Hormonal Factors in the Treatment of Anorexia Nervosa: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Hormonal Factors in the Treatment of Anorexia Nervosa (NCT01121211), a randomized phase 2 study. The primary outcome is Change From Baseline in Weight.

2. Clinical Question

3. Trial Design

The registry identifies this as a randomized, parallel-group phase 2 trial of testosterone and placebo for anorexia nervosa. The listed enrollment is 90. The study is marked completed, with a start date of 2010-04 and a primary completion date of 2016-02. Those dates describe the trial record and do not establish the duration of follow-up for every participant.

The two recorded interventions were testosterone and placebo, each described as lasting 24 weeks. Allocation was randomized, and the masking entry is quadruple. The registry does not explain how randomization was generated or concealed, who was masked, or how the masking was maintained. Details about eligibility criteria, recruitment, treatment delivery, adherence, and concomitant care are not reported in the registry record.

The primary outcome was mean change from baseline in weight, measured in kilograms at baseline and 24 weeks. The stated population was all subjects who received at least one dose of study medication. The posted outcome summaries list 42 subjects in the testosterone arm and 46 in the placebo arm. These are the outcome-specific counts in the record; the registry does not explain why they differ from the listed enrollment.

The trial record also lists anxiety, depression, eating disorder, and anorexia nervosa among its conditions. Its summary says that investigators were interested in weight, disordered eating, depression, and anxiety. However, the registry lists no posted secondary outcome results. It therefore does not support a detailed account of how those other interests were measured or what their results were.

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: All subjects who received at least one dose of study medication

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 Change From Baseline in Weight, measured over Baseline, 24 Weeks.

The primary endpoint is change from baseline in weight, reported in kilograms, with measurements specified at baseline and 24 weeks. Weight is a continuous measure, so the posted group summaries use means and standard deviations. A mean expresses the average change in a group, while a standard deviation describes the spread of individual values around that average. Neither measure alone shows the full distribution of participants’ changes.

Change from baseline is used to describe how the measured value differs over the stated interval, rather than relying only on a single measurement at follow-up. Comparing the mean changes between randomized groups gives an estimate of the group-level contrast over that period. In this record, the testosterone group’s mean change was 0.1 kg, with a standard deviation of 2.7 kg; the placebo group’s mean change was 1.5 kg, with a standard deviation of 3.2 kg.

Weight is one measurable aspect of a condition that the registry describes as involving broader interests, including disordered eating, depression, and anxiety. A weight endpoint cannot, on its own, capture those other outcomes or establish how participants felt or functioned. The primary endpoint description does not report a threshold for a meaningful individual change, and such a threshold is not reported in the registry record.

7. Results Posted in the Registry

• Testosterone: mean 0.10 versus 1.50 on control; difference -1.40 (95% CI -2.63 to -0.17), standardized effect d = -0.47; z-test p = 0.026.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous: it measures change in weight in kilograms. The registry reports each arm’s mean and standard deviation, then gives a mean difference between arms. The contrast is defined in the order testosterone minus placebo. As a result, a negative contrast means the testosterone group had a lower mean change than the placebo group. The reported estimate is -1.4 kg.

What does the mean difference represent?

A group mean is the arithmetic average of participants’ measured changes within that group. The mean difference compares the two group averages on the same measurement scale. In this case, the arm means are 0.1 kg for testosterone and 1.5 kg for placebo. The posted estimate summarizes their contrast; it is not a statement about the change experienced by every individual. Participant-level results can vary around a group mean, as the reported standard deviations indicate.

In a randomized study, comparison between groups is intended to assess differences associated with assignment while reducing some forms of systematic group imbalance. Randomization does not guarantee identical groups, and its successful implementation cannot be evaluated fully from a brief registry record. The method used to generate the allocation sequence, concealment procedures, and any baseline balance details are not reported in the registry record.

What do the standard deviations tell the reader?

A standard deviation describes how dispersed observations are around their group mean. It is not the same as a confidence interval, and it is not a measure of uncertainty in the mean difference alone. The posted standard deviations are 2.7 kg in the testosterone arm and 3.2 kg in the placebo arm. Their presence indicates variation among observed changes, but they do not reveal the shape of each group’s distribution, whether there were outliers, or whether the same participants contributed measurements at both times.

The comparison also reports a pooled standard deviation of 2.972 and a standardized effect estimate of -0.471. A standardized effect expresses a difference relative to a measure of variability, which can help put an effect on a scale without kilograms. It does not change the outcome’s clinical interpretation, and its value depends on the standardization method. The registry record does not provide a fuller explanation of the standardized effect calculation.

