Progesterone Treatment for Cocaine-dependent Women: A Pilot Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Progesterone Treatment for Cocaine-dependent Women: A Pilot Study (NCT00632099), a randomized phase 2, phase 3 study. The primary outcome is Cocaine Abstinence Based on Urine Toxicology Results.

2. Clinical Question

3. Trial Design

The registry identifies this as a randomized, triple-masked, parallel-group pilot study. Its phase is listed as both phase 2 and phase 3. The two arms were oral micronized progesterone and matched placebo. The active arm description specifies oral progesterone up to 400 mg/day. The registry records 21 participants enrolled: 10 assigned to matched placebo and 11 assigned to oral micronized progesterone.

The study focused on women with cocaine dependence. The brief description says participants were to reside inpatient for one week before randomization to ensure cocaine abstinence. The rationale given for this period was that time to cocaine relapse was one of the primary outcome measures. The posted primary outcome, however, is the number of participants abstinent by urine toxicology during the final three weeks of the trial, described as weeks 7–9.

The sponsor was New York State Psychiatric Institute. The registry lists the study as completed, with a start date of 2008-02 and primary completion in 2012-09. These dates describe the trial record and do not establish when results were analyzed or posted. The registry reports one outcome measure and one analysis. It does not report the trial’s original sample-size assumptions or a detailed account of the allocation procedure, medication adherence, treatment discontinuation, or analysis population. Those details are not reported in the registry record.

Because this was a small pilot with 21 enrolled participants, the results are best read as a limited estimate from the enrolled groups. The randomized design supports a between-arm comparison, but the small arm counts mean that each participant represents a substantial part of the observed count. The record’s trial phase labels and pilot description do not, on their own, establish the intended role of this study in a broader development program.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

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 Cocaine Abstinence Based on Urine Toxicology Results, measured over during last 3 weeks of the trial.

The primary endpoint is cocaine abstinence based on urine toxicology results during the last three weeks of the trial, specified as weeks 7–9. It is a binary participant-level outcome: each participant either met the stated abstinence criterion for that period or did not. The posted arm results therefore report counts of participants with the event, rather than a continuous measure such as the number of days abstinent.

Urine toxicology provides an objective test of cocaine exposure at the times samples are collected. A measure limited to the final three weeks answers a focused question about abstinence late in the trial. It does not describe the full course of use across the trial, nor does it directly measure the timing of a return to use. The study description separately identifies time to cocaine relapse as a primary outcome measure and explains the pre-randomization inpatient period in relation to that measure. The posted outcome analyzed is the late-trial abstinence count; a separate time-to-relapse result is not reported in the registry record.

The endpoint is relevant to the stated treatment question because it records whether participants were abstinent in a defined treatment period. Its meaning depends on the urine-testing schedule, rules for classifying missing specimens, and handling of participants who left treatment or lacked an assessment. Those operational details are not reported in the registry record. Consequently, the posted counts show the recorded group outcomes, while the record does not fully describe how every possible testing or follow-up circumstance was handled.

7. Results Posted in the Registry

• Oral Micronized Progesterone: 27.3% (3/11) versus 10.0% (1/10) on control; risk difference 17.3 percentage points (95% CI -15.0 to 49.5), risk ratio 2.73 (0.34–22.16); unadjusted two-proportion z-test p = 0.314.

8. Statistical Interpretation

9. Statistical Methods Explained

The posted endpoint is binary, and the arm results are counts of participants meeting the abstinence criterion. The statistical comparison is therefore about the proportion with the event in each group. The registry’s reported sponsor method is Fisher Exact. The recalculated comparison gives an unadjusted risk difference, confidence interval, and p-value from the posted summary counts. These methods answer related but not identical questions and should be read with their distinct purposes in mind.

What does a risk difference express?

A risk difference subtracts the event proportion in one group from the event proportion in the other. For this outcome, the event is being cocaine-abstinent during the specified final three-week period. The recalculated point estimate is positive, meaning the observed event proportion is higher in the progesterone arm than in the placebo arm. Unlike a ratio, a risk difference expresses the comparison on the absolute probability scale. It is tied to the outcome and time window used in the trial, rather than being a general measure of treatment effect across other settings.

