A Study of Armour® Thyroid Compared to Synthetic T4 (Levothyroxine) in Previously Hypothyroid Participants: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Armour® Thyroid Compared to Synthetic T4 (Levothyroxine) in Previously Hypothyroid Participants (NCT04124705), a randomized phase 2 study. The primary outcome is Percentage of Participants With a Sustained TSH Response.

2. Clinical Question

The study addresses dose conversion from stable levothyroxine therapy to Armour® Thyroid. It enrolled participants who were already euthyroid on T4 replacement, so the registry’s question is about maintaining a target TSH response during a treatment change, rather than initiating thyroid replacement in untreated participants. Armour® Thyroid and levothyroxine were compared in a randomized, double-blind, parallel-group design. The registry records 284 participants enrolled. The primary outcome timeframe varied according to the length of each participant’s titration period: the end of titration was Week 18, 24, 30, or 36, and the end of stabilization was Week 30, 36, 42, or 48, respectively. The clinical question is therefore whether participants met the specified TSH range at both designated assessments following titration and stabilization. The registry record does not report a non-inferiority margin, so the clinical threshold for judging whether a difference was acceptable cannot be established from this record alone.

3. Trial Design

This was a Phase 2, randomized, double-blind, parallel-group study sponsored by Allergan. The registry lists the study as completed. The recorded start date is 2019-10-11, and the primary completion date is 2021-06-22. Enrollment was 284. Participants were assigned to Armour® Thyroid or levothyroxine. In the Armour® Thyroid group, the starting dose corresponded to the participant’s pre-randomized dose of synthetic T4; in the levothyroxine group, participants received their pre-randomized dose. Both descriptions state that dose titration was based on TSH during the titration period.

The titration period could end at Week 18, 24, 30, or 36, and the stabilization assessment could occur at Week 30, 36, 42, or 48 depending on titration length. The primary outcome was assessed in all randomized participants who took at least one dose of study intervention, described in the registry as the intent-to-treat population. The registry also describes non-responder imputation incorporating multiple imputation for missing TSH values due to COVID-19 (NRI-C).

The posted primary outcome results list 141 participants in the Armour® Thyroid group and 143 in the levothyroxine group. The registry’s reported analysis is a stratified response-rate difference with a 95% confidence interval and is labeled as a non-inferiority analysis. The registry record does not report the stratification factors, the non-inferiority margin, the randomization ratio, or the detailed statistical analysis plan. It also does not report the original sample-size justification or whether interim analyses were planned or conducted. These omissions limit assessment of how closely the posted analysis can be matched to the trial’s original design assumptions.

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: Eligible participants were randomized equally to receive either their same dose of levothyroxine or a matching dose of Armour Thyroid at Baseline. The randomization was stratified by age (\< 65 years, ≥ 65 years).

Titration Period

Reasons for not completing this period:

Stabilization Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants who took at least 1 dose of study intervention (intent-to-treat population). The non-responder imputation incorporating multiple imputation to handle missing data due to coronavirus disease 2019 (COVID-19) (NRI-C) was performed for participants who had missing TSH values either at the end of the Titration Period or at the end of the Stabilization 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 Percentage of Participants With a Sustained TSH Response, measured over End of the titration period (Week 18, 24, 30, or 36) and end of the stabilization period (Week 30, 36, 42, or 48, depending on the length of the titration period)..

The primary endpoint is the percentage of participants with a sustained TSH response. A response requires TSH values within the normal reference range of 0.45 to 4.12 mIU/L, inclusive, at both the end of the titration period and the end of the stabilization period. The endpoint therefore reflects response at two assessments rather than a single measurement. The assessment weeks varied with titration length: titration ended at Week 18, 24, 30, or 36, followed by stabilization assessment at Week 30, 36, 42, or 48.

