Introduction
For individuals with severe-to-profound sensorineural hearing loss (SNHL), cochlear implantation (CI) is widely recognised as a safe and effective standard of care. As of July 2022, more than one million cochlear implants had been performed worldwideReference Zeng1. Over the last four decades, CI research has expanded well beyond initial device development to encompass device engineering, surgical techniques, coding strategies, rehabilitation, and objective assessment. A recent bibliometric analysis identified 13,934 CI-related articles (1980–2022) with an average annual growth of 13.81%. Contributions are highly concentrated, with the United States leading (5,247 publications)Reference Abari, Tekin, Bahşi and Topsakal2. Given this concentration, the U.S.–based registry ClinicalTrials.gov offers both broad coverage and analytic leverage for field-level assessment. Moreover, because prospective registration is required by major journals, ClinicalTrials.gov also captures a substantial share of non-U.S. trials that register to meet editorial requirements, extending its relevance beyond the United StatesReference De Angelis, Drazen, Frizelle, Haug, Hoey and Horton3.
Within this context, analysing CI trials registered on ClinicalTrials.gov provides a denominator-aware view of where the field is investing effort (study designs, sponsors, populations, geographies) and how efficiently that activity converts into peer-reviewed evidence (registry-to-publication yield). Moreover, given that publication and outcome-reporting biases make statistically “positive” findings more likely to be fully reported, this further underscores the need for registry–publication–linked evaluations to reduce evidence distortion and more faithfully depict the CI evidence pipelineReference Dwan, Gamble, Williamson and Kirkham4.
The value of this registry–publication framework has been demonstrated within otolaryngology. In otitis media, for example, Oca et al. (2025) reported that fewer than one in five completed ClinicalTrials.gov–registered trials were subsequently published, with pronounced variation across intervention domainsReference Oca, Rajan and Qian5. Extending this framework to CI can guide funding allocation and trial-design optimisation, strengthen results reporting and overall transparency, and, ultimately, more effectively accelerate progress in clinical and rehabilitation research for CI users.
This study aims to estimate the proportion of ClinicalTrials.gov–registered CI trials with primary completion on or before October 10, 2022, provide a 2022–2025 publication window, and elucidate factors associated with nonpublication and registry–publication discrepancies; in addition, we profile CI trials registered or completed during 2022–2025 to contextualise recent shifts in study focus, sponsorship, design, and geography.
Materials and Methods
To identify trials for inclusion, we queried the NIH ClinicalTrials.gov registry using the advanced search (accessed October 10, 2025) to capture studies related to cochlear implantation. ClinicalTrials.gov was selected as the primary data source because it is the largest single clinical trial registry worldwide; moreover, International Committee of Medical Journal Editors (ICMJE) registration requirements mean that many non-U.S. trials are also registered on the platform to satisfy editorial policiesReference De Angelis, Drazen, Frizelle, Haug, Hoey and Horton3. To maximise coverage of CI-related research, we populated the Other Terms field with “Cochlear Implant” OR “Cochlear Implants” OR “Cochlear Implantation” and Study Type was set to Interventional and Observational. The historical cohort was restricted to records with recruitment status of Completed or Unknown and primary completion before October 10, 2022, thereby allowing a 3-year publication window. Trials classified as Unknown were included because their registered completion dates had passed and non-update of the record does not preclude actual trial completion. This query returned 253 records (206 Completed, 47 Unknown). Manual verification excluded 155 records lacking direct relevance to cochlear implantation (e.g., studies of other implantable hearing devices, vestibular assessments, or unrelated conditions incidentally captured by the full-text search), yielding a final analytical cohort of 98 trials (80 Completed, 18 Unknown; Supplemental Data). In addition, trials registered between October 10, 2022, and October 10, 2025, were retrieved to compare recent trends. For each eligible record, we extracted intervention type, trial phase, principal-investigator location, age categories, enrolment, study status, conditions, and the National Clinical Trial (NCT) number.
