by averaging the results in
0 20 20 20 40 80 100
Ceravital, Total (n=38)
columns 5 and 6. The eighth column, an average for partial and total prostheses, is the average of columns 4 and 7. There was a trend toward better hearing results with Plastipore than with Ceravital. The greatest difference between materials occurred when an airbone gap of 15 dB or less was considered the criterion for success. In column 8, 21 % (8 of 38) of results with a Ceravital prosthesis were successful, compared with 40% (17 of 42) of results successful with Plastipore (Xl = 3.50, p = .061). There were no "dead ears," ie, ears with no hearing at the limits of the audiometric output. Table 3 shows the air-bone gaps at 3 years as a function of stimulus frequency. Both groups had a similar pattern of result, with the smallest mean airbone gap occurring at 2,000 Hz. At 4,000 Hz, the Plastipore group had a mean hearing level that was 6.04 dB better than the Ceravital group (26.07 dB HL versus 32.11 dB HL; ANOVA, F-statistic = 4.39, p = .036). Surgeon I experienced 13 of the 14 extrusions. The extrusions were similarly distributed between the materials, Ceravital and Plastipore, and between the prosthesis types, partial and total. TABLE 3. AIR-BONE PURE-TONE AVERAGE (dB) AT 3-YEAR FOLLOW-UP Material Ceravital (n = 38)
250
Frequency (Hz) 500 1,000 2,0004,000*
Mean 37.37 27.89 30.66 16.84 32.11 SD 19.20 16.05 15.65 15.70 14.69 Plastipore (0 = 42) Mean 33.81 25.24 27.74 16.43 26.07 SD 17.59 17.11 15.19 15.03 10.96 •ANOVA for differences between materials at 4,000 Hz: F-statistic = 4.39, p= .036.
Mangham & Lindeman, Ceraoital Versus Plastipore
Fig 1. Ceravital total prosthesis as received from manufacturer is shown on left. Total prosthesis removed from middle ear of I8-year-old girl 3 years after original operation is shown on right. Rule is in millimeters.
. Differences Due to Surgical Technique. As shown 10 Table 2, surgeon 1 had better hearing results than surgeon 2, an unexpected finding. In an exploratory analysis, we found that surgeon 1 had hearing results that were 15 dB better than the hearing results for surgeon 2 in the case of one-stage ossicular chain reconstructions (surgeon 1, mean = 19.00 dB; surgeon 2, mean = 34.39 dB; ANOVA, F-statistic = 4.53, p = .025). Further, surgeon 1 did all 29 of the second-stage reconstructions, with surgeon 2 electing to revise the one-stage cases that Were failures from a hearing point of view rather than do a planned second stage. Other differences between surgeons were secondary to the effect of staging operations.
. We performed exploratory analyses on all hearI?g-result differences greater than 5 dB. There were SIx such analyses; therefore, we had about a 30% probability of finding that one of these analyses would have a p value of .05 or less by chance.
115
Fig 2. Cera vital partial prosthesis as received from manufacturer is shown on left. Partial prosthesis removed from middle ear of 72-year-old woman 3 years after original operation is shown on right. Rule is in millimeters.
group with extensive experience with Plastipore. By contrast, neither of us began using Ceravital before preparing for this study. There may be technical factors that are required for success with Ceravital that cannot be gained from talking to the manufacturer's representatives or from experience in the laboratory. A surgeon starting to use Ceravital should consider getting training from an individual or group with good results from this prosthesis. A factor that may lead to long-term failure with Ceravital prostheses is the variable rate of prosthesis absorption. In animal studies, areas of lysis of the prosthesis material occur that are as deep as 250 I'm at 1 year and as deep as 450 I'm after 2 years.' We had two patients who had a progression in hearing loss at about 3 years related to resorption of the prosthesis in the area in which the prosthesis was making contact with the sound-conducting mechanism (Figs 1 and 2).
We rejected the null hypothesis that differences between Ceravital and Plastipore were insignificant, on the basis of our finding that the Plastipore group was twice as likely as the Ceravital group to have a postoperative air-bone gap of 15 dB or less. F~rther, the Plastipore group had a 6-dB smaller au-bone gap at 4,000 Hz than the Ceravital group.
