Prognostic Factors for Short-term Outcomes After Ossiculoplasty Using Multivariate Analysis With Logistic Regression

Objective  To investigate prognostic factors for short-term hearing outcomes after ossiculoplasty.

Design  Retrospective study.

Setting  Tertiary referral and academic center.

Patients  Seven hundred twenty patients who underwent ossiculoplasty performed by a single surgeon from January 1, 1989, through December 31, 2006, and who were followed up for longer than 1 year.

Main Outcome Measures  Hearing outcomes were considered successful if the postoperative air-bone gap was 20 dB or less. The prognostic factors were analyzed using multivariate analysis with logistic regression.

Results  Hearing outcomes were successful in 505 patients (70.1%). Presence of the stapes superstructure, presence of the malleus handle, normal mucosa, normal stapes mobility, and use of local anesthesia were significantly favorable predictive factors.

Conclusions  Multivariate analysis should be performed to investigate prognostic factors of favorable short-term hearing outcomes after ossiculoplasty. Better knowledge of these predictive factors may contribute to the surgeon's judgment and the information given to patients.

The purpose of ossiculoplasty is to improve the hearing level. Many factors, such as eustachian tube function, the severity of disease, the presence of residual ossicular chain, the surgeon's skill, and the length of follow-up, affect the outcome of ossiculoplasty. In this article, we report prognostic factors for short-term hearing outcomes after ossiculoplasty performed by a single surgeon (Y.M.). We used multivariate analysis with logistic regression because hearing outcomes after ossiculoplasty can be affected by many factors and confounding factors should be excluded. Furthermore, we compared the significant prognostic factors we identified with those listed in previous reports.

From January 1, 1989, through December 31, 2006, 746 patients underwent ossiculoplasty performed by a single surgeon (Y.M.) at Osaka Rosai Hospital or Osaka University Hospital, Osaka, Japan, or Hyogo College of Medicine Hospital, Nishinomiya City, Japan. Of these, 720 patients (96.5%) were followed up for longer than 1 year. The pathogenesis of these 720 cases consisted of 458 cholesteatomas (63.6%), 126 cases of chronic otitis media with perforation (17.5%), 35 atelectatic ears (4.9%), and 101 other diseases such as tympanosclerosis and middle ear anomaly (14.0%). Table 1 lists the demographic and clinical characteristics of the patients.

Until 1991, we preferred hydroxyapatite as the material for ossiculoplasty, but some extrusions of hydroxyapatite occurred during follow-up. Therefore, we have preferred autologous materials since 1992. The first choice is to use autologous ossicles, but, in patients with cholesteatoma, the ossicles may be missing or the use of ossicles affected by cholesteatoma may carry a risk of implanting cholesteatoma. The second choice is autologous auricular cartilage. However, auricular cartilage may sometimes be very thin in small women or in children. The third choice in such cases is autologous cortical bone. In our series, autologous auricular cartilage was used in most of the patients (521 [72.4%]).

During these 18 years, various techniques had been performed. However, in most of the patients who had the stapes superstructure, autologous incus was used for the incus transposition technique and cartilage was used as a double cartilage block (as described by Luetje and Denninghoff¹). In patients without the stapes superstructure, we used incus or cartilage to create a slender prosthesis. We attached greater importance to placing the prosthesis vertical to the footplate rather than using the malleus handle. A 1-stage operation was performed in 516 patients (71.7%), and a 2-stage operation was performed in 204 (28.3%).

Among the 720 patients undergoing ossiculoplasty, the stapes superstructure was present in 509 (70.7%) and the malleus handle was present in 552 (76.7%). Five hundred sixty patients (77.8%) underwent primary surgery, and 160 (22.2%) underwent revision surgery. General anesthesia was used in most patients (593 [82.4%]).