Trials were assigned—based on the primary objective recorded at registration—to four different categories: Device, Procedure, Evaluation & Prognosis, and Rehabilitation Training. Device denotes studies centred on the CI system (hardware/software), such as the safety/effectiveness of alternative electrode arrays or evaluations/adjustments of signal-coding strategies to improve speech recognition. Procedure comprises surgeon-facing work on operative technique and perioperative management, emphasising greater precision or less invasiveness in implantation. Evaluation & Prognosis covers patient-focused, observational research linking baseline characteristics to postoperative outcomes (e.g., subgroup prognostics, longitudinal changes in cognition or quality of life). Rehabilitation Training refers to interventional studies testing specific rehabilitation strategies and their impact on recovery and functional performance. When records contained overlapping elements, categorisation followed the registered primary aim; if uncertainty persisted, the primary outcome and stated intent were used as tiebreakers, with final assignments harmonised by manual review. Trial titles, NCT identifiers, and category assignments are reported in the Supplemental Data.
Trials that listed at least one for-profit entity as a sponsor or collaborator were classified as industry-sponsored; all others were grouped as non-industry (e.g., academic, governmental, foundation). Because clinical phase is missing or inconsistently reported for many CI registrations, we did not stratify analyses by phase. To capture geography, studies were categorised by the principal investigator’s (PI) location as United States versus outside the United States. Age focus was coded into adult-only, paediatric-only, or mixed cohorts based on the target population specified at registration. Finally, each record retained the registry’s study design designation (interventional vs observational), and comparisons by design were conducted to assess whether these two study types differed statistically in our outcomes of interest.
All registry records were exported from ClinicalTrials.gov as a comma-separated values (CSV) file, and analyses were conducted in R 4.3.1. Associations between candidate covariates and publication status were estimated using multivariable logistic regression, with results reported as odds ratios (ORs) accompanied by 95% confidence intervals (CIs) and two-sided P values. Statistical significance was defined a priori as a 95% CI not encompassing 1.00 with α = 0.05. This study qualified for exemption from Institutional Review Board review because it analysed existing, de-identified registry data and did not involve any patient-identifiable information.
Results
Characteristics of Registered Trials before 2022 October
Of the 98 trials in the analytical cohort (Table 1), 34 (34.7%) were industry-sponsored and 64 (65.3%) non-industry. The majority adopted an interventional design (79/98, 80.6%), a predominance more pronounced among industry-sponsored trials (91.2%) than non-industry trials (75.0%). Device studies were the most frequent (43/98, 43.9%), followed by Evaluation & Prognosis (31/98, 31.6%), Rehabilitation Training (13/98, 13.3%), and Procedure (11/98, 11.2%). Industry sponsorship was heavily concentrated in device-oriented research—27/34 (79.4%) of industry-sponsored trials—while non-industry efforts were more evenly distributed across all four categories (Table 1).
Characteristics of CI Clinical Trials Registered on ClinicalTrials.gov and Completed Before October 2022, Stratified by Sponsor Type
Table 1 Long description
The table summarizes 98 completed clinical trials and reports, for all trials and by sponsor type, the number registered and the share that were published. Overall, 26 of 98 trials were published, for a publication rate of 26.5 percent. Non-industry sponsors registered 64 trials and published 19, for 29.7 percent, while industry sponsors registered 34 trials and published 7, for 20.6 percent. By category, devices were most common with 43 trials and 20.9 percent published; procedures had 11 trials and 36.4 percent published, with no industry-sponsored procedure trials. Evaluation and prognosis trials totaled 31 with 29.0 percent published, and rehabilitation training totaled 13 with 30.8 percent published; in both areas, non-industry publication rates exceeded industry, where industry had no published evaluation and prognosis trials and no published rehabilitation training trials. By principal investigator location, international trials were more common than United States trials, and publication rates were slightly higher internationally overall; non-industry international trials published at 31.0 percent versus 20.0 percent for industry. By age focus, adult-only trials were most frequent with 63 trials; industry adult-only trials had a higher publication rate than non-industry adult-only trials, while mixed child and adult trials showed higher publication for non-industry than industry. Interventional studies dominated with 79 trials and had a higher publication rate than observational studies overall, and within both study types non-industry trials published more often than industry. Percentages reflect the share published within each subgroup and may be influenced by small subgroup sizes.
