IGF-1 and Bone Loss in Women With Anorexia Nervosa: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for IGF-1 and Bone Loss in Women With Anorexia Nervosa (NCT01406444), a randomized na study. The primary outcome is Postero-anterior Spine Bone Mineral Density by DXA.

2. Clinical Question

3. Trial Design

The registry record describes a randomized, double-blind, parallel-group trial sponsored by Massachusetts General Hospital. It lists an enrollment of 148 and a completed status. The recorded start was 2011-10 and primary completion was 2019-05. The phase is listed as NA.

Three treatment groups are described: sequential rhIGF-1 followed by risedronate, risedronate alone, and placebo. The sequential regimen is described as rhIGF-1 for 6 months followed by risedronate for 6 months. The risedronate-only and placebo regimens each span 12 months. The primary endpoint timeframe is 12 months.

The primary outcome is postero-anterior spine bone mineral density by DXA. The posted arm summaries for this outcome include 23 participants in the sequential-treatment group, 24 in the risedronate group, and 14 in the placebo group. Those outcome counts differ from the enrollment figure. The registry record does not explain the difference or report the analysis populations used to produce those counts. Adverse-event records list 33 participants at risk in each active-treatment group and 16 in the placebo group; those denominators also differ from the outcome counts.

Randomization and double masking describe important aspects of the planned comparison, but the registry record does not report details such as the randomization procedure, how masking was maintained, or how allocation was concealed. Original design assumptions, including the trial's own sample-size justification, are not reported in the registry record.

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: 148 subjects were consented and screened, 90 subjects were randomized, and 82 subjects completed the baseline visit. Of the 90 who were randomized, 8 subjects did not attend their scheduled baseline visits and did not receive study medication.

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 Postero-anterior Spine Bone Mineral Density by DXA, measured over 12 Months.

The primary endpoint is postero-anterior spine bone mineral density measured by dual-energy X-ray absorptiometry, or DXA, at 12 months. Bone mineral density is a continuous measure: the posted results report group least-squares means in g/cm2, together with standard errors.

DXA provides a measure of bone mineral density at a specified skeletal site. The registry's summary describes bone thinning as relevant to anorexia nervosa and identifies bone density as one of the outcomes investigators sought to assess. A continuous endpoint can retain information about the measured value across participants rather than reducing results to a simple event count. The meaning of a group mean is different from the experience of any one participant; it describes a group-level summary.

The primary measure concerns the postero-anterior spine. The registry also posts lateral spine bone mineral density as a secondary outcome. These are distinct DXA measures, so their numerical values should be interpreted on their own scales rather than treated as interchangeable. The record does not report a threshold for a clinically important change, nor does it state how changes in bone density relate to fractures or other clinical outcomes in this trial.

7. Results Posted in the Registry

• rhIGF-1 Followed by Risedronate: mean 0.89 versus 0.87 on control; difference 0.02 (95% CI -0.02 to 0.05), standardized effect d = 0.28; z-test p = 0.374.
• Risedronate: mean 0.89 versus 0.87 on control; difference 0.01 (95% CI -0.02 to 0.05), standardized effect d = 0.21; z-test p = 0.496.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Lateral Spine Bone Mineral Density by DXA

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • rhIGF-1 Followed by Risedronate: mean 0.69 versus 0.67 on control; difference 0.03 (95% CI -0.00 to 0.06), standardized effect d = 0.57; z-test p = 0.063.
  • Risedronate: mean 0.68 versus 0.67 on control; difference 0.01 (95% CI -0.02 to 0.04), standardized effect d = 0.21; z-test p = 0.481.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between rhIGF-1 Followed by Risedronate and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Risedronate and the control. The interval gives the range of effects that remain compatible with the data.

The registry posts one secondary outcome: lateral spine bone mineral density by DXA at 12 months. Its general pattern can be considered alongside the primary postero-anterior spine measure, but it remains a distinct endpoint measured at a different spine orientation. Agreement across related measurements can add context, while differences in their estimates or statistical support can reflect endpoint-specific information and uncertainty.

The registry reports sponsor analyses for the secondary outcome, including comparisons involving the sequential-treatment group. It also posts unadjusted recalculations for both active-treatment groups versus placebo. These analyses should not be treated as independent confirmation of the primary result: they concern a related measure in the same trial and use multiple comparisons. The record does not report a multiplicity adjustment or a prespecified hierarchy for interpreting the secondary analyses.

Unadjusted secondary results carry less weight as confirmatory evidence when several outcomes or contrasts are tested without a reported plan for controlling the overall chance of false-positive findings. A nominal p-value can be informative as part of the evidence, but it should be interpreted in light of the number of analyses and whether they were prespecified. The registry does not report those details. The secondary findings are therefore best viewed as supportive or exploratory context, rather than as a stand-alone resolution of the primary result's discrepancy.

