Phenobarbital for Acute Alcohol Withdrawal: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phenobarbital for Acute Alcohol Withdrawal (NCT01184417), a randomized na study. The primary outcome is Number of Patients Requiring Continuous Lorazepam Infusion.

2. Clinical Question

The trial’s stated rationale was to assess whether intravenous phenobarbital added to symptom-guided, standardized lorazepam care would be associated with less need for intensive care unit admission and continuous lorazepam infusion, without increased adverse events. The posted primary endpoint addresses the continuous-infusion part of that rationale. It does not, by itself, answer the question about intensive care unit admission, and the results for that outcome are not reported in the registry record.

The primary result is a comparison of the observed number of participants who required continuous infusion in each assigned group. It is a binary outcome: each participant either required an infusion or did not. This result should be interpreted as an estimate from this trial, rather than as a universal estimate for all people experiencing alcohol withdrawal. The registry record does not report a more detailed description of the enrolled population.

3. Trial Design

The registry record describes a completed, randomized, parallel-group trial with quadruple masking. It reports enrollment of 102 participants, with 51 assigned to the phenobarbital group and 51 to the placebo group. The listed phase is NA. The trial began in 2009-01, and primary completion was recorded in 2010-03.

The phenobarbital group received 10 mg/kg intravenous phenobarbital in 100 ml saline. The placebo group received 100 ml saline. Both groups were described in the record as receiving the standardized institutional alcohol withdrawal protocol, with lorazepam boluses guided by the acute alcohol withdrawal score. The record does not give further detail about the randomization sequence, how treatment allocation was concealed, or what the four masking roles were.

The primary endpoint was the number of participants requiring continuous lorazepam infusion. The record defines this as a yes-or-no outcome. Participants who were refractory to the maximum dose of lorazepam allowed by the protocol could be placed on a continuous intravenous infusion. The endpoint timeframe is listed as 1 year, although the record does not explain the relationship between that timeframe and the trial’s other follow-up procedures.

Other listed outcomes include length of stay, endotracheal intubation, use of a bedside sitter, seizures, and mortality. The registry record lists five secondary outcomes, but results for these outcomes are not reported in the registry record. Details of the trial’s original sample-size rationale, planned statistical methods, follow-up schedule, and analysis populations are also not reported in the registry record. The enrollment total is not a substitute for those design details.

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 Patients Requiring Continuous Lorazepam Infusion, measured over 1 year.

The primary endpoint counts participants who required continuous intravenous lorazepam infusion. The registry record describes standard care in which participants received lorazepam boluses based on their acute alcohol withdrawal score. Boluses of 1, 2 or 4 mg IV could be given serially up to every 15 minutes. Participants refractory to the protocol’s maximum permitted dose, up to 4mg lorazepam IV q 15 mins, were placed on a continuous infusion.

This endpoint is binary at the participant level: a participant either required continuous infusion or did not. The trial result therefore compares event counts between groups, rather than measuring the duration or dose of the infusion. It reflects escalation to continuous lorazepam within the described care protocol. It does not directly measure all aspects of withdrawal severity, total medication exposure, or intensive care unit admission.

As a general statistical point, binary endpoints can be summarized as the number with an event in each group, as well as a difference or ratio between group risks. Counting events makes the outcome straightforward to interpret, while comparative estimates quantify how the observed event frequency differs between groups. The endpoint was relevant to the trial’s stated question about the need for continuous infusion. However, the registry record does not report the detailed rationale for selecting it as the primary endpoint or the original statistical analysis plan.

7. Results Posted in the Registry

• Phenobarbital Group: 3.9% (2/51) versus 31.4% (16/51) on control; risk difference -27.5 percentage points (95% CI -41.3 to -13.6), risk ratio 0.12 (0.03–0.52); unadjusted two-proportion z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a binary event count: each participant either required continuous lorazepam infusion or did not. The registry record reports the number of participants with the event and the group size. These counts are the direct observations. Comparative statistics then describe the direction and magnitude of the contrast between the assigned groups.

What does the risk difference describe?

The risk difference is the event probability in one group minus the event probability in the comparison group, using the ordering defined for the posted estimate. It is an absolute measure: it describes how far apart the groups are in their event probabilities. In this trial, the reported estimate is negative, consistent with the lower observed event count in the phenobarbital group. A risk difference near zero would indicate little absolute separation in observed event frequency; values farther from zero indicate greater separation. This explanation is general, while the direction of the posted estimate comes from the registry result.

