Aging and Cognitive Health Evaluation in Elders (ACHIEVE): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Aging and Cognitive Health Evaluation in Elders (ACHIEVE) (NCT03243422), a randomized na study. The primary outcome is Change in Global Cognitive Function.

2. Clinical Question

3. Trial Design

ACHIEVE was a randomized, parallel-group trial with no masking listed in the registry. The record gives an enrollment of 977 and a completed status. The study summary describes a trial nested within the ARIC study infrastructure and says participants would be randomized 1:1 to the hearing intervention or successful-aging health education. That summary describes planned follow-up every 6 months for 3 years; the primary outcome timeframe is also listed as 3 years. Whether every participant completed the planned follow-up is not reported in the registry record.

The two interventions were non-drug programs. The hearing arm was described as best-practices hearing rehabilitative treatment, while the comparator involved individual sessions on healthy-aging topics. The registry lists the trial phase as NA. The sponsor is Johns Hopkins Bloomberg School of Public Health. The recorded start date is 2018-01-04, and the primary completion date is 2022-11-30.

The record lists a continuous primary outcome, change in global cognitive function, and describes its analysis as a linear mixed-effects model. The reported analysis is identified as a mixed models analysis. Details such as the model's covariates, covariance structure, and the analysis population are not reported in the registry record. The registry also does not report a sample-size justification or a detailed statistical analysis plan. Random allocation can support a comparison between groups, but understanding how the reported estimate was generated also depends on those analytic details.

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

Reasons for not completing this period:

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.

6. Primary Endpoint

The registry lists the primary outcome as Change in Global Cognitive Function, measured over 3 years.

The primary endpoint was mean change in global cognitive function from baseline to year 3, expressed in standard deviations. Global cognitive function was defined using a factor score derived from a neurocognitive testing battery. The listed tests cover memory, language, and executive function, among other cognitive tasks. Factor analytic methods use items from the battery to produce one score rather than treating each test as the sole measure of global function. The score's stated range is -5 to 4, and lower values denote worse cognitive function.

A global measure can summarize performance across several cognitive domains, making it relevant to a trial asking about overall cognitive change rather than a single task. It also has limits: a combined score may not reflect every domain's pattern equally, and a group mean does not describe each participant's experience. The registry says the battery is administered annually and that the primary change is estimated with a linear mixed-effects model. The endpoint is therefore a model-based mean change over the stated timeframe, not a direct count of participants who experienced a particular clinical event. The record does not report a threshold for a clinically important change in this score.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a score, and the summary measure is a mean. The registry describes mean change in global cognitive function in standard deviations from baseline to year 3. A mean represents the center of a distribution, but it does not show how widely individual participants' changes vary. The arm-level confidence intervals quantify uncertainty around the reported means; the comparison estimate and its confidence interval address the randomized contrast.

What does a confidence interval convey?

In general, a confidence interval expresses the precision of an estimate under the statistical model and its assumptions. A narrower interval indicates greater precision than a wider interval. It is not a guarantee that the true value lies within the particular interval, and it is not a range containing a stated proportion of individual participant scores. The primary result's 95% confidence interval includes zero. That means the posted analysis does not rule out zero difference at the corresponding confidence level, while the interval also gives a range of nonzero values compatible with the analysis.

How should the p-value be read?

In general, a p-value describes how unusual data at least as discrepant as those observed would be under a specified null hypothesis and model. It is not the probability that the null is true, and it does not measure the size or clinical importance of an effect. The primary p-value is 0.96. Considered alongside the estimate and interval, it indicates little evidence against the null hypothesis in this analysis; it is not, by itself, evidence that two interventions are equivalent. An equivalence conclusion would require a prespecified equivalence margin and an analysis designed for that question. Such a margin is not reported in the registry record.

Why use a mixed-effects model?

In general, mixed-effects models can account for repeated measurements from the same participant, whose observations are not independent of one another. They can estimate mean change over time while representing within-person correlation. The registry identifies a linear mixed-effects model for the primary endpoint, but does not report its full specification. Details about fixed effects, random effects, covariance structure, covariates, and the handling of measurements not observed at every time are not reported in the registry record. It is therefore not possible to evaluate those choices from the record alone.

Why distinguish the arm changes from the comparison?

The arm-level entries describe mean change within each group. The primary question concerns the contrast between randomized groups. In general, a comparison should be interpreted using the estimate and uncertainty from the reported between-group analysis, rather than by comparing whether separate within-group intervals include zero. The registry's mixed-model result is the reported primary analysis. A two-arm re-calculation from the posted summary statistics is not estimable, as the record indicates that values are missing or not estimable for a continuous endpoint comparison. No independent re-calculated comparison should be substituted for the reported analysis.