This endpoint is used to capture whether the target TSH range was maintained across the transition from dose adjustment to stabilization. Requiring both measurements makes the definition more demanding than requiring an in-range result at only one of the two assessments. At the same time, a percentage within range does not describe the size or direction of TSH deviations among participants who did not meet the response definition, nor does it alone characterize symptoms or other clinical outcomes. The registry describes the primary population as randomized participants who took at least one dose of study intervention and reports NRI-C for missing TSH values related to COVID-19. Details of the imputation procedure beyond that description are not reported in the registry record.

7. Results Posted in the Registry

• Levothyroxine: 76.2% (109/143) versus 61.0% (86/141) on control; risk difference 15.2 percentage points (95% CI 4.6 to 25.9), risk ratio 1.25 (1.06–1.47); unadjusted two-proportion z-test p = 0.006.

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.

Percentage of Participants With a Titration TSH Response

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Levothyroxine: 90.9% (130/143) versus 80.9% (114/141) on control; risk difference 10.1 percentage points (95% CI 2.0 to 18.1), risk ratio 1.12 (1.02–1.24); unadjusted two-proportion z-test p = 0.015.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Levothyroxine than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry posts one secondary endpoint: the percentage of participants with a TSH response at the end of titration. The response definition is linked to the same normal reference range, but the secondary outcome concerns the end of titration rather than requiring response at both titration and stabilization. The posted group results and their comparison are reported with the secondary outcome statistics.

At a broad level, the pattern across the primary and secondary outcomes is consistent: the response percentage is higher in the levothyroxine group than in the Armour® Thyroid group in the posted summaries, and the unadjusted comparisons point in the same direction. The secondary endpoint is relevant as an earlier treatment-period measure, while the primary endpoint asks whether the response was sustained across both assessment periods. These endpoints are related but not interchangeable; the earlier outcome cannot establish that a response persisted through stabilization.

The secondary result should carry less inferential weight than the primary result. Secondary outcomes can be useful for understanding other aspects of the response pattern, but testing several outcomes increases the chance of apparently notable findings when each is assessed independently. Whether the trial controlled the overall type I error across endpoints, or specified a testing hierarchy, is not reported in the registry record. The unadjusted secondary p-value should therefore be treated as exploratory evidence rather than as a stand-alone confirmatory conclusion.

The registry reports a stratified non-inferiority analysis for the secondary endpoint as well as an unadjusted re-calculation. The non-inferiority margin and detailed analysis plan are not reported, so the sponsor analysis cannot be independently evaluated against the intended criterion. The secondary outcome adds context about response at the end of titration, but it does not change the primary endpoint’s focus on sustained response and does not resolve the missing details about multiplicity or non-inferiority criteria.

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

• Percentage of Participants With a Titration TSH Response

9. Statistical Interpretation

10. Statistical Methods Explained

What does a response percentage represent?

A response percentage is the proportion of participants in a group who meet the endpoint definition. For this primary endpoint, response means having TSH within the stated normal reference range at both the end of titration and the end of stabilization. The reported percentages are 60.76% for Armour® Thyroid and 75.92% for levothyroxine. This binary outcome simplifies a continuous laboratory measure into response or non-response. That can make a group comparison easier to interpret, but it also discards information about how far an individual TSH result lies from the range and about changes that do not alter response classification.

What does the risk difference mean?

The risk difference compares the response proportions by subtracting one group’s proportion from the other’s. It is expressed on the proportion scale: a value of 0 means equal response proportions, while positive or negative values indicate which group has the higher proportion according to the chosen group order. The unadjusted re-calculation uses levothyroxine relative to Armour® Thyroid and gives 0.152, with a 95% confidence interval of 0.046 to 0.259. The direction depends on the order used, so the group order must accompany any interpretation of the sign. The registry’s stratified analysis reports -15.16, which is expressed in the opposite direction. The two results should not be compared by sign alone without attending to group order and analysis method.

What does a confidence interval add?