The estimate is unadjusted: it compares the posted arm summaries without adjustment for participant characteristics or other variables. The registry does not report an adjusted model for this endpoint. In a small randomized trial, an unadjusted comparison is a direct summary of the observed group contrast, but the estimate can be affected substantially by the outcomes of a small number of participants. The posted counts make that limitation especially relevant.

What does Fisher Exact test?

Fisher Exact is a statistical test for association in a two-by-two table of categorical outcomes. Here, the two categories are abstinent or not abstinent, compared across the two treatment groups. The test evaluates how compatible the observed allocation of outcomes is with a null hypothesis of no association between assigned group and endpoint, under its statistical framework. It is often used for small samples because it calculates probabilities from the possible tables conditional on the observed margins, rather than relying on a large-sample approximation for the test.

The registry reports a Fisher Exact analysis with p < 0.05, but does not provide the associated effect estimate or confidence interval. The recalculated unadjusted p-value is 0.314. A p-value is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of a difference. It describes how unusual data at least as incompatible with the null as the observed data would be under the assumptions of the test. A small p-value does not by itself quantify precision; that is why effect estimates and confidence intervals are also important.

How should the confidence interval be read?

A confidence interval gives a range of values compatible with the observed data under the method’s assumptions. The recalculated interval for the risk difference extends across zero. Zero represents no difference in event proportions, so an interval that includes zero indicates that the recalculation does not rule out no difference at the interval’s stated confidence level. The interval also extends to values in the direction of a higher abstinence proportion in the progesterone group. The range is broad, which signals limited precision.

The registry does not state the confidence level for the recalculated interval in its outcome record. The interval should therefore be interpreted as the interval generated by the recalculation rather than attributed to a confidence level not stated in the registry record. More generally, confidence intervals reflect sampling uncertainty under the method’s assumptions; they do not account automatically for unreported missing-data rules, departures from the analysis plan, or limitations in how the endpoint was measured.

Why are risk ratios and odds ratios different?

A risk ratio compares the event proportions as a ratio, while an odds ratio compares the odds of the event. Both are relative measures, whereas the risk difference is an absolute measure. When outcomes are uncommon, these relative measures can still be highly uncertain if the sample and event counts are small. The recalculated relative estimates have confidence intervals that span the null value for a ratio, which is one, and cover a wide range. The registry does not provide a sponsor-reported estimate for these measures.

For any measure, the choice of scale changes the interpretation. A relative ratio does not state the absolute difference in event proportions, and an absolute risk difference does not state the relative change. Neither scale turns the trial’s small number of observed events into a precise estimate. The outcome definition, follow-up window, and analysis population also matter to interpretation, and the registry does not describe several of those analytic details.

Why does over-dispersion matter in general?

Over-dispersion is a general concern in count models when the observed variability in counts exceeds what the model assumes. It can make model-based standard errors too small if it is ignored. It is not a specific feature reported for this binary participant-count endpoint, and the registry does not describe a count model or an over-dispersion assessment. For a binary endpoint, the central issues in the posted analysis are the small number of participants and events, uncertainty in the group comparison, and the handling of missing or unobserved outcomes.

These methods explain different aspects of the result: the counts state what was observed, the risk difference states the absolute contrast, the confidence interval describes uncertainty in that contrast, and the p-value summarizes compatibility with a null hypothesis under a test. None substitutes for the others. Their interpretation is also limited by what the registry does not report about analysis populations, missing data, and the sponsor’s detailed statistical procedure.

10. Confidence Intervals and Precision

The recalculated risk difference has a confidence interval that extends from a negative value to a positive value. Because zero lies within that interval, the posted summary data do not exclude no difference between the groups. The interval also permits a range of effects in either direction, so the point estimate should not be read as a precise measure of the size of a treatment-group contrast.