What do the secondary analysis and planning illustrations add?

The registry reports a time-to-event analysis for cognitive impairment, with a hazard ratio and confidence interval. In general, a hazard ratio compares event rates over time under the model; it is not the same quantity as a risk difference at a fixed time. The registry also contains an unadjusted re-calculation for this secondary outcome. Such a re-calculation is based on posted summary statistics and does not replace the registry's reported analysis. The sample-size illustrations are planning calculations based on the observed effect and stated assumptions. They describe hypothetical future study sizes for target power levels, not the power achieved by this trial or an analysis of its observed power. The record says these illustrations use a two-sided significance level of 0.05, equal allocation, and a normal approximation.

Why does over-dispersion matter in general?

For count or event data, a model that assumes less variability than is actually present can make uncertainty appear too small. Over-dispersion means that observed variation exceeds what a model's assumptions anticipate; accounting for it can affect standard errors and confidence intervals. This is a general statistical consideration, not a claim that over-dispersion was present or modeled in this trial. The registry does not report an over-dispersion assessment for the primary continuous outcome.

10. Analyses Reported by the Sponsor

The registry reports a mixed-model analysis for the primary outcome: an estimate of 0.002 standard deviations, a 95% confidence interval from -0.077 to 0.081, and p = 0.96. The arm-level mean changes and their confidence intervals are also posted, but they do not by themselves provide the uncertainty for a between-arm contrast. The record states that a two-arm continuous comparison could not be recalculated from the posted summary values because values are missing or not estimable. Accordingly, there is no independent continuous-endpoint re-calculation to compare with the reported analysis, and the sponsor's reported analysis remains the relevant primary result.

For cognitive impairment, the registry reports a regression, Cox analysis with a hazard ratio of 0.90, a 95% confidence interval from 0.61 to 1.33, and p = 0.59. A separate unadjusted re-calculation is based on the posted group counts. It is a simple summary-statistic comparison, not the same as the reported time-to-event model. The two analyses can differ because they use different methods and may address the data differently. The re-calculation should be read as supplementary, not as a replacement for the registry's reported analysis. The record does not describe the full specification of the sponsor's models.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 20 outcome measures and 7 analyses, as well as four named secondary outcomes. When multiple outcomes or analyses are examined, the chance of obtaining at least one apparently notable result can increase. In general, a prespecified multiplicity strategy can help control the interpretation of multiple tests. Whether the trial specified a hierarchy of outcomes, adjusted significance thresholds, or another multiplicity procedure is not reported in the registry record. The reported p-values should not be assumed to have been adjusted.

The record identifies the primary endpoint and gives its timeframe and model family, but does not report a full statistical analysis plan or the analysis population. It does not state whether the primary analysis followed an intention-to-treat principle, how participants were handled if they did not receive their assigned intervention, or what exclusions were made. These omissions limit assessment of whether the reported analysis preserves the intended randomized comparison.

The trial summary describes follow-up every 6 months for 3 years, while the cognitive battery is described as annual. The record does not report the completeness of follow-up, the amount or pattern of missing outcome data, or reasons for missingness. Although the primary model is identified as a linear mixed-effects model, the missing-data assumptions and any sensitivity analyses are not reported in the registry record. In general, mixed models can use available repeated measurements, but their interpretation depends on assumptions about why measurements are missing and on the model specification. The presence of a model description alone does not establish that missingness could not have affected the estimate.

Interim analyses and stopping rules are not reported in the registry record. A sample-size justification for the trial is also not reported. The registry's planning illustrations are based on observed effects and are not evidence of the trial's design rationale or observed power. These gaps do not negate the posted results, but they leave important aspects of the planned and implemented analysis unclear.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent non-serious adverse event terms

The serious adverse event entries report zero affected participants in each arm. The record lists 487 participants at risk in the successful-aging health education group and 490 in the hearing intervention group for these entries. It also reports other adverse events affecting 17 participants in the education group and 41 in the hearing group. Deaths are listed for 20 and 21 participants, respectively.

These are counts as recorded, not a comparison adjusted for follow-up time, baseline health, or other factors. The registry does not report event-level descriptions for serious adverse events, causes of death, timing, severity details for other adverse events, or whether events were judged related to an intervention. It also does not report a statistical comparison of these safety entries. Counts alone cannot establish that an intervention caused or prevented an event, and the record does not clarify whether categories overlap. They should be read as descriptive safety information rather than as a causal estimate.

