A Study to Evaluate the Efficacy, Safety, and Tolerability of SAGE-324 in Participants With Essential Tremor: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Evaluate the Efficacy, Safety, and Tolerability of SAGE-324 in Participants With Essential Tremor (NCT04305275), a randomized phase 2 study. The primary outcome is Change From Baseline Compared to Placebo in TETRAS Performance Subscale Part 4 Upper Limb Tremor Score on Day 29.

2. Clinical Question

The primary question is about a between-group difference in change, rather than whether either group changed from baseline on its own. The registry reports a least-squares mean difference for SAGE-324 relative to matched placebo. The measure is a clinical rating of upper-limb tremor; the posted result therefore describes the trial’s measured score, not every aspect of essential tremor or participants’ experience.

3. Trial Design

This was a phase 2 randomized, parallel-group, placebo-controlled study of SAGE-324 in participants with essential tremor. The registry lists quadruple masking and also describes the study as double-blind. It reports Sage Therapeutics as the sponsor, an enrollment of 69, and a completed status. The recorded study start was 2020-05-19 and the primary completion date was 2021-02-01.

Participants assigned to the active treatment arm received SAGE-324 60 mg as oral tablets once daily in the morning for 28 days. The comparator arm received matched placebo tablets on the same daily morning schedule for 28 days. The registry’s primary outcome was the change from baseline to Day 29 in the TETRAS Performance Subscale Part 4 upper-limb tremor score.

The registry identifies a Full Analysis Set for the primary outcome. It includes randomized participants who received any amount of SAGE-324 or placebo and had a baseline and at least one postbaseline efficacy assessment. The outcome results list 21 participants analyzed in the SAGE-324 group and 33 in the placebo group. The safety results use different at-risk counts, listed as 34 for SAGE-324 and 35 for placebo. These are outcome-specific analysis counts, and should not be treated as interchangeable.

Other registry outcomes include TETRAS upper-limb tremor, activities-of-daily-living, and total performance scores; a Kinesia ONE accelerometer score; and treatment-emergent adverse events. The registry record does not report the trial’s original sample-size justification or detailed design assumptions. The numerical planning illustration is based on the posted effect and is not evidence of the original enrollment target or design rationale.

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.

Before assignment: A total of 153 participants were screened, of which 69 participants were randomized to receive SAGE-324 or placebo.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set included all randomized participants who received any amount of SAGE-324 or placebo and had a baseline and at least one postbaseline efficacy assessment. 'Overall number of participants analyzed' indicates the number of participants with data available for outcome measure analysis.

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.

Population: Safety Set included all participants who were administered SAGE-324 or placebo.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline Compared to Placebo in TETRAS Performance Subscale Part 4 Upper Limb Tremor Score on Day 29, measured over Baseline, Day 29.

The primary endpoint is change from baseline to Day 29 in the TETRAS Performance Subscale Part 4 upper-limb tremor score, compared between SAGE-324 and matched placebo. TETRAS is a clinical evaluation of essential tremor. Part 4 combines ratings from three maneuvers for each arm: limbs extended forward, the wing-beating position with elbows flexed, and a kinetic finger-nose-finger maneuver.

Each item is scored from 0 to 4. Scores for the items are combined for each arm, and the total upper-limb score across both arms ranges from 0 to 24. Higher scores indicate more severe tremor; a negative change from baseline indicates improvement. The endpoint therefore expresses change in a clinical severity score over the treatment period, rather than a count of events or a time-to-event outcome.

A continuous score can retain information about the degree of change, rather than reducing each participant’s result to a simple improved/not-improved category. The registry reports least-squares means and a least-squares mean difference from a mixed model repeated measures analysis. In general, least-squares means are model-based group estimates, while the treatment contrast quantifies the estimated difference between groups under that model. The endpoint is used to assess whether the average change in the measured upper-limb tremor score differs between randomized groups.

