A Study of the Relationship Between Disulfiram and Cocaine Self-administration.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of the Relationship Between Disulfiram and Cocaine Self-administration. (NCT00729300), a randomized phase 1, phase 2 study. The primary outcome is Cocaine Self-administration.

2. Clinical Question

The registry’s stated rationale refers to prior suggestions that disulfiram might be beneficial for cocaine dependence and to an assumption about disulfiram, dopamine availability, and the aversive aspects of stimulants. Those statements describe the motivation for the research; they are not results of this trial. The primary posted result concerns choices in a laboratory procedure, rather than a clinical measure of treatment for cocaine dependence. The registry record does not report a clinical outcome of abstinence, longer-term use, or treatment response.

3. Trial Design

The registry identifies this as a randomized, triple-masked crossover study, with Phase 1 and Phase 2 listed. The interventions were placebo and disulfiram. In general, a crossover design evaluates more than one condition within a participant, while randomization assigns an order or sequence. The registry record does not report the sequence, the number or timing of study periods, or whether a washout interval was used. It also does not describe who was masked.

Enrollment was 52. The study started in 2006-04, reached primary completion in 2016-06, and is listed as completed. The primary outcome analysis population was study completers; the outcome description explicitly says that only data from completers were considered. The registry record does not report why participants did not complete the study, how many were excluded from the primary outcome, or whether completion was defined in a particular way.

The primary endpoint was continuous: the mean number of cocaine choices per participant under each drug condition. Results are posted for placebo and a disulfiram treatment group. Each condition’s reported outcome includes 17 participants. The registry record does not say whether the displayed summaries represent every participant who completed both crossover conditions, or explain how the crossover structure was handled in the statistical analysis.

The record identifies Baylor College of Medicine as sponsor. It does not report the original design assumptions, including the planned sample size rationale, or a detailed statistical analysis plan. The current registry status is completed, but the record does not give details about study conduct beyond the listed design, enrollment, dates, interventions, and outcome information.

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

Analysis population for the primary outcome, as defined in the registry: Only data from study completers was considered

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: Only data from study completers was considered

6. Primary Endpoint

The registry lists the primary outcome as Cocaine Self-administration, measured over 4 days.

The primary endpoint is titled “Cocaine Self-administration.” It measures the mean number of cocaine choices per participant during 4 days under each drug condition. In the laboratory procedure, participants chose between previously sampled cocaine doses and cash of increasing monetary value. The registry reports the mean number of choices for placebo and disulfiram.

This endpoint quantifies choice behavior within the specific laboratory procedure. It can therefore be used to compare the average number of cocaine choices recorded under the two conditions. A continuous measure such as this retains information about the magnitude of participants’ recorded choices, rather than reducing the outcome to a yes-or-no category. In general, comparing condition means is one way to describe how the average value of a continuous outcome differs across conditions.

The measure is not itself a direct assessment of long-term cocaine use, abstinence, or clinical recovery. The registry record does not report those outcomes as primary measures. Nor does the outcome description report separate results by genotype, despite the study summary’s stated interest in C/C DBH genotype and plasma activity. The posted endpoint should therefore be interpreted narrowly: it describes cocaine choices in a human laboratory model over the stated timeframe, among study completers.

7. Results Posted in the Registry

• Disulfiram Treatment Group: mean 4.80 versus 2.50 on control; difference 2.30 (95% CI 0.45 to 4.15), standardized effect d = 0.84; z-test p = 0.015.

8. Statistical Interpretation

9. Statistical Methods Explained

The endpoint is a continuous measure: a mean number of cocaine choices per participant. A mean describes the average outcome in a condition. The posted values are 2.5 for placebo and 4.8 for disulfiram. The registry also reports a standard error for each mean: 0.5 for placebo and 0.8 for disulfiram. A standard error describes the estimated precision of a mean, rather than the spread of individual participants’ outcomes. It generally becomes smaller when a mean is estimated more precisely. It should not be read as a range containing most participants’ values.

What does the mean difference describe?

The posted comparison reports a mean difference of 2.3, defined in the comparison as disulfiram relative to placebo. This is the difference between the two reported condition means as presented in the registry. The comparison is unadjusted. No covariate-adjusted estimate or model description is reported in the registry record.

