This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Clinical Investigation of New CI Delivery Models in an Adult Nucleus CI Population (NCT03304106), a non-randomized na study. The primary outcome is Change in Monosyllabic Word Recognition Scores (CNC) in Quiet.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03304106 |
| Phase | NA |
| Condition | Sensorineural Hearing Loss, Bilateral, Cochlear Implant |
| Allocation | Non-Randomized |
| Design model | Parallel |
| Masking | None (open label) |
| Enrollment | 102 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Cochlear |
| Results posted | Yes |
2. Clinical Question
Population
Adults with cochlear implants or receiving a new cochlear implant were studied. The registry lists bilateral sensorineural hearing loss and cochlear implant as the conditions. The primary outcome population was participants with a complete data set; eligibility criteria beyond the stated study population are not reported in the registry record.
Intervention
The intervention was a commercially available cochlear implant and AI system. AI technology was used to assist audiologists in evaluating and programming both newly implanted and existing cochlear implant recipients. The registry does not describe the system’s algorithms, the programming procedures, or how audiologists used its recommendations.
Comparator
The parallel groups were newly implanted CI recipients and existing CI recipients. This was a nonrandomized study with no masking, and the groups had different clinical histories and different outcome timeframes. They therefore should not be treated as like-for-like control groups.
Primary question
The primary outcome was change in monosyllabic word recognition scores in quiet, measured using Consonant-Nucleus-Consonant words and expressed as percentage of correct words. The registry describes comparison from preoperative assessment to six months for Group 1, and from Visit A to Visit B after one month for Group 2. It does not report the groups’ numerical changes or a statistical comparison. The record therefore describes the question being studied, but does not establish the size or precision of any observed change.
3. Trial Design
This completed study, NCT03304106, is titled “Clinical Investigation of New CI Delivery Models in an Adult Nucleus CI Population.” Cochlear is listed as the sponsor. The registry records an enrollment of 102 and a study start date of 2017-10-19, with primary completion on 2019-09-18. The phase is listed as NA.
The design was parallel and nonrandomized, with no masking. It included two groups: newly implanted CI recipients and existing CI recipients. In both groups, AI technology was used to assist an audiologist’s evaluation and programming. The registry describes the newly implanted group as receiving standard-of-care commercial cochlear implants. It does not give further detail about how participants entered each group, how programming was performed, or whether the AI system was used in exactly the same way in both groups.
The groups were assessed over different intervals. Group 1’s primary outcome was described as preoperative to postoperative performance after six months of cochlear implant use. Group 2’s was described as Visit A to Visit B after one month of use. Because allocation was not randomized and the groups differed in implant status and follow-up interval, their outcomes are not a direct randomized comparison of two programming strategies. Differences between groups could reflect factors beyond the AI-assisted programming approach.
The registry does not report a conventional control group without AI assistance. It also does not report allocation procedures, masking procedures beyond the stated absence of masking, or details of the clinical sites and participant flow. These omissions limit how fully the study conduct can be reconstructed from the registry record.
4. Participant Flow and Accounting
The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.
Overall Study
| Stage | Newly Implanted CI Recipients | Existing CI Recipients |
|---|---|---|
| STARTED | 44 | 58 |
| COMPLETED | 31 | 56 |
| NOT COMPLETED | 13 | 2 |
Analysis population for the primary outcome, as defined in the registry: All participants with complete data set
5. Baseline Characteristics
The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.
