The Effect of Glutamatergic Modulation on Cocaine Self-administration: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Effect of Glutamatergic Modulation on Cocaine Self-administration (NCT02596022), a randomized phase 2 study. The primary outcome is Number of Choices to Self-administer Cocaine (Out of 5 Choices).

2. Clinical Question

The registry describes a laboratory task intended to assess choices between immediate cocaine and a delayed monetary reward. At each opportunity, the options were cocaine 25 mg now or $11 later. The choices were spaced 15 minutes apart over the session. This outcome represents behavior within that task and at the stated assessment time; it is not a direct measure of all cocaine use or of longer-term behavior.

The scientific rationale in the registry summary is that glutamate modulation might affect the balance between immediate drug use and delayed non-drug rewards. The trial tested that question by comparing two drug conditions in a crossover design. The posted outcome is the number of cocaine choices out of 5, reported as a mean with a standard deviation for each condition. The comparison therefore describes the average task response under the two conditions, not a diagnosis, a broader measure of motivation, or a long-term clinical outcome.

3. Trial Design

This was a Phase 2, randomized, quadruple-masked, crossover trial listed as completed. The registry records 18 participants enrolled and identifies cocaine dependence as the condition under study. The sponsor was New York State Psychiatric Institute. The recorded start was 2013-06 and the primary completion was 2015-06.

In a crossover design, participants receive more than one study condition in a specified order. The registry describes two sequences: CI-581a followed 2 weeks later by CI-581b, and CI-581b followed 2 weeks later by CI-581a. Because the same participants may contribute observations under both conditions in a crossover trial, the comparison can be informed by within-participant as well as between-participant variation. The statistical model and details of how the order or repeated observations were handled are not reported in the registry record, so the posted comparison should not be assumed to reflect a particular crossover model.

Random allocation indicates that assignment to a sequence was randomized. The record identifies masking as quadruple, but does not specify which roles were masked. The active control is identified as CI-581b in the summary. The posted outcome lists 18 observations for CI-581a and 18 for CI-581b, while the overall enrollment is 18. These condition-level counts are consistent with observations under both conditions, but further detail about completion, exclusions, or analysis populations is not reported in the registry record.

The primary outcome was the mean number of choices to self-administer cocaine, out of 5 choices, assessed 24 hours post-infusion. The registry reports one outcome measure and no posted secondary outcomes. It lists no posted analysis entries. Details of the original sample-size justification, planned analysis method, handling of carryover or period effects, and any interim analyses are not reported in 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

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 Number of Choices to Self-administer Cocaine (Out of 5 Choices), measured over 24 hours post-infusion.

The primary endpoint counts the number of task opportunities on which a participant chose cocaine 25 mg now rather than $11 later. Participants made 5 choices, with choices spaced 15 minutes apart over the session. The assessment was scheduled for 24 hours post-infusion. The registry reports the outcome as a mean and standard deviation for each condition.

This endpoint is used as a behavioral measure of the choices made in the laboratory task. It maps directly onto the registry’s question about the relative salience of immediate cocaine and a delayed monetary reward. A larger mean indicates that more cocaine choices were recorded on average under that condition in this task. It does not, by itself, establish what participants would choose in other settings or over a longer period.

The result is a count constrained by the task’s 5 opportunities, rather than a continuous measure such as a concentration. The condition means were 1.61 for CI-581a and 4.33 for CI-581b. The standard deviations were 1.3 and 1.8, respectively. These summary statistics describe the observed task responses but do not show the full distribution of individual choices, the number of participants with each possible response, or how responses changed within individuals across conditions. Those details are not reported in the registry record.

7. Results Posted in the Registry

• CI-581b: mean 4.33 versus 1.61 on control; difference 2.72 (95% CI 1.69 to 3.75), standardized effect d = 1.73; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a continuous measure in the registry’s terminology: a mean number of cocaine choices. Although each participant’s response is a count of choices, the posted group-level summaries are means and standard deviations. The mean describes the average number of choices in a condition. The standard deviation describes how dispersed participant values were around that mean; it is not the uncertainty in the mean itself. The posted standard deviations were 1.3 for CI-581a and 1.8 for CI-581b.

The registry reports a mean difference of 2.72, comparing CI-581b with CI-581a. A mean difference preserves the original outcome scale, so its interpretation is in choices on the laboratory task. It is positive in the stated comparison because the CI-581b mean, 4.33, is higher than the CI-581a mean, 1.61. The estimated difference summarizes the contrast between the condition averages; by itself it does not show how every participant responded or how much individual responses changed.