A confidence interval gives a range of values compatible with the data and the method used to estimate the result. For a response percentage, it describes uncertainty in that group’s estimated response. For a risk difference, it describes uncertainty in the contrast. An interval is not a prediction that a particular proportion of individual outcomes will fall inside its limits. Nor does an interval alone establish clinical importance. Here, the 95% intervals are reported for each group’s primary response estimate and for the re-calculated risk difference. The registry also reports a 95% interval for its stratified non-inferiority analysis.

Why can the simple re-calculation differ from the registry analysis?

A simple unadjusted comparison uses the posted group response summaries without adjustment for other variables. A stratified analysis can account for defined strata when estimating a group contrast. Such methods may produce different estimates or intervals because they use different information or weighting. The registry labels its reported estimate as a stratified response-rate difference, but does not identify the strata or method details. The unadjusted re-calculation is therefore a transparent summary of the posted group rates, not a substitute for the sponsor’s analysis. The discrepancy flag is false, but that does not supply missing methodological details.

How should the p-value be read?

The unadjusted p-value of 0.006 summarizes how incompatible the observed comparison is with a specified null hypothesis of no difference, assuming the statistical model and test are appropriate. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The registry’s reported stratified analysis has no posted p-value. Because multiple endpoints and analyses are reported, a p-value from an individual comparison should also be interpreted in light of the multiplicity plan, which is not reported in the registry record.

What is special about non-inferiority analysis?

A non-inferiority analysis asks whether a treatment’s result is not worse than a comparator by more than a prespecified acceptable margin. The margin is central: an interval must be assessed against it to evaluate the non-inferiority question. The registry labels its stratified analysis as non-inferiority and gives an estimate and 95% confidence interval, but the non-inferiority margin is not reported in the registry record. The record also does not report a p-value for that analysis. The label and interval can be described, but the conclusion under the trial’s own criterion cannot be independently assessed without the margin and relevant analysis details.

Why does over-dispersion matter in general?

General statistical explanation: over-dispersion occurs when observed variability is greater than a model assumes. For a binary response, a model that understates variability can produce confidence intervals that are too narrow and tests that are too optimistic. The registry record does not report an over-dispersion assessment or identify whether it was relevant to the sponsor’s analysis. This concept is a general consideration, not a claim about a problem in this trial.

What does the missing-data description tell us?

The primary population is described as randomized participants who took at least one dose of study intervention. The registry says non-responder imputation incorporating multiple imputation was performed for participants with missing TSH values due to COVID-19, called NRI-C. This indicates that missing values related to COVID-19 were addressed through a named procedure rather than simply omitted, but the record does not provide enough detail to reconstruct that procedure, assess its assumptions, or determine how other missing values were handled. The extent and pattern of missing outcome data are not reported in the registry record.

11. Confidence Intervals and Precision

The group response estimates are 60.76% for Armour® Thyroid, with a 95% confidence interval of 52.6 to 68.92, and 75.92% for levothyroxine, with a 95% confidence interval of 68.65 to 83.19. These intervals communicate uncertainty in each estimated percentage. Their limits are not guarantees about the true value, nor do the intervals describe variation among individual participants.

For the unadjusted risk difference, the estimate is 0.152 and the 95% confidence interval is 0.046 to 0.259, with levothyroxine compared with Armour® Thyroid. The interval lies on the positive side of zero, indicating that the unadjusted estimate is compatible with a higher response proportion for levothyroxine under this group ordering. This is a statement about the posted statistical contrast, not about clinical importance or the non-inferiority criterion. The registry’s separate stratified estimate is -15.16, with a 95% confidence interval from -26.10 to -4.23; it uses the opposite group direction and is labeled as a non-inferiority analysis.