In general, a difference in means expresses how far apart the average outcomes are on the original measurement scale. Here, that scale is cocaine choices per participant. An unadjusted difference is a descriptive comparison that does not account for other variables in a statistical model. Because the registry classifies the design as crossover, within-participant dependence may be relevant to an appropriate analysis. The record does not state whether the posted comparison uses paired data or otherwise accounts for the crossover structure. It also does not report period, sequence, or carryover analyses.

What does the confidence interval add?

The interval for the mean difference runs from 0.451 to 4.149. In general, a confidence interval gives a range of estimates compatible with the observed data and the assumptions of the procedure used. It communicates precision as well as direction and magnitude. Its width reflects uncertainty: a narrower interval is more precise than a wider one on the same scale. The interval is not a range of individual responses, and it does not mean that there is a particular probability that the fixed true difference lies inside these specific endpoints.

The registry record does not identify the method used to calculate this interval. In particular, it does not explain whether the calculation incorporated the crossover pairing. The interval should therefore be read as the interval attached to the posted comparison, without assuming a method not described in the record.

How should the p-value be read?

The posted unadjusted p-value is 0.015. In general, a p-value measures the compatibility of the observed data with a specified null hypothesis under a statistical model. It does not give the probability that the null hypothesis is true, nor does it quantify the practical importance of an estimate. “Unadjusted” means the value is reported without a stated correction for multiple comparisons. The registry record does not report a multiplicity plan or a more detailed statistical analysis plan.

The registry reports a pooled standard deviation of 2.75 for the comparison. A standard deviation describes variability among observations, unlike a standard error, which describes precision of an estimated mean. The comparison also gives a standardized effect size, but that value is not needed to interpret the mean difference in its original outcome units. Standardized measures can help express a difference relative to variability; they do not replace the original-scale result or its uncertainty.

Why does over-dispersion matter in general?

For count outcomes, variability can exceed what a simple count model would expect under its assumptions; this is often called over-dispersion. General statistical practice checks whether the model’s variability assumptions fit the data, because underestimated variability can make standard errors too small and intervals too narrow. The registry record does not report an over-dispersion assessment or a count-model analysis. The posted endpoint is summarized as a mean with standard errors, and the registry’s reported comparison is a mean difference. No count distribution or model should be inferred from those summaries alone.

What does the crossover label establish?

The registry classification establishes that the trial is recorded as crossover, but does not supply the operational details needed to reconstruct the analysis. In general, crossover comparisons can benefit from comparisons within the same participant, because stable characteristics are shared across conditions. Such analyses must also account for the design’s periods and sequence where relevant. The record does not report the treatment sequence, period structure, washout, paired estimates, or a model for participant-level correlation. The displayed condition summaries and unadjusted comparison cannot answer those questions by themselves.

10. Confidence Intervals and Precision

The posted mean difference is 2.3 cocaine choices per participant, with a confidence interval from 0.451 to 4.149. The interval expresses uncertainty in the estimated comparison on the endpoint’s original scale. Its lower and upper limits are both above zero, while the interval also includes values that differ substantially in magnitude. It therefore communicates an estimated direction alongside uncertainty about the size of the difference.

A confidence interval is not a prediction interval for individual participants. It does not show how much participants’ choices varied, nor does it guarantee that a future study’s estimate will fall within these limits. The standard errors reported for the condition means, 0.5 for placebo and 0.8 for disulfiram, describe precision of those means rather than individual variability. The pooled standard deviation reported for the comparison is 2.75 and describes variability on a different statistical scale.

The registry record does not report the method used for the interval or whether it accounts for paired observations in this crossover design. This matters because the precision of a crossover comparison can depend on the within-participant relationship between condition outcomes. Without a reported analysis method, the interval’s treatment of that relationship cannot be assessed from the record.

The unadjusted p-value is 0.015, but a p-value and an interval answer different questions. The p-value summarizes compatibility with a null model under test assumptions; the interval presents a range of estimated differences consistent with the procedure. Neither alone establishes clinical importance. The laboratory outcome’s meaning is bounded by the procedure and the completer analysis population, and the registry record does not report a clinical threshold for interpreting the magnitude.

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 participants per arm, not total events. They are illustrations based on the observed effect, not the trial’s original design assumptions. The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each planning scenario targets a stated fraction of the observed effect. The registry record does not report the trial’s own original sample-size assumptions or justification.