An unexpected finding was a difference between the hearing results and extrusion rates of surgeon 1 and surgeon 2. Our explanations were related to staging the operation and the length of the prosthesis (the operation was staged according to guidelines described by Sheehy" and Sheehy and Crabtree"). Had we known about the confounding effect of differences in staging, we would have controlled for this variable by having the surgeons use the same indications for staging.
Our results with Plastipore were comparable to those achieved by Gersdorff et aP; however, our results with Ceravital were not as good as the results of the Gersdorff group. One explanation for our Poorer results with Cera vital is that we had several years' prior experience with Plastipore, including one author's having fellowship training with a
Our other unexpected finding, the high extrusion rate for surgeon 1, was probably related to prosthesis length. Our speculation was that surgeon 1 tended to use a longer prosthesis than surgeon 2. A longer prosthesis would be less likely to lateralize from the stapes footplate or stapes capitulum and less likely to medialize from the eardrum or mal-
DISCUSSION
Mangham & Lindeman, Ceravital Versus Plastipore
116
leus. The longer prosthesis would be more likely to extrude through the eardrum, so there is a penalty with these materials for using a long prosthesis.
these two prostheses. Further, the 3-year data tend to underestimate the difference between the Plastipore and the Ceravital group.
Sharp corners predisposed a prosthesis to extrusion. When the platform of a Plastipore prosthesis was cut into a T shape, similar to the Fisch design, the prosthesis tended to extrude at the corners. The Ceravital prosthesis tended to extrude at a corner or at one of the edges of the groove cut in the lateral surface of the platform for the malleus handle.
Another cause of differences among published results would be the bias introduced by the surgeon. In this study, the decision to use Ceravital or Plastipore was made by a random assignment of prosthesis material. By contrast, Niparko et al" used a nonrandom assignment of prosthesis material: "Ceravital. ..was used when an incus autograft was not obtainable or not suitable for reconstruction." Other possible causes of bias include the staging of the operation and whether the malleus was present or absent, as described above, as well as the severity of disease. 5
Results in this study should be compared with caution to results of other studies because of confounding variables such as differences in patient self-selection for follow-up. In this study, for example, the nine patients with Plastipore prostheses who returned for 6-month hearing tests but did not return for 3-year hearing tests had a better average hearing result at 6 months by 5.9 dB than their 42patient cohort that did not drop out of the study. The 11 patients with Ceravital prostheses who dropped out had a better average hearing result at 6 months by 1.4 dB than their 38-patient cohort that did not drop out. This suggests that our 3-year data underestimate the hearing results achieved with ACKNOWLEDGMENTS -
SUMMARY
The hearing results with Plastipore were better than the results with Ceravital. Technical factors such as staging the operation in cases of mucosal loss, the height of the prosthesis, and the shape of the prosthesis platform were as important in determining the final hearing result and extrusion rate as was the material used.
Brent Blumenstein, PhD, and Gloria Bailey, PhD, provided consultation for study design and statistical analysis.
REFERENCES 1. Wehrs RE. Homograft ossicles in tympanoplasty. Laryn-
5. Bellucci RJ. Dual classification of tympanoplasty. Laryngoscope 1973;83:1754-8.
2. Teichgraeber JF, Spector M, Per-Lee JH,Jackson RT. Tissue response to Plasti-Pore and Proplast otologic implants in the middle ears of cats. Am J OtoI1983;5:127-36.
6. Sheehy JL. Plastic sheeting in tympanoplasty. Laryngoscope 1973;83:1144-59.
goscope 1982;5:540-6.
3. Gersdorff MCH, Maisin JP, Munting E. Comparative study of the clinical results obtained by means of Plastipore and ceramic ossicular prosthesis and bone allografts. Am J Otol 1986; 7:294-7. 4. Brackmann DE, Sheehy JL, Luxford WM. TORPs and PORPs in tympanoplasty: a review of 1042 operations. Otolaryngol Head Neck Surg 1984;92:32-7.