Precision depends on the amount of information, the observed response pattern, and the analysis method. The registry does not report the original precision target or the trial’s sample-size justification. The reported group denominators are 141 for Armour® Thyroid and 143 for levothyroxine. A confidence interval’s width should be interpreted together with the estimand and method: the group-rate intervals and contrast interval answer different questions. The registry does not provide the stratification factors or enough information to recreate the sponsor’s interval. The reported intervals are therefore best read as the stated uncertainty estimates for their respective analyses, rather than as interchangeable measures of precision.

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 for 80% power and participants per arm for 90% power. They are planning illustrations based on the observed effect, not the trial’s original design assumptions. The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect. The trial’s own original design assumptions are not reported in the registry record.

For each stated effect fraction, the 80% power column gives the number of participants per arm indicated by the planning calculation for an 80% chance of detecting an effect of that size under those assumptions. The 90% power column gives the corresponding per-arm planning figure for a 90% chance. These are conditional design calculations, not guarantees about what a future study will find. The displayed planning figures are based on the primary comparison’s observed effect, and the secondary endpoint has its own planning illustration based on that endpoint’s observed effect.

The rows represent planning for 100%, 75%, and 50% of the observed effect. A smaller assumed effect requires more participants per arm to detect it at the same stated power, which is why the planning figures increase as the assumed effect becomes smaller. The accompanying curve illustrates that relationship across assumed effect sizes. These figures are not observed power: they do not estimate the probability that the completed trial detected the effect after observing its results. They are prospective planning illustrations under the listed assumptions, and should not be read as the trial’s actual sample-size justification.

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

Worked calculation for this trial

1. Control proportion p₀ = 0.610; treatment proportion p₁ = 0.762; average p̄ = 0.686.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6563 = 1.2863.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6474 = 0.5449.
4. n = (1.2863 + 0.5449)² / (0.1523)² = 144.6, rounded up to 145 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 <- 0.609929078014184                          # control proportion
p1 <- 0.762237762237762                          # 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(4124705)
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 = 145 (80% power), 194 (90% power)
 75% of observed effect: n per arm = 266 (80% power), 356 (90% power)
 50% of observed effect: n per arm = 617 (80% power), 825 (90% power)
power.prop.test: 144.6 
Simulated power at n = 145 per arm: 0.802 (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.

14. Analyses Reported by the Sponsor

The registry reports a sponsor analysis described as a stratified response-rate difference. For the primary endpoint, the estimate is -15.16, with a 95% confidence interval from -26.10 to -4.23, and the analysis is labeled as non-inferiority. The reported group order is Armour® Thyroid followed by levothyroxine. The registry does not post a p-value for this analysis.

The simple re-calculation from the posted response summaries uses levothyroxine relative to Armour® Thyroid and gives a risk difference of 0.152, with a 95% confidence interval of 0.046 to 0.259 and an unadjusted p-value of 0.006. The estimates use opposite group directions, so their signs differ. The simple calculation is unadjusted; the sponsor’s is identified as stratified. The registry record does not report the stratification factors, detailed method, or non-inferiority margin, so the sponsor estimate cannot be fully reconstructed or assessed against its decision criterion from this record alone.

The discrepancy flag is false, but the two analyses remain distinct because their methods and group ordering differ. The re-calculation is a plain comparison of the posted response summaries and should not be presented as replacing the sponsor’s analysis. The registry record’s non-inferiority label is not by itself enough to establish whether the prespecified criterion was met; interpreting that requires the missing margin and analysis details.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint and one secondary endpoint. The primary endpoint assesses sustained TSH response at the end of titration and the end of stabilization; the secondary endpoint assesses TSH response at the end of titration. Because more than one outcome is reported, interpretation of individual p-values depends in part on whether and how the statistical plan addressed multiplicity. The registry record does not report a multiplicity adjustment, a testing hierarchy, or whether the secondary outcome was intended to be confirmatory. The posted unadjusted p-values should therefore not be treated as though a family-wise testing plan were documented.