The 80% power column gives the participants per arm indicated by the planning calculation for an 80% chance of detecting the specified effect under those assumptions. The 90% power column gives the corresponding participants per arm for a 90% chance. These are conditional planning quantities: they depend on the effect size chosen and on the stated assumptions. They are not observed power.

For 100% of the observed effect, the figures are 23 participants per arm for 80% power and 31 for 90% power. For 75% of the observed effect, they are 40 per arm and 54 per arm. For 50% of the observed effect, they are 90 per arm and 121 per arm. Smaller effects require more participants to detect under the same planning assumptions, which is why the listed requirements increase as the targeted fraction decreases.

The accompanying planning curve represents the same type of relationship between the target effect and required sample size; it should be interpreted as a planning illustration, not as a record of observed enrollment or a reconstruction of the trial’s design process. The registry classifies the study as crossover, whereas the stated illustration assumes equal allocation between two arms. The record does not explain how this planning framework maps to the trial’s crossover implementation.

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 = 2.300; pooled standard deviation σ = 2.750.
2. n = 2 × (1.96 + 0.842)² × 2.750² / 2.300² = 22.4, rounded up to 23 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 <- 2.3                      # observed difference in means
sigma <- 2.75045450789501                     # 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(729300)
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 = 23 (80% power), 31 (90% power)
 75% of observed effect: n per arm = 40 (80% power), 54 (90% power)
 50% of observed effect: n per arm = 90 (80% power), 121 (90% power)
power.t.test: 23.4 
Simulated power at n = 23 per arm: 0.794 (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 posts one primary outcome measure, Cocaine Self-administration, and reports no posted secondary outcomes. It lists no analyses. The comparison’s p-value is explicitly unadjusted. The registry record does not report whether multiple comparisons were planned, whether any multiplicity adjustment was considered, or how outcome selection and analysis were governed by a prespecified statistical plan. The absence of posted secondary results does not establish that no other outcomes or analyses were considered during the study.

The primary outcome description states that only study completers were considered. The registry record does not report the number of enrolled participants with missing primary outcome data, the reasons for noncompletion, or how missing observations were handled. It also does not say whether completion required outcome data under both conditions. These omissions prevent assessment of whether missing data could have influenced the reported condition means or comparison.

The analysis population is therefore described as completers, but its exact definition and relationship to randomization are not reported. The record does not provide an intention-to-treat analysis, a sensitivity analysis for missing data, or a comparison of completers with noncompleters. No conclusion about the effect of missingness can be drawn from the outcome summaries alone.

Although the registry identifies a randomized crossover design, the statistical analysis plan, sequence handling, period effects, and within-participant correlation method are not reported in the registry record. Interim analyses and stopping rules are also not reported in the registry record. These details matter for interpreting precision and the role of the unadjusted p-value, but should not be inferred from the design label or posted result.

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 adverse-event record reports serious adverse events affecting 0 participants in the placebo group and 0 in the disulfiram group. Other adverse events affected 0 participants in each group, and deaths were 0 in each group. For the serious and other adverse-event summaries, 26 participants are listed as at risk in each group. No serious adverse-event terms are listed.

These figures describe the events recorded in the registry for the stated groups. They do not establish that no adverse events of any kind occurred outside the categories or reporting covered by these entries. The registry record does not provide event definitions, severity grading beyond the serious-event category, timing, exposure duration, or additional safety analyses. The at-risk counts in the safety record should not be treated as the primary endpoint analysis counts; the posted primary outcome gives 17 participants for each condition.

Zero recorded events in these summaries do not prove that an event cannot occur or quantify risk beyond the observation recorded. They should be read as the registry’s event counts, with the limits of the available reporting.

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

The posted result shows a higher mean number of cocaine choices in the disulfiram condition than in the placebo condition in this laboratory self-administration procedure: 4.8 compared with 2.5. The reported unadjusted mean difference is 2.3 choices per participant, with a confidence interval from 0.451 to 4.149 and an unadjusted p-value of 0.015. The result describes the completer analysis population and the stated 4-day endpoint timeframe.

That finding does not establish how disulfiram affects cocaine use outside the laboratory task. A choice between previously sampled cocaine doses and cash is a specific behavioral measure, not a direct measure of longer-term use, abstinence, or treatment success. The registry record does not report those clinical outcomes as results of this primary endpoint. It also does not post a genotype-specific outcome, despite the study summary’s stated interest in C/C DBH genotype and plasma activity.