The posted 95% confidence interval extends from 1.694 to 3.746. In general, a confidence interval communicates uncertainty around an estimate by giving a range of values that are compatible with the observed data under the analysis assumptions and procedure. A narrower interval indicates greater precision than a wider interval on the same scale. The interval is not a range containing a stated proportion of individual participant outcomes, and it does not mean that the true value has a probability of being within the already-calculated interval.

The unadjusted p-value is <0.001. In general, a p-value evaluates how surprising the data, or data more extreme, would be under a null hypothesis and the assumptions of the statistical test. The registry calls this value unadjusted. It should not be interpreted as a measure of effect size or as the probability that the result occurred by chance. The precise test and its assumptions are not reported in the registry record.

The record also reports a pooled standard deviation of 1.57 and d = 1.732. A standardized difference puts a mean contrast in relation to a pooled measure of variability, rather than keeping it in the original choice-count units. This can give a scale-free description of magnitude, but it depends on the variability used for standardization and is less directly tied to the task’s choices than the mean difference. The registry record does not describe the detailed calculation method for this standardized estimate.

Randomization and masking describe design features, not statistical analyses. Randomization assigned participants to a sequence, while the registry labels the trial quadruple-masked. The specific people or roles masked are not reported. The crossover design means that sequence and repeated measurements matter: observations from the same participant are related, unlike observations from unrelated participants. A suitable crossover analysis generally needs to account for that dependence and may need to consider period or carryover effects. This is a general methodological point; which model or adjustments were used in this trial is not reported in the registry record.

Over-dispersion is a general concern when count data vary more than a simple count model expects. It can affect standard errors and confidence intervals if a model does not allow for the observed variability. The registry presents means and standard deviations for a bounded choice count, but does not report a count model, an over-dispersion assessment, or any diagnostics. It is therefore not possible to determine from the registry whether over-dispersion was evaluated or whether a particular modeling approach was appropriate for these data.

The posted comparison includes an unadjusted p-value. Whether the investigators planned multiple statistical tests, adjusted for multiplicity, or selected a single primary comparison in advance is not reported in the registry record. Only one outcome measure is listed and no secondary outcome measures are posted, but this does not establish the full analysis plan or the number of analyses considered. The registry lists no analysis entries, so details such as covariate adjustment, the handling of missing data, and the definition of the analysis population are not available.

General statistical interpretation also depends on the distinction between an estimate and its uncertainty. The mean difference is the point estimate; the confidence interval describes precision under the method used; and the p-value addresses compatibility with a null hypothesis under a test. None alone gives a complete account of design quality, bias, measurement validity, or the relevance of the task to outcomes outside the laboratory. The registry’s numerical summaries are informative about the posted outcome but do not replace details of the trial protocol and statistical analysis plan.

10. Confidence Intervals and Precision

The posted mean difference for CI-581b relative to CI-581a is 2.72 choices, with a 95% confidence interval from 1.694 to 3.746. This interval communicates uncertainty around the estimated contrast on the original outcome scale. The interval lies on the positive side of zero, consistent with the higher reported mean for CI-581b. The unadjusted p-value is <0.001.

Precision is different from the size of the observed contrast. The mean difference describes the estimated separation between condition means; the confidence interval describes how uncertain that estimate is under the analysis assumptions. The interval has limits that are separated from the point estimate on each side, so the exact magnitude remains uncertain even though the direction is supported by the interval. The confidence interval does not show the range of individual participant responses.

The condition-level standard deviations, 1.3 for CI-581a and 1.8 for CI-581b, describe variation among observed values. They should not be confused with the confidence interval for the mean difference. In a crossover trial, within-participant association can also affect the precision of a comparison. The registry does not report the within-participant correlation, the detailed model, or the method used to calculate the interval, so the contribution of those features cannot be assessed from the record.

Interpretation should remain tied to the measured task and its assessment time: 24 hours post-infusion, with choices between cocaine 25 mg now and $11 later. The confidence interval concerns the posted contrast for this endpoint, not other outcomes or longer-term responses. Additional detail about individual data and the analysis procedure is not reported in the registry record.

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 estimates are participants per arm, not total events. They use the observed effect as a starting point and show how the illustrative participant count changes when the effect to detect is set to 100%, 75%, or 50% of that observed effect. For 100% of the observed effect, the estimates are 6 participants per arm for 80% power and 8 participants per arm for 90% power. For 75% of the observed effect, they are 10 and 13 participants per arm, respectively. For 50% of the observed effect, they are 21 and 29 participants per arm, respectively.