13. What the Result Does, and Does Not, Mean

The primary result is an estimate near zero, with a 95% confidence interval that includes zero, from the registry's mixed-model analysis of change in global cognitive function. It indicates that this analysis did not detect a clear difference between the randomized groups on the primary measure. The interval conveys the uncertainty around that estimate; it does not establish exact equality, exclude every potentially meaningful difference, or show that either group had no cognitive change over time.

The score is a standardized factor score derived from a neurocognitive testing battery, not a direct measure of every aspect of daily function. The registry does not report a threshold for a clinically important difference. Therefore, statistical uncertainty cannot be translated from the record into a definitive judgment about the practical importance of every value in the interval. Nor does the result establish the effects of the interventions in populations beyond the older adults with hearing loss described in the trial summary.

The comparison is between a hearing intervention and successful-aging health education, not between an intervention and no contact. Both groups could experience changes over time for many reasons, and within-group changes do not by themselves identify an intervention effect. The randomized between-group analysis is the relevant comparison for the primary question, subject to the trial's unreported analysis details and missing-data assumptions.

The findings are not medical advice and should not be used to make an individual treatment decision. They describe the outcome estimates and uncertainty recorded for this trial. They do not establish the reason for any participant's cognitive trajectory, prove equivalence, or settle every question about cognition, safety, and hearing rehabilitation.

14. Why This Trial Matters Statistically

This trial illustrates why a clinical result should be read through its estimand, measurement scale, and uncertainty rather than through a p-value alone. Its primary outcome combines information from a battery of cognitive tests into a global factor score, and its reported analysis estimates change over time using a mixed-effects model. Readers need to distinguish the arm-level changes from the between-group result, since the randomized contrast answers a different question from whether either arm changed from baseline.

The reported estimate is close to zero and its confidence interval includes zero. That combination supports a restrained interpretation: the analysis does not show a clear difference, but it is not an equivalence test. A clinical importance threshold would help connect the statistical interval to practical relevance; none is reported. The result also highlights the value of complete reporting about model specification, analysis populations, missing data, and multiplicity. Without those details, readers cannot fully evaluate the assumptions underlying even a clearly stated estimate.

Finally, the registry includes a secondary time-to-event analysis and safety counts. These require different interpretations from the continuous primary endpoint: a hazard ratio is not a mean change, and adverse-event counts alone do not establish causation. Distinguishing outcome types and analysis methods helps prevent an isolated secondary result or descriptive count from overshadowing the primary evidence. This is especially important when the record lists many outcome measures and analyses but does not report a multiplicity strategy.

15. 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 registry record does not report the complete statistical analysis plan, the full mixed-effects model specification, the analysis population, or how missing cognitive measurements were handled. Follow-up completion and the reasons for missing data are also not reported. These details matter because conclusions from longitudinal models depend on their assumptions and implementation. The record does not report a clinically important difference threshold for the global cognitive score, so the clinical relevance of the confidence interval cannot be determined from the stated endpoint scale alone.

The trial was unmasked, as recorded. The registry does not describe whether outcome assessors were masked, so that detail is not reported in the registry record. The comparator involved health education rather than no intervention, and the record does not report adherence, intervention exposure, or treatment fidelity. Those omissions limit interpretation of how the implemented interventions may relate to the observed outcomes.

Detailed results are not reported in the registry record for all named secondary cognitive domains. The multiplicity strategy is not reported, and the secondary cognitive-impairment analysis uses a different outcome type and method from the primary analysis. Safety information is presented as counts without event descriptions, attribution, or an inferential comparison. Finally, the registry does not report a sample-size justification or a threshold for equivalence. These limitations call for caution in interpreting the breadth and practical meaning of the findings, without changing what the posted primary estimate itself states.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Aging and Cognitive Health Evaluation in Elders (ACHIEVE) — NCT03243422.
• Registry record last updated 2024-02-01; status last verified 2024-01.
• Results first posted 2024-02-01.
• 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.

19. Record Summary

ACHIEVE (NCT03243422) is a completed randomized parallel-group trial sponsored by Johns Hopkins Bloomberg School of Public Health. The registry lists an enrollment of 977, no masking, and a comparison between a hearing intervention and successful-aging health education. The primary endpoint was mean change in global cognitive function over 3 years, measured with a factor score derived from neurocognitive testing and analyzed using a linear mixed-effects model.

The reported primary estimate was 0.002 standard deviations, with a 95% confidence interval from -0.077 to 0.081 and p = 0.96. The registry also lists secondary cognitive outcomes and reports a time-to-event analysis for cognitive impairment. It records serious adverse event counts, other adverse event counts, and deaths by group. Details about the full analysis plan, missing-data handling, multiplicity, and analysis populations are not reported in the registry record.