7. Reck R. Bioactive glass-ceramics in ear surgery: animal studies and clinical results. Laryngoscope 1984;94(suppl 33). 8. Sheehy JL, Crabtree JA. Tympanoplasty: staging the operation. Laryngoscope 1973;83:1594-621. 9. Niparko JK, Kemink JL, Graham MD, Kartush MD. Bioactive glass ceramic in ossicular reconstruction: a preliminary report. Laryngoscope 1988;98:822-5.
CERAVITAL VERSUS PLASTIPORE IN TYMPANOPLASTY: A RANDOMIZED PROSPECTIVE TRIAL CHARLES
A.
MANGHAM, MD
ROGER C. LINDEMAN, MD SEATTLE, WASHINGTON
T~e hearing results a~d extrusion rates. for ossicular chain reconstruction using a new material called Ceravital, a bioactive glass ceramic, were compared with those for Plastipore, a porous polyethylene that is our standard for synthetic materials. Our null hypothesis was that the results of the two materials were similar. A Ceravital or a Plastipore prosthesis was randomly assigned to 112 consecutive patients. We had 6-month follow-up on 100 patients and 3-year follow-up on 80 patients. The average air-bone gap at 4,000 Hz was 6 dB less w~th Plasti~ore than with Ceravital (p = .036). Patients were .twice ~ likely t~ have an air-bone gap of 15 dB or lesswith Plastipore than with Ceravital (40% versus 21%; p= .061). Two of the 38 patients with Ceravital prostheses had a late hearing-result failure due to resorption of the prosthesis material. The materials had similar extrusion rates. We rejected our null hypothesis. The Plastipore group had better hearing results than did the Ceravital group.
KEY WORDS - ceramics, Ceravital, ossicular chain reconstruction, Plastipore, polyethylene, prosthesis, tympanoplasty.
INTRODUCTION
suIts for Ceravital and Plastipore partial prostheses, but that the results for Ceravital total prostheses were "unquestionably better" than the results for Plastipore prostheses. We did a cross-tabulation of their hearing results by material (Ceravital or Plastipore) controlled for prosthesis (partial or total) that showed an insignificant difference between Ceravital and Plastipore total prostheses (Gersdorff et al,? Table 7; degrees of freedom = 3, X2 =4.07, P = .255). A more accurate conclusion is that should their results with total prostheses remain constant with additional subjects, then the differences between prostheses would be significant.
Over the years, the standard for ossicular chain reconstruction has been the sculpted ossicle (for an example, see the article by Wehrs'). This is still the choice when the patient's own ossicle can be used; however, frequently the patient's disease process makes autologous ossicles unsuitable. The requirement to eliminate the risk of transmitted diseasessuch as acquired immunodeficiency syndrome makes homograft ossicles difficult to acquire in our practice. For this reason, like many other surgeons, we have turned to alloplastic materials for ossicular reconstruction. Plastipore, the trade name for porous polyethylene introduced in the 1970s, is our standard for alloplastic implant materials against which the performance of newer alloplastic materials may be compared. Plastipore fixes to middle ear structures by tissue growth into the pores of the material. Some of this ingrowth is an inflammatory response that may be responsible for some hearing result failures and prosthesis extrusion." Ceravital, the trade name for a specific glass ceramic, fixes to the middle ear by a chemical reaction between the glass ceramic and bone. The prospect of greater biologic compatibility leading to better hearing results and fewer prosthesis extrusions with Ceravital encouraged us to consider reevaluating our use of Plastipore. The purpose of this randomized prospective study is to compare the hearing results and extrusion rates of Plastipore with the hearing results and extrusion rates of Ceravital. Our null hypothesis is that no difference exists between materials.
METHODS
Subjects. The Research Advisory Committee of the Virginia Mason Research Center reviewed the research plan and the Institutional Review Board approved the written informed consent before patients were enrolled into the study. On the basis of study design considerations discussed below, we planned on having 120 surgical patients: 80 with a partial prosthesis, needed when the incus is absent and the stapes superstructure is present, and 40 with a total prosthesis, needed when both the incus and stapes superstructure are absent (Table 1). Beginning in April 1984, patients were accrued from our surgical practices. A third surgeon was added in TABLE 1. ENROLLMENT IN STUDY Material
Plastipore
When our study began in 1983, there were no published reports comparing results with Plastipore and Ceravital. In 1986, Gersdorff et al" reported that there was no significant difference between re-
Ceravital Total
Prosthesis Planned
Partial Total Partial Total
40 20 40 20 120
Actual
40 16 40 16 112
6-Month 3-Year Follow- Followup up
38 13 35 14 100
31 11 26 12 80
From the Virginia Mason Research Center, Seattle, Washington. Supported by grants from Xomed, Inc, Jacksonville, Florida, and the Virginia Mason Research Center, Seattle, Washington. Presented at the meeting of the American Otological Society, Inc. San Francisco, California, April 2-3, 1989.