In plain language, the 80% power column gives an illustrative per-arm sample size for a design intended to have an 80% chance of detecting the stated effect under the calculation assumptions, if that effect is real and the assumptions hold. The 90% power column gives the corresponding illustration for a higher chance of detecting it. These are planning calculations, not observed power for this completed trial.

The assumptions listed are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. Smaller target effects require more participants in these illustrations. The registry’s own original sample-size justification and design assumptions are not reported in the registry record. Because the trial is described as crossover, the posted arm-based planning illustration should not be taken as evidence that the original trial was planned using this same model or these same assumptions.

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 = 2.720; pooled standard deviation σ = 1.570.
2. n = 2 × (1.96 + 0.842)² × 1.570² / 2.720² = 5.2, rounded up to 6 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 <- 2.72                      # observed difference in means
sigma <- 1.57003184681076                     # 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(2596022)
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 = 6 (80% power), 8 (90% power)
 75% of observed effect: n per arm = 10 (80% power), 13 (90% power)
 50% of observed effect: n per arm = 21 (80% power), 29 (90% power)
power.t.test: 6.4 
Simulated power at n = 6 per arm: 0.772 (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 one primary outcome measure and no posted secondary outcome measures. It lists no analysis entries. The comparison reports an unadjusted p-value of <0.001. Whether other endpoints or statistical tests were considered, whether a multiplicity strategy was specified, and whether the posted result was selected from a larger set of analyses are not reported in the registry record. The absence of posted secondary outcomes does not establish that no other analyses were planned or performed.

Missing-data handling is also not reported in the registry record. The posted condition summaries each list 18 observations, while the enrollment is 18. These counts alone do not describe whether all participants contributed complete measurements to both conditions, whether any values were excluded, or how missing observations were treated. The method for determining the analysis population, including any distinction between randomized participants and those included in the outcome analysis, is not reported.

For a crossover trial, the treatment sequence, period, and possible carryover can affect the comparison. The record gives the two sequences and says the second drug followed 2 weeks after the first, but it does not describe whether period or carryover effects were assessed or adjusted for. It also does not identify the statistical model used for the repeated observations. These details matter because measurements from the same participant are generally related, and analysis should reflect the design.

The record does not report a prespecified statistical analysis plan, covariates, sensitivity analyses, interim analyses, or procedures for handling protocol deviations. These omissions limit assessment of whether the reported estimate and unadjusted p-value reflect the planned primary analysis, and how robust the result may be to alternative reasonable assumptions. No conclusions about those procedures can be drawn from the posted summary alone.

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.

The serious adverse event figures are reported by treatment sequence. In the CI-581a-first, followed by CI-581b group, 10 participants were at risk and 0 were affected by a serious adverse event; deaths were 0. In the CI-581b-first, followed by CI-581a group, 8 participants were at risk and 0 were affected by a serious adverse event; deaths were 0.

The record also reports 0 affected by other adverse events among 10 at risk in the first sequence group, and 0 affected among 8 at risk in the second sequence group. These figures are counts in the registry’s safety reporting categories. They should be read with their stated at-risk group sizes and sequence labels, rather than as evidence that all possible harms were ruled out.

The registry does not list serious adverse event terms. It also does not provide further detail about event definitions, collection procedures, duration of safety follow-up, or whether safety reporting was analyzed by treatment period as well as by sequence. The reported zeros therefore describe the posted safety counts; they do not establish that no unrecorded or unreported events occurred.

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

The posted results show that the mean number of cocaine choices in the laboratory task was higher after CI-581b than after CI-581a: 4.33 compared with 1.61. The reported mean difference was 2.72 choices, with a 95% confidence interval from 1.694 to 3.746 and an unadjusted p-value of <0.001. Those figures describe the recorded endpoint at 24 hours post-infusion.

The result does not, by itself, show what participants would do outside the task. The choices were made between cocaine 25 mg now and $11 later, across 5 opportunities spaced 15 minutes apart. A response to that specific task is not interchangeable with longer-term cocaine use, craving, functioning, or other clinical outcomes. No secondary outcome measures are posted, so the record does not provide results for those broader outcomes.

The difference also does not establish that either study drug improves health or produces a clinical benefit. The comparison is between the two study conditions in a randomized, quadruple-masked crossover trial, with CI-581b described as an active control. The registry’s posted data do not provide a clinical interpretation or an assessment of the practical importance of the difference. The effect’s relevance beyond this laboratory endpoint is therefore not established by these results.

