American Trial Using Tranexamic Acid in Thrombocytopenia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for American Trial Using Tranexamic Acid in Thrombocytopenia (NCT02578901), a randomized phase 3 study. The primary outcome is Bleeding Within 30 Days.

2. Clinical Question

A-TREAT asked whether tranexamic acid (TXA), compared with placebo, was associated with less bleeding among patients with thrombocytopenia related to primary bone marrow disorders or to chemotherapy, immunotherapy and/or radiation therapy. The primary question concerned the proportion of participants with bleeding of WHO grade 2 or above during the 30 days after activation of study drug.

The trial compared TXA with placebo, administered intravenously or orally after participants met the inclusion and exclusion criteria. The registry describes the placebo as intravenous normal saline or oral placebo pills. The comparison is therefore between assignment to the active study drug and assignment to a placebo, not between different doses or treatment schedules.

The registry also posts outcomes for platelet transfusions and days alive and without WHO grade 2 bleeding. These offer additional descriptions of the study period, but they do not replace the prespecified primary outcome as the main clinical question presented in the record.

3. Trial Design

The American Trial Using Tranexamic Acid in Thrombocytopenia (A-TREAT; NCT02578901) is recorded as a phase 3, randomized, quadruple-masked, parallel-group trial. The University of Washington is the sponsor. The registry lists an enrollment of 330, a start date of 2016-06, primary completion in 2020-03, and a completed status.

In a randomized trial, allocation is intended to make treatment groups comparable on average at baseline. Masking is intended to limit the influence of knowing the assigned intervention on trial conduct or outcome assessment. The registry labels masking as quadruple but does not identify the four roles covered by that label in the record. The parallel model means the record describes separate groups assigned to TXA or placebo rather than a crossover comparison.

The primary outcome is recorded for 145 participants in the TXA group and 144 in the placebo group. The secondary outcome records use different group sizes: 165 in each group for platelet transfusions, and 136 for TXA and 132 for placebo for days alive and without WHO grade 2 bleeding. The registry does not report why the analyzed or measured group sizes differ across outcomes, nor does it identify an analysis population for each outcome.

The record lists two interventions, TXA and placebo, and gives intravenous or oral administration descriptions. It does not report the original sample-size justification, the intended enrollment rationale, or the trial’s own statistical analysis plan. Those details cannot be inferred from the randomized design or the posted outcomes.

4. Participant Flow and Accounting

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

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Bleeding Within 30 Days, measured over 30 days after activation of study drug.

The primary endpoint is the number and proportion of participants with bleeding of WHO grade 2 or above over the 30 days after activation of study drug. The registry describes the outcome as a participant count and specifies the time window. In the posted results, 73 of 145 participants in the TXA group and 78 of 144 in the placebo group had the event.

A binary endpoint classifies each participant according to whether the defined event occurred during the specified period. It captures the occurrence of bleeding at or above the stated grade threshold, rather than the number of bleeding episodes, their duration, or a continuous measure of bleeding severity. These are distinct aspects of clinical experience and should not be treated as interchangeable.

As a general statistical consideration, an event proportion is useful for comparing how often a defined outcome occurs in each randomized group. The time window and grade threshold make the comparison interpretable, while the participant-level definition avoids counting multiple episodes as multiple participants. The endpoint is relevant to the stated prevention question because it directly records whether qualifying bleeding occurred during follow-up.

The registry does not report additional detail about how bleeding was assessed, how events were adjudicated, or how missing endpoint observations were handled. Those features therefore cannot be used to further interpret the outcome from the registry record alone.

7. Results Posted in the Registry

• Tranexamic Acid (TXA): 50.3% (73/145) versus 54.2% (78/144) on control; risk difference -3.8 percentage points (95% CI -15.3 to 7.7), risk ratio 0.93 (0.75–1.16); unadjusted two-proportion z-test p = 0.515.

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.

Number of Platelet Transfusions

• 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 deviations.
• Result:
  • Tranexamic Acid (TXA): mean 7.70 versus 7.60 on control; difference 0.10 (95% CI -1.93 to 2.13), standardized effect d = 0.01; z-test p = 0.923.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Tranexamic Acid (TXA) and the control. The interval gives the range of effects that remain compatible with the data.

Number of Days Alive and Without WHO Grade 2 Bleeding

• 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 deviations.
• Result:
  • Tranexamic Acid (TXA): mean 28.10 versus 27.70 on control; difference 0.40 (95% CI -0.61 to 1.41), standardized effect d = 0.09; z-test p = 0.440.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Tranexamic Acid (TXA) and the control. The interval gives the range of effects that remain compatible with the data.

The two secondary outcomes offer complementary descriptions of the same 30-day period: platelet transfusion use and days alive without WHO grade 2 bleeding. Across these measures, the posted group averages are close, and the unadjusted intervals include no difference. This broad pattern does not suggest a clear separation between groups on the secondary outcomes.

These outcomes relate to the primary endpoint but do not measure precisely the same thing. Platelet transfusions describe supportive treatment use; days alive and without qualifying bleeding combine survival and bleeding-free time as defined in the outcome title. A result on one measure cannot automatically be substituted for a result on the primary measure of whether a participant had qualifying bleeding. Differences in outcome definition, measurement, and available participant counts also matter when comparing results across outcomes.

The secondary outcomes should be interpreted with less weight than the primary comparison. The registry identifies the bleeding outcome as primary and lists no reported analyses for the secondary outcomes. It also does not report whether the trial used a multiplicity plan. When several outcomes are examined, the chance of finding an apparently notable result by chance can increase; as a general statistical principle, adjustments or a prespecified hierarchy can help account for this. No such approach is documented in the registry record.

In this case, the unadjusted secondary comparisons do not show clear evidence of differences, but their interpretation remains limited by the absence of reported analysis methods and the different participant counts across outcomes. They can be read as additional descriptive context, not as independent confirmation of the primary result and not as proof that the outcomes are equivalent between groups.

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

• Number of Platelet Transfusions
• Number of Days Alive and Without WHO Grade 2 Bleeding

9. Statistical Interpretation

10. Statistical Methods Explained

For the primary binary outcome, the displayed comparison is a risk difference: the event proportion in the TXA group minus the event proportion in the placebo group. A negative value therefore points toward a lower observed event proportion in the TXA group, while a positive value points toward a higher one. The observed counts were 73 of 145 for TXA and 78 of 144 for placebo. The unadjusted p-value is 0.515.

As a general statistical explanation, a risk difference expresses an absolute contrast in event probability. It differs from a risk ratio, which expresses a relative contrast, and an odds ratio, which compares event odds. Each measure answers the comparison on a different scale. A ratio can be difficult to interpret when considered alone; an absolute difference can make the scale of the group contrast more direct. The registry record does not report a sponsor analysis using these effect measures, so the posted comparison should not be mistaken for a complete account of the trial’s