How should the p-value be understood?

The unadjusted p-value is 0.026. In general, a p-value measures how unusual data at least as incompatible with a null model as the observed data would be, assuming that model and the analysis conditions hold. It is not the probability that the treatment has no effect, and it does not express the probability that the finding will replicate. The term unadjusted matters because the record does not report whether the primary comparison was selected from a broader set of planned tests or whether any correction for multiple testing was applied.

The registry lists 2 outcome measures but no posted secondary outcome results, and it contains no analysis records. The complete statistical analysis plan, including any covariate adjustment, distributional checks, handling of protocol deviations, or prespecified testing hierarchy, is not reported in the registry record. The primary comparison can therefore be described from its posted summary, but its full analytic context cannot be reconstructed.

Why can over-dispersion matter in general?

For a continuous outcome, the spread of the observations affects the uncertainty of estimated means and contrasts. Greater variability generally makes an estimate less precise when other design features are held constant. If a statistical model assumes less variability than the data actually contain, standard errors may be too small and confidence intervals too narrow. Conversely, a model that accounts for appropriate variability can better reflect uncertainty. This is a general statistical point, not a claim that over-dispersion was present in this trial. The registry record does not report diagnostic checks or a model-based assessment of the outcome distribution.

The posted comparison includes a confidence interval, which communicates uncertainty around the estimated mean difference. The interval is based on a statistical procedure, but the record does not describe all its assumptions or adjustments. It should be interpreted alongside the effect estimate, the observed within-group variation, the outcome population, and the limitations in reporting rather than as a complete account of the analysis.

What does the analysis population mean?

The primary outcome population is defined as all subjects who received at least one dose of study medication. That definition tells the reader which general group the outcome is intended to represent, but the registry does not provide a participant flow diagram or explain the handling of subjects without an observed endpoint. It also does not state whether the analysis followed intention-to-treat principles, used a per-protocol population, or applied another approach. Those labels should not be inferred from the available wording.

Overall, the posted statistics support a description of an unadjusted mean contrast in weight change and its reported uncertainty. They do not supply enough methodological detail to verify all assumptions, reproduce the analysis, or determine how alternative analysis choices might affect the result.

10. Confidence Intervals and Precision

The estimated mean difference in weight change was -1.4 kg for testosterone minus placebo. Its 95% confidence interval runs from -2.634 kg to -0.166 kg. In general, a confidence interval is a range produced by a statistical method that reflects sampling uncertainty under that method’s assumptions. It is not a range containing a stated proportion of individual participants’ changes, and it is not a guarantee that the true effect lies inside the interval.

Both reported endpoints are below zero, indicating that the interval is on the negative side of no mean difference. The width of the interval also matters: estimates within it differ in magnitude, so the point estimate alone should not be treated as perfectly precise. The interval describes uncertainty in the estimated group contrast, not uncertainty about every aspect of the trial or the probability of a clinical outcome.

The group standard deviations were 2.7 kg for testosterone and 3.2 kg for placebo. These describe variation in individual observations, whereas the confidence interval describes uncertainty around the comparison estimate. The outcome counts were 42 and 46, respectively. Precision depends on both variability and the amount of information available, but the registry does not report enough detail about missing measurements or the analysis procedure to assess those factors fully.

The interval is conditional on the statistical method used. Details such as any adjustment for baseline values, covariates, or repeated measurements are not reported in the registry record. The interval should therefore be read as the uncertainty attached to the posted comparison, not as a complete measure of uncertainty across all reasonable analyses or outcomes.

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 describe participants per arm, not total events. They are illustrations based on the observed mean difference, not the original trial’s documented design calculation. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect used for each illustration is the stated fraction of the observed effect. The trial’s own original sample-size assumptions are not reported in the registry record.

For planning for 100% of the observed effect, the figures are 71 participants per arm for 80% power and 95 participants per arm for 90% power. For planning for 75% of the observed effect, they are 126 participants per arm for 80% power and 169 participants per arm for 90% power. For planning for 50% of the observed effect, they are 283 participants per arm for 80% power and 379 participants per arm for 90% power.

In plain language, the 80% power column gives an illustrative group size for a design with an 80% chance of detecting the chosen effect under the stated assumptions if that effect were the true one. The 90% power column uses the higher target chance of detection. These are planning probabilities, not guarantees about a future result. Smaller effects require more participants to detect under the same assumptions, which is why the smaller fractions of the observed effect correspond to larger group-size illustrations.