The trial enrolled 21 participants, with 10 in the placebo arm and 11 in the progesterone arm. Only 1 participant in the placebo arm and 3 in the progesterone arm met the abstinence endpoint. With such small counts, changing the outcome classification of a single participant could materially affect the observed proportions and the estimated contrast. The confidence interval reflects substantial uncertainty in the comparison, not a guarantee that the true value falls within the interval in any particular trial.

The recalculated risk ratio and odds ratio also have broad confidence intervals spanning their respective null value of one. This is consistent with limited precision on both absolute and relative scales. The registry’s sponsor analysis does not include an effect estimate or interval, so its reported p-value alone cannot show the magnitude or precision of the sponsor’s estimated difference. The registry does not state a confidence level for the recalculated interval.

Precision concerns are separate from the question of whether the endpoint is clinically important. A narrow interval would indicate a more precise estimate, while a broad interval leaves more uncertainty about the range of plausible effects. Here, the available participant and event counts support only a limited estimate. The record does not report additional outcome data that would resolve that uncertainty.

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 event counts. They are illustrations based on the observed effect, not the trial’s original design assumptions. The calculations use 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 trial’s own original sample-size justification and design assumptions are not reported in the registry record.

For planning at 100% of the observed effect, the figures are 79 participants per arm for 80% power and 105 participants per arm for 90% power. For planning at 75% of the observed effect, they are 128 participants per arm for 80% power and 171 participants per arm for 90% power. For planning at 50% of the observed effect, they are 258 participants per arm for 80% power and 344 participants per arm for 90% power.

In plain language, the 80% power column gives the estimated number per group for a design with an 80% chance of detecting the specified effect under the stated assumptions, if that effect is the true one. The 90% power column gives the corresponding planning target for a higher chance of detection. These are design calculations, not a measure of the chance that the observed result is true. Smaller effects need more participants to detect, which is why the planning figures rise as the assumed fraction of the observed effect becomes smaller.

This is not observed power. The figures do not describe how much power the completed trial had after its outcomes were known, and they do not validate the observed effect as the effect to expect in another study. Because the observed estimate comes from a small pilot, planning directly from it can be uncertain. The figures are best read as conditional illustrations under the listed assumptions, not as a definitive sample-size recommendation.

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 = 0.100; treatment proportion p₁ = 0.273; average p̄ = 0.186.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5507 = 1.0793.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5370 = 0.4519.
4. n = (1.0793 + 0.4519)² / (0.1727)² = 78.6, rounded up to 79 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 <- 0.1                          # control proportion
p1 <- 0.272727272727273                          # 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(632099)
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 = 79 (80% power), 105 (90% power)
 75% of observed effect: n per arm = 128 (80% power), 171 (90% power)
 50% of observed effect: n per arm = 258 (80% power), 344 (90% power)
power.prop.test: 78.6 
Simulated power at n = 79 per arm: 0.821 (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. Analyses Reported by the Sponsor

The registry reports that the sponsor’s analysis compared the two groups using Fisher Exact and gives p < 0.05. It does not report an effect estimate, confidence interval, test statistic, or additional details for that analysis. The result therefore communicates a thresholded significance finding but does not show the size or precision of the sponsor’s estimated group difference.

The unadjusted recalculation from the posted arm counts gives a risk difference with an interval that crosses zero and a p-value of 0.314. This is not the same result as the sponsor’s reported p-value. The registry record does not explain the difference, so the two results should be presented as separate analyses rather than blended into a single conclusion. They may differ in analytic choices, but the registry does not identify a reason, and no explanation should be inferred.

The discrepancy flag in the registry data is false. The sponsor result remains the study’s reported analysis, while the recalculation is a separate unadjusted comparison based on the posted summary statistics. The recalculation does not replace the sponsor’s analysis. Because the registry lacks the sponsor’s effect estimate and detailed methods, readers cannot assess from the record alone which specific calculation or analysis population produced the reported p-value. A p-value by itself also cannot establish the magnitude or practical importance of the endpoint difference.