The result pertains specifically to this scale, time point, analysis population, and comparison. It does not by itself describe every symptom or functional outcome, and it should be read alongside the uncertainty around the estimated group difference.

7. Results Posted in the Registry

• SAGE-324 60 mg: mean -2.31 versus -1.24 on control; difference -1.07 (95% CI -2.11 to -0.03), standardized effect d = -0.55; z-test p = 0.044.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a numerical score, and the reported group summaries are least-squares means with standard errors. In general, a group mean describes an average, while a standard error describes uncertainty in estimating that average. A standard error is not the range of individual participant scores. The registry reports standard errors of 0.401 for SAGE-324 and 0.349 for placebo.

What does the reported model do?

The registry says the primary analysis used a mixed model repeated measures (MMRM) approach. In general, repeated-measures models are designed for outcome measurements recorded across more than one time point. They estimate group patterns while accounting for the relationship among repeated observations from the same participant. A mixed model can use available repeated outcome data under its modeling assumptions; that does not mean missing outcomes are automatically harmless or that the assumptions have been verified by the registry summary.

The posted primary contrast is a least-squares mean difference. In general, this is a model-based comparison of estimated group means. It is not necessarily identical to subtracting unadjusted group averages, because a model can incorporate the structure specified in the analysis. The registry reports a difference of -1.07 for SAGE-324 relative to placebo. Given the score direction, the negative estimate indicates a greater average reduction in the SAGE-324 group under the reported analysis.

How should the p-value be understood?

The sponsor’s MMRM result reports p = 0.049. In general, a p-value measures how unusual data at least as incompatible with a specified null hypothesis would be if that null hypothesis and the analysis assumptions held. It is not the probability that the null hypothesis is true, nor the probability that the finding occurred “by chance.” The value should be considered together with the estimated effect, its confidence interval, the endpoint definition, and the analysis context. The registry record does not report whether this p-value was adjusted for multiple outcome testing.

Why is there also a recalculation?

A simple unadjusted recalculation can be made from the posted summary statistics, using an approximation based on the reported group means and standard errors. It produces the same mean difference, -1.07, and p = 0.044. Such a calculation does not reproduce the MMRM’s full data structure, repeated-measures modeling, or any model terms not recoverable from the posted summaries. It is therefore a secondary check, not the sponsor’s primary analysis. The recalculated standard deviation is described as approximate in the underlying calculation, so the result should not be interpreted as a more exact analysis.

In general, over-dispersion means that observed variability is greater than assumed by a model. When relevant to a particular type of outcome, ignoring over-dispersion can make uncertainty appear too small. This endpoint is a continuous score and the registry reports a repeated-measures model; the registry record does not report an over-dispersion assessment or indicate that this was a modeling issue for the primary analysis.

What does the analysis population mean?

The stated Full Analysis Set includes randomized participants who received any amount of assigned study treatment and had baseline and at least one postbaseline efficacy assessment. The group counts for the primary outcome are 21 and 33. In general, defining an analysis population makes clear which participants contribute to a result. The registry definition is not the same as saying that every randomized participant necessarily contributed a Day 29 score. The record does not report detailed participant flow or the reasons outcome data may be unavailable.

The registry record also does not report the detailed MMRM specification, such as its covariance structure, the exact fixed effects, or how missing observations were handled in the fitted analysis. It does not report interim analyses or the trial’s original sample-size justification. These omissions limit independent reproduction of the sponsor’s model-based result from the summary alone.

10. Confidence Intervals and Precision

The registry reports a 95% confidence interval for the least-squares mean difference from -2.14 to 0.00. The estimated difference is -1.07, on the TETRAS upper-limb score scale. Because the scale’s negative change indicates improvement, negative between-group differences correspond to a greater reduction in the SAGE-324 group under the stated contrast. The interval conveys uncertainty about the size of that difference, not a range containing a fixed proportion of individual participants’ outcomes.