Geographically, the majority of trials (62/98, 63.3%) were conducted exclusively outside the United States; sponsor-stratified geographic distributions are detailed in Table 1. The age focus skewed toward adults: 63 trials (64.3%) enrolled adults only, 23 (23.5%) included both children and adults, and 12 (12.2%) enrolled children only. Industry-sponsored trials showed a stronger adult emphasis (adult-only 70.6%; mixed 26.5%; paediatric-only 2.9%), whereas non-industry trials were relatively more paediatric-oriented (paediatric-only 17.2%), though adult-only enrolment remained predominant (60.9%; Table 1).
Characteristics of Published Trials
Overall, only 26.5% (26/98) of CI trials reached publication. Published trials encompassed the full breadth of CI research domains, including device optimisation (e.g., electrode–array configurations, signal-coding, and EAS platforms), procedural refinements (minimally invasive/robotic trajectories, round-window access, perioperative cochlear-preservation strategies), patient-centred assessment/prognosis (baseline predictors, quality-of-life/health-utility), and rehabilitation interventions (structured auditory training, working-memory–oriented programmes, tele-rehabilitation).
Across study categories, Procedure trials had the highest publication proportion (36.4%) and Device the lowest (20.9%); industry-funded trials were published less frequently than non-industry trials (20.6% vs 29.7%). Publication proportions across geography, age focus, and study design are presented in Table 1. See also the Supplemental Data for trial-level details and NCT identifiers.
Predictors of Publication
Table 2 presents the multivariable logistic regression results. After adjusting for study category, design, sponsor type, PI geography, and log-transformed enrolment, none of the covariates reached statistical significance at α = 0.05. No category contrast differed significantly from Device (all p > 0.05; Table 2). Trials with an interventional design showed suggestive but non-significant higher odds of publication than observational trials (OR = 5.16; 95% CI, 0.80–33.19; p = 0.084). Similarly, enrolment showed a positive but non-significant association (per log unit: OR = 2.15; 95% CI, 0.92–4.99; p = 0.076). Neither sponsor type (p = 0.414) nor geography (p = 0.282) was significantly associated with publication. Given the modest cohort size (n = 98), these null findings should be interpreted with caution, as the analysis may have been underpowered to detect moderate effects.
Multivariable Logistic Regression Analysis of Factors Associated with Publication of CI Clinical Trials Registered on ClinicalTrials.gov
Table 2 Long description
The table reports adjusted odds ratios with 95 percent confidence intervals and p values for factors associated with whether registered cochlear implant clinical trials were published. Industry funding and device category are the reference groups; compared with industry funding, non-industry funding had higher odds of publication (odds ratio 1.68; confidence interval 0.49 to 5.77; p 0.414). Compared with device trials, procedure (1.75; 0.33 to 9.38; p 0.512), evaluation and prognosis (0.62; 0.13 to 2.87; p 0.540), and rehabilitation training (1.46; 0.24 to 9.04; p 0.685) showed no clear differences. International principal investigator location had higher odds than United States location (1.82; 0.61 to 5.43; p 0.282). Relative to children-only trials, mixed child and adult trials had lower odds (0.27; 0.04 to 2.02; p 0.203) and adult-only trials were similar (1.19; 0.24 to 6.01; p 0.831). Interventional studies had higher odds than observational studies (5.16; 0.8 to 33.19; p 0.084), and larger enrolment was associated with higher odds (2.15; 0.92 to 4.99; p 0.076). All confidence intervals are wide and all p values exceed 0.05, so none of the associations are statistically significant and estimates are imprecise.
Abbreviations: CI, confidence interval; OR, odds ratio; Enrolment represents the number of enrolled participants (per log unit)
Changes in research trends
Table 3 compares trial characteristics between the pre-2022 cohort (n = 98) and post-2022 registrations (n = 69). The research portfolio shifted away from device-focused trials (from 43.9% to 31.9%) toward rehabilitation (from 13.3% to 20.3%) and procedure (from 11.2% to 17.4%), while evaluation & prognosis remained essentially stable (from 31.6% to 30.4%). Trials were more often international (from 63.3% to 73.9%). Study design tilted modestly toward observational approaches (from 19.4% to 29.0%); age focus showed no material change (Table 3).
