The evaluation of canal wall up cholesteatoma surgery with the Glasgow Benefit Inventory

Patients aged 15 years or older who had primary surgery for congenital or acquired cholesteatoma at a tertiary-stage hospital, Linköping University Hospital, and a secondary-stage hospital, Vrinnevi Hospital, Norrköping, between January 2005 and December 2009 were included. Exclusion criteria were isolated atticoantrotomies, isolated myringo-ossiculoplasties, and canal wall down (CWD) procedures. Further, exclusion criteria were children under the age of 15 years at the time of surgery and patients who were not able to answer the questionnaire because of mental disorders, dementia, or poor knowledge of the Swedish language, needing an interpreter. The GBI questionnaire, information about the study, and a prepaid self-addressed envelope were sent to all patients included in the study (n = 47) in 2012. A written informed consent was received from the participants or the guardian if the patient was under the age of 18 years. Non-respondents were contacted twice to be reminded of the study. Cholesteatoma surgery using canal wall up (CWU) with obliteration was performed as described in our previous study [15].

The GBI is an 18-item questionnaire for measuring changes in health-related benefits post-ORL interventions. The questions are answered using a 5-point Likert scale, ranging from a large deterioration (1 point) to a large improvement (5 points) in health status. To avoid response bias, half of the questions are graded from a large deterioration to a large improvement and half in the opposite order. The GBI contains a total score (18 questions) and three sub scores: general benefit (12 questions), social support (3 questions), and physical benefit (3 questions). When calculating the scores, the mean score for each group is subtracted by 3 and multiplied by 50. The worst possible change becomes − 100, and the best possible change becomes + 100 [13].

Audiograms were accessed from patient files. The pure tone average (PTA) was defined as the mean value of hearing thresholds at the frequencies 0.5, 1.0, 2.0, and 3.0 kHz, as proposed by the British Society of Audiology [16]. A hearing threshold of 30 dB PTA was used as the level for socially adequate hearing [17]. Air conduction thresholds exceeding the maximum output level of clinical audiometers were registered as 130 dB HL to enable the assessment of possible postoperative deterioration and avoid unnecessary dropouts [16]. Bone conduction thresholds exceeding the maximum output level of clinical audiometers were registered as missing. The audiometry was performed with clinical audiometers (Madsen OB822, Otometrics, Denmark), at ear, nose and throat clinics in Linköping and Norrköping. The audiometers were regularly calibrated according to international standards [18]. Audiograms selected as preoperative were the ones performed closest to surgery. One-year postoperative hearing results were based on audiograms obtained closest to 1 year after surgery, with the earliest accepted time being 9 months postoperatively. Three-year postoperative hearing results were based on audiograms obtained closest to 3 years after surgery, with the earliest accepted time being 1.5 years postoperatively. The gain was calculated as preoperative–postoperative PTA.

Statistical analyses were performed using IBM^® SPSS^® (version 23), Microsoft Excel, and Medlog^® database program. Linear regression analysis was used to determine the association between GBI score and age, gender, years since surgery, 1-year postoperative gain, PTA, and mean PTA ≤ 30 dB. Audiometric results and GBI scores are reported as mean ± standard deviation (SD).

In all patients, a CWU procedure with obliteration was performed. There were no findings of residual or recurrent cholesteatomas in the study group. Partial ossicular replacement prosthesis (PORP) and total ossicular replacement prosthesis (TORP) were used for ossicular reconstruction in 26 and 4 patients, respectively. The autologous incus or cortical bone was chosen for ossicular reconstruction. Fascia or perichondrium was used as a graft material for the tympanic membrane. Ossicular reconstruction was not performed in four patients. In one of these four patients, an ossicular chain defect was detected, but reconstruction was not possible because of a facial nerve dehiscence. Surgery was performed by a senior surgeon on 26 (76%) of the patients and by a junior surgeon on the rest of the patients under the supervision of a senior surgeon (see Table 1).