14. Multiplicity, Missing Data and Analysis Populations

The registry posts one outcome measure and one analysis for this results record. It does not report secondary outcomes. The brief study description mentions time to cocaine relapse as one of the primary outcome measures, while the posted results identify cocaine abstinence based on urine toxicology during the last three weeks of the trial. A separate posted result for time to relapse is not reported in the registry record.

Multiplicity arises when several outcomes, time points, subgroup comparisons, or statistical tests are examined. If many tests are conducted, the chance of at least one apparently significant finding can increase unless the analysis plan addresses that issue. The registry’s one posted outcome and one posted analysis do not establish that only one outcome or test was considered during the study. Any prespecified hierarchy, multiplicity adjustment, or broader testing plan is not reported in the registry record.

The registry does not describe missing-data handling for urine toxicology results. It does not state how missed tests were classified, whether participants with incomplete testing were excluded, or whether imputation or sensitivity analyses were used. These choices can affect a binary abstinence outcome, because the endpoint depends on how results across the final three weeks were combined into a participant-level classification. The available counts alone do not reveal whether every randomized participant had complete outcome data.

The analysis population is also not reported in the registry record. The results list arm sizes of 10 and 11, matching the 21 participants enrolled, but the record does not specify whether the analysis followed an intention-to-treat principle, included only participants who completed treatment, or used another population definition. The analysis plan, interim analyses, and stopping rules are not reported in the registry record. Without these details, the posted results can be described, but the full analytic context and any protection against selective reporting cannot be assessed from the registry record.

15. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: uveitis (9.1%).

The registry reports serious adverse events affecting 0 of 10 participants in the matched-placebo group and 1 of 11 participants in the oral micronized progesterone group. These are participant counts, not event rates per person-time. The registry lists Uveitis at 9.1% among the top serious adverse event terms, but does not provide additional context about the event in the results record. Death counts are not reported in the registry record.

The numbers are small, so the comparison is descriptive and does not precisely characterize the occurrence of serious adverse events. A count of zero in one arm does not establish that the underlying risk is zero, and one affected participant in the other arm does not establish a stable rate. The record does not give event onset, duration, severity details beyond the serious classification, relatedness to treatment, or outcome. It also does not provide enough information to determine whether the listed serious term accounts for the single affected participant or how events were classified across participants.

The registry separately records other adverse events affecting 8 of 10 placebo participants and 9 of 11 progesterone participants. Those figures are not serious-event counts and should not be conflated with the serious adverse event summary. The safety data are limited to the posted counts and term; fuller safety assessment and interpretation are not reported in the registry record.

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

The posted results show more participants meeting the urine-toxicology abstinence endpoint in the progesterone arm than in the matched-placebo arm during the final three weeks of the trial: 3 of 11 compared with 1 of 10. The recalculated point estimate is in the direction of a higher abstinence proportion in the progesterone group. However, the confidence interval for the recalculated risk difference crosses zero, and the unadjusted p-value is 0.314. The small number of participants and outcomes leaves substantial uncertainty about the size and direction of the underlying difference.

The registry separately reports a sponsor Fisher Exact result of p < 0.05. Because the record does not explain how this result relates to the unadjusted recalculation, both findings need to be reported distinctly. The sponsor’s p-value does not provide an effect size or interval in the registry, and the recalculation does not reproduce that reported threshold. The record does not provide enough detail to resolve the difference.

These data do not establish that progesterone caused a change in abstinence, nor do they show that any observed difference would be reproduced in a larger study. They also do not establish a conclusion about time to relapse, despite the study description’s reference to that outcome, because a separate time-to-relapse result is not reported in the registry record. The endpoint is restricted to urine-toxicology abstinence during weeks 7–9; it does not summarize the whole treatment period or longer-term outcomes.

The trial is a pilot with 21 enrolled participants. Its results offer a small-sample comparison rather than a precise estimate for a broader population. Interpretation is further limited by the lack of registry entries on missing urine samples, analysis population, adherence, and the sponsor analysis procedure. The posted data should therefore be understood as a record of observed outcomes and reported analyses, not as a definitive estimate of efficacy or a basis for medical decisions.