112
Mangham & Lindeman, Ceravital Versus Plastipore
the second year of patient accrual. The partialprosthesis group was filled after 15 months. We terminated enrollment with 112 patients after the total prosthesis group was still unfilled after 24 months. The six patients enrolled by the third surgeon were eliminated because the surgeon requested to withdraw from the study. Six of the remaining 106 patients were eliminated because they failed to return for follow-up. We had 6-month follow-up on the remaining 100 patients and 3-year follow-up on 80 patients. Excluding patients who had revision operations, the average follow-up for the Ceravital group was 36.7 months, and it was 36.5 months for the Plastipore group; thus, the variable of duration of follow-up was equalized. Most of this report involves the 69 patients who had 80 ossicular prosthesis operations and both 6-month and 3-year follow-up. Four of the 69 had ossicular prostheses in both ears. Seven of the 69 had an ossicular prosthesis replaced at a revision operation during the enrollment period. The 69 patients included 32 males who were an average of 45.5 years of age and 37 females who were an average of 51.1 years of age. There were only two patients younger than 16 years of age; therefore, there should be little bias due to differences between results in children and results in adults. Study Design. In a sample-size analysis we assumed a worst-case scenario, ie, that preoperative and postoperative hearing results had a correlation coefficient no greater than 0.1. In order to detect a true difference of 10 dB between groups 80 % of the time with a risk of falsely detecting this difference no more than 5 % of the time, we found that we needed 37 subjects in each group. To detect a 5-dB difference, we would need 151 in each group. Although detecting a 5-dB difference was desirable, it Was impractical because we anticipated that we would enroll subjects at a rate of 75 per year and we wanted to finish enrollment in 18 months. In a post hoc analysis, the measured correlation between the preoperative pure-tone average air-bone gap (PTA AB gap) and the 3-year postoperative PTA AB gap Was 0.20; hence, our actual sample size requirement was 34 in each group to detect a lO-dB difference between groups.
Patients with partial prostheses have a greater chance of meeting a criterion for success than patients with total prostheses." To equally distribute the prosthesis type, we stratified the subjects into those with a partial prosthesis and those with a total prosthesis prior to randomizing between Ceravital and Plastipore. We assumed that other variables that affect hearing results such as stage of disease, and whether the operation was one-stage, twostage, or a revision, would be equally distributed between the two materials by the randomization process.s
113
Procedure. The surgeon told the circulating nurse whether he would need a partial or total prosthesis. The nurse then opened the appropriate envelope, labeled partial or total, and told the surgeon whether Ceravital or Plastipore was written on a card inside. The surgeon then used the material directed by the contents of the envelope and used the surgical technique appropriate for the prosthesis material. In general, Ceravital prostheses had bone pate placed on the lateral surface of the prosthesis platform, and Plastipore prostheses had cartilage - some homograft, some autograft - interposed between the platform of the prosthesis and the undersurface of the eardrum. After the operation, the surgeon filled out a data collection sheet. Data from these sheets were then entered into a data base management package (dBase III, Ashton-Tate, Culver City, Calif) run on a microcomputer (Compaq Deskpro, Compaq Computer Corporation, Houston, Tex) and analyzed with a statistical package (ySTAT, Ming Telecomputing, Inc, Lincoln Center, Mass).