The record does not report a clinical-importance threshold for lateral spine bone mineral density, nor does it describe whether the secondary measure was intended to test a distinct clinical hypothesis. Those omissions limit conclusions about how much weight to assign the secondary endpoint beyond its role as an additional measurement.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Lateral Spine Bone Mineral Density by DXA

9. Statistical Interpretation

10. Statistical Methods Explained

The registry reports a least-squares mean for each group on the primary continuous endpoint. A mean summarizes the central value of a measurement across a group. A least-squares mean is a model-based group estimate; it may account for features of the fitted model. The registry record does not provide enough detail to determine which covariates or other terms were included in that model.

What does the standard error describe?

The posted spread is a standard error. In general, a standard error describes the precision of an estimated mean, rather than the spread of individual participants' measurements. A smaller standard error indicates a more precisely estimated mean, all else equal. It should not be interpreted as the range containing most individual values. The registry does not post standard deviations for the arms.

What is the sponsor's reported model?

For the sponsor-reported comparisons, the registry names a linear random effects model and gives a p-value for each reported superiority analysis. In general, a random effects model can represent variation associated with grouping or repeated structure, depending on its specification. The exact structure, fitted coefficients, adjustment variables, and model diagnostics are not reported in the registry record. The reported p-values therefore cannot be independently reconstructed from the model description alone.

What does the recalculation do?

The unadjusted recalculations compare posted group means and use the posted standard errors to obtain approximate uncertainty for mean differences. The pooled standard deviations used for standardized effect summaries are approximate because they are derived from the standard errors. These calculations are not the sponsor's linear random effects model. They do not incorporate unreported covariates, repeated-measure structure, or other possible adjustments. The confidence intervals and p-values from these recalculations should be understood as simple, unadjusted summaries of the posted arm-level results.

In general, a confidence interval describes a range of estimates compatible with the observed data under the calculation's assumptions. Its width conveys precision: a wider interval leaves more uncertainty about the contrast's size. A p-value describes how compatible the data are with a specified null hypothesis under a statistical model; it is not the probability that the null hypothesis is true, and it does not measure clinical importance. The registry's sponsor analyses and the recalculations have different methods, so their inferential results need not agree.

Why can over-dispersion matter?

General explanation: over-dispersion occurs when observed variability is greater than a model assumes. If a model does not account for that extra variability, its standard errors can be too small and its intervals too narrow. This is a general statistical consideration, not a finding about this trial: the registry record does not report an over-dispersion assessment or enough model detail to determine whether it was relevant.

What does a superiority analysis test?

In general, a superiority analysis evaluates whether the data support a difference in a specified direction or a difference from a null value, according to the analysis plan. The registry labels the reported comparisons as superiority analyses, but does not provide the full hypothesis specification, decision threshold, or multiplicity procedure. A reported p-value should therefore be read alongside the estimated contrast and its precision, where those are available. For the sponsor's primary comparison, the registry reports a p-value but no model-based effect estimate or interval.

What can the posted analysis not establish?

The record does not report the handling of missing outcome data, the analysis population definition, the method used for any imputation, or sensitivity analyses. It also does not report the original sample-size assumptions or interim analyses. Those details can affect how a trial's estimates and tests should be understood. Their absence does not show that a particular method was or was not used; it means the registry record does not supply that information.

11. Confidence Intervals and Precision

The primary unadjusted recalculation for sequential rhIGF-1 followed by risedronate versus placebo gives a mean difference of 0.017 g/cm2 and a 95% confidence interval from -0.020 to 0.054. This interval spans zero and also includes positive values. It therefore reflects substantial uncertainty about the contrast's magnitude and direction when judged from the recalculation alone. The associated unadjusted p-value is 0.374.

For risedronate versus placebo on the primary endpoint, the recalculated mean difference is 0.013 g/cm2, with a 95% confidence interval from -0.024 to 0.051 and p = 0.496. This interval also spans zero. Neither recalculated interval is a sponsor-reported model interval; the registry provides no confidence interval for the sponsor's primary comparison.

In general, an interval's width reflects statistical precision, while its position reflects the range of contrasts compatible with the data under the calculation's assumptions. A confidence interval is not a guarantee that the true value lies in the displayed range, and it does not determine whether a difference matters clinically. The registry record does not state a clinically important difference threshold for bone mineral density.