There was an overall hearing improvement in the study group (n = 34) at 1 and 3 years postoperatively. Audiograms were possible to achieve in 33 patients at 1-year and in 27 patients at 3-year postoperative follow up. The mean PTA for air conduction was 47 dB (SD 20) preoperatively, and 29 dB (SD 18) at 1 year and 33 dB (SD 21) at 3 years postoperatively. In the PORP subgroup (n = 26), 1- and 3-year follow up audiograms were achieved in 25 and 21 patients, respectively, with a mean PTA for air conduction being 45 dB (SD 19) preoperatively, 25 dB (SD 10) at 1 year and 28 dB (SD 13) at 3 years postoperatively. A follow up with audiogram in the TORP (n = 4) group as well as in the group of “no reconstruction of ossicular chain” (n = 4) was achieved in four patients at the one year postoperative control, and in three patients at the three year postoperative control. In the TORP subgroup, the mean PTA for air conduction was 61 dB (SD 20) preoperatively, and 52 dB (SD 33) at 1 year and 39 dB (SD 16) at 3 years postoperatively. In the group without ossicular reconstruction, the mean PTA for air conduction was 42 dB (SD 25) preoperatively, and 35 dB (SD 19) at 1 year and 51 dB (SD 34) at 3 years postoperatively. Gain at 1 year postoperatively was overall improved. The highest mean gain, 20 dB (SD 16), was seen in the PORP subgroup. The proportion of patients within PTA ≤ 30 dB increased from 9 (26%) to 20 (61%) at 1 year postoperatively and to 15 (60%) at 3 years postoperatively. An interaural difference of ≤ 15 dB was obtained in 18 (58%) of the patients at 1 year postoperatively, compared to 7 (21%) of the patients preoperatively. As described above, the restoration of hearing was not possible in one of the patients. The same patient suffered from deteriorated hearing at 2 years postoperatively, without any known reason. In one patient, an ossiculoplasty revision was performed after the 1-year and before the 3-year postoperative audiograms were recorded. There were no cases of total hearing loss after surgery. Four patients presented a negative overall GBI score. Of these, the mean PTA for air conduction of ≤ 30 dB was achieved in two of the patients, whereas two of them had a mean of > 30 dB at 1 and 3 years postoperatively. The interaural PTA air conduction difference was between 19 and 34 dB in four patients with a negative GBI score.

In the non-respondent group (n = 13), the mean PTA for air conduction was 39 dB (SD 22) preoperatively, compared to 38 dB (SD 26; n = 13) and 42 dB (SD 27; n = 9) at 1 and 3 years postoperatively. The mean gain was 3 dB (SD 13; n = 11) at 1 year postoperatively and 2 dB (SD 14; n = 9) at 3 years postoperatively. The proportion of patients with the mean PTA for air conduction of ≤ 30 dB was 6 (46%; n = 13) preoperatively, compared to 7 (58%; n = 12) and 4 (44%; n = 9) at 1 and 3 years postoperatively. No cases of total hearing loss after surgery were seen.

The means and SDs of GBI scores were reported. The overall GBI score was 21 (SD 22). In the group of 4 patients under the age of 18, the total GBI score was 39 (SD 26). In the PORP, TORP, and no ossicular reconstruction subgroups, the total scores were 24 (SD 23), 19 (SD 8), and 8 (SD 11), respectively. The general, social, and physical health subscale scores for the total study group were 26 (SD 27), 10 (SD 21), and 13 (SD 24), respectively, as provided in Table 2. Twenty-nine (85%) of the patients benefitted from surgery, one (3%) patient had no change, and four (12%) expressed deterioration after surgery (see Table 3). One patient scored between 1 and 2 in question 1, and we chose to use the lower score for calculations. There was no effect on the GBI score of age, gender, years since surgery or 1-year postoperative gain, PTA, or mean PTA ≤ 30 dB, as mentioned in Table 4.

A poor response rate is not seldom the case in retrospective questionnaire studies. There is, thus, a risk that the participants in the study would only represent a smaller part of the entirety. From the consecutive group of patients with performed cholesteatoma surgery using CWU in the present study, there was an equal participation of males 16 (47%) and females 18 (53%) and also the spread of ages was equal (15–76 years). As we routinely have a 5-year follow-up for adults and a 10-year follow-up for children (<18 years) after cholesteatoma surgery, the majority of patients in the study group were still regularly offered clinical check-ups. This fact might have been an incitement to participation for some patients. The response rate was 72%, which is the same percentage as presented in the study by Robinson et al. [13], for the ear surgery group aiming at eradicating ear activity.

Robinson et al. [13] studied 138 patients for a maximum of 6 years after ear surgery. Despite the limitation of a retrospective questionnaire as the memories of patients of surgery and its affects would fade with time, it seems that it did not affect the overall score [13]. This was in line with our data, where the longest time from surgery to answering the questionnaire was 7 years. The results of the GBI questionnaire give information about the present status of the patients and have to be interpreted in relation to the “time factor”. Maile et al. [14] had a follow-up at “the postoperative clinical appointment” and suggested a later GBI follow-up to their study group to quantify long-term changes in HRQoL following surgery [14].