The registry’s crossover classification is important to interpretation, but the analysis details needed to understand participant-level pairing are not reported. The posted summaries do not establish whether the comparison accounts for order, period, or within-participant correlation. The record also does not report missing-data handling or the details of the statistical analysis plan. These gaps limit independent assessment of the inferential result.

The figures do not show that the result is clinically important, explain its mechanism, or demonstrate an effect for every participant. They do not establish that the laboratory difference would persist in other settings or populations. They describe an observed comparison in a particular study population and procedure, with uncertainty represented by the posted interval. The result should be interpreted within those boundaries rather than as a general statement about treatment or clinical outcomes.

16. Why This Trial Matters Statistically

This study illustrates why the endpoint and design matter when interpreting a treatment comparison. The primary measure is a count-like behavioral outcome reported as a mean number of choices, and the trial is classified as crossover. For a crossover study, within-participant comparisons can be important because the same person may contribute outcomes under both conditions. Yet the registry record does not describe whether the reported comparison incorporates that pairing. A design label alone is not enough to determine the analysis used.

The reported condition means and their standard errors are not interchangeable. The means describe average choices, while standard errors describe precision of those averages. The comparison’s confidence interval gives a complementary view of uncertainty on the original outcome scale. Together, these statistics help distinguish the estimated difference from its precision, while the unadjusted p-value provides a separate measure of compatibility with a null model.

The record also demonstrates the importance of defining the analysis population. The primary result uses completers, while the registry does not report the missing-data counts or reasons for noncompletion. Without these details, readers cannot judge the possible influence of attrition. Nor can they tell how the crossover design was implemented or whether the analysis included appropriate design terms.

Finally, a laboratory self-administration measure answers a narrower question than a clinical treatment outcome. Keeping that distinction clear prevents an observed difference in experimental choices from being treated as evidence about longer-term clinical outcomes that are not reported. The registry’s missing analysis details and genotype-specific results further define what can and cannot be inferred.

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 most important design-specific limitation in interpreting this record is the lack of detail about how the crossover design was carried out and analyzed. The registry identifies the study as randomized, triple-masked, and crossover, but does not report condition sequence, study periods, washout, period effects, carryover assessment, or a paired analysis. The posted unadjusted comparison therefore cannot be independently linked to a particular crossover model.

The primary outcome population is limited to study completers. The registry does not report how many participants lacked outcome data, why they did not complete, whether completion required data from both conditions, or how missing outcomes were handled. The posted condition summaries each include 17 participants, but the record does not explain how those outcome counts relate to enrollment or the crossover analysis population.

The endpoint is a laboratory choice measure assessed over 4 days. It does not directly represent longer-term clinical cocaine use or treatment outcomes. The study summary’s interest in C/C DBH genotype and plasma activity is not accompanied by a posted genotype-specific result, so that aspect of the research question cannot be evaluated from the primary outcome data.

The registry record does not report a detailed statistical analysis plan, original sample-size rationale, multiplicity approach, or interim-analysis information. It also does not explain how the posted confidence interval was calculated. These gaps constrain evaluation of prespecification, model assumptions, and precision. The adverse-event entries report zero affected participants for the listed categories, but do not provide event definitions or further safety analyses.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: A Study of the Relationship Between Disulfiram and Cocaine Self-administration. — NCT00729300.
• Registry record last updated 2023-07-19; status last verified 2023-06.
• Results first posted 2023-07-19.
• 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 identifies NCT00729300 as “A Study of the Relationship Between Disulfiram and Cocaine Self-administration,” with the official title “Laboratory Models of Cocaine Self-administration.” Baylor College of Medicine is listed as sponsor. The condition is Cocaine Addiction, and the interventions are placebo and disulfiram.

The study is listed as randomized, triple-masked, and crossover, with Phase 1 and Phase 2. Enrollment was 52. The recorded start was 2006-04, primary completion was 2016-06, and status is completed.

The primary outcome is the mean number of cocaine choices per participant during 4 days, with the primary analysis described as including study completers. The posted means are 2.5 for placebo and 4.8 for disulfiram. The posted mean difference is 2.3, with a confidence interval from 0.451 to 4.149 and an unadjusted p-value of 0.015. No secondary outcome results or analyses are listed.