In general, a confidence interval is a range of estimates compatible with the observed data under the statistical model and its assumptions. A wider interval indicates less precision than a narrower interval on the same scale. The registry interval extends from a potentially more substantial difference to 0.00 as its reported upper endpoint. This means the estimate is not highly precise about the magnitude, even though the point estimate is separated from zero.

The reported p-value is 0.049. The fact that this is close to a conventional threshold should not be treated as a sharp division between a definitive and a non-definitive finding. The confidence interval, point estimate, analysis methods, and context all matter. The upper endpoint is printed as 0.00, and the registry’s reported values are rounded; that displayed endpoint and the p-value should be read as the posted results, not used to infer unreported precision.

The outcome analysis includes 21 participants in the SAGE-324 group and 33 in the placebo group. The registry also reports standard errors of 0.401 and 0.349 for the group estimates. Precision depends on factors including sample size, variability, and the analysis model; the group counts and standard errors give context for the uncertainty but do not remove it. The registry record does not report additional precision measures such as a prespecified clinically important difference.

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 sample-size figures are planning illustrations based on the observed effect, not the trial’s original design. Each row uses a different fraction of the observed effect: 100%, 75%, or 50%. For this continuous outcome, the displayed figures are participants per arm, not total event counts. The rows give the number of participants per arm for 80% power and the number of participants per arm for 90% power.

In plain terms, the 80% power column describes the planned number per arm under the stated assumptions for a design with an 80% chance of detecting the specified effect if that effect and the assumptions were correct. The 90% power column describes the corresponding planning number for a design with a 90% chance under those conditions. Higher planned power requires more participants for the same effect size. The figures use two-sided significance level 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect.

The illustration shows that planning for a smaller effect requires more participants per arm. For example, the 75% row lists 92 participants per arm for 80% power and 124 for 90% power; the 50% row lists 207 and 277, respectively. These are prospective planning calculations, not counts from the completed trial. They are not observed power.

The registry record does not report the original sample-size justification or the assumptions used when the trial was designed. These calculations should therefore be read as illustrations based on the observed result, not as evidence that the trial was initially planned to detect a particular effect or that its enrollment target was selected by this method.

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 = -1.070; pooled standard deviation σ = 1.942.
2. n = 2 × (1.96 + 0.842)² × 1.942² / -1.070² = 51.7, rounded up to 52 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 <- -1.07                      # observed difference in means
sigma <- 1.94223441902916                     # 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(4305275)
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 = 52 (80% power), 70 (90% power)
 75% of observed effect: n per arm = 92 (80% power), 124 (90% power)
 50% of observed effect: n per arm = 207 (80% power), 277 (90% power)
power.t.test: 52.7 
Simulated power at n = 52 per arm: 0.800 (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. Analyses Reported by the Sponsor

The sponsor’s reported primary analysis is an MMRM comparison of least-squares means. It gives a difference of -1.07, a 95% confidence interval from -2.14 to 0.00, and p = 0.049. This is the trial’s reported model-based result. The record does not flag a discrepancy between the sponsor’s analysis and the simple recalculation.

The unadjusted recalculation from the posted summary statistics also gives a difference of -1.07, with p = 0.044. It uses an approximation from group summaries and does not reproduce the full repeated-measures analysis. Its role is to provide a basic numerical check; it should not be described as the sponsor’s analysis or as a substitute for the MMRM result.

The reported estimate is the same in both summaries, while the p-values differ because the calculations are not identical in method. For interpretation of the trial result, the sponsor’s own MMRM analysis takes precedence when a discrepancy is flagged; the registry does not flag one. Regardless, the recalculation does not provide a separate confirmation from independent participant-level data. It is based on the same posted outcome summaries.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists the primary endpoint and several secondary outcomes, including additional TETRAS scores, a Kinesia ONE accelerometer score, and treatment-emergent adverse events. When multiple outcomes or time points are tested, the chance of obtaining at least one apparently unusual result can increase. In general, multiplicity procedures may control that risk by adjusting inference or by specifying a testing hierarchy. The registry record does not report whether the primary p-value or secondary analyses were adjusted for multiple comparisons, nor does it report a testing hierarchy. The reported p-value should therefore be read as posted, without assuming an unreported adjustment.