Characteristics of CI Clinical Trials on ClinicalTrials.gov (Pre-2022 VS Post-2022), Stratified by Sponsor Type, Study Category, Age Group, and Study Design
Table 3 Long description
The table compares characteristics of 98 pre-2022 and 69 post-2022 clinical trials, reporting counts and percentages by sponsor type, study category, principal investigator location, age focus, and study type, along with percent-point change and a p value for each section. Funding was stable: industry-sponsored trials were 34.7 percent pre-2022 and 37.7 percent post-2022, while non-industry trials were 65.3 percent and 62.3 percent. By category, device trials decreased from 43.9 percent to 31.9 percent, while rehabilitation training increased from 13.3 percent to 20.3 percent; procedure trials rose from 11.2 percent to 17.4 percent, and evaluation and prognosis stayed near 31 percent. Principal investigator location shifted toward international sites, from 63.3 percent to 73.9 percent, with United States-based trials dropping from 36.7 percent to 26.1 percent. Age focus changed little: children-only trials were about 12 to 13 percent, mixed child and adult about 22 to 24 percent, and adult-only about 62 to 64 percent. Study type showed fewer interventional trials after 2022, from 80.6 percent to 68.1 percent, and more observational trials, from 19.4 percent to 29.0 percent. Reported p values for each section were all above 0.05, so differences should be interpreted as descriptive rather than statistically significant.
Overall sponsorship mix was stable (industry from 34.7% to 37.7%). At the category level, industry sponsorship increased significantly only among evaluation & prognosis trials (from 12.9% to 38.1%). Changes in the remaining categories did not reach statistical significance: device (from 62.8% to 72.7%) and procedure (from 0.0% to 16.7%) showed numerical increases, while rehabilitation training declined (from 23.1% to 0.0%); these non-significant shifts should be interpreted with caution given the small subgroup sizes.
Discussion
Despite the exponential growth in CI research over the past two decades, speech-recognition outcomes for many CI users appear to plateau around ∼80%Reference Ma, Fried, Nguyen, Schvartz-Leyzac, Camposeo and Meyer6. Several bibliometric syntheses have quantified the publication side of this expansion and reflected on prevailing methods and trajectories in CIReference Abari, Tekin, Bahşi and Topsakal2, Reference Ko, Tan, Chu and Chan7, Reference Chen and Chang8. However, publication-based samples cannot recover the appropriate denominator of registered clinical activity, nor can they localise pipeline attrition and delays along the path from trial registration to citable evidence. In the field of otology, approximately 37.7% of registered trials remained unpublished even years after completion, underscoring persistent shortfalls in dissemination and transparencyReference van Heteren, van Beurden, Peters, Smit and Stegeman9.
In this registry–publication linkage focused on cochlear implants, only 26.5% of completed ClinicalTrials.gov–registered trials had an identifiable peer-reviewed publication. To our knowledge, this is the first CI-specific assessment of publication yield from the registry base. Although the proportion exceeds that reported for otitis media cohorts (18%)Reference Oca, Rajan and Qian5, it remains considerably lower than benchmarks in related subspecialties—approximately 53.7% for otology and 40% for rhinosinusitisReference Yang, Tsai, Mukdad and St John10, Reference Ross, George, Wayant, Hamilton and Vassar11. The gap is equally striking when viewed against broader medical benchmarks: a cross-specialty meta-analysis reported an overall publication rate of approximately 54% across diverse clinical fieldsReference Schmucker, Schell, Portalupi, Oeller, Cabrera and Bassler12, and an analysis of all ClinicalTrials.gov-registered trials estimated that roughly 46% reached publicationReference Ross, Tse, Zarin, Xu, Zhou and Krumholz13. Against both ENT-specific and cross-disciplinary references, the 26.5% yield for CI trials appears notably low.
We applied a conservative 3-year allowance after primary completion, consistent with prior evaluationsReference Meller, Sambo, Nguyen, Robbins and Granet14, to accommodate typical timelines for manuscript preparation, peer review, and indexing. Even under this permissive window, the modest publication yield indicates a material dissemination gap in the CI evidence pipeline, particularly for a high-cost intervention that requires long-term follow-up and rehabilitation.