The mobility of the stapes or the footplate was classified as normal or deteriorated (including fixed) in the surgical records, which noted the subjective estimation of the surgeon during the surgery, and was classified as normal in 577 patients (80.1%). The mucosa of the tympanic cavity was identified as normal or diseased (ie, thick or edematous) in the surgical records and was specified as normal in 544 patients (75.6%).

Audiometric data were calculated according to the 1995 American Academy of Otolaryngology–Head and Neck Surgery guidelines.² However, thresholds at 3 kHz are not usually measured in Japan. Therefore, the mean of thresholds at 2 and 4 kHz were used to represent thresholds at 3 kHz. Postoperative hearing outcomes were categorized in 4 groups of 10-dB increments, and the hearing outcome was considered successful if the postoperative air-bone gap was 20 dB or less. The air conduction threshold 1 year after ossiculoplasty was used as the postoperative air-conduction threshold to determine the short-term outcome.

We used logistic regression analysis to predict which factors affect the short-term hearing outcomes of success vs failure after ossiculoplasty. The variables included in the logistic regression analysis were sex (male vs female), patient age (<16 vs ≥16 years), stapes superstructure (present vs absent), mucosa (normal vs diseased), stapes mobility (normal vs deteriorated), anesthesia (local vs general), type of surgery (primary vs revision), operation type (1-stage vs 2-stage), malleus handle (present vs absent), kind of prosthesis (cartilage vs others), and cholesteatoma (present vs absent). We used SPSS version 11.0J statistical software (SPSS Inc, Chicago, Illinois) to perform the analysis. P < .05 was considered significant.

Successful hearing outcomes were achieved in 505 patients (70.1%). Table 2 shows univariate analysis of short-term hearing outcomes after ossiculoplasty using logistic regression analysis. The presence of the stapes superstructure and the malleus handle, normal mucosa, primary surgery, and a 1-stage operation were significantly favorable predictors of ossiculoplasty outcome. Sex and age were excluded from multivariate analysis. Using backward selection, the presence of the stapes superstructure, malleus handle, and normal mucosa, and cholesteatoma; primary surgery; the operation type; stapes mobility; and the type of material and anesthesia used were tested by multivariate analysis. Cholesteatoma presence was dropped from the model first, then primary surgery, operation type, and material type used. In the final model, the presence of the stapes superstructure, the malleus handle, and normal mucosa; normal stapes mobility; and the use of local anesthesia remained significant predictive factors (Table 3).

Many reports have detailed the prognostic factors of tympanoplasty including ossiculoplasty.^3-11 Bellucci³ classified all cases into those with a good prognosis (group 1), a fair prognosis (group 2), a poor prognosis (group 3), and a very poor prognosis (group 4) according to the degree of otorrhea and eustachian tube function. The author concluded that excellent outcomes could only be obtained in group 1; in the other 3 groups, outcomes were less successful. Austin⁴ advocated the following prognostic stratification: (1) disease categories, (2) disease stage categories, and (3) disease descriptors. Kartush⁵ proposed the Middle Ear Risk Index, which is a scoring system based on otorrhea, perforation, cholesteatoma, ossicular chain, middle ear granulation/effusion, and previous surgery. Becvarovski and Kartush⁶ reported that smoking is a risk factor not only for preoperative middle ear disease but also for long-term graft failure. They added smoking to the revised Middle Ear Risk Index.

Black⁷ reported the prognosis of 535 ossiculoplasties after an investigation of 5 types of factors: surgical, prosthetic, infection, tissue, and eustachian tube (the SPITE factors). He concluded that there were 12 prognostic factors, including (1) complex surgery, (2) requirement of major scutum repair plus myringoplasty, (3) absence of the malleus handle, (4) absence of the stapes superstructure, (5) a 50-dB air-bone gap, (6) unremitting otorrhea, (7) chronic myringitis, (8) the general condition of the patient, (9) meatoplasty involving the tympanic membrane, (10) presence of damaged or diseased mucosa, (11) presence of effusion, and (12) severe pars tensa collapse. However, the author performed only univariate analysis using the χ² test.

