xRead - Episodic Vertigo (January 2026)

422

D. LI ET AL.

any disagreement was resolved through discussion and arbitra tion by a third reviewer (Zhang or Chen). We screened the titles and abstracts first, and full texts were reviewed where the abstracts were ambiguous. The included studies were further reviewed to confirm that no relevant study was missed. Figure 1 shows the flow chart of the selection process. Outcomes The primary outcome of this study was the effectiveness of treatments, which was defined as the negative nystagmus rates of the Dix–Hallpike test recorded at 1 week and/or 1 month of follow-up after treatment. We excluded studies that only reported subjective responses to treatments as the primary outcomes (17). For studies that applied both the Dix–Hallpike test and vertigo symptoms as the main outcomes, we only extracted the objective results of the Dix–Hallpike test for analysis. We defined the follow-up time of 3 days to 2 weeks as the 1-week follow-up; a follow-up time of 4 weeks was set as the 1-month follow-up. As the recurrence rates of treatments and the dizziness handicap inventory (DHI) score were only reported in a few articles, we did not analyze DHI in this network meta-analysis. Network and Pairwise Subgroup Meta-analysis We synthesized this network meta-analysis by using the protocol proposed by Cipriani et al. (18), and a random-effect model based on a Bayesian evidence framework was adopted for this network meta-analysis. Arm-based data were used during network meta-analysis. The odds ratios (ORs) and 95% credible interval (CrI) for dichotomous outcomes were estimated by binomial likelihood. We statistically evaluated the inconsistencies between direct and indirect comparisons by the node-split method (19). Deviance Information Criterion (DIC) was used to assess the goodness-of-fit of the model, which is particularly useful in Bayesian model selection problems (20,21). The square matrix was applied to present relative effect between any pair of treatments of NMA (22), and the cumula tive ranking probabilities of each treatment graphed by the surface area under the cumulative ranking curves (SUCRA) was used to rank the treatment outcomes (18,23). The larger the SUCRA value, the better the rank of the intervention in the network. We performed two subgroups of pairwise meta-analysis based on disease duration and age to explore the inconsistency between indirect and direct comparisons of the Epley versus a sham maneuver and the Epley versus Semont maneuver, respec tively. A random-effects model was used to calculate the pooled point estimate. The intervention effects were measured by RR with 95% CI for dichotomous outcome variables. I 2 value ranging from 50% to 75% was considered to have substantial heterogeneity, while a value from 75% to 100% was regarded as considerable heterogeneity (14). Sensitivity analysis was per formed to evaluate the robustness of the meta-analysis by omitting one single study at a time and then estimating the combined effects of the remaining studies. Quality Assessment and Bias Identification We assessed the risk of bias for each study in accordance with the Cochrane Handbook for Systematic Reviews of Inter ventions (24) by considering seven domains, such as selection bias (random sequence generation and allocation concealment), performance bias, detection bias, attrition bias, reporting bias and other bias, and ranked the overall scores into three different levels via two independent reviewers (Li and Yang). Any

affected side is the Dix–Hallpike test, which was imple mented by Dix and Hallpike in 1952 (4). Patients with BPPV endure mildly to severely impaired quality of life before interventions (5). Several treatments, that is, the Brandt-Daroff exercise and the Epley and Semont maneuvers, have been proposed for the management of PC-BPPV (6–8), among which the Epley particle repositioning maneuver is considered the most popular (9). However, there are still some contro versies regarding their treatment differences (10). BPPV may resolve spontaneously in most patients 1 month after onset (11,12). Therefore, it is crucial to evaluate the efficacy of each treatment for PC-BPPV patients. Compared with the traditional meta-analysis, network meta-analysis (NMA) has more advantages. NMA can be used to evaluate multiple treatments simultaneously in a single analysis by comparing direct and indirect eviden ces, enabling it to compare the relative effects between any pair of interventions and estimate relative rankings of interventions (13). Therefore, we conducted a network meta-analysis to evaluate the shorter-term (1 week) and longer-term (1 month) outcomes of various treatments for PC-BPPV patients and try to estimate relative rank ings for those included treatments. This review was registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the regis tration number CRD42021232096. We strictly followed the methods recommended by the Cochrane Collaboration (14) and conformed to the reporting standards of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis for Network Meta-Analysis (PRISMA-NMA) (15). Inclusion and Exclusion Criteria Randomized case-control studies, which compared particle repositioning maneuvers, vestibular rehabilitation, or pharma cological interventions to the placebo/sham maneuver or other treatments for PC-BPPV patients ( > 18 years old) diagnosed by the Dix–Hallpike test, were included. We excluded studies in which the participants suffered from subjective positional vertigo, pure vertical nystagmus, or central neuropathy. Literature Search Strategy After consulting with experienced information experts, we developed a comprehensive literature search strategy. The data bases of Ovid MEDLINE, Ovid EMBASE, EBSCO, ProQuest, SpringerLink, Cochrane Library on Wiley, and Web of Science were retrieved from the date of their inception to December 1, 2020, and only papers available in full text in English were considered. The MeSH terms ‘‘Benign Paroxysmal Positional Vertigo,’’ ‘‘Randomized Controlled Trial,’’ and related free words were combined to search relevant studies, and syntax was adjusted according to the needs of each database. We additionally con ducted searches for published, unpublished, and ongoing random ized controlled trials (RCTs) in international trial registers (16), and manual searches were also performed for specific studies. Literature Screening and Data Extraction Literature screening was independently performed by two reviewers (Li and Yang) according to the eligibility criteria, and METHODS

Otology & Neurotology, Vol. 43, No. 4, 2022

Copyright © 2022 Otology & Neurotology, Inc. Unauthorized reproduction of this article is prohibited.