The recalculated intervals are based on small posted outcome groups and approximate variance information derived from standard errors. They should be read as unadjusted summaries, not as replacements for the sponsor's reported model. The model-based p-value and the recalculated interval answer related but not identical analytical questions because their methods differ.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm, not total events. They are illustrations based on the observed effect, rather than the trial's original design calculations. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the listed fraction of the observed effect. The trial's own original design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the illustrative number of participants per arm under which the calculation would have an 80% chance of detecting the specified effect if the assumptions hold. The 90% power column gives the corresponding planning number for a 90% chance. These are design probabilities under assumed conditions, not guarantees for a future study. They do not describe the probability that the observed result is true or that a treatment effect exists.

The rows use the full observed effect and smaller fractions of that effect. For a smaller effect, the illustrative requirement is larger: a study generally needs more participants to distinguish a subtler contrast from sampling variation. The planning figures differ between contrasts because their observed effects and variability differ. They should be treated as scenario-based planning information, not as a restatement of how this trial was designed or conducted.

These figures are not observed power. Observed power is not an appropriate substitute for examining the effect estimate and its uncertainty after a study. The confidence intervals and the analysis method remain essential for understanding the precision of the posted results.

13. 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 = 0.017; pooled standard deviation σ = 0.061.
2. n = 2 × (1.96 + 0.842)² × 0.061² / 0.017² = 201.6, rounded up to 202 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 <- 0.017                      # observed difference in means
sigma <- 0.0609294674193038                     # 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(1406444)
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 = 202 (80% power), 270 (90% power)
 75% of observed effect: n per arm = 359 (80% power), 480 (90% power)
 50% of observed effect: n per arm = 807 (80% power), 1,080 (90% power)
power.t.test: 202.6 
Simulated power at n = 202 per arm: 0.797 (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.

14. Analyses Reported by the Sponsor

The registry's sponsor-reported primary analysis compares rhIGF-1 followed by risedronate with placebo using a linear random effects model. It reports a superiority p-value of 0.03, without a model-based effect estimate or confidence interval. The unadjusted recalculation from posted arm summaries gives a different result: a mean difference of 0.017 g/cm2, a 95% confidence interval from -0.020 to 0.054, and p = 0.374.

The difference between these results is a genuine discrepancy in the posted information. The recalculation is a simple unadjusted analysis and does not reproduce the sponsor's named model. The record does not provide the model details needed to reconcile the two results. In this circumstance, the sponsor's own analysis takes precedence over the recalculation, while the discrepancy remains important context for readers.

Precedence does not make the missing model estimate or interval available. The p-value alone does not show the size or precision of the sponsor's estimated contrast. The registry also does not report a sponsor analysis for every primary-endpoint comparison listed in the recalculations. Reported model results and recalculated summary-statistic results should therefore be identified separately rather than combined as though they came from one analysis.

15. Multiplicity, Missing Data and Analysis Populations

The registry record does not report whether the trial adjusted for multiplicity across the primary and secondary outcomes, treatment comparisons, or statistical tests. Multiplicity matters because, in general, conducting several tests increases the chance that at least one result appears statistically notable by chance. An adjustment or a prespecified testing hierarchy can address that issue, but neither is described in the record. The reported p-values should not be assumed to have been adjusted.

Missing-data methods are also not reported in the registry record. It does not explain why the primary outcome counts differ from the overall enrollment figure, whether participants lacked a 12-month measurement, or how any unavailable measurements were handled. The record does not specify whether the analyses used all randomized participants, only participants with available measurements, or another analysis population. It also does not report imputation procedures or sensitivity analyses for missing outcomes.

The registry does not provide the original sample-size justification, the assumptions used to plan enrollment, or information about interim analyses. Those details should not be inferred from the planning illustrations, which are based on observed effects and are not the trial's own design calculations.

Other design and analysis details are absent as well. The randomization procedure, allocation concealment, masking procedures, covariate adjustment, model specification, and diagnostic checks are not reported. These omissions do not establish that the procedures were inadequate or unused; they limit what can be independently assessed from the registry record. In particular, the lack of model detail prevents a direct reconstruction of the sponsor's primary analysis and prevents a full explanation of the difference from the unadjusted recalculation.

16. 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 non-serious adverse event terms

The serious adverse-event fields report zero affected participants in each group: 0 among 33 at risk for rhIGF-1 followed by risedronate, 0 among 33 for risedronate, and 0 among 16 for placebo. Deaths are also reported as 0 in each group. These figures describe the registry's posted counts during the recorded safety assessment; they do not establish that no other harms occurred.

The record also gives counts for other adverse events: 19 in the sequential-treatment group, 18 in the risedronate group, and 5 in the placebo group. The registry fields do not identify the event terms in the posted information, nor do they clarify whether these counts represent unique participants or events. No serious adverse-event terms are listed.

Safety denominators differ from the posted primary-outcome counts. The registry record does not explain the difference, define the safety observation period in detail, or provide information about event severity beyond the serious-event classification. The counts should be read as descriptive registry data, not as a complete account of comparative safety.

