*2.7. Statistical Tests*

Data were managed with Excel software (Office 2010, Microsoft Inc. Redmond, VI, USA) and Graphpad prism (v.6, Graphpad Inc., San Diego, CA, USA). Continuous variables were presented as mean ± standard error of mean (SEM) [min.-max.] and nominal variables were noted as *n* (%). Comparison of continuous parameters in 2 groups was studied by paired or unpaired t-tests. Continuous variables in multiple groups were tested by oneor two-way ANOVA. Music test scores were compared to the random level (score 5 out of 10) for each test by a one-sample t-test. A *p*-value < 0.05 was considered as significant. Linear regression analysis was conducted by F-test for the slope of the regression line and R for goodness of fit. Correlations were considered significant when R > 0.5 and *p* < 0.05. Test–retest reliability was tested by Cronbach's alpha. A value in the range of [0.8–0.9] was considered as good and >0.9 as excellent. To control for the effect of the usual strategy used by the patients in their music test performances, a mixed-model analysis was used

to compare the results of the 4 music tests with the CystalisXDP versus MPIS program as a function of their usual strategy. A separate model was mixed employed for the global music score.

**Table 3. Musical Questionnaire part 2: Music Perception with Cochlear Implant.** Numbers indicate the number of choices among propositions, positive responses or Likert scores (mean ± standard error of mean, range, *n* = 21). MPIS (*n* = 6) and Crystalis XDP (*n* = 15) refer to the usual strategies used by the patients. CI: cochlear implant.


The population size was estimated for test–retest reliability by setting α = 0.05, β = 0.1, k (number of test items) = 4, the value of Cronbach's alpha at null hypothesis = 0, and the expected value of Cronbach's alpha = 0.75. The required number was evaluated as 17 subjects according to Bonnett [42] and increased to 21 to account for potential loss to follow-up at the retest.

#### **3. Results**

#### *3.1. CI and Sound Processing Strategies*

The number of active electrodes was 15 ± 0.7 (*n* = 21): 11 ± 2.5 for patients with Digisonic DX10 (*n* = 5), 18 ± 1.5 for unilateral Digisonic SP (*n* = 12), 12 on each side for binaural CI (*n* = 3), and 16 and 18 for the bilateral Digisonic SP. Before inclusion, 6 patients were fitted with the standard program (MPIS) and 15 already used CrystalisXDP. Patients using CrystalisXDP before inclusion performed similarly to those with a standard program as assessed by WDS (78 ± 6.5% *n* = 15, versus 60 ± 13.9, *n* = 6, not significant, unpaired t-test followed by Bonferroni). Speech performances were not related to the number of active electrodes in this group (WDS: 83 ± 5.8%, *n* = 15, versus 48 ± 10.5, *n* = 6, respectively, not significant, unpaired t-test, followed by Bonferroni correction).

#### *3.2. Musical Experience*

At inclusion, the questionnaire revealed that music was important in the daily life of this group (average Likert score 3.6 ± 1.20, with 18 patients (86%) scoring >3 out of 5, Table 2). The implantation did not change the frequency of music listening (response to "How often?", not significant, chi-2 test), or the type of music (not significant, chi-2 test). The majority (18, 86%) continued to listen for pleasure (Table 3) and practiced active music listening (17, 81%). While most declared being capable of recognizing a known melody (18, 86%), the musical style (15, 71%), and even the lyrics (15, 71%), only a few declared being capable of detecting a wrong note (6, 29%), singing in tune (5, 24%) or singing in public (2, 10%) underlining the inherent CI limitations in frequency discrimination.

CI negatively impacted music activities in this group. After implantation, many patients stopped musical activity such as music lessons (6 out of 7), playing an instrument (3 out of 6) or singing (3 out of 9). However, most declared training themselves with music after CI (13, 62%).