What does a relative measure add?

A risk ratio compares the event probabilities as a ratio, while an odds ratio compares the odds of an event. Both are relative measures and can express the contrast differently from the risk difference. For a binary outcome, a ratio may look large or small depending on the baseline event probability, so the event counts and the absolute difference remain important context. The record includes relative estimates, but this explanation does not treat them as evidence of a mechanism or as a substitute for the observed counts.

How should the p-value be read?

The registry record reports an unadjusted p < 0.001. As a general concept, a p-value measures how unusual data at least as discrepant as those observed would be if a specified null model were true and the analysis assumptions held. A small p-value is evidence against that null model, but it does not give the probability that the treatment effect is real. It also does not show whether the observed difference is important for every clinical decision. The record does not identify the statistical test used to obtain the p-value, so a more specific account of its assumptions is not reported in the registry record.

What does the confidence interval contribute?

A confidence interval gives a range of effect estimates compatible with the data and the method used to construct the interval. It conveys uncertainty around the estimated group contrast, not a range that is guaranteed to contain the true effect. The posted interval for the risk difference lies below zero, indicating that the interval’s estimates share the same direction as the point estimate. Its width also matters: a wider interval reflects less precision than a narrower interval, all else being equal. The interval construction method is not reported in the registry record.

What does randomization contribute?

Random assignment is intended to make the groups comparable, on average, with respect to factors that could influence the outcome. It provides a basis for comparing the assigned groups, but it does not eliminate uncertainty or guarantee that every characteristic was balanced in a particular trial. The registry record identifies the allocation as randomized, but further details about sequence generation and concealment are not reported in the registry record.

What cannot be inferred from the posted analysis?

The result does not identify why the groups differed in infusion use, and it does not establish the findings for outcomes whose results are not posted. Nor does the p-value or confidence interval describe the completeness of follow-up, adherence to assigned treatment, or the robustness of missing-data decisions. The record does not report the original statistical analysis plan, model specification, or any prespecified subgroup analyses. Those details cannot be reconstructed from the event counts and comparative estimate alone.

The trial’s primary outcome is a count of participants with an event, rather than a count of repeated infusion episodes or a duration measure. This distinction matters: the analysis concerns whether an individual reached the threshold for continuous infusion, as defined in the protocol description. It does not quantify how long an infusion lasted or how much lorazepam was delivered after infusion began.

The posted comparative result is unadjusted. An adjusted analysis could account for specified baseline variables, but whether one was planned or performed is not reported in the registry record. No adjustment should be assumed. Likewise, the record does not explain whether the estimate was calculated under an intention-to-treat, per-protocol, or other analysis population. These omissions limit the ability to reproduce or fully assess the statistical analysis, even though the group counts and the principal comparison are posted.

10. Confidence Intervals and Precision

Precision concerns how much uncertainty surrounds an estimated group difference. The risk-difference confidence interval is reported entirely below zero, so its range does not include no difference. This is consistent with the negative point estimate and the reported unadjusted p < 0.001. Exact interval endpoints are not repeated in this explanation; the direction and exclusion of zero are the key features for interpreting the posted interval.

A confidence interval is not a guarantee that the true effect lies within its endpoints. In general, it is constructed by a method that would cover the true parameter at a stated rate over repeated use, if its assumptions hold. The registry record does not report the confidence level or construction method, so those details are not inferred. The interval should be read as a measure of uncertainty around the estimate under the method used for the posted result.

The event counts were 2 of 51 in the phenobarbital group and 16 of 51 in the placebo group. Equal group sizes make the group counts directly comparable, but do not remove sampling uncertainty. A confidence interval is useful because the observed difference is an estimate based on this trial’s participants; results in another sample could vary. The interval’s direction is clear in the posted analysis, while the record does not provide enough information to assess sensitivity to alternative analysis choices.

Precision and statistical significance are related but distinct. A result can have an interval that excludes zero while still leaving uncertainty about the exact magnitude of the group difference. The interval does not establish the effect on other outcomes, including intensive care unit admission, and it does not address unreported analysis or follow-up details.

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 expressed as participants per arm, not as total events. They illustrate the group size that a hypothetical study would require under the stated planning assumptions, using different fractions of the observed effect. The displayed values are 30 participants per arm for 80% power and 39 per arm for 90% power when planning for 100% of the observed effect; 61 and 81 participants per arm when planning for 75% of the observed effect; and 153 and 205 participants per arm when planning for 50% of the observed effect.

