Brain Energy for Amyloid Transformation in Alzheimer's Disease Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Brain Energy for Amyloid Transformation in Alzheimer's Disease Study (NCT03472664), a randomized na study. The primary outcome is Cerebrospinal Fluid Abeta42.

2. Clinical Question

3. Trial Design

BEAT-AD is a randomized, double-masked, parallel-group study. The registry lists an enrollment of 84 and a completed status. The study description says volunteers with mild cognitive impairment were assigned to one of two diet interventions. The registered intervention names are Modified Mediterranean Ketogenic Diet and American Heart Association Diet.

The summary describes a 16-week diet period and an 8-week follow-up after the diet ended, for a 24-week study period. It says measures, clinic visits, and phone sessions occurred at baseline and throughout the study. Participants were to resume their normal diet after completing the diet period. The registry record gives a start date of 2018-07-16 and a primary completion date of 2024-07-19.

The primary outcome reported for statistical interpretation is cerebrospinal fluid Aβ42 at baseline. The registry gives a mean and standard deviation for each group, with 33 analyzed in the ketogenic-diet group and 39 in the low-fat-diet group. It explicitly notes that differences in the number analyzed arose because of failed lumbar punctures. These analyzed counts should not be confused with the registered enrollment figure.

The registry lists 33 outcome measures and 28 posted secondary outcomes. Secondary measures include cerebrospinal fluid and plasma biomarkers. The registry lists zero analyses. Details of the randomization procedure, the people masked, allocation concealment, adherence assessment, and the trial’s original statistical analysis plan are not reported in the registry record. The posted baseline comparison is an unadjusted recalculation from the group summary statistics, rather than a description of a reported longitudinal treatment-effect analysis.

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:

Analysis population for the primary outcome, as defined in the registry: discrepancies in number of participants analyzed due to failed lumbar puncture

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 Cerebrospinal Fluid Abeta42, measured over baseline.

The registered primary measure is cerebrospinal fluid Aβ42, also written Aβ42, in pg/ml. The registry describes Aβ42 as a biomarker reflecting pathological aggregation of amyloid in the brain. It is a continuous laboratory measure, so the posted results summarize it with a mean and standard deviation for each group.

Biomarkers such as cerebrospinal fluid Aβ42 can be used to characterize biological processes relevant to Alzheimer disease. In this record, the measure is specified at baseline. Baseline values provide a description of the groups at the start of the study; they are not a measure of change caused by the diet. Although the study summary says measures were collected throughout the 24-week study, a primary follow-up value or change-from-baseline result for Aβ42 is not reported in the registry record.

The registry reports a mean of 1046.0 pg/ml for the Modified Mediterranean Ketogenic Diet group and 956.1 pg/ml for the American Heart Association Diet group. The standard deviations are 547.1 and 418.1, respectively. These group summaries describe the observed baseline values among those analyzed. The registry notes that the numbers analyzed differed because of failed lumbar punctures, so the baseline comparison also reflects which participants had an available cerebrospinal fluid measurement.

The endpoint’s biological relevance does not make a baseline difference a treatment effect. To assess whether the diets altered Aβ42, results comparing measurements over time would be needed; such results are not reported in the registry record.

7. Results Posted in the Registry

• American Heart Association Diet: mean 956.15 versus 1,046 on control; difference -90.00 (95% CI -318.18 to 138.18), standardized effect d = -0.19; z-test p = 0.439.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is continuous, and the registry reports each group’s mean and standard deviation. A mean represents the average measured value in a group. A standard deviation describes how spread out individual values are around that average; it is not the uncertainty in the mean itself. The posted baseline summaries are 1046.0 pg/ml with a standard deviation of 547.1 for the ketogenic-diet group, and 956.1 pg/ml with a standard deviation of 418.1 for the low-fat-diet group.

What does the mean difference describe?

The reported contrast is the mean in the American Heart Association Diet group minus the mean in the Modified Mediterranean Ketogenic Diet group. Its estimate is -90.0 pg/ml. A negative value means the comparator group’s observed mean is lower than the ketogenic-diet group’s observed mean at baseline. This direction is only a description of the baseline data; it does not describe a diet effect.

The pooled standard deviation, 481.4 pg/ml, combines the within-group variability for the purpose of the posted standardized comparison. The standardized mean difference is -0.187. Standardization expresses a mean difference relative to a measure of variability, which can help compare differences measured on different scales. It does not change the fact that these are baseline data, and it does not establish whether a difference matters clinically.

