HSC Section 8_April 2017

CROS VS BAHD FOR SINGLE-SIDED DEAFNESS

All reported p values for parametric post hoc measures were adjusted using Bonferroni corrections. Confidence intervals (CI) were adjusted for within-subjects testing using the Cousineau- Morey method with Baguley’s corrections (30).

QuickSIN was administered at 50 dB HL in soundfield. Again, two listening conditions were assessed and counterbalanced across participants: the S0Npe (sentences at 0 degree azimuth, noise at 90 degrees azimuth to poor ear) and the S0Nbe (sen- tences at 0 degree azimuth, noise at 90 degrees azimuth to the better ear) condition. For each condition, two different lists were presented and the scores averaged. CROS Hearing Aid Fitting Participants were randomly assigned to be fitted with the CROS hearing aid either at the end of the first visit or at the end of the second visit. They were given a 2-week trial with the CROS and asked to refrain from wearing their BAHD during that time. At the end of the first visit, four participants were fitted with the CROS hearing aid and were subsequently tested with this device on their second visit, at which time they were instructed to use the BAHD for the next 2 weeks. This protocol was reversed in the other four participants, with testing after each device experience. CROS hearing aids were fitted with a retainer earhook on the poorer ear and a slim tube with an open dome on the better ear. The hearing aid’s response was verified with real-ear measures (Audioscan Verifit) using Dillon’s (26) recommended approach for probe microphone verification of CROS hearing aids. Dur- ing real-ear verification, the hearing aid’s response was fine- tuned as needed. Adaptive directional microphone and noise reduction algorithm were active. Aided Measures With CROS and BAHD Participants were tested with either the CROS or BAHD, depending on which device they had been instructed to use during the previous 2 weeks. Similar to the baseline measures, aided soundfield thresholds were obtained with warble tones delivered at 90 degrees azimuth to the better ear in one condition and at 90 degrees azimuth to the poorer ear in the other condi- tion. Aided word recognition testing and aided QuickSIN were administered using the same protocol as for unaided measures. Moreover, two self-assessment questionnaires, the Bern Benefit in Single-Sided Deafness Questionnaire (BBSS) and the Speech Spatial Qualities Questionnaire (SSQ), were given to assess the self-perceived benefits provided by the device that was worn the previous 2 weeks. The BBSS (27) is a 10-item questionnaire where participants rate the benefit derived from their device in different situations, with ratings that range from j 5 (‘‘Much Easier Without the Aid’’) to +5 (‘‘Much Easier With the Aid’’). The SSQ (28) requires participants to rate their perceived hear- ing ability for 49 scenarios using a 10-point scale, ranging from ‘‘Not at all’’ to ‘‘Perfectly.’’ Participants were asked to fill out a diary during the 2-week period for the CROS and for the BAHD; they were instructed to indicate the date the device was worn, hours of device use per day, and specific situations in which the device was worn. Fi- nally, at the end of the study, the researcher verbally asked participants whether the CROS or BAHD was preferred and questioned them on the specific reasons for their choice. Statistical Analyses Statistical analyses were conducted in R (29). An alpha of 0.05 was assumed for all analyses. Shapiro-Wilks tests were used to assess normality and Mauchly’s test was used to assess sphericity in all repeated measures analyses. No violations of sphericity were detected. Repeated measures ANOVAs were used to com- pare conditions for normally distributed data, otherwise statistical analysis was performed using Friedman’s non-parametric test.

RESULTS

Head Shadow Figure 1 shows mean head-shadow values (i.e., thresh- old increases for sounds presented to the poorer ear) at each frequency in all three conditions, with error bars showing 95% confidence intervals. Significant devia- tions from normality were detected for the unaided con- dition (W = 0.912, p = 0.013) and the CROS condition (W = 0.896, p = 0.005), so a Friedman’s nonparametric ANOVA was used to test for within-subject device- dependent differences in head shadow across all fre- quencies. There was a significant main effect of device ( W 2 (2) = 21.769, p = 0.00002). Post hoc analyses con- ducted using the Friedman post hoc analysis (31) showed significant reductions in head shadow for both the BAHD (mean reduction of 25 dB) and the CROS (mean reduc- tion of 32 dB), relative to the unaided condition. The 7-dB advantage for the CROS over the BAHD did not exceed the critical difference (19.15 dB) required for significance. Word Recognition Scores The mean word recognition percentages for each de- vice condition are displayed in Figure 2. Error bars show 95% confidence intervals. Note that the scores for the quiet condition are close to 100%, suggesting a ceiling effect. This observation was indeed the case, as 10 of the scores (42%) were perfect. The quiet condition was therefore excluded from further statistical analysis. Data in the remaining conditions were normally distributed and were analyzed with a repeated measures ANOVA. The ear to which the noise was presented (better or poorer) had a significant effect on word recognition scores (F(1,7) = 32.82, p = 0.0007, G 2 = 0.329), but there was no signifi- cant main effect of device ( p 9 0.05). The interaction of

FIG. 1. Mean head shadow in the three device conditions. Error bars denote 95% confidence intervals.

Otology & Neurotology, Vol. 36, No. 5, 2015

150