| Characteristic | Unit | Group 1 | Group 2 | Total |
|---|---|---|---|---|
| Participants analysed | 31 | 56 | 87 | |
| Age, Continuous | years | 63.52 (23 to 90) | 56.98 (13 to 89) | 59.31 (13 to 90) |
| Sex: Female, Male: Female | Participants | 15 | 22 | 37 |
| Sex: Female, Male: Male | Participants | 16 | 34 | 50 |
| Race and Ethnicity Not Collected | Participants | — | — | 0 |
| Duration of Hearing Loss | years | 28.29 (3 to 56) | 28.39 (1 to 74) | 28.35 (1 to 74) |
| Duration of Device Use | months | 0 (0 to 0) | 79.30 (1 to 305) | 51.04 (0 to 305) |
| Ear Tested: Left | participants | 15 | 29 | 44 |
| Ear Tested: Right | participants | 16 | 27 | 43 |
| Type of Sound Processor: Kanso | Participants | 0 | 1 | 1 |
| Type of Sound Processor: N6 | Participants | 0 | 49 | 49 |
| Type of Sound Processor: N7 | Participants | 0 | 6 | 6 |
6. Primary Endpoint
The registry lists the primary outcome as Change in Monosyllabic Word Recognition Scores (CNC) in Quiet, measured over Pre-operative/Visit A; Group 1: 6 months; Group 2: 1 month.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | Standard Deviation |
| Unit | percentage of correct words |
| Endpoint classification | Continuous Endpoint |
The primary endpoint was change in monosyllabic word recognition scores in quiet, assessed with Consonant-Nucleus-Consonant (CNC) words. The outcome is expressed as the percentage of correct words, and the registry specifies a mean with standard deviation as the summary measure. The stated outcome population was participants with a complete data set.
Word recognition is a direct performance measure: participants are asked to recognize spoken monosyllabic words, and the score reflects the proportion answered correctly. The registry states that a higher percentage corresponds to a better score. As a continuous measure, the score can represent a range of performance rather than sorting participants only into categories such as success or failure. That can make it useful for describing changes in speech recognition over time.
The timing differs by group. For newly implanted recipients, the described comparison is from preoperative assessment to six months after implantation. For existing recipients, it is from Visit A to Visit B after one month of implant use. These are not identical observation periods or starting points. The endpoint description therefore identifies what was measured and when, but numerical means, standard deviations, changes, and uncertainty estimates are not reported in the registry record. The magnitude and precision of the outcome cannot be assessed from the recorded endpoint description alone.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry identifies a clinically interpretable primary measure: CNC word recognition in quiet, scored as percentage correct. It also states the direction of the scale: higher scores indicate better word recognition. However, the registry record does not report numerical values for the primary endpoint. In particular, it does not give group means, standard deviations, mean changes, confidence intervals, or p-values. Consequently, the direction, magnitude, and precision of the observed changes cannot be described from this record.
The study’s structure matters when interpreting any contrast between the groups. Newly implanted recipients and existing recipients were not randomly assigned to comparable strategies. They had different implant histories, and the recorded follow-up intervals differed: six months for Group 1 and one month for Group 2. A difference between their outcomes would not by itself isolate an effect of AI-assisted programming. The registry explicitly cautions that the arms are not a like-for-like control comparison.
The outcome is described as a change within each group, but the registry does not provide the underlying baseline and follow-up values or the change estimates. It is therefore not possible to determine whether scores changed, how variable those changes were, or whether estimates were compatible with a range of effects. A confidence interval, if reported, would convey the uncertainty around an estimate; no such interval is reported in the registry record. A p-value would address a specified statistical test, but neither test results nor an analysis plan are reported.
The record lists no analyses for this outcome. Thus there is no sponsor-reported statistical result to interpret or compare with a recalculation. The enrollment figure is not a substitute for outcome data: a sample size alone does not show what the observed effect was or how precisely it was estimated.
Safety information is reported separately from the primary outcome. The registry lists serious adverse events and other adverse events by group, as well as zero deaths in each group. These counts describe reported events in the recorded groups; they do not establish that an event was caused by the intervention or that the groups had comparable risk. Taken together, the record documents the study question, design, endpoint, and selected safety counts, but it does not provide enough primary-outcome information to characterize an efficacy estimate.
9. Statistical Methods Explained
What does a mean and standard deviation describe?
General statistical explanation: for a continuous outcome, the mean describes the central value of a set of scores. The standard deviation describes how spread out individual scores are around that mean. They answer different questions: a mean alone does not show how much participants’ scores varied. The registry specifies these summary measures for the CNC word-recognition outcome, but does not report their numerical values.