In our study, device trials constituted the largest category yet yielded the lowest publication proportion and were also the most heavily industry sponsored. This paradox—devices are most investigated but least often published—is also consistent with broader otology/neurotology landscapes in which devices dominate the intervention mixReference Oca, Rajan and Qian5, Reference Altshuler, Viswanathan, Dasani, Webb, Ramsey and Patel15. Several structural factors plausibly underpin this gap. First, evidentiary horizons for device innovation are longer: outcomes such as durability, reliability, and speech-in-noise benefit often require extended follow-up and multi-factor modelling before effects are conclusive. Second, regulatory and intellectual-property pathways specific to medical devices can prioritise submissions and iterations over early journal publication, thereby deferring disseminationReference Faris and Shuren16. Third, consistent with field critiques and our data, CI development within the prevailing electrical stimulation/standard processing paradigm appears to be approaching a practical performance plateauReference Carlyon and Goehring17. Relative to trials completed before October 10, 2022, the post-2022 registration portfolio contained a smaller device share. Published device reports over the past two decades largely reflect incremental refinements in electrode geometry, coding strategies, and noise-reduction algorithms, with limited gains in complex listening environments. Looking ahead, optical/optogenetic stimulation and intraneural or focused multipolar interfaces may offer avenues for step-change improvements; however, current progress largely remains in animal modelsReference Carlyon and Goehring17 or early-clinical stageReference Thomas, Zuniga, Sondh, Leber, Solzbacher and Lenarz18.
In comparing trials registered after 2022 with those completed before October 2022, a notable shift is the marked expansion of industry sponsorship in Evaluation & Prognosis studies. Within this domain, post-operative speech recognition remains the dominant focus of industry-funded work (5/8 trials, 62.5%), underscoring the centrality of speech‐in‐noise and related perceptual outcomes as decision-critical endpoints. By contrast, Rehabilitation Training shows a different pattern: despite a decline in industry participation, the overall share of rehabilitation trials increased—a trend largely driven by hospitals and research institutes—reflecting sustained interest in active intervention to accelerate CI recovery. Clinically, because the fidelity of information transmission with CI is inherently constrained, patients commonly require 1–2 years of structured rehabilitation to achieve stable conversational functionReference Moberly, Vasil, Baxter and Ray19, Reference Holden, Finley, Firszt, Holden, Brenner and Potts20. Consistent with this need, we observe a growing number of intervention-oriented studies (e.g., music-based training, game-based/computerised training) that integrate targeted practice into the CI pathway and report encouraging early gainsReference Veltman, Maas, Beijk, Groenhuis, Versnel and Vissers21, Reference Gnadlinger, Werminghaus, Selmanagić, Filla, Richter and Kriglstein22. Despite the recent proliferation of rehabilitation-training studies, the evidence base remains constrained by protocol heterogeneity (content, intensity, duration), small samples, and brief follow-up windows—often only weeks to a few months—conditions that limit both inferential robustness and external generalisability. Consistent with this landscape, the positive coefficient for enrolment size in our regression model did not attain statistical significance (p = 0.076)—a finding more plausibly attributable to the limited power of a 98-trial cohort than to a genuine absence of effect, and one that echoes broader concerns about underpowered designs and selective dissemination in otologyReference van Heteren, van Beurden, Peters, Smit and Stegeman9. Looking ahead, progress in this field will hinge on multicentre, adequately powered pragmatic trials with extended follow-up periods to evaluate the durability of rehabilitation effects and their real-world impact.