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

The posted results describe group-level bone mineral density measurements at 12 months in a randomized, double-blind, parallel-group trial. They do not establish what an individual participant experienced. The sponsor's primary model analysis reports a superiority p-value for sequential rhIGF-1 followed by risedronate versus placebo, but the registry does not provide the associated model estimate or confidence interval. The simple unadjusted recalculation differs, and the registry record does not provide enough model detail to reconcile the discrepancy.

A statistical result is not, by itself, a measure of clinical importance. The record does not report a threshold for a meaningful difference in bone mineral density, and it does not report fracture outcomes in the results described. The posted endpoint therefore cannot be translated into a conclusion about bone strength or fracture risk from these data alone.

The results also do not establish whether the observed group differences would persist beyond the 12-month timeframe. The registry record does not report longer-term results for the measures discussed. Nor does it report the analysis population and missing-data handling needed to assess how incomplete outcome information may have affected the estimates.

Interpretation should keep three kinds of information distinct: the arm-level means, the sponsor's model-based tests, and the unadjusted recalculations. The means show the posted group summaries; the sponsor analysis reflects a model whose details are not reported; and the recalculation supplies a transparent but limited comparison based on posted summaries. None of these alone resolves every uncertainty about effect magnitude, clinical relevance, or generalizability.

18. Why This Trial Matters Statistically

This trial illustrates why a p-value should not be read without knowing the analysis that produced it. The sponsor reports a primary superiority result from a linear random effects model, while a simple unadjusted recalculation from the posted arm summaries yields a different inferential result. Without the sponsor's model estimate, interval, and specification, readers cannot fully trace the source of that difference.

It also illustrates the value of reporting effect estimates and uncertainty alongside tests. A p-value does not communicate the magnitude of a contrast or its precision. Confidence intervals from a clearly described analysis help readers assess which effect sizes are compatible with the data, while an explicit analysis description supports independent review.

The trial has multiple groups and a secondary DXA endpoint, so interpretation also depends on how the analysis plan handled multiple comparisons. That information is not reported in the registry record. The distinction between a confirmatory primary question and additional secondary comparisons matters when interpreting nominal statistical evidence.

Finally, the posted participant counts vary across enrollment, outcome summaries, and safety records. Transparent reporting of analysis populations and missing-data methods would help readers understand which participants contributed to each result. These are general reporting and inference issues that apply well beyond this trial.

19. 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 leaves important details unavailable for independent assessment. The sponsor's primary analysis is identified as a linear random effects model, but the fitted model, adjustment variables, estimate, and confidence interval are not reported. This limits interpretation of the sponsor's p-value and prevents direct reconstruction. The unadjusted recalculation is based on arm summaries and approximate variance information, so it should not be mistaken for the sponsor's analysis.

Enrollment is listed as 148, while the primary endpoint summaries include 23, 24, and 14 participants across groups. The safety records list 33, 33, and 16 at risk. The registry record does not explain these differing counts, define the analysis populations, or report how missing endpoint data were handled.

The record does not report the original sample-size justification, randomization procedure, masking procedures, multiplicity strategy, interim analyses, or detailed statistical analysis plan. It also does not state a threshold for a clinically important difference in bone mineral density. These gaps constrain assessment of both the design and the meaning of the statistical findings.

The available outcome is a 12-month DXA measure. The posted results do not establish effects on fractures or longer-term outcomes. Safety reporting is limited to the posted counts, with no serious adverse-event terms listed and no detailed event descriptions in the information summarized. These limitations call for restrained interpretation rather than assumptions about unreported procedures or outcomes.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: IGF-1 and Bone Loss in Women With Anorexia Nervosa — NCT01406444.
• Registry record last updated 2020-07-17; status last verified 2020-07.
• Results first posted 2020-07-17.
• 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.

23. Record Summary

The registry describes a completed randomized, double-blind, parallel-group trial sponsored by Massachusetts General Hospital, with 148 enrolled participants and a primary completion date of 2019-05. The treatment groups were sequential rhIGF-1 followed by risedronate, risedronate alone, and placebo.

The primary outcome was postero-anterior spine bone mineral density by DXA at 12 months. The sponsor reported a superiority analysis for sequential treatment versus placebo with p = 0.03, but did not report the model-based estimate or confidence interval. An unadjusted recalculation from the posted summaries differs; the discrepancy means the sponsor's own analysis takes precedence, while the recalculation remains a limited complementary description.

The registry also posts lateral spine bone mineral density as a secondary outcome. Multiplicity handling, missing-data methods, analysis populations, and the original sample-size rationale are not reported in the registry record. Serious adverse-event and death counts are reported as zero in all three groups.