Erratum to: Stability, survival, and tolerability of a 4.5-mm-wide bone-anchored hearing implant: 6-month data from a randomized controlled clinical trial

Erratum to: Eur Arch Otorhinolaryngol DOI 10.1007/s00405-015-3593-x

Despite stringent data recording and accurate data monitoring, to our regret, one patient was mistakenly included in the control population, while she should have been included in the test population. Therefore, population sizes and, thus, the data analysis were not correct. Upon this coincidental identification, we have verified all data once again, which revealed no other incorrectness. Although this concerned only one patient and no significant impact was expected, it was inevitable to re-open the locked database for the statisticians to run new analyses. Indeed, insignificant differences in results were found and the primary and secondary outcomes remained unchanged. Consequently, the conclusions drawn in the original manuscript remained the same. Nevertheless, we report the exact newly computed results below.

Table 1 was updated to describe the correct populations. The baseline demographics remained largely the same, with no statistically significant differences between test (n = 39 patients) and control (n = 20 patients) populations. The mean AUC for ISQ low remained 64.4 (SD 2.9; range 55.5–70.1) for test implants and was 59.3 (SD 2.1; range 55.5–62.5) for control implants, a statistically significant difference of 5.1 ISQ points (95 % CI 3.6–6.6; p < 0.0001). For ISQ high, a difference of 3.3 (95 % CI 1.8–4.7; p < 0.0001) was observed, with a mean AUC of 65.8 (SD 2.7; range 57.0–70.5) for the test implant and 62.5 (SD 2.8; range 56.9–66.8) for the control implant. The mean increase in ISQ low was 5.1 (SD 4.9; range −4 to 29) in the test group and 3.3 (SD 3.8; range −3 to 13) in the control group. The mean difference in the increase in ISQ low between both groups was statistically significant (95 % CI −0.7 to 4.4; p = 0.046). The analysis of soft tissue statuses revealed findings of Holgers grade 0 in 87.1 % (test) and 88.4 % (control) of visits, Holgers grade 1 in 11.8 % (test) and 9.5 % (control) of visits, Holgers grade 2 in 1.1 % (test) and 1.4 % (control) of visits, Holgers grade 3 in 0.0 % (test) and 0.7 % (control) of visits, and no Holgers grade 4 cases over all of the visits. The only complication rate that differed from the original analyses was postoperative wound dehiscence, which occurred in three test implants (7.7 %) and two control implants (10.0 %). All these corrected outcomes are summed up in Table 3. The GBI comparison remained largely the same and not statistically significantly different in the total score (33.1 test versus 36.5 control), general subscale (46.9 test versus 50.7 control), and social subscale (11.0 test versus 10.0 control), but did change to a statistically significant difference in the physical subscale (1.28 test versus 6.67 control; p = 0.037). Table 2 was updated to include these corrected GBI outcomes. Despite this difference being statistically significant, we do not consider this difference clinically important, as both scores are low and clinical outcomes are comparable.

Table 1 Corrected patient demographics and baseline characteristics

Table 2 Corrected subjective benefit as measured by the GBI

Table 3 Overview of corrected outcomes