Dronabinol Treatment for Marijuana Addiction: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Dronabinol Treatment for Marijuana Addiction (NCT00217971), a randomized phase 2 study. The primary outcome is Proportion of Patients Abstinent From Marijuana During Weeks 7 and 8 of the Trial.

2. Clinical Question

The central statistical question is whether the observed difference in abstinence proportions is distinguishable from chance variation in this trial. The posted estimate is a risk difference, with dronabinol compared against placebo. Its confidence interval includes values on both sides of no difference, and the unadjusted p-value is 0.72. These results do not establish a clear difference between the arms for this endpoint.

The question is narrow: it concerns self-reported abstinence in the specified weeks, not all cannabis use throughout the trial or outcomes beyond the primary endpoint. The registry record does not report a primary-outcome analysis population beyond the endpoint description, and it does not post analyses for secondary outcomes. Those omissions limit how broadly the primary comparison can be interpreted.

3. Trial Design

The registry describes a randomized, parallel-group, placebo-controlled phase 2 study of dronabinol for marijuana addiction. It identifies the study as double-blind and triple-masked. The sponsor was New York State Psychiatric Institute. The trial status is completed; the recorded start was 2005-03 and primary completion was 2009-12.

The intervention arm received dronabinol, described as 20mg bid for a daily maximum dose of 40mg. The comparator arm received placebo. The registry lists enrollment of 156. For the primary endpoint, the arm records list 79 participants in the dronabinol group and 77 in the placebo group. These arm counts are reported as recorded; the registry does not explain how the outcome counts relate to any analysis population or missing observations.

The primary outcome was the proportion of patients abstinent from marijuana during weeks 7 and 8. The description says daily Timeline Followback self-report data were used to assess abstinence during that timeframe. The posted counts are 14 for dronabinol and 12 for placebo. The registry also posts a comparison as a risk difference and an unadjusted p-value.

The record does not report the trial’s original sample-size assumptions, the detailed randomization procedure, the operational details of masking, or the statistical analysis plan. It also does not post analyses for secondary outcomes. These omissions do not change the basic design classification, but they restrict assessment of the intended analysis and how the posted outcome analysis was conducted.

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: The trial included a one-week placebo lead-in phase. Individuals who were able to significantly decrease their marijuana use during the first week of the study were not randomized. Participants who reported marijuana use less than twice a week during the placebo lead-in phase were considered placebo responders.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Proportion of Patients Abstinent From Marijuana During Weeks 7 and 8 of the Trial, measured over weeks 7 and 8.

The primary endpoint is the proportion of patients abstinent from marijuana during weeks 7 and 8 of the trial. It is a binary outcome: for the specified period, a participant is classified as abstinent or not abstinent. The endpoint description identifies daily Timeline Followback self-report data as the source used to assess abstinence.

A binary endpoint gives a direct count of participants meeting a defined outcome, which can then be expressed as a proportion within each randomized arm. Comparing those proportions addresses whether the frequency of abstinence differed between the dronabinol and placebo groups during the selected weeks. The registry posts 14 participants with the outcome in the dronabinol arm and 12 in the placebo arm, alongside their respective arm counts.

The endpoint is limited to the stated weeks and to self-reported data. It should not be read as a measure of abstinence across the entire trial, as a measure of changes in use outside those weeks, or as an independently verified outcome. Whether participants with incomplete reports were treated as abstinent, not abstinent, or excluded is not reported in the registry record. The endpoint’s clinical interpretation also depends on the trial population and analysis procedures, which are not fully described in the posted record.

7. Results Posted in the Registry

• Dronabinol: 17.7% (14/79) versus 15.6% (12/77) on control; risk difference 2.1 percentage points (95% CI -9.5 to 13.8), risk ratio 1.14 (0.56–2.30); unadjusted two-proportion z-test p = 0.720.

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a binary primary outcome and a risk-difference comparison. The outcome is based on whether a participant was abstinent during weeks 7 and 8, using daily Timeline Followback self-report data. A risk difference compares the proportions meeting the endpoint in two groups by subtracting one group’s proportion from the other’s. In this comparison, the stated contrast is dronabinol relative to placebo.

What does a risk difference describe?

In general, a risk difference is an absolute contrast between two event proportions. A value of zero represents equal proportions. Positive and negative values indicate opposite directions of the comparison, depending on which arm is placed first. Here, the posted estimate is accompanied by a confidence interval that includes zero, so the data are compatible with no difference as well as with differences in either direction. The point estimate alone would not express the uncertainty around the comparison.