The hearing results were reported as pure-tone averages (PTAs) of the difference between bone conduction thresholds and air conduction thresholds for pure tones of 500,1,000, and 2,000 Hz. We used bone conduction and air conduction threshold data from the same audiogram, rather than using the best preoperative bone conduction score compared with the postoperative air conduction score as suggested by Sheehy" (called residual conduction deficit in Gersdorff et al"), Average bone conduction scores improved after operation by 2.3 dB for the 6-month data and 1.0 dB for the 3-year data. In other words, Sheehy's method would have resulted in slightly better hearing results, because Sheehy's method would have us using slightly worse bone conduction scores to subtract from our postoperative air conduction scores. We considered the possibility that there was a correlation between a patient's air conduction threshold and the air-bone gap after operation. This might occur, for example, if patients with more severe disease had worse air conduction thresholds and their severity of disease limited the benefits of operation and left them with large air-bone gaps. We found no correlation between air conduction threshold and air-bone gap. We decided not to report word recognition scores. We found that having eight different audiologists giving live-voice tests at intensity levels that differed among preoperative tests and postoperative tests introduced variability for which we could not develop controls. Some surgeons differentiate between a protrusion of the prosthesis, in which a part of the prosthesis can be seen through a defect in the eardrum, and an extrusion, in which the whole platform of the prosthesis can be seen through a hole in the drum. The rationale is that protrusions may have spontaneous
Mangham & Lindeman, Ceravital Versus Plastipore
114
TABLE 2. HEARING RESULTS AT 3 YEARS EXPRESSED AS PERCENTAGE OF PATIENTS Postop Air-Bone Gap (dB)
Partial Ceravital, Sl (n =20)
Partial Ceravital, S2 (n=6)
Partial Ceravital, Subtotal (n =26)
Total Ceravital, Sl (n=5)
Total Ceravital, S2 (n= 7)
Total Ceravital, Subtotal (n=12)
:55 :510 :515 :520 :525 :530 :550
0 10 25 65 65 85 100
17 17 17 17 33 50 100
4 12 23 54 58 77 100
0 0 20 20 20 40 100
0 0 14 29 29 86 100
0 0 17 25 25 67 100
3 8 21* 45 47 74 100
Postop Air-Bone Gap (dB)
Partial Plastipore, Sl (n =22)
Partial Plastipore, S2 (n=9)
Partial Plastipore, Subtotal (n = 31)
Total Plastlpore, Sl (n=5)
Total Plastipore, S2 (n=6)
Total Plastipore, Subtotal (n =11)
Plastipore, Total (n =42)
0 0 33 33 50 67 100
0 9 27 27 45 73 100
2 19 40* 50 64 79 100
5 0 3 :55 23 27 11 :510 45 50 33 :515 58 64 :520 44 82 44 71 :525 91 81 :530 56 100 100 :550 100 51 - surgeon 1, 52 - surgeon 2. "Difference betweenCeravital and Plastipore for s 15 dB, x' = 3.50, p = .061.
eardrum healing, whereas extrusions do not heal. We labeled both protrusions and extrusions as "extrusions" if there was any prosthesis showing at any time during the 3-year follow-up. This resulted in an overestimate of extrusion rate by about 20 %, since some of the protrusions have since healed. To determine if differences in technique could account for the difference in surgeons' results, we had to identify the surgeons. In order to make the reader unaware of which surgeon used which technique, we have randomly assigned one of the surgeons the code name "surgeon 1," and the other the name "surgeon 2."
RESULTS Differences Due to Material and Type oj Prosthesis. The 3-year hearing results are shown in Table 2.