The endpoint is a percentage-correct score. The registry describes the primary measure as change over time, so an analysis could summarize each group’s score change using a mean and standard deviation. General statistical explanation: a change score depends on the measurements being compared and on when they were collected. Here the stated intervals differ by group, so a numerical comparison across groups would require careful attention to both the differing follow-up periods and the nonrandomized design. The methods used to estimate change are not reported in the registry record.
What would a confidence interval add?
General statistical explanation: a confidence interval gives a range of values compatible with an estimate under a statistical model and reflects uncertainty in that estimate. It is not a guarantee that the true effect lies within the particular interval. For a mean change, an interval would help readers judge precision alongside the point estimate. No confidence interval or other interval estimate for the primary endpoint is reported in the registry record, so its precision cannot be evaluated.
What does a p-value address?
General statistical explanation: a p-value is calculated under a specified statistical model and null hypothesis. It describes how unusual the observed data, or data more extreme, would be if that null hypothesis and model were correct. It does not give the probability that the null hypothesis is true, and it does not describe the size or clinical importance of an effect. The registry lists no analyses, p-values, or statistical test for the primary outcome.
Why does the allocation design matter?
General statistical explanation: random allocation can help balance measured and unmeasured characteristics between groups on average. Without randomization, group membership may be associated with participant characteristics, clinical history, or other factors that also relate to the outcome. This study was nonrandomized, and its two groups were newly implanted and existing CI recipients. The registry itself states that the arms are not a like-for-like control comparison. Statistical adjustment might address measured differences if appropriate data and a prespecified model were available, but such an approach is not described in the registry record. It would not automatically remove bias from unmeasured factors.
The parallel design means the groups are followed as separate groups rather than as a crossover sequence. It does not, on its own, make the groups comparable. The absence of masking is also recorded, but the registry does not describe which people knew the treatment or programming details, or how that might have affected measurement or care. No further masking procedures are reported.
How should the planning illustration be read?
General statistical explanation: sample-size calculations are planning tools that describe the number of participants that would be needed under stated assumptions to detect a specified effect with a chosen error rate and power. They do not establish that an effect was observed, and they are not observed power. The registry’s stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect defined as a stated fraction of the observed effect. The record does not report a numerical observed effect or the resulting planning counts in the information available for interpretation. Without those values, the illustration cannot be evaluated as a numerical design target. It is not the trial’s own design justification.
General statistical explanation: over-dispersion occurs when variability is greater than a statistical model expects. It matters because an analysis that assumes too little variability can understate uncertainty. The primary outcome is continuous and is described with a standard deviation; the registry does not report an over-dispersion assessment or a model that would make such an assessment relevant. The analysis distribution, handling of repeated measurements, adjustment factors, and methods for uncertainty estimation are not reported in the registry record.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists one primary outcome measure: change in CNC word-recognition score in quiet. Listing one primary measure does not establish that no other outcomes or comparisons were examined. The record does not describe a multiplicity strategy, a hierarchy of outcomes, or whether any adjustment was made for multiple testing. Interim analyses are not reported in the registry record.
The stated primary-outcome population is all participants with a complete data set. This indicates that complete data were required for inclusion in that outcome population, but the record does not report how many participants had complete data, how many were excluded from the endpoint analysis, or why any data were incomplete. It also does not describe imputation, sensitivity analyses, or other missing-data methods. The enrollment figure should not be assumed to equal the number analyzed for the primary endpoint.