How should the p-value be read?

The registry reports an unadjusted p-value of 0.72. In general, a p-value measures how compatible the observed data, or data more extreme, are with a specified null model. A large p-value is not proof that the arms have the same outcome, and it does not measure the probability that the observed effect is due to chance. It should be considered alongside the estimate, confidence interval, outcome definition, and analysis details. The record does not report whether any adjusted or alternative hypothesis tests were conducted.

Why does the confidence interval matter?

In general, a confidence interval describes uncertainty in an estimated quantity under the assumptions of the analysis. Its width indicates how precisely the effect is estimated, while its position indicates which effect values are compatible with the data at the stated confidence level. The registry comparison’s interval crosses zero. That means the posted result does not isolate a direction of difference. The exact interval limits are not repeated in this explanation; the important interpretation is that the range includes both negative and positive risk differences.

What does binary-outcome analysis require?

For a binary endpoint, each participant contributes an outcome classification, such as abstinent or not abstinent, according to the endpoint definition and analysis rules. The arm’s proportion depends on both the number meeting the endpoint and the denominator used. The registry posts counts and arm sizes, but does not state how incomplete or missing self-reports were handled, whether all randomized participants were included, or whether the denominator changed for the primary analysis. Those details matter because different missing-data rules can affect the estimated proportions.

What does randomization contribute?

In general, random assignment is intended to make comparison groups similar on average with respect to measured and unmeasured baseline characteristics. It supports a causal interpretation when the trial is conducted and analyzed according to appropriate procedures, but it cannot by itself address missing outcome data, deviations from assigned treatment, or selective reporting. The registry identifies the study as randomized and parallel-group. The exact randomization sequence, allocation concealment procedures, and analysis population are not reported in the registry record.

What can be assessed from the posted analysis?

The record reports one primary outcome measure, an arm comparison, and no posted analyses. It does not report the detailed statistical plan, prespecified covariates, interim analyses, or procedures for multiplicity. Consequently, the reader can interpret the posted risk difference, interval, and unadjusted p-value, but cannot verify from the registry record whether these match all prespecified analyses. The record also does not identify an adjusted effect estimate or a secondary analysis that could provide another view of the endpoint.

General statistical interpretation also distinguishes statistical uncertainty from clinical importance. A confidence interval can include zero while still allowing effects that might matter in context; deciding the importance of any such effect requires a relevant clinical threshold, which is not reported in the registry record. Conversely, a p-value alone cannot establish that an effect is important. The posted results should therefore be read as an uncertain estimate for the defined endpoint, not as a definitive statement about all outcomes or treatment effects.

10. Confidence Intervals and Precision

The precision of the primary comparison is limited by the uncertainty represented in its confidence interval. The interval includes zero and extends to both negative and positive risk differences. Thus, the results are compatible with no difference and with effects in either direction. The posted result does not pin down a single direction or a narrow range of possible effects.

In general, a confidence interval is a range of estimates compatible with the data under the analysis assumptions. A narrower interval indicates greater precision; a wider interval indicates more uncertainty. It does not guarantee that the true effect lies in the range, nor does it correct for limitations in outcome measurement, missing-data handling, or analysis choices. The exact confidence level and interval endpoints are recorded with the posted comparison, but the interpretation focuses on the interval crossing zero.

The arm counts are 14 and 12, with arm sizes of 79 and 77. These counts provide the observed data context, but they do not make the effect estimate precise by themselves. Precision depends on the comparison and its uncertainty, not simply on whether the observed counts look similar. The unadjusted p-value of 0.72 is consistent with substantial uncertainty about the direction and magnitude of the contrast.

The registry does not report the original sample-size justification or a prespecified precision target. The observed result therefore cannot be judged against a documented design goal for confidence-interval width. Nor do the posted data establish equivalence: that would require a prespecified equivalence margin and a suitable analysis, neither of which is 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 figures are participants per arm, not total events. They are illustrations calculated from the observed effect, rather than the trial’s original enrollment plan. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect.

For planning for 100% of the observed effect, the figures are 4,770 participants per arm for 80% power and 6,385 participants per arm for 90% power. For planning for 75% of the observed effect, they are 8,371 participants per arm for 80% power and 11,206 participants per arm for 90% power. For planning for 50% of the observed effect, they are 18,588 participants per arm for 80% power and 24,883 participants per arm for 90% power.