These calculations are not observed power. They do not describe the power achieved by this completed trial, and they do not establish what effect the original investigators planned to detect. The illustrations use the observed effect as a reference and should be read as conditional planning scenarios.

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 = -1.400; pooled standard deviation σ = 2.972.
2. n = 2 × (1.96 + 0.842)² × 2.972² / -1.400² = 70.7, rounded up to 71 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 <- -1.4                      # observed difference in means
sigma <- 2.97213806058245                     # 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(1121211)
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 = 71 (80% power), 95 (90% power)
 75% of observed effect: n per arm = 126 (80% power), 169 (90% power)
 50% of observed effect: n per arm = 283 (80% power), 379 (90% power)
power.t.test: 71.7 
Simulated power at n = 71 per arm: 0.798 (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

Multiplicity arises when several outcomes, time points, or comparisons are tested. In general, as the number of statistical tests grows, the chance of obtaining at least one apparently unusual result by chance can increase. Researchers may address this through a prespecified hierarchy, adjusted significance thresholds, or other methods. The registry record does not report a multiplicity strategy, and the posted p-value is identified as unadjusted. It is therefore not possible to determine from the record whether the result was adjusted for testing across outcomes or analyses.

The registry lists 2 outcome measures, but no secondary outcome results are posted. It reports no analysis records. The complete protocol and statistical analysis plan are not reported in the registry record, so the intended number of comparisons, any outcome hierarchy, and any planned interim analyses cannot be established. Interim analyses are not reported in the registry record; their presence or absence should not be inferred.

The primary endpoint population is stated as all subjects who received at least one dose of study medication. The outcome summaries report 42 participants for testosterone and 46 for placebo. The record does not describe missing-data handling, explain whether observations were excluded, or report the number of participants with missing weight measurements. It also does not say whether the analysis used imputation, a complete-case approach, or a model that used partial observations.

These details matter because missing outcomes can affect both the amount of information and the balance of groups represented in an analysis. Statistical methods can address missingness under particular assumptions, but no method or assumption should be attributed to this trial without documentation. The registry also does not specify whether the primary comparison was adjusted for baseline weight or other variables.

Consequently, the posted contrast can be described as a result for the stated treated population, but the record does not permit a full assessment of multiplicity control, analysis-population choices, or the sensitivity of the estimate to missing data. Those uncertainties are about reporting and analytic context; they are distinct from the numerical estimate itself.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hospitalization due to exacerbation of gi disorders (4.7%), hospitalization due to anorexia nervosa (2.3%), hospitalization due to depression exacerbation (2.3%).

The serious adverse-event summaries list 4 affected participants among 43 at risk in the testosterone group and 1 among 47 at risk in the placebo group. The record also reports 0 deaths in each group. These counts describe the entries in the registry and should not be read as proof that treatment caused or prevented an event.

The registry lists the most frequent serious-event terms as “Hospitalization due to exacerbation of GI disorders” at 4.7%, “Hospitalization due to anorexia nervosa” at 2.3%, and “Hospitalization due to depression exacerbation” at 2.3%. These are the percentages attached to the listed terms in the record. The record does not provide enough detail to infer how events were adjudicated, whether one participant could contribute to more than one term, or whether events were considered related to study medication.

For other adverse events, 33 affected participants are reported in each group, with 43 at risk for testosterone and 47 at risk for placebo. Serious and other adverse-event summaries are distinct categories in the registry. Their counts are descriptive; the record does not report a comparative safety analysis, event severity details beyond the serious category, or follow-up beyond the recorded trial period.

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

The result describes a difference in average weight change between the two randomized groups over the stated 24-week treatment period. The mean change was 0.1 kg with testosterone and 1.5 kg with placebo, and the estimated contrast was -1.4 kg. The confidence interval and unadjusted p-value add information about statistical uncertainty, but they do not translate directly into an individual outcome or establish clinical importance.

The result does not show that every participant in one group changed in the same way. Group means can coexist with a range of individual experiences, as the reported standard deviations illustrate. Nor does a mean weight contrast answer whether disordered eating, depression, anxiety, quality of life, or other outcomes changed. Although those topics appear in the trial summary, secondary results are not posted in the registry record.

The result also does not establish what would happen beyond the recorded interval or in populations whose characteristics are not described. The registry identifies a phase 2 trial and provides limited information about eligibility, analysis procedures, and missing data. Those omissions restrict conclusions about how broadly the finding applies and how robust it might be to alternative analytic choices.