The primary population is described as all randomized participants who took at least one dose of study intervention, called the intent-to-treat population in the record. For participants with missing TSH values due to COVID-19, the record states that non-responder imputation incorporating multiple imputation (NRI-C) was performed. It does not describe the imputation model, the assumptions made about missingness, or how the procedure was combined with non-responder imputation. The number and pattern of missing TSH measurements, reasons for missingness other than COVID-19, and handling of those other missing data are not reported in the registry record.

The registry record does not report whether sensitivity analyses assessed the robustness of the primary result to alternative missing-data assumptions. It also does not report a detailed analysis population definition beyond randomized participants who took at least one dose, or the handling of participants who did not complete titration or stabilization. These omissions matter because a percentage response can depend on how missing outcomes and incomplete follow-up are classified or imputed.

The sponsor analysis is labeled as stratified and non-inferiority, but the stratification factors, non-inferiority margin, and full analysis plan are not reported. Interim analyses and their potential effect on inference are also not reported in the registry record. The posted estimates can be described, but these missing elements limit independent evaluation of multiplicity control, missing-data robustness, and the sponsor’s non-inferiority conclusion.

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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: sepsis (1.4%), coronary artery disease (0.7%), small intestinal obstruction (0.7%), non-cardiac chest pain (0.7%), otitis media (0.7%).

The registry reports serious adverse events affecting 4 participants in the Armour® Thyroid group and 7 in the levothyroxine group. The respective numbers at risk were 141 and 143. Other adverse events affected 15 participants in the Armour® Thyroid group and 9 in the levothyroxine group. One death was reported in each group. These counts describe events recorded in the trial; they do not establish that a study treatment caused an event or death.

The listed serious-event terms include sepsis at 1.4%, and coronary artery disease, small intestinal obstruction, non-cardiac chest pain, otitis media, and seizure, each at 0.7%. These are the registry’s listed terms and percentages, not a complete account of every clinical detail. The record does not provide event narratives, timing, severity grades beyond the serious-event classification, or causal assessments in the posted safety summary.

Safety counts should be interpreted alongside the number at risk and the study’s follow-up, which differed in duration according to the titration and stabilization schedule. These figures are descriptive and do not support a comparative safety conclusion on their own. The registry record does not report enough detail to assess exposure-adjusted event rates or to determine whether the observed differences arose by chance.

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

The posted results describe the proportion meeting a defined TSH response criterion in each randomized group. The primary endpoint requires an in-range TSH value at both the end of titration and the end of stabilization. The response percentages were higher for levothyroxine than for Armour® Thyroid, and the unadjusted risk difference is positive when levothyroxine is compared with Armour® Thyroid. The confidence intervals communicate uncertainty around those estimates.

These results do not, by themselves, establish whether any observed difference is clinically important. The registry labels its own stratified analysis as non-inferiority, but the non-inferiority margin is not reported in the registry record. Without that margin and full analysis details, readers cannot determine from the record whether the sponsor’s criterion for non-inferiority was met. The unadjusted re-calculation is not a substitute for that criterion.

The endpoint is a binary summary of TSH values at specified visits. It does not describe every TSH measurement during the study, how far out-of-range values were from the target, or other outcomes not posted in the registry record. The study enrolled participants who were already euthyroid on stable levothyroxine therapy; its findings should not be generalized automatically to people outside that described population. The registry record also does not provide enough detail to assess the effect of missing data beyond its description of NRI-C for COVID-19-related missing TSH values.

The safety summary reports serious and other adverse-event counts and deaths in the two groups. Those counts are descriptive and do not establish causality or comparative safety. Taken together, the registry results support a careful description of observed response and safety data in this trial population, with uncertainty and analysis limitations kept explicit. They do not justify broader claims about treatment effects, patient experience, or outcomes that were not reported.