In general, power is the chance that a planned statistical test will reject its null hypothesis if a specified effect is present, under the assumptions used for the calculation. The 80% and 90% columns describe different planned chances of detecting the selected effect, not the probability that the trial’s conclusion is correct. A higher target power generally requires a larger planned study, all else equal. Smaller effects also require more participants to detect with the same power, which is reflected by the larger figures for smaller fractions of the observed effect.

These values are planning illustrations, not observed power. The registry record does not report the trial’s original design assumptions or sample-size justification, so the displayed calculations should not be read as evidence about why the original enrollment was selected.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.156; treatment proportion p₁ = 0.177; average p̄ = 0.167.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5269 = 1.0327.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5267 = 0.4432.
4. n = (1.0327 + 0.4432)² / (0.0214)² = 4,769.4, rounded up to 4,770 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
p0 <- 0.155844155844156                          # control proportion
p1 <- 0.177215189873418                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(217971)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 4,770 (80% power), 6,385 (90% power)
 75% of observed effect: n per arm = 8,371 (80% power), 11,206 (90% power)
 50% of observed effect: n per arm = 18,588 (80% power), 24,883 (90% power)
power.prop.test: 4769.4 
Simulated power at n = 4,770 per arm: 0.805 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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 and no secondary outcome analyses. It does not report a multiplicity procedure, such as an adjustment for testing several outcomes or contrasts. Because only one analysis is posted, the record does not establish whether additional planned comparisons were conducted but not reported, or whether any multiplicity strategy was specified in advance. The statistical plan is not reported in the registry record.

In general, multiplicity arises when several hypotheses are tested. Repeated testing can increase the chance of at least one apparently notable result even when none of the tested effects is real. Prespecified approaches can address that issue, but the registry record does not identify such an approach for this trial. The posted p-value is described as unadjusted.

Missing-data handling is also not reported. The endpoint is based on daily self-report, but the record does not explain what happened when a report was absent, incomplete, or inconsistent, or whether those observations were included in the analysis. It does not state whether participants were analyzed according to randomized assignment, according to treatment received, or under another analysis-population definition. No analysis population is named for the primary endpoint.

These omissions matter because the denominator and treatment of missing reports can affect a binary endpoint’s estimated proportion. They also make it harder to determine whether the posted comparison follows an intention-to-treat approach or another prespecified strategy. The registry provides the arm counts and the comparison, but not enough detail to reconstruct the analysis or assess the sensitivity of the result to alternative missing-data assumptions.

The registry record does not report interim analyses, stopping rules, or a detailed protocol for handling deviations from assigned treatment. No inference should be made about whether such procedures were used. The posted results can be described as recorded, while the absence of these details limits independent assessment of the full analysis process.

14. Safety Results

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

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: asthma (1.3%), diabetes (1.3%), concussion (1.3%), stomach flu (1.3%).

The registry posts serious adverse-event counts of 3 in the dronabinol group and 1 in the placebo group. The corresponding groups at risk are 79 and 77. It also lists other adverse-event counts of 8 for dronabinol and 10 for placebo, with those same arm sizes at risk. These are counts, not rates; no percentages should be inferred from them.

The listed serious adverse-event terms are asthma, diabetes, concussion, and stomach flu, each recorded as 1.3%. The registry does not provide enough detail in this record to connect each listed term to a specific participant or to assess timing, severity beyond the serious classification, or causal attribution. The death counts are not reported in the registry record.

In general, adverse-event counts require context about the collection period, definitions, completeness of reporting, and whether events were considered related to study treatment. Those details are not reported in the registry record. The counts describe the posted safety data but do not, on their own, establish comparative risk or causation.

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

The posted primary result describes self-reported abstinence during weeks 7 and 8. The endpoint counts were 14 in the dronabinol arm and 12 in the placebo arm. The registry’s risk-difference comparison has a confidence interval that includes zero, and the unadjusted p-value is 0.72. Taken together, the posted analysis does not establish a clear difference between the arms for this endpoint.

That interpretation is not the same as showing the treatments are equivalent. A non-significant comparison can reflect uncertainty, and the interval includes values indicating effects in either direction. An equivalence conclusion would require a predefined margin for what counts as sufficiently similar and an analysis designed to test that question. Such a margin and analysis are not reported in the registry record.