Many factors can affect outcomes after ossiculoplasty. Therefore, multivariate analysis should be preferred. For example, in this series, primary surgery and use of a 1-stage operation were significant predictive factors in univariate analysis but were no longer significant in multivariate analysis. Table 4 demonstrates why use of a 1-stage operation dropped out on multivariate analysis. The stapes superstructure was present in 76.2% of patients undergoing a 1-stage operation but was present in only 23.8% of patients undergoing a 2-stage operation. Thus, there was a bias in the ratio of patients with the stapes superstructure present, which was a confounding factor between the operation type and hearing outcomes. To avoid such a confounding factor, multivariate analysis is recommended.

Meanwhile, Mills,⁸ Albu et al,⁹ Dornhoffer and Gardner,¹⁰ and Yung and Vowler¹¹ reported prognostic factors of ossiculoplasty using multivariate analysis. Mills⁸ reported that the loss of the stapes arch was a factor causing a significantly worse outcome, but only 55 ossiculoplasties were analyzed. Albu et al⁹ examined prognostic factors in 544 ossiculoplasties and concluded that the presence of the malleus handle and the mucosal status were the most important predictors in simple, granulating chronic otitis media and in the use of the canal wall-up technique in cholesteatoma. Dornhoffer and Gardner¹⁰ reported the prognostic factors of 200 ossiculoplasties and concluded that mucosal status, presence of the malleus handle, otorrhea, mastoidectomy, and revision surgery were significant prognostic factors. They advocated the Ossiculoplasty Outcome Parameter Staging index, which is a scoring system based on these factors. Yung and Vowler¹¹ reported the long-term outcomes of 145 ossiculoplasties and concluded that the presence of the malleus handle was the only significant factor.

In our patient series, mucosal status and the presence of the stapes superstructure and the malleus handle were significant predictors, which is similar to most of the previous reports. Stapes mobility was also a significant predictor; stapes mobility is subjectively estimated by the surgeon during surgery but is very important for ossiculoplasty. We tried to remove granulation or calcification around the stapes superstructure to mobilize it. Performing stapes surgery in cases of tympanosclerotic stapes fixation has been controversial.^12-15 In this series, we performed only mobilization in such cases, not stapes surgery. Further study is necessary to determine how to deal with stapes showing deteriorated mobility. The merit of local anesthesia, which was also a significantly favorable predictor in our patients, is that hearing can be confirmed during surgery. If possible, local anesthesia should be recommended for patients with several poor prognostic factors. Further study is necessary to confirm whether our model shows good fit.

In conclusion, the presence of the stapes superstructure, malleus handle, and normal mucosa; normal stapes mobility; and the use of local anesthesia are significant predictive factors after performing multivariate analysis with logistic regression. Multivariate analysis is preferred for investigating prognostic factors of ossiculoplasty. Better knowledge of these predictive factors may contribute to the surgeon's judgment and the information given to patients.

Correspondence: Yasuo Mishiro, MD, Department of Otolaryngology, Hyogo College of Medicine, 1-1, Mukogawa-cho, Nishinomiya, Hyogo 663-8501, Japan (ymishiro@hyo-med.ac.jp).

Submitted for Publication: January 7, 2009; final revision received March 21, 2009; accepted April 12, 2009.

Author Contributions: Drs Mishiro, Sakagami, and Kakutani had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mishiro and Sakagami. Acquisition of data: Mishiro and Adachi. Analysis and interpretation of data: Mishiro and Kakutani. Drafting of the manuscript: Mishiro. Critical revision of the manuscript for important intellectual content: Sakagami, Adachi, and Kakutani. Statistical analysis: Mishiro and Kakutani. Administrative, technical, and material support: Mishiro and Adachi. Study supervision: Sakagami.

Financial Disclosure: None reported.