Precision and design information constrain interpretation. The confidence interval conveys uncertainty around the mean contrast, but the detailed analysis method, missing-data handling, analysis population, and treatment of crossover period or carryover effects are not reported in the registry record. The enrolled sample was 18, and participant characteristics beyond the listed condition are not reported. These limitations leave uncertainty about how broadly the task result applies and how fully the posted estimate accounts for the design.

The most defensible reading is narrow: the registry reports different average task choices under the two conditions, with the numerical contrast and uncertainty stated above. It does not settle questions about behavior over time, other populations, mechanisms, or clinical outcomes.

16. Why This Trial Matters Statistically

This trial illustrates why the endpoint and study design need to be read together. The outcome is not a general measure of substance use; it counts choices in a structured task contrasting immediate cocaine with delayed money. The task gives a clearly specified behavioral outcome, but its interpretation should remain limited to that setting and assessment time.

The crossover design makes within-participant information potentially important. Each sequence included both CI-581a and CI-581b, so the comparison may be influenced by how repeated measurements and treatment order were handled. A crossover study can reduce the influence of stable differences between participants when its analysis uses the paired structure appropriately. It also requires attention to period and carryover effects. The registry does not report the model or those assessments, making the distinction between design and analysis especially relevant.

The posted confidence interval and p-value answer different statistical questions. The interval describes uncertainty in the mean contrast on the choice scale, while the p-value reports compatibility with a null hypothesis under test assumptions. Neither establishes the clinical meaning of the contrast or its relevance beyond the task. The interval is useful precisely because it shows that a single point estimate does not capture all uncertainty.

The planning illustrations also demonstrate how sample-size targets depend on the effect selected for detection: smaller effects require more participants under the listed assumptions. Those calculations are not the original design justification and are not observed power. Keeping planning calculations separate from the trial’s realized results avoids treating retrospective illustrations as evidence of the study’s design or success.

Finally, registry reporting completeness affects what can be evaluated. The numerical summary permits interpretation of the direction, magnitude, and stated precision of the primary outcome, but does not explain the analysis population, missing data, multiplicity, or crossover model. Those distinctions are central to statistical interpretation even when an estimate and a small p-value are available.

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 enrolled sample was 18, and the registry provides no further participant characteristics beyond cocaine dependence. This limits assessment of how representative the participants were or how the result might apply to other groups. The posted condition-level counts are 18 for each drug condition, but the record does not describe completion by participant, exclusions, or whether every participant contributed to both conditions.

The endpoint is a task-specific choice count, based on 5 opportunities between cocaine 25 mg now and $11 later. It was measured 24 hours post-infusion. The record does not report longer-term outcomes or other behavioral and clinical measures, and no secondary outcomes are posted. The result therefore has a narrow measurement scope.

Although the trial is identified as randomized, quadruple-masked, and crossover, the registry does not specify the masked roles or the statistical model. It does not report whether period effects, carryover, within-participant dependence, or missing observations were addressed. The interval and p-value are reported, but the analytic procedures that produced them are not described in the record.

The safety summary reports no affected participants in the listed serious or other adverse event categories and no deaths in the sequence groups. Further details about event ascertainment, timing, definitions, and follow-up are not reported in the registry record. These limitations mean the posted safety counts should not be treated as a comprehensive account of all possible harms.

The registry reports no analysis entries, no posted secondary outcomes, and no original sample-size justification. The prespecified multiplicity plan, any interim analyses, and the handling of protocol deviations are not reported. These gaps restrict evaluation of analysis choices and of the result’s robustness.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: The Effect of Glutamatergic Modulation on Cocaine Self-administration — NCT02596022.
• Registry record last updated 2018-06-15; status last verified 2018-06.
• Results first posted 2017-10-17.
• 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.

21. Record Summary

The registry record describes a completed, randomized, quadruple-masked, Phase 2 crossover trial of CI-581a and CI-581b in participants with cocaine dependence. It records 18 participants enrolled, with a start of 2013-06 and primary completion in 2015-06. The sponsor was New York State Psychiatric Institute.

The primary outcome was the mean number of cocaine choices out of 5, measured 24 hours post-infusion. The reported means were 1.61 for CI-581a and 4.33 for CI-581b. The mean difference for CI-581b relative to CI-581a was 2.72, with a 95% confidence interval from 1.694 to 3.746 and an unadjusted p-value of <0.001.

The record lists no posted secondary outcomes and no analysis entries. It reports 0 serious adverse events, 0 other adverse events, and 0 deaths in each of the two sequence groups, with 10 and 8 participants at risk, respectively. Details of the statistical model, missing-data procedures, analysis populations, and original sample-size assumptions are not reported in the registry record.