18. Why This Trial Matters Statistically

This study illustrates why the definition of a clinical endpoint matters. A single in-range TSH result and a sustained response requiring in-range values at both the end of titration and stabilization answer different questions. The primary outcome is designed to represent response across two treatment stages, while the secondary outcome focuses on the end of titration. Reading the outcomes together helps distinguish an early response measure from the sustained-response measure without treating them as interchangeable.

It also demonstrates why the estimand and group order should be stated when reporting a difference. The simple re-calculation compares levothyroxine with Armour® Thyroid, while the sponsor’s reported stratified estimate lists the groups in the opposite order. The resulting signs point in opposite directions numerically, even though the group pattern is consistent when each is interpreted with its stated ordering. Identifying the method as unadjusted or stratified is equally important because estimates and intervals can depend on the analysis approach.

The non-inferiority label makes the missing margin particularly consequential. A non-inferiority result is judged against a prespecified acceptable difference; an estimate and confidence interval alone do not communicate that criterion. The registry record’s omission of the margin means the sponsor’s stated analysis cannot be fully evaluated from the posted details. Similarly, the description of NRI-C identifies an approach to COVID-19-related missing TSH values but does not provide enough detail to assess its assumptions or robustness.

Finally, the planning illustrations distinguish prospective sample-size planning from post hoc descriptions of the completed study. They show how planning requirements change with the effect size assumed, under stated calculation assumptions, and are not observed power. Together, these features underscore the value of transparent endpoint definitions, analysis methods, missing-data plans, and decision thresholds when interpreting clinical trial results.

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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

Several trial-specific details needed for a full appraisal are not reported in the registry record. The non-inferiority margin is absent, despite the sponsor analysis being labeled as non-inferiority. The stratification factors and detailed statistical method are also not reported, limiting independent reconstruction of that analysis. The record does not provide a p-value for the sponsor’s analysis or identify the complete prespecified analysis plan.

The missing-data description is partial. NRI-C is named for TSH values missing due to COVID-19, but the imputation model, assumptions, extent of missingness, and handling of other missing outcomes are not reported. The record does not state whether sensitivity analyses tested alternative missing-data assumptions. Attrition and completion details for the variable titration and stabilization schedules are also not reported.

The original sample-size justification and design assumptions are not reported. The posted planning calculations are illustrations based on observed effects and stated assumptions, not evidence of the trial’s original power target. The registry record also does not report interim analyses or a multiplicity plan. These gaps complicate interpretation of the secondary endpoint’s unadjusted result and of the nominal p-value for the primary re-calculation.

The safety summary is limited to event counts, selected serious-event terms, and deaths. It does not provide event narratives, treatment attribution, or exposure-adjusted rates. The primary outcome itself is a binary TSH response measure and does not capture the magnitude of out-of-range values or outcomes not reported in the registry. These limitations do not invalidate the posted estimates, but they constrain the questions those estimates can answer.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of Armour® Thyroid Compared to Synthetic T4 (Levothyroxine) in Previously Hypothyroid Participants — NCT04124705.
• Registry record last updated 2024-06-05; status last verified 2024-05.
• Results first posted 2024-06-05.
• 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

NCT04124705 is a completed, randomized, double-blind, parallel-group Phase 2 study sponsored by Allergan. It enrolled 284 previously hypothyroid participants who were euthyroid on T4 replacement therapy and compared Armour® Thyroid with levothyroxine. The primary endpoint was sustained TSH response, defined by TSH within the normal reference range at both the end of titration and the end of stabilization. Posted response percentages were 60.76% for Armour® Thyroid and 75.92% for levothyroxine.

The unadjusted re-calculation gives a risk difference of 0.152 for levothyroxine relative to Armour® Thyroid, with a 95% confidence interval of 0.046 to 0.259. The registry also reports a stratified analysis labeled as non-inferiority, but its margin and key method details are not reported in the registry record. NRI-C is described for COVID-19-related missing TSH values. Safety counts are descriptive; they do not establish causality. Interpretation is limited by missing analysis, multiplicity, and missing-data details.