The first column shows seven outcomes for differences between air conduction and bone conduction PTA scores ranging from 5 dB or less to 50 dB or less. The second column shows the results for surgeon 1 using partial prostheses. In column 2, data rows 1 through 7 show the results with use of partial Ceravital prostheses, and data rows 8 through 14 show the results of partial Plastipore prostheses. The third column shows the results for surgeon 2 using partial prostheses. The fourth column has the average hearing results with partial prostheses for surgeon 1 and surgeon 2 obtained by averaging the results in columns 2 and 3. The fifth column shows the results for surgeon I using total Ceravital prostheses and total Plastipore prostheses. The sixth column shows the results for surgeon 2 using total prostheses. The seventh column has the average hearing results with total prostheses for surgeon 1 and surgeon 2, obtained by averaging the results in
0 20 20 20 40 80 100
Ceravital, Total (n=38)
columns 5 and 6. The eighth column, an average for partial and total prostheses, is the average of columns 4 and 7. There was a trend toward better hearing results with Plastipore than with Ceravital. The greatest difference between materials occurred when an airbone gap of 15 dB or less was considered the criterion for success. In column 8, 21 % (8 of 38) of results with a Ceravital prosthesis were successful, compared with 40% (17 of 42) of results successful with Plastipore (Xl = 3.50, p = .061). There were no "dead ears," ie, ears with no hearing at the limits of the audiometric output. Table 3 shows the air-bone gaps at 3 years as a function of stimulus frequency. Both groups had a similar pattern of result, with the smallest mean airbone gap occurring at 2,000 Hz. At 4,000 Hz, the Plastipore group had a mean hearing level that was 6.04 dB better than the Ceravital group (26.07 dB HL versus 32.11 dB HL; ANOVA, F-statistic = 4.39, p = .036). Surgeon I experienced 13 of the 14 extrusions. The extrusions were similarly distributed between the materials, Ceravital and Plastipore, and between the prosthesis types, partial and total. TABLE 3. AIR-BONE PURE-TONE AVERAGE (dB) AT 3-YEAR FOLLOW-UP Material Ceravital (n = 38)
250
Frequency (Hz) 500 1,000 2,0004,000*
Mean 37.37 27.89 30.66 16.84 32.11 SD 19.20 16.05 15.65 15.70 14.69 Plastipore (0 = 42) Mean 33.81 25.24 27.74 16.43 26.07 SD 17.59 17.11 15.19 15.03 10.96 •ANOVA for differences between materials at 4,000 Hz: F-statistic = 4.39, p= .036.
Mangham & Lindeman, Ceraoital Versus Plastipore
Fig 1. Ceravital total prosthesis as received from manufacturer is shown on left. Total prosthesis removed from middle ear of I8-year-old girl 3 years after original operation is shown on right. Rule is in millimeters.
. Differences Due to Surgical Technique. As shown 10 Table 2, surgeon 1 had better hearing results than surgeon 2, an unexpected finding. In an exploratory analysis, we found that surgeon 1 had hearing results that were 15 dB better than the hearing results for surgeon 2 in the case of one-stage ossicular chain reconstructions (surgeon 1, mean = 19.00 dB; surgeon 2, mean = 34.39 dB; ANOVA, F-statistic = 4.53, p = .025). Further, surgeon 1 did all 29 of the second-stage reconstructions, with surgeon 2 electing to revise the one-stage cases that Were failures from a hearing point of view rather than do a planned second stage. Other differences between surgeons were secondary to the effect of staging operations.
. We performed exploratory analyses on all hearI?g-result differences greater than 5 dB. There were SIx such analyses; therefore, we had about a 30% probability of finding that one of these analyses would have a p value of .05 or less by chance.
115
Fig 2. Cera vital partial prosthesis as received from manufacturer is shown on left. Partial prosthesis removed from middle ear of 72-year-old woman 3 years after original operation is shown on right. Rule is in millimeters.
group with extensive experience with Plastipore. By contrast, neither of us began using Ceravital before preparing for this study. There may be technical factors that are required for success with Ceravital that cannot be gained from talking to the manufacturer's representatives or from experience in the laboratory. A surgeon starting to use Ceravital should consider getting training from an individual or group with good results from this prosthesis. A factor that may lead to long-term failure with Ceravital prostheses is the variable rate of prosthesis absorption. In animal studies, areas of lysis of the prosthesis material occur that are as deep as 250 I'm at 1 year and as deep as 450 I'm after 2 years.' We had two patients who had a progression in hearing loss at about 3 years related to resorption of the prosthesis in the area in which the prosthesis was making contact with the sound-conducting mechanism (Figs 1 and 2).
We rejected the null hypothesis that differences between Ceravital and Plastipore were insignificant, on the basis of our finding that the Plastipore group was twice as likely as the Ceravital group to have a postoperative air-bone gap of 15 dB or less. F~rther, the Plastipore group had a 6-dB smaller au-bone gap at 4,000 Hz than the Ceravital group.
An unexpected finding was a difference between the hearing results and extrusion rates of surgeon 1 and surgeon 2. Our explanations were related to staging the operation and the length of the prosthesis (the operation was staged according to guidelines described by Sheehy" and Sheehy and Crabtree"). Had we known about the confounding effect of differences in staging, we would have controlled for this variable by having the surgeons use the same indications for staging.