Statistical evidence and clinical importance are different questions. The p-value reflects compatibility with a null model under particular assumptions; the confidence interval expresses uncertainty around the estimated contrast. Neither determines whether a change is meaningful to patients. A threshold for meaningful change is not reported in the registry record, and no such threshold can be inferred from the summary statistics alone.

In short, the posted data show a lower mean change in weight in the testosterone arm than in the placebo arm for the stated analysis population. That is a description of this measured outcome in this trial, not a conclusion about all aspects of anorexia nervosa, all possible benefits and harms, or effects in other settings.

16. Why This Trial Matters Statistically

This trial illustrates how a continuous clinical outcome can be summarized with group means, standard deviations, a mean difference, a confidence interval, and a p-value. Each statistic answers a different question. The means describe average changes, the standard deviations describe within-group spread, the contrast summarizes the between-group difference, and the interval conveys uncertainty around that contrast.

The example also shows why direction and contrast definition matter. Because the estimate is defined as testosterone minus placebo, its negative sign means the testosterone group had a smaller mean change than the placebo group. Reversing the order would reverse the sign without changing the underlying comparison. Readers should attend to the stated ordering before interpreting a positive or negative estimate.

Statistical significance does not settle the question of clinical meaning. The unadjusted p-value is 0.026, while the confidence interval communicates a range of compatible contrast estimates. Neither value establishes whether the magnitude matters to participants, and the registry does not report a threshold for meaningful change. The absence of posted secondary results further limits what can be said about the other outcomes mentioned in the study summary.

The record is also a useful reminder that precision and completeness of reporting are connected. Outcome-specific participant counts are available, but missing-data methods, multiplicity procedures, and a detailed analysis plan are not. Without those details, readers can describe the posted estimate but cannot fully reconstruct the analysis or evaluate its sensitivity to alternative assumptions.

Finally, planning illustrations based on an observed effect are different from a trial’s original sample-size justification. The record does not report the original design assumptions. Clearly separating retrospective planning scenarios from prospective design choices helps prevent the mistaken interpretation that a completed trial’s size was necessarily chosen to detect the effect later observed.

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 does not report detailed eligibility criteria or participant characteristics. This limits assessment of which people the results may represent. It also does not describe recruitment setting, treatment dose, adherence, or concomitant care, so those aspects of the trial cannot be evaluated from the posted summary.

The primary outcome population is defined as subjects who received at least one dose, while the outcome-specific counts are 42 for testosterone and 46 for placebo. The listed enrollment is 90, but the record does not explain the relationship between enrollment and the outcome counts. It also does not report how many weight measurements were missing, how missing observations were handled, or whether any participants were excluded from the analysis.

Statistical methods are incompletely documented. The record provides the estimate, confidence interval, standard deviations, and unadjusted p-value, but not a complete analysis plan. Details of baseline adjustment, model specification, assumption checks, multiplicity handling, and any interim analyses are not reported in the registry record. The listed outcome measures have no posted secondary results, limiting consideration of other outcomes named in the trial summary.

The follow-up window for the primary endpoint is baseline to 24 weeks. The posted result therefore does not establish what happened after that interval. Safety reporting is also limited: the registry gives affected counts and selected serious-event terms, but does not detail event adjudication, attribution, or the complete clinical context for each event. These reporting limitations should be kept separate from the numerical result, which remains a description of the comparison recorded for the primary outcome.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Hormonal Factors in the Treatment of Anorexia Nervosa — NCT01121211.
• Registry record last updated 2020-09-02; status last verified 2020-08.
• Results first posted 2017-04-26.
• 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

NCT01121211 is a completed randomized, parallel-group phase 2 trial titled “Hormonal Factors in the Treatment of Anorexia Nervosa.” It compared testosterone with placebo, each described as treatment for 24 weeks, and was recorded as quadruple masked. Listed enrollment was 90.

The primary endpoint was change from baseline in weight, measured in kilograms at baseline and 24 weeks, among subjects who received at least one dose. Mean change was 0.1 kg for testosterone and 1.5 kg for placebo. The estimated mean difference, testosterone minus placebo, was -1.4 kg, with a 95% confidence interval from -2.634 kg to -0.166 kg and an unadjusted p-value of 0.026.

The record lists serious adverse events in both arms and reports 0 deaths in each. It does not post secondary outcome results or provide a complete account of missing-data procedures, multiplicity handling, or the statistical analysis plan.