In plain terms, the 80% power column represents the planned chance of detecting the specified effect under the assumed model if that effect were the true one. The 90% power column uses a higher target chance of detection. These are planning quantities, not evidence about the chance that the trial’s result is correct. This is not observed power.

The assumptions for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect equal to the stated fraction of the observed effect. The original trial’s own sample-size assumptions are not reported in the registry record. The planning figures are therefore not a reconstruction of the trial’s design rationale.

Smaller effects require more participants to detect under otherwise similar assumptions. That pattern is reflected in the larger planning figures for smaller fractions of the observed effect. These figures are illustrative and depend on the assumptions used; they should not be read as evidence that a particular sample size would guarantee a result.

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 = 0.314; treatment proportion p₁ = 0.039; average p̄ = 0.176.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5391 = 1.0567.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5030 = 0.4233.
4. n = (1.0567 + 0.4233)² / (-0.2745)² = 29.1, rounded up to 30 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 <- 0.313725490196078                          # control proportion
p1 <- 0.0392156862745098                          # 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(1184417)
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 = 30 (80% power), 39 (90% power)
 75% of observed effect: n per arm = 61 (80% power), 81 (90% power)
 50% of observed effect: n per arm = 153 (80% power), 205 (90% power)
power.prop.test: 29.1 
Simulated power at n = 30 per arm: 0.862 (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 record lists one primary endpoint and five secondary outcomes. The listed secondary outcomes include length of stay, endotracheal intubation, use of a bedside sitter, seizures, and mortality. Results for those secondary outcomes are not reported in the registry record. The record also notes eight outcome measures and four analyses, but does not explain how these relate to outcome definitions or the posted primary comparison.

When several outcomes or analyses are examined, the chance of at least one apparently notable result can increase. Statistical methods can address this multiplicity, for example through prespecified testing sequences or adjusted significance thresholds. Whether any such approach was planned or used is not reported in the registry record. The posted primary p-value is described as unadjusted; no multiplicity adjustment should be assumed.

The registry record does not report a statistical analysis plan, the handling of missing primary outcome data, or the analysis population. It is therefore not possible to determine whether all randomized participants were included in the primary analysis, whether any participants lacked outcome data, or how any missing data were handled. The fact that the groups each have 51 participants in the posted primary result does not itself describe the analysis population rules or confirm completeness of follow-up.

Further details about protocol deviations, treatment adherence, exclusions after randomization, and prespecified subgroup analyses are not reported in the registry record. These items matter because they can affect which participants contribute to an estimate and how well the analysis follows the randomized comparison. Without those details, the posted result can be described, but its sensitivity to alternative analysis decisions cannot be assessed.

These reporting gaps do not alter the recorded event counts or comparative result. They do limit the ability to judge whether the reported inferential analysis followed a prespecified plan, how multiplicity was addressed, and whether the result is robust to missing-data or population choices.

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.

The registry record reports zero serious adverse events among 51 participants at risk in the phenobarbital group and zero among 51 participants at risk in the placebo group. It also reports zero other adverse events among 51 participants at risk in each group. These figures describe the counts recorded in the registry; they do not establish that no adverse events of any kind occurred beyond those categories or under other definitions.

Deaths are not reported as a numerical result in the adverse-event record. The registry lists mortality as a secondary outcome, but its outcome result is not reported in the registry record. The record also does not provide event definitions, the collection period for these adverse-event counts, or detail about how events were assessed.

Zero recorded events in both groups means that these posted counts do not show a difference between groups for the listed adverse-event categories. It does not demonstrate equivalent safety or rule out uncommon events. Safety interpretation depends on how events were collected and classified, and those details are not reported in the registry record. The counts should be considered alongside the trial’s limited enrollment and the absence of posted results for several listed safety-related outcomes.

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

The posted primary result shows fewer participants requiring continuous lorazepam infusion in the phenobarbital group than in the placebo group: 2 of 51 compared with 16 of 51. The risk-difference estimate is negative, its confidence interval lies below zero, and the registry record reports an unadjusted p < 0.001. These statistics describe a clear difference in the observed primary endpoint in this randomized trial.

The result does not establish what happened for every outcome named in the trial rationale. Intensive care unit admission is mentioned in the stated rationale, but its result is not reported in the registry record. Results for the listed secondary outcomes, including length of stay, endotracheal intubation, bedside sitter use, seizure, and mortality, are also not reported in the registry record. The primary endpoint should not be treated as a stand-in for those outcomes.