ClinicalTrials.gov was selected as the primary data source because it is the largest single trial registry, broadly covers all CI research directions, and contains a substantial proportion of international trials (63.3% of our pre-2022 cohort originated outside the United States), conferring good global representativeness. A supplementary cross-search of the WHO ICTRP—a meta-registry aggregating records from 17 primary registries worldwide—confirmed that ClinicalTrials.gov contributed over half of all CI-relevant records, with no regional registry exceeding 12%. In line with most registry-to-publication studies in otolaryngology and related fieldsReference Lewis, Zolkind, Kallogjeri and Piccirillo23–Reference Vorland, Brown, Kilicoglu, Ying and Mayo-Wilson25, a multi-registry analysis was not undertaken, primarily because cross-platform heterogeneity in data standards, intervention-type restrictions (e.g., the EU Clinical Trials Register excludes device and surgical trials by regulation), incomplete deduplication mechanisms, and the absence of a standardised identifier equivalent to the NCT number for bibliographic linkage currently preclude reliable cross-registry publication ascertainment. Trials were harmonised into four categories based on the registered primary objective—Device, Procedure, Evaluation & Prognosis, and Rehabilitation—to enhance cross-study comparability, although this schema inevitably introduces some subjectivity.
Publication ascertainment used NCT identifiers, titles, and registry cross-links across multiple databases; outputs with unstable indexing or inconsistent labelling (e.g., local/non-English journals, conference proceedings, theses, preprints, or retitled publications) may have been missed. Some long-duration studies may also fall outside the 3-year publication window and were therefore excluded. Model-based inference is further constrained by the modest number of published trials, yielding wide confidence intervals and several non-significant contrasts (e.g., category effects vs Device; observational vs interventional), which raises the possibility of type-II error. Accordingly, findings regarding sample size and study design should be interpreted as suggestive signals rather than definitive conclusions. Furthermore, registry-based analyses capture only trials that investigators chose to register; approximately half of published randomised trials remain unregistered, with prospective registration rates lowerReference Trinquart, Dunn and Bourgeois26. Because unregistered trials are typically smaller and carry higher risk of biasReference Lindsley, Fusco, Li, Scholten and Hooft27, the 26.5% publication yield reported here likely represents an upper bound, and the true scale of CI evidence loss may be greater—further underscoring the imperative for stricter prospective registration and results-reporting enforcement.
Notwithstanding these limitations, this investigation offers a rigorous, registry-anchored account of publication patterns in CI trials. Key strengths include a pre-specified, standardised analytic workflow; comprehensive modelling of dissemination correlates across sponsor, design, topical category, geography, and registered sample size; and temporal benchmarking that contrasts pre-2022 completed trials with 2022–2025 registrations to delineate recent shifts in the CI trial landscape. To our knowledge, this is the first study to quantify the publication yield of ClinicalTrials.gov–registered CI trials, addressing a substantive evidentiary gap in the otolaryngology literature.
• Registry-to-publication studies in otolaryngology have shown that a substantial proportion of completed trials remain unpublished, with publication rates varying widely across subspecialties; however, no study has specifically quantified the publication yield of registered cochlear implant trials.
• In this registry-based analysis of 98 completed ClinicalTrials.gov-registered cochlear implant trials, only 26.5% reached peer-reviewed publication—a rate lower than reported benchmarks in otology (53.7%) and across clinical disciplines (approximately 54%).
• Device trials were the most prevalent category yet had the lowest publication rate (20.9%); post-2022 registrations indicate a shift away from device-focused research towards rehabilitation training (from 13.3% to 20.3%) and procedural studies (from 11.2% to 17.4%).
• These findings highlight a marked dissemination gap in the cochlear implant evidence pipeline and underscore the need for stricter prospective registration and results-reporting enforcement to reduce evidence waste in this field.
Conclusion
In summary, leveraging ClinicalTrials.gov, we quantified the publication yield of CI trials and delineated its correlates: only about one quarter of completed trials progressed to peer-reviewed publication; interventional designs and larger planned enrolments showed positive trends toward higher publication likelihood, although neither association reached conventional statistical significance, possibly owing to the limited cohort size. Recent registration patterns indicate a relative shift toward patient-related assessment and rehabilitation, suggesting a broadening, patient-oriented CI research agenda whose effect on dissemination will require continued longitudinal tracking.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0022215126105003.
Acknowledgements
The authors are grateful to all partners for their contribution to the article.
Funding statement
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Ethics announcement
This study qualified for exemption from Institutional Review Board review because it analysed existing, de-identified registry data and did not involve any patient-identifiable information.
Competing Interests
This study has no affiliations with or involvement in any organisation or entity with any financial interest in the subject matter or materials discussed in this manuscript.