What does the confidence interval add?

A confidence interval describes the statistical uncertainty around an estimate under the method used. The 95% interval for the unadjusted mean difference extends from -318.2 to 138.2 pg/ml. It includes zero and also allows values on either side of zero. General interpretation: an interval that includes zero does not rule out no difference, while its breadth indicates that appreciably different baseline contrasts remain compatible with the data. The interval is not a range containing 95% of individual participants’ Aβ42 values.

How should the p-value be understood?

The unadjusted p-value is 0.439. In general, a p-value measures how unusual data at least as discrepant as those observed would be under a specified null model. It is not the probability that the null hypothesis is true, nor does it quantify the size or clinical importance of a difference. Here it accompanies an unadjusted comparison of baseline group means. It is not a test of change during the diet period.

The posted standard errors for the group means are 95.25 and 66.95. A standard error describes uncertainty in an estimated mean, rather than variation among individual observations. The registry record does not report the trial’s original statistical model or analysis plan. The comparison should therefore be read as the unadjusted recalculation based on the posted summary statistics, not as an adjusted analysis or a reconstruction of an unreported protocol.

Why do analyzed group sizes matter?

The registry reports 33 analyzed participants for the ketogenic-diet group and 39 for the low-fat-diet group. It notes that differences in the number analyzed were due to failed lumbar punctures. Missing outcome measurements can affect both precision and representativeness if participants with measurements differ from those without them. General statistical practice may use methods to address missingness, but the method used for this outcome is not reported in the registry record. The available summaries cannot show whether the missing measurements changed the estimated group contrast.

The registry lists 33 outcome measures and 28 posted secondary outcomes. When many outcomes are examined, the chance of at least one apparently unusual result can increase. General approaches to this issue include defining a primary outcome in advance and adjusting inference for multiple comparisons. Whether any such approach was part of this trial’s analysis is not reported in the registry record. No secondary-outcome estimates are included in the registry entries considered for this interpretation.

Finally, the registry describes a randomized, double-masked, parallel-group study, but those design labels do not replace a result for change over time. For a continuous outcome measured repeatedly, an analysis might compare follow-up values, changes from baseline, or trajectories using a specified model. Which approach was planned or used for Aβ42 is not reported in the registry record. No such model should be inferred from the baseline means alone.

10. Confidence Intervals and Precision

The baseline mean difference is -90.0 pg/ml, with a 95% confidence interval from -318.2 to 138.2 pg/ml. The interval is wide relative to the estimated difference and crosses zero. It therefore leaves uncertainty both about the direction and the size of the baseline contrast. The unadjusted p-value of 0.439 is consistent with the interval’s failure to exclude zero, but neither statistic establishes equivalence between the groups.

Precision is also informed by the spread of the observations. The standard deviations are 547.1 pg/ml in the ketogenic-diet group and 418.1 pg/ml in the low-fat-diet group. The pooled standard deviation is 481.4 pg/ml. The posted standard errors of the means are 95.25 and 66.95. These summaries convey distinct things: standard deviations describe variation among observed values, while standard errors and confidence intervals describe uncertainty in estimated quantities.

The analyzed group sizes are 33 and 39. The registry attributes differences in the number analyzed to failed lumbar punctures. The confidence interval reflects the precision of the posted unadjusted calculation using the available summaries; it does not account for uncertainty about how missing measurements might have affected the comparison. The registry record does not report a missing-data analysis or an adjusted estimate.

Most importantly, the confidence interval concerns a baseline difference. It does not quantify uncertainty in a treatment-related change, because a change estimate is not reported in the registry record. Narrower uncertainty around a baseline contrast would still not answer whether either diet altered Aβ42 during the intervention.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are illustrative calculations based on the observed baseline effect, not the trial’s own design assumptions. They use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to the stated fraction of the observed effect. These figures are participants per arm for 80% power or participants per arm for 90% power; they are not total event counts.

For planning for 100% of the observed effect, the illustration gives 450 participants per arm for 80% power and 602 participants per arm for 90% power. For planning for 75% of the observed effect, it gives 799 participants per arm for 80% power and 1,069 participants per arm for 90% power. For planning for 50% of the observed effect, it gives 1,797 participants per arm for 80% power and 2,406 participants per arm for 90% power.