The registry lists zero analyses. No statistical analysis plan, model specification, covariate adjustment, treatment of repeated measures, or rule for excluding observations is reported in the registry record. It is therefore not possible to determine whether the endpoint description corresponds to a prespecified analysis or to assess the effects of missingness and analytic choices on the reported results. Numerical primary-outcome results are also not reported, so there is no estimate or uncertainty measure to interpret.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Group 1 | 5 | 31 | 16.1% |
| Group 2 | 1 | 56 | 1.8% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Group 1 | 0 of 31 |
| Group 2 | 0 of 56 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Group 1 | 17 of 31 |
| Group 2 | 6 of 56 |
Most frequent serious adverse event terms
| Event term | Organ system | Group 1 | Group 2 |
|---|---|---|---|
| Hospitalization secondary to losing consciousness from sinus bradycardia | Cardiac disorders | 1 of 31 | 0 of 56 |
| Hospitalization | General disorders | 1 of 31 | 0 of 56 |
| Parkinson's disease diagnosis | Nervous system disorders | 1 of 31 | 0 of 56 |
| Hospitalization for sepsis | Infections and infestations | 1 of 31 | 0 of 56 |
| Dizziness | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Vertigo and BPPV | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Abnormal heart rhythm requiring placement of pacemaker | Cardiac disorders | 0 of 31 | 1 of 56 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Group 1 | Group 2 |
|---|---|---|---|
| Dizziness and imbalance | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Tongue numbness/mouth numbness | Injury, poisoning and procedural complications | 1 of 31 | 0 of 56 |
| Headaches | Nervous system disorders | 1 of 31 | 0 of 56 |
| Tinnitus | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Vertigo | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Change in taste | Injury, poisoning and procedural complications | 1 of 31 | 0 of 56 |
| Loss of hearing | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| MAP too loud, unable to wear within 48hours of activation | Injury, poisoning and procedural complications | 1 of 31 | 0 of 56 |
| Swelling at CI site (Ear not specified) | Ear and labyrinth disorders | 1 of 31 | 0 of 56 |
| Altered taste post-op | Injury, poisoning and procedural complications | 1 of 31 | 0 of 56 |
Serious events with the highest arm-level rate: hospitalization secondary to losing consciousness from sinus bradycardia (3.2%), hospitalization (3.2%), parkinson's disease diagnosis (3.2%), hospitalization for sepsis (3.2%), dizziness (3.2%).
The registry reports serious adverse events for 31 participants at risk in Group 1, with 5 affected, and for 56 participants at risk in Group 2, with 1 affected. For other adverse events, it reports 31 at risk and 17 affected in Group 1, and 56 at risk and 6 affected in Group 2. Deaths are reported as 0 in each group. These are the recorded counts; they should not be read as evidence that the groups were comparable, because allocation was nonrandomized and the groups differed in implant status.
The registry lists several serious adverse-event terms, each with a reported percentage of 3.2%: hospitalization secondary to losing consciousness from sinus bradycardia, hospitalization, Parkinson's disease diagnosis, hospitalization for sepsis, dizziness, and vertigo and BPPV. The record does not state how these terms map to the group counts or provide further event-level context. The listed terms should not be added together or assumed to represent distinct participants.
General statistical explanation: adverse-event counts describe events observed and recorded during a study; by themselves, they do not establish causation. The registry does not provide enough detail to determine whether events were considered related to the device or AI-assisted programming, how events were ascertained, or whether follow-up was equivalent between groups. These counts are descriptive safety information, not a randomized safety comparison.
12. What the Result Does, and Does Not, Mean
The registry record establishes that a completed, nonrandomized, parallel study investigated AI assistance in audiologists’ evaluation and programming of newly implanted and existing cochlear implant recipients. Its primary endpoint was change in CNC word recognition in quiet, and it describes different measurement intervals for the two groups. It also records selected adverse-event counts.
What the record does not establish is the numerical result for the primary endpoint. No group means, standard deviations, changes, confidence intervals, or p-values are reported. Therefore, the direction and size of any change, and the precision of an estimate, cannot be determined from the record. The enrollment total does not fill this gap: without outcome results, it cannot tell readers how the observed scores behaved.
The study design also limits what a comparison between groups could mean. Participants were not randomized, and newly implanted and existing recipients are not interchangeable groups. Their outcome intervals also differ. Any difference between group outcomes could reflect those distinctions or other factors, not solely the use of AI-assisted programming. The registry specifically indicates that the arms are not a like-for-like control comparison.
General statistical explanation: a study can document a measurement and report event counts without providing enough information to estimate a treatment effect. Interpreting an effect requires an estimate, an appropriate comparison, and information about uncertainty and analysis choices. Those elements are not reported for the primary endpoint in the registry record.