The result does not prove the same event pattern would occur in every setting or population. The registry record gives limited detail about the enrolled population and does not report the original statistical plan, missing-data handling, or analysis population. These gaps restrict how thoroughly the comparison can be evaluated.

Nor does the primary result alone establish an overall benefit-risk balance. The adverse-event counts recorded are zero for serious and other adverse events in both groups, but event definitions and collection details are not reported. The mortality result is not reported in the registry record. Absence of a posted event count is not evidence that an outcome did not occur.

In statistical terms, the findings concern the observed frequency of a defined binary endpoint under the trial’s protocol. The estimates quantify uncertainty in that group comparison, but do not answer questions about mechanisms, longer-term outcomes, or outcomes that were not posted. No treatment recommendation follows from the statistical result alone.

16. Why This Trial Matters Statistically

This trial illustrates how a binary clinical endpoint can be presented in several complementary ways. The event counts show what happened in each assigned group, the risk difference expresses the absolute contrast, and the confidence interval conveys uncertainty around that contrast. The unadjusted p-value addresses compatibility with a no-difference model under the analysis assumptions. None of these measures alone provides a complete account; their interpretations are strongest when read together with the endpoint definition and group sizes.

The endpoint definition is especially important. Continuous lorazepam infusion was not simply a general measure of withdrawal severity; it represented escalation after participants were refractory to the maximum bolus dose allowed by the described protocol. That makes the result specific to an operational threshold embedded in the care pathway. Different protocols or thresholds could define a different event.

The result also illustrates the distinction between a trial rationale and a posted outcome result. The stated rationale includes intensive care unit admission, but the registry record reports a primary comparison for continuous infusion. Without posted results for the other outcomes, the primary finding cannot answer the broader set of questions in the rationale.

Finally, the record shows why statistical interpretation depends on reporting as well as calculation. The counts and primary comparison are available, while original design assumptions, analysis-population rules, missing-data methods, and multiplicity procedures are not reported in the registry record. Recognizing both what the comparison establishes and what remains undocumented is central to a careful reading of clinical trial evidence.

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 registry record provides limited detail about the trial’s statistical design. The original sample-size justification, planned analysis, confidence-interval method, randomization procedure, allocation concealment, and analysis population are not reported in the registry record. Missing-data handling and the extent of follow-up are also not reported. These omissions prevent a full assessment of whether the posted analysis followed a prespecified plan.

The endpoint is a protocol-defined escalation event, not a direct measure of all treatment use or severity. It records whether a participant required continuous lorazepam infusion, but does not report infusion duration, total lorazepam exposure, or the reason for each individual escalation. The registry record lists a timeframe of 1 year without explaining the timing of endpoint assessment in more detail.

Several outcomes relevant to the stated rationale or listed as secondary outcomes lack posted results. In particular, intensive care unit admission is not reported as a result, and the registry record does not report results for length of stay, endotracheal intubation, bedside sitter use, seizure, or mortality. Thus the primary endpoint cannot serve as a complete summary of the trial’s outcomes.

The safety record reports zero serious and zero other adverse events in both groups, but does not give event definitions, assessment procedures, or collection details. Death counts are not reported. These limitations constrain interpretation of safety and the ability to assess uncommon outcomes. The registry record also gives limited information about the enrolled population, so the extent to which the findings apply beyond the trial participants cannot be established from this record alone.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Phenobarbital for Acute Alcohol Withdrawal — NCT01184417.
• Registry record last updated 2012-03-09; status last verified 2012-03.
• Results first posted 2012-03-09.
• 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, parallel-group trial of phenobarbital versus placebo for acute alcohol withdrawal. Enrollment was 102, with 51 participants in each group. The phenobarbital group received 10 mg/kg IV phenobarbital in 100 ml saline; the placebo group received 100 ml saline. Both groups were described as receiving the standardized institutional lorazepam-based alcohol withdrawal protocol.

The primary outcome was the number of participants requiring continuous lorazepam infusion. The posted counts were 2 of 51 in the phenobarbital group and 16 of 51 in the placebo group. The estimated risk difference was negative, its confidence interval lay below zero, and the unadjusted result was p < 0.001.

The registry record does not report results for the listed secondary outcomes, the original statistical plan, missing-data handling, or analysis populations. Adverse-event counts were zero for serious and other events in each group; death counts are not reported.