17. Why This Trial Matters Statistically

This trial illustrates why an observed direction and statistical certainty are different things. The progesterone arm had more participants recorded as abstinent than the placebo arm, but the recalculated interval for the absolute difference extends across zero. A positive point estimate describes the direction in these enrolled participants; it does not establish that the underlying difference is positive or precisely quantify its size.

It also demonstrates why effect estimates and uncertainty intervals add information beyond a p-value. The registry’s sponsor analysis reports Fisher Exact p < 0.05, whereas the unadjusted recalculation gives p = 0.314 and an interval crossing zero. The record does not reconcile these analyses. A p-value cannot show the magnitude of a group contrast, and a threshold statement without an estimate and interval is incomplete for assessing precision. Keeping the reported sponsor result distinct from the recalculation preserves the analysis history without implying an explanation that the record does not give.

The small event counts make the analysis especially sensitive to individual outcomes. The participant counts and wide uncertainty help explain why pilot results are useful for understanding the data structure and generating planning illustrations, but are not a substitute for a more precise estimate. The sample-size illustrations also show the general statistical relationship that smaller target effects require more participants, while depending on assumptions that may not hold in another study.

Finally, endpoint and analysis details matter. A final-period urine-toxicology endpoint is not interchangeable with time to relapse, and the registry does not report how missing tests or analysis populations were handled. Statistical interpretation requires attention to what was measured, who was analyzed, how the test was conducted, and how much uncertainty remains—not only whether a reported p-value crosses a conventional threshold.

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 principal limitation is the small enrollment: 21 participants divided between two arms, with only 1 and 3 participants meeting the posted abstinence endpoint. Estimates from sparse counts are imprecise and sensitive to the outcome status of individual participants. The trial is described as a pilot, and its results should not be treated as a stable estimate of the effect expected in a larger population.

The registry’s outcome record gives a late-trial abstinence measure for weeks 7–9, while the study description also identifies time to cocaine relapse as a primary outcome measure. A separate time-to-relapse result is not reported in the registry record. The posted abstinence measure therefore cannot answer every question described in the trial rationale. The testing schedule and participant-level rule for determining abstinence across the period are not reported in the registry record.

Important analytic details are absent, including missing urine-test handling, the analysis population, adherence, discontinuations, and whether the analysis followed a prespecified plan. The registry reports a sponsor Fisher Exact p-value but not the sponsor’s effect estimate, interval, or detailed analysis procedure. The unadjusted recalculation differs in its p-value, and the registry does not explain why. This limits the ability to reconcile the two analyses.

The record includes one posted outcome and one analysis but does not describe any multiplicity strategy, interim analyses, or stopping rules. The original sample-size justification and design assumptions are not reported in the registry record. Safety data are also sparse and lack event-level detail, relatedness, timing, and outcomes. These limitations restrict conclusions about precision, generalizability, the analytic process, and the complete safety profile.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Progesterone Treatment for Cocaine-dependent Women: A Pilot Study — NCT00632099.
• Registry record last updated 2014-05-12; status last verified 2014-04.
• Results first posted 2014-05-12.
• 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

NCT00632099 is a completed randomized, triple-masked, parallel-group pilot study sponsored by New York State Psychiatric Institute. The registry lists phase 2 and phase 3. It enrolled 21 women with cocaine dependence, assigning 10 to matched placebo and 11 to oral micronized progesterone, up to 400 mg/day.

The posted primary outcome is the number of participants abstinent by urine toxicology during the final three weeks of the trial, identified as weeks 7–9. The recorded counts were 1 in the placebo arm and 3 in the progesterone arm. The recalculated unadjusted comparison has a positive point estimate, a confidence interval crossing zero, and p = 0.314. The registry reports a sponsor Fisher Exact analysis with p < 0.05; the record does not explain the difference between those results.

The registry does not report the original sample-size justification, missing-data approach, analysis population, or detailed sponsor analysis. The small number of participants and outcomes limits precision, and the posted late-trial abstinence result does not provide a separate time-to-relapse analysis.