The result also does not answer whether outcomes differed at other times or for other measures. The primary endpoint concerns only weeks 7 and 8, and the registry posts no secondary outcome analyses. It is based on daily self-report, rather than a measurement method described as independent verification. The record does not report how missing or incomplete self-reports were treated.

The trial was randomized, parallel-group, and placebo-controlled, which defines the intended comparison. Still, details needed to assess the analysis fully—including the primary analysis population, missing-data rules, and original statistical plan—are not reported in the registry record. The result should therefore be understood as the posted comparison for a particular endpoint and timeframe, with uncertainty and reporting limitations, rather than a general conclusion about all effects of dronabinol or placebo.

The safety counts are descriptive entries in the registry. They do not establish that a particular adverse event was caused by treatment, and the record does not supply the details needed for a broader safety comparison. Statistical interpretation should remain separate from clinical decision-making; these findings alone are not a basis for medical advice.

16. Why This Trial Matters Statistically

This trial illustrates how a randomized comparison can produce an estimate that remains uncertain even when an endpoint and arm counts are reported. The primary question is framed as a difference in proportions, and the risk-difference interval includes zero. The result is not simply a contest between the two observed counts: the denominators, effect estimate, and uncertainty all matter.

It also demonstrates why a p-value should not be treated as a yes-or-no verdict. The unadjusted p-value of 0.72 does not show that the arms are identical, nor does it quantify the probability that the result occurred by chance. The confidence interval conveys a more useful description of the range of effect values compatible with the analysis, although the interval does not isolate a direction.

The outcome definition is important to interpretation. Abstinence was assessed using daily Timeline Followback self-report during weeks 7 and 8, so the posted comparison is about that particular measure and period. It is not a direct summary of all marijuana use during the trial, and no secondary outcome analyses are posted to broaden the statistical picture.

Finally, the registry record shows the value of transparent reporting. Details about missing data, analysis populations, multiplicity, and the original sample-size plan are not reported. Without them, readers can interpret the reported comparison but cannot fully evaluate whether it reflects every planned analysis or how robust it is to alternative analytic choices.

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 primary endpoint is limited to self-reported abstinence during weeks 7 and 8. Although the registry says daily Timeline Followback self-report data were used, it does not report procedures for checking those reports or how missing daily information affected classification. The outcome may therefore be difficult to interpret fully without the underlying assessment and analysis details.

The primary-outcome population is not further specified. The registry lists arm sizes, but it does not explain whether all randomized participants contributed to the analysis, whether any were excluded, or what denominator was used if outcome data were missing. The exact approach to nonadherence or deviations from assigned treatment is also not reported in the registry record.

The posted results include one outcome measure and no secondary analyses. The statistical plan, multiplicity procedures, interim analyses, and original sample-size justification are not reported. This limits assessment of whether the reported comparison was selected and conducted as prespecified, and whether other planned analyses exist outside the posted results.

The reported confidence interval includes zero and values in both directions, so the primary estimate is not precise enough to establish a clear direction of difference. The registry does not provide a stated threshold for clinical importance or an equivalence margin. The study’s phase 2 designation and recorded enrollment do not, by themselves, establish what magnitude of effect the trial was intended to detect.

Safety reporting also has limits in this record. It gives serious and other adverse-event counts and several serious-event terms, but does not report causal attribution, timing, or death counts. These constraints should remain part of any interpretation of the posted efficacy and safety information.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Dronabinol Treatment for Marijuana Addiction — NCT00217971.
• Registry record last updated 2019-04-24; status last verified 2019-04.
• Results first posted 2011-11-01.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

NCT00217971 is a completed, randomized, parallel-group, placebo-controlled phase 2 study sponsored by New York State Psychiatric Institute. It is described as double-blind and triple-masked. The recorded start was 2005-03, and primary completion was 2009-12. The registry lists enrollment of 156.

The intervention was dronabinol, described as 20mg bid for a daily maximum dose of 40mg; the comparator was placebo. The primary endpoint was the proportion abstinent from marijuana during weeks 7 and 8, assessed using daily Timeline Followback self-report. The posted counts were 14 in the dronabinol arm and 12 in the placebo arm.

The registry reports a risk-difference comparison with an unadjusted p-value of 0.72. Its confidence interval includes zero and extends in both directions, so the result does not establish a clear difference for this endpoint. One outcome measure is posted, with no secondary analyses. Missing-data procedures, the analysis population, and the original sample-size assumptions are not reported in the registry record.