The safety counts can be read as recorded participant counts in each group, with zero deaths reported in both. They do not establish that the device or AI system caused or prevented any event. The most defensible interpretation is limited to what the registry records about study design, measurement, and safety counts, while recognizing that the primary-outcome results are absent.
13. Why This Trial Matters Statistically
This study is statistically notable because the clinical question involves a measurable performance outcome, but the available registry record does not report the numerical primary results. The endpoint is clearly defined as CNC word recognition in quiet and the direction of the scale is stated. Yet a well-described endpoint cannot answer how scores changed without the corresponding estimates and uncertainty.
The design also illustrates why group labels do not necessarily define a valid treatment comparison. Both groups received AI assistance, but one consisted of newly implanted recipients and the other of existing recipients. Their baseline circumstances and follow-up intervals differ, and allocation was nonrandomized. A group contrast would therefore mix differences in patient history and observation period with any difference potentially related to the intervention. The registry’s warning that the arms are not a like-for-like control comparison is important for readers considering causal interpretation.
General statistical explanation: an estimate and its precision are both central to interpretation. A point estimate alone can be uncertain, while a p-value alone does not convey effect size. In this record, neither is reported for the primary endpoint. This makes clear the distinction between registering a research question and publishing an interpretable statistical result.
The adverse-event counts are another example of why denominators and study design matter. The recorded group counts help describe what was observed, but nonrandom allocation and absent event-level context prevent a causal or comparative safety conclusion. The study can still be useful as a record of an investigation and its planned outcome framework, but the available results do not support a quantitative conclusion about changes in word recognition.
14. Limitations
- The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
- Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.
The primary limitation for interpreting the study is that numerical results for the primary endpoint are not reported in the registry record. Means, standard deviations, changes, confidence intervals, and p-values are absent. The study’s stated outcome is therefore identifiable, but the result and its precision are not.
The nonrandomized allocation limits causal interpretation. Newly implanted and existing CI recipients are different groups, and the registry notes that they are not a like-for-like control comparison. The endpoint intervals also differ: six months for Group 1 and one month for Group 2. The record does not describe adjustment for baseline differences or other potential confounders.
The primary-outcome population is limited to participants with complete data. The number with complete data, extent of missingness, reasons for missing data, and methods for handling incomplete observations are not reported in the registry record. Nor are detailed eligibility criteria, recruitment procedures, study sites, or participant flow reported in the information available.
The record does not describe the AI system’s operation, programming workflow, audiologist training, or adherence to the intervention. It states that there was no masking but does not provide further detail about who was aware of group or programming information. Statistical methods, analysis models, covariate adjustment, multiplicity handling, and interim analyses are also not reported. Finally, the adverse-event information lacks event-level context and does not establish attribution. These limitations constrain interpretation of both the clinical outcome and safety counts.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Clinical Investigation of New CI Delivery Models in an Adult Nucleus CI Population — NCT03304106.
- Registry record last updated 2020-10-29; status last verified 2020-10.
- Results first posted 2020-10-29.
- Data retrieved from ClinicalTrials.gov on 2026-11-09.
- Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT03304106 is a completed, nonrandomized, parallel investigation sponsored by Cochlear. The registry records enrollment of 102, a start date of 2017-10-19, and primary completion on 2019-09-18. The study examined AI-assisted evaluation and programming for newly implanted and existing cochlear implant recipients, without masking.
The primary endpoint was change in CNC monosyllabic word recognition in quiet, expressed as percentage of correct words. The described follow-up was six months for Group 1 and one month for Group 2. The registry lists the primary outcome population as participants with a complete data set, but does not report numerical endpoint results or statistical analyses.
Serious and other adverse-event counts are recorded by group, with zero deaths reported in each. Because the groups were nonrandomized and differ in implant status, they are not a like-for-like control comparison. The record documents the study question and selected safety counts, but does not report enough primary-outcome information to assess the magnitude or precision of a change in word recognition.