Treatments for hearing loss in osteogenesis imperfecta: a systematic review and meta-analysis on their efficacy

Abstract

About 70% of people with osteogenesis imperfecta (OI) experience hearing loss. There is no cure for OI, and therapies to ameliorate hearing loss rely on conventional treatments for auditory impairments in the general population. The success rate of these treatments in the OI population with poor collagenous tissues is still unclear. Here, we conduct a systematic review and meta-analysis on the efficacy of treatments addressing hearing loss in OI. This study conforms to the reporting standards of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA). Data sources include published articles in Medline via PubMed, Web of Science, Scopus, and Embase, from their inception to November 2020. Studies included individuals with OI undergoing a hearing loss treatment, having pre- and postoperative objective assessment of hearing function at a specified follow-up length. Our search identified 1144 articles, of which 67 were reviewed at full-text screening. A random-effects meta-analysis was conducted on the selected articles (n = 12) of people with OI that underwent stapes surgery. Success was assessed as the proportion of ears with a postoperative Air-Bone Gap (ABG) ≤ 10 dB. A systematic review was conducted on the remaining articles (n = 13) reporting on other treatments. No meta-analysis was conducted on the latter due to the low number of articles on the topic and the nature of single case studies. The meta-analysis shows that stapes surgeries have a low success rate of 59.08 (95% CI 45.87 to 71.66) in the OI population. The systematic review revealed that cochlear implants, bone-anchored hearing aids, and other implantable hearing aids proved to be feasible, although challenging, in the OI population, with only 2 unsuccessful cases among the 16 reviewed single cases. This analysis of published data on OI shows poor clinical outcomes for the procedures addressing hearing loss. Further studies on hearing loss treatments for OI people are needed. Notably, the mechanisms of hearing loss in OI need to be determined to develop successful and possibly non-invasive treatment strategies.

Figure 1

PRISMA flow diagram. A total of 1144 articles were retrieved from the search process. After removing duplicates, 198 studies were screened at the abstract level. After removing articles (n = 131) that deviated from the objective of this study, 67 articles were reviewed at full text. 42 articles did not satisfy the inclusion criteria and were excluded from subsequent screening. Among the articles left, 13 met the inclusion criteria for the systematic review, and 12 different studies met the inclusion criteria for the meta-analysis.

Figure 2

Efficacy of stapes surgery in the osteogenesis imperfecta population in the short term. Studies included for meta-analysis are shown in the table together with the overall result with its corresponding sample size, effect size represented as a proportion (proportion of ears with a postoperative Air Bone Gap (ABG) ≤ 10 dB), 95% confidence interval (CI), and weight for random effect model. The forest plot graphically represents the results. There, each study is represented with a box. The size of each box represents the weight of each study whereas the lines determine their 95% CI. The overall result is shown by a diamond in which the middle indicates the estimate, and the sides represent its 95% CI. Our meta-analysis shows an overall result of 59.01% (95% CI 45.87 to 71.66). The test for heterogeneity table shows Cochrane’s Q test, degrees of freedom (DF), significance level, inconsistency value ($I^2$), and its corresponding 95% CI. Significative heterogeneity was found in our results (P < 0.0001) and $I^2$ of 83.19% (95% CI 72.01 to 89.9). The funnel plot was used to detect publication bias of meta-analysis by plotting the standard error over the proportion (effect size). The diagonal lines represent the 95% CI around the summary effect size. Studies are represented with dots. Publication bias were also assessed with Egger’s and Begg’s test, represented with an intercept, a 95% CI and a significance level; and Kendall’s Tau and significance level, respectively. Results indicate low to no risk of bias across studies (Egger’s test, P = 0.42; and Begg’s test, P = 0.53).