Smyth and Patterson [20] identified higher satisfaction among patients after myringo-ossiculoplasties in case of smaller interaural hearing differences, greater gains, and better postoperative hearing thresholds. According to the Belfast rule of thumb, patients would not consider their hearing to be improved unless the hearing threshold is ≤ 30 dB or within 15 dB of the contralateral ear [12]. Nadol et al. [7] saw a strong correlation between the preoperative and postoperative hearing results and the CES scores. We did not see a significant effect on the overall GBI score of postoperative hearing data (Table 4), but four patients with a negative total GBI score had a greater interaural difference than that of the mean of the total group. A correlation between hearing amelioration and GBI results could have been expected. The modest GBI scores, despite a low frequency of recurrence and residual cholesteatomas, a lack of discharge problems and only one ossicular revision in the study group, could possibly partly be explained by the fact that the symptoms preoperatively were scanty.

Hendry et al. [21] recommended the reporting of GBI scores of the percentage of patients for whom there was no or negative benefit with surgery. As seen above, four (12%) of the patients had a negative total score. The first patient, with − 28 as the total GBI score, had acute surgery for labyrinthitis and a labyrinthine fistula caused by cholesteatoma. PTA air conduction was 44 dB at 1 year postoperatively, and a subsequent sudden hearing loss occurred at 2.5 years postoperatively. MRI and CT scans were performed with no detected residual or recurrent disease.

The second patient with a deteriorated GBI score (− 8) had an attic cholesteatoma, a history of Eustachian tube dysfunction, and ear problems since childhood. There were no surgical complications. At 1-year postoperative control, PTA air conduction was 20 dB with a gain of 15 dB, compared to preoperative hearing. The questionnaire was filled in 4 years postoperatively. In this case, it is difficult to explain why the patient’s GBI score was low, but it might mirror the difficult task to convey realistic expectations of surgery.

The third patient with a deteriorated GBI score (− 6) had an attic cholesteatoma. Because of a difficult anaesthetic situation, the surgery had to be cancelled and postponed to a few days later for awake intubation. At 1 year postoperatively, the patient had a conductive hearing loss with a mean PTA for air conduction of 37 dB because of the fixation of ossicular reconstruction and retraction problems contributing to a difficult hearing situation.

The fourth patient with a deteriorated GBI score (− 5) also with an attic cholesteatoma showed a mean PTA for air conduction of 19 dB at 1 year postoperatively. By the time of filling in the questionnaire, tinnitus had newly developed in the contralateral ear. An MRI was performed, but it did not show any detectable pathology.

It was not possible to address the negative GBI scores to a specific postoperative parameter in this study, but hearing problems and possibly interaural difference in hearing thresholds could partly be responsible for the deterioration. The number of labyrinthine fistulas peroperative were too small in this study group for drawing further conclusions. One of the patients with labyrinthine fistula showed negative GBI scores, one had positive GBI scores, and the third unfortunately did not answer the questionnaire.

The Swedish translated version of the GBI questionnaire is currently being validated by a Swedish research group, but the conclusion is not yet published. However, the GBI questionnaire is used in the clinic by several members of the Swedish Association for Ear Surgery.

The sample size is small, and in the study group, there was only one revision surgery and no residual or recurrent disease, which makes generalization difficult. Data from the current 34 patients are in line with the group of 230 patients from our earlier study [15], showing a low residual and recurrent rate. In the study by Maile et al. [14], 31 patients with cholesteatoma were included, which is a sample size similar to ours. As there is a lack of publications in the field of HRQoL and GBI for patients with cholesteatoma, comparison with other studies is limited. In the subgroups of “no reconstruction of ossicular chain”, and the TORP group only contains four patients, and at the 3 year follow up only three patients present hearing data. This could explain the ameliorated hearing between one and 3 year postoperative control.

There is always a risk of positive bias after interventions, which is not unique to our study, as patients tend to appreciate the surgical attention and caretaking [22]. To dampen this effect, an impartial questionnaire administration could possibly be an advantage.

Another general problem with questionnaires is exclusion on the basis of language skills. Patients with poor knowledge of the Swedish language were excluded from the study, which has to be considered when interpreting these data. However, the number of non-Swedish-speaking patients excluded from the study was low in this study group.

The study group is part of the cohort earlier described [15]. Therefore, these data are useful as a complement to the previously reported hearing data and surgical outcome. The GBI is a non-disease-specific tool for the measurement of HRQoL in contrast to, for example, the CES, but to its advantage, it is possible to make comparisons between different interventions [23].