Our results with Plastipore were comparable to those achieved by Gersdorff et aP; however, our results with Ceravital were not as good as the results of the Gersdorff group. One explanation for our Poorer results with Cera vital is that we had several years' prior experience with Plastipore, including one author's having fellowship training with a
Our other unexpected finding, the high extrusion rate for surgeon 1, was probably related to prosthesis length. Our speculation was that surgeon 1 tended to use a longer prosthesis than surgeon 2. A longer prosthesis would be less likely to lateralize from the stapes footplate or stapes capitulum and less likely to medialize from the eardrum or mal-
DISCUSSION
Mangham & Lindeman, Ceravital Versus Plastipore
116
leus. The longer prosthesis would be more likely to extrude through the eardrum, so there is a penalty with these materials for using a long prosthesis.
these two prostheses. Further, the 3-year data tend to underestimate the difference between the Plastipore and the Ceravital group.
Sharp corners predisposed a prosthesis to extrusion. When the platform of a Plastipore prosthesis was cut into a T shape, similar to the Fisch design, the prosthesis tended to extrude at the corners. The Ceravital prosthesis tended to extrude at a corner or at one of the edges of the groove cut in the lateral surface of the platform for the malleus handle.
Another cause of differences among published results would be the bias introduced by the surgeon. In this study, the decision to use Ceravital or Plastipore was made by a random assignment of prosthesis material. By contrast, Niparko et al" used a nonrandom assignment of prosthesis material: "Ceravital. ..was used when an incus autograft was not obtainable or not suitable for reconstruction." Other possible causes of bias include the staging of the operation and whether the malleus was present or absent, as described above, as well as the severity of disease. 5
Results in this study should be compared with caution to results of other studies because of confounding variables such as differences in patient self-selection for follow-up. In this study, for example, the nine patients with Plastipore prostheses who returned for 6-month hearing tests but did not return for 3-year hearing tests had a better average hearing result at 6 months by 5.9 dB than their 42patient cohort that did not drop out of the study. The 11 patients with Ceravital prostheses who dropped out had a better average hearing result at 6 months by 1.4 dB than their 38-patient cohort that did not drop out. This suggests that our 3-year data underestimate the hearing results achieved with ACKNOWLEDGMENTS -
SUMMARY
The hearing results with Plastipore were better than the results with Ceravital. Technical factors such as staging the operation in cases of mucosal loss, the height of the prosthesis, and the shape of the prosthesis platform were as important in determining the final hearing result and extrusion rate as was the material used.
Brent Blumenstein, PhD, and Gloria Bailey, PhD, provided consultation for study design and statistical analysis.
REFERENCES 1. Wehrs RE. Homograft ossicles in tympanoplasty. Laryn-
5. Bellucci RJ. Dual classification of tympanoplasty. Laryngoscope 1973;83:1754-8.
2. Teichgraeber JF, Spector M, Per-Lee JH,Jackson RT. Tissue response to Plasti-Pore and Proplast otologic implants in the middle ears of cats. Am J OtoI1983;5:127-36.
6. Sheehy JL. Plastic sheeting in tympanoplasty. Laryngoscope 1973;83:1144-59.
goscope 1982;5:540-6.
3. Gersdorff MCH, Maisin JP, Munting E. Comparative study of the clinical results obtained by means of Plastipore and ceramic ossicular prosthesis and bone allografts. Am J Otol 1986; 7:294-7. 4. Brackmann DE, Sheehy JL, Luxford WM. TORPs and PORPs in tympanoplasty: a review of 1042 operations. Otolaryngol Head Neck Surg 1984;92:32-7.
7. Reck R. Bioactive glass-ceramics in ear surgery: animal studies and clinical results. Laryngoscope 1984;94(suppl 33). 8. Sheehy JL, Crabtree JA. Tympanoplasty: staging the operation. Laryngoscope 1973;83:1594-621. 9. Niparko JK, Kemink JL, Graham MD, Kartush MD. Bioactive glass ceramic in ossicular reconstruction: a preliminary report. Laryngoscope 1988;98:822-5.