The primary outcome timeframe is baseline to Day 29. The secondary outcomes include several assessment days, so there are multiple reported opportunities to assess change. The registry record does not provide enough information to determine how all outcome and time-point comparisons were handled together. The number of analyses reported is 26, but the record does not describe a multiplicity strategy for those analyses.

The stated Full Analysis Set includes randomized participants who received any amount of treatment and had baseline and at least one postbaseline efficacy assessment. The registry reports 21 analyzed participants for SAGE-324 and 33 for placebo on the primary outcome. It does not report detailed reasons for missing outcome measurements, participant flow at each assessment, or the exact missing-data assumptions used in the MMRM. In general, repeated-measures models can use observed outcomes under specified assumptions, but that does not guarantee that missingness has no effect on the estimate.

The registry record does not report interim analyses or the trial’s detailed statistical analysis plan. It also does not report whether sensitivity analyses were conducted to assess how alternative assumptions about missing outcomes might affect the result. These gaps limit evaluation of the robustness of the primary finding and of the interpretation of secondary results.

15. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: mental status changes (5.9%), dehydration (2.9%), headache (2.9%), transient ischaemic attack (2.9%), angina pectoris (2.9%).

The registry lists serious adverse events among participants at risk in each safety group. For SAGE-324, 34 participants were at risk and 3 were affected; for matched placebo, 35 were at risk and 1 was affected. The registry lists zero deaths in each group. These figures describe the recorded serious adverse-event counts for the listed safety groups; they do not by themselves establish whether an event was caused by study treatment.

The registry also names serious-event terms and percentages: Mental status changes, 5.9%; Dehydration, 2.9%; Headache, 2.9%; Transient ischaemic attack, 2.9%; Angina pectoris, 2.9%; and Dyspnoea, 2.9%. The record does not provide enough detail in these figures alone to determine the clinical course, timing, relatedness assessment, or distribution of every listed term by treatment group. The terms and percentages should be read as registry safety reporting, not as a comparative risk analysis.

The at-risk counts differ from the primary efficacy analysis counts. Safety and efficacy populations can be defined differently, so the denominators should not be interchanged. These serious-event figures also do not summarize all treatment-emergent adverse events or establish the overall safety profile. The registry lists treatment-emergent adverse events as a secondary outcome, but interpretation of safety requires the full context of event types, severity, exposure, and follow-up.

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

The registry results show that the average TETRAS upper-limb tremor score decreased from baseline in both groups at the primary assessment. The estimated reduction was larger in the SAGE-324 group than in the matched-placebo group, and the reported MMRM contrast was -1.07. On this score, negative change represents improvement, so the direction of the estimate favors a greater reduction in the active-treatment group for this measured outcome.

The result does not mean that every participant had a reduction, or that the group difference describes every individual’s experience. It is a group-level estimate from the defined Full Analysis Set. The confidence interval, from -2.14 to 0.00, shows uncertainty around the estimated difference and includes the rounded boundary value of no difference. The p-value of 0.049 is close to a conventional threshold and should not be treated as proof of a particular magnitude or as a measure of clinical importance.

The finding is specific to the TETRAS Performance Subscale Part 4 upper-limb tremor score at Day 29, in a phase 2 study with the stated treatment schedule and analysis population. It does not by itself establish persistence beyond the assessments reported, effects on every secondary outcome, or a broader effect on all aspects of essential tremor. Nor do the serious adverse-event counts alone establish whether treatment caused or prevented any event.

The posted result can support a statistical description of the comparison under the sponsor’s stated model. It cannot answer unreported questions about the original sample-size rationale, missing-data sensitivity analyses, multiplicity adjustment, or the clinical importance threshold for this score. Those items are not reported in the registry record. Interpretation should distinguish the estimated score contrast from claims that go beyond the endpoints, time points, and analyses recorded.