In plain language, 80% power is the probability a study would detect the specified effect under the calculation’s assumptions if that effect were the true one; 90% power describes a higher probability of detecting it under the same kind of assumptions. Detecting a smaller effect generally requires more participants, which is reflected in the larger planning figures for smaller fractions of the observed effect.

These are planning illustrations, not observed power and not a record of the trial’s original sample-size justification. The endpoint is a baseline comparison, and the observed effect used for the illustration is not a treatment-related change. The trial’s own original design assumptions are not reported in the registry record.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -90.000; pooled standard deviation σ = 481.401.
2. n = 2 × (1.96 + 0.842)² × 481.401² / -90.000² = 449.1, rounded up to 450 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -90.0000000000001                      # observed difference in means
sigma <- 481.401396932509                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(3472664)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 450 (80% power), 602 (90% power)
 75% of observed effect: n per arm = 799 (80% power), 1,069 (90% power)
 50% of observed effect: n per arm = 1,797 (80% power), 2,406 (90% power)
power.t.test: 450.1 
Simulated power at n = 450 per arm: 0.798 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists 33 outcome measures and 28 posted secondary outcomes. The primary measure discussed in the posted results is baseline cerebrospinal fluid Aβ42. Examining multiple outcomes can create a multiplicity issue: as the number of statistical tests grows, the chance of at least one apparently notable result can also grow. General statistical approaches include prespecifying a primary outcome, organizing secondary outcomes into families, or adjusting inference for multiple tests. Whether the trial used any such approach is not reported in the registry record.

The registry lists zero analyses. It does not report an analysis plan, a multiplicity adjustment, or an inferential method for the secondary outcomes. The unadjusted p-value of 0.439 belongs to the posted baseline mean comparison; it is not described as adjusted for other outcomes or analyses. No claims about secondary biomarker or clinical results can be made from the available outcome details.

For the primary measure, the registry explicitly notes differences in the number analyzed due to failed lumbar punctures. The group summaries include 33 and 39 analyzed participants. The registry record does not report how missing cerebrospinal fluid results were handled, whether missingness was assessed by assigned group, or whether any imputation or sensitivity analysis was performed. General concern about missing data is that estimates may be affected if the chance of having an observed measurement relates to the outcome or other participant characteristics.

The registry record also does not state the precise analysis population used for the primary comparison, such as whether it followed assignment regardless of adherence or included only participants with available measurements. It does not report a participant flow for the primary endpoint or explain whether other exclusions were made. These omissions limit assessment of how the analyzed groups relate to the randomized groups and how robust the result is to missing data. No particular handling method or analysis population should be assumed.

14. 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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cholelithiasis (2.5%), diverticulitis (2.5%), gastric ulcer (2.5%), hydronephrosis (2.5%), pulmonary nodule excision (2.5%).

The registry reports serious adverse events among 44 participants at risk in the Modified Mediterranean Ketogenic Diet group, with 1 affected, and among 40 participants at risk in the American Heart Association Diet group, with 6 affected. It reports zero deaths in each group. These are the posted serious-event counts and denominators; they should be read as registry safety summaries, not as proof that the diets caused or prevented any event.

The registry also lists other adverse events among 44 participants at risk in the ketogenic-diet group, with 30 affected, and among 40 at risk in the low-fat-diet group, with 15 affected. Serious and other adverse-event categories are reported separately. The record lists cholelithiasis, diverticulitis, gastric ulcer, hydronephrosis, pulmonary nodule excision, and small bowel obstruction among the leading serious-event terms, each at 2.5%.

The registry record does not provide enough detail to determine event timing, whether events were judged related to the diet, how repeated events in the same participant were handled, or how adverse-event ascertainment was conducted. Counts across categories should not be added to estimate unique participants. These summaries describe what was recorded and do not establish comparative safety.

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

The central statistical finding is a baseline comparison of cerebrospinal fluid Aβ42, not an estimate of change after either diet. The ketogenic-diet group mean was 1046.0 pg/ml and the low-fat-diet group mean was 956.1 pg/ml. The unadjusted difference, expressed as low-fat diet minus ketogenic diet, was -90.0 pg/ml, with a 95% confidence interval from -318.2 to 138.2 pg/ml and an unadjusted p-value of 0.439.

The interval includes zero and is broad, so the baseline data do not pin down the direction or size of the difference precisely. This does not establish that the groups are equivalent. Nor does the baseline comparison establish that either diet altered Aβ42. To answer a treatment-effect question, an estimate comparing follow-up results or change over time would be needed; that result is not reported in the registry record.