17. Why This Trial Matters Statistically

This trial illustrates why a treatment comparison should be read through its endpoint definition and score direction. The primary measure is a clinical rating with higher scores indicating more severe tremor, while a negative change from baseline indicates improvement. Without that direction, the negative group means and negative between-group estimate could be misread.

It also shows the value of separating a model-based result from a simple summary-statistic recalculation. The sponsor’s MMRM uses a repeated-measures framework and reports a least-squares mean difference; the recalculation is an approximate, unadjusted check using posted summaries. Their shared estimate does not make the methods interchangeable, and their p-values are not identical. Understanding what each analysis uses is essential to interpreting its inferential result.

Precision is another central issue. The estimated difference is -1.07, while the 95% confidence interval ranges from -2.14 to 0.00. The point estimate alone does not convey that uncertainty, and the p-value alone does not describe the range of plausible effect sizes. A close-to-threshold p-value should not substitute for considering the interval, the score scale, and the endpoint’s meaning.

Finally, the record makes clear that efficacy and safety summaries may use different analysis denominators, and that trial planning illustrations are distinct from the original design rationale. These distinctions help prevent overinterpretation of a completed phase 2 comparison. The result is best understood as evidence about a defined score at a defined assessment, accompanied by limitations in the registry’s reporting about multiplicity, missing data, and design assumptions.

18. 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’s primary outcome result is reported at Day 29, and the record does not establish how durable the observed score difference may be beyond the posted assessment schedule. Although secondary outcomes include other time points, a broader interpretation requires their results and the analysis context rather than extrapolation from the primary endpoint alone.

The primary efficacy analysis includes 21 participants in the SAGE-324 group and 33 in the placebo group, while the registry enrollment is 69. The safety at-risk counts are 34 and 35. These figures describe different registry populations or outcome records; the record does not provide detailed participant flow or reasons for participants not appearing in the primary analysis. No totals or participant-level transitions should be inferred from these separate counts.

The registry record does not report the detailed MMRM specification, missing-data sensitivity analyses, multiplicity adjustment, interim analyses, or original sample-size justification. The primary confidence interval reaches 0.00 at its rounded upper endpoint, and the reported p-value is close to a conventional threshold. Together, these points make effect-size uncertainty and analysis assumptions important to the interpretation.

The registry describes the study as double-blind in its summary and lists quadruple masking in its design fields. Both descriptions are part of the record; it does not explain the distinction. The reported outcomes and safety counts also do not supply a clinical importance threshold for the TETRAS score or establish causal attribution for individual adverse events. These limitations constrain conclusions to the specific comparison and reported measures.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study to Evaluate the Efficacy, Safety, and Tolerability of SAGE-324 in Participants With Essential Tremor — NCT04305275.
• Registry record last updated 2024-04-17; status last verified 2024-03.
• Results first posted 2024-04-17.
• 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.

22. Record Summary

NCT04305275 is a completed randomized phase 2 study of SAGE-324 60 mg versus matched placebo in participants with essential tremor. The registry reports Sage Therapeutics as sponsor, an enrollment of 69, and a parallel-group, placebo-controlled design. Treatment was taken orally once daily in the morning for 28 days.

The primary outcome was change from baseline to Day 29 in the TETRAS Performance Subscale Part 4 upper-limb tremor score. The sponsor’s MMRM result reports a least-squares mean difference of -1.07, a 95% confidence interval from -2.14 to 0.00, and p = 0.049. A negative change indicates improvement on this scale.

The primary analysis population and safety counts differ, and the registry record does not report several details needed for fuller assessment, including multiplicity handling, missing-data sensitivity analyses, or original sample-size assumptions. The posted result should be interpreted as a model-based comparison for this endpoint and assessment, with the stated uncertainty and reporting limitations.