The study summary describes additional interests, including cognitive function, cerebral blood flow, proteins, and hormones in body fluids. The registry lists several secondary biomarker outcomes, but estimates for those outcomes are not included in the record details. Their absence means no conclusion about those measures can be drawn from the posted primary comparison.

The analyzed counts differ, and the registry attributes this to failed lumbar punctures. The available data therefore describe participants with analyzed baseline measurements. Without a reported missing-data method, analysis population, or longitudinal analysis, it is not possible to determine how missing measurements affect conclusions about the enrolled or randomized groups.

The result is best understood as a descriptive, imprecise baseline contrast. It does not demonstrate a beneficial or harmful effect, does not establish a lack of effect, and does not answer whether the interventions changed a biomarker or clinical outcome.

16. Why This Trial Matters Statistically

This trial illustrates why the timing of an endpoint matters as much as its label. A biomarker can be relevant to a disease process, but a baseline measurement is a description of participants before the intervention period. A baseline contrast is not a treatment-effect estimate. Statistical interpretation must match the question the reported measurement can answer.

The example also shows why an estimate should be read together with its uncertainty. The mean difference is -90.0 pg/ml, but its 95% confidence interval ranges from -318.2 to 138.2 pg/ml. That interval indicates substantial uncertainty around the baseline contrast. A point estimate alone could give an overly definite impression; the interval makes clear that materially different values remain compatible with the calculation.

Missing outcome measurements are another important part of the interpretation. The registry notes failed lumbar punctures as the reason for differences in analyzed numbers. A complete-case summary can be informative, but assessment of its relationship to the randomized groups depends on the missing-data process and the method used to handle it. Those details are not reported in the registry record.

Finally, the record lists many outcomes, while the posted analysis information is limited. When multiple outcomes are assessed, prespecified priorities and multiplicity procedures can affect how findings should be interpreted. The registry does not report whether such procedures were used. These statistical considerations do not diminish the value of collecting biological and clinical measurements; they clarify which conclusions the available estimates can support and which questions remain unanswered.

17. 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 main limitation for interpreting the primary result is that the posted measure is at baseline. Although the study summary describes repeated assessments across the 24-week study, a follow-up Aβ42 estimate or change-from-baseline analysis is not reported in the registry record. The posted comparison therefore cannot quantify a diet-related change.

The registry notes that the analyzed counts differ because of failed lumbar punctures. It reports group summaries for 33 participants in the ketogenic-diet group and 39 in the low-fat-diet group, but does not report a missing-data method, imputation, sensitivity analysis, or detailed flow for this endpoint. The effect of unavailable measurements on the comparison cannot be assessed from the reported summaries.

The registry lists 33 outcome measures and 28 posted secondary outcomes, but lists zero analyses. The statistical analysis plan, multiplicity handling, and analysis population are not reported in the registry record. This limits evaluation of whether the posted calculation was prespecified, adjusted for other factors, or part of a broader testing strategy.

The record describes randomization and double masking but does not report details of the randomization process or identify who was masked. It also does not provide adherence results for the diet interventions in the registry record. The baseline comparison does not resolve these design or implementation questions.

Finally, the planning figures are illustrative and based on the observed baseline effect, rather than the trial’s original design assumptions. They should not be treated as evidence about the trial’s achieved power or as a justification for its enrollment.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Brain Energy for Amyloid Transformation in Alzheimer's Disease Study — NCT03472664.
• Registry record last updated 2026-08-03; status last verified 2026-05.
• Results first posted 2026-08-03.
• 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.

21. Record Summary

The registry record describes BEAT-AD as a completed, randomized, double-masked, parallel-group study sponsored by Wake Forest University Health Sciences, with enrollment of 84. It compares a Modified Mediterranean Ketogenic Diet with an American Heart Association Diet in adults with mild cognitive impairment. The study summary describes a 16-week diet period and an 8-week follow-up.

The primary measure considered is baseline cerebrospinal fluid Aβ42 in pg/ml. The group means are 1046.0 and 956.1, and the unadjusted mean difference is -90.0, with a 95% confidence interval from -318.2 to 138.2 and an unadjusted p-value of 0.439. The registry attributes differences in the analyzed counts to failed lumbar punctures.

The posted baseline comparison does not estimate change during treatment. A longitudinal primary-outcome result, missing-data method, multiplicity approach, and original statistical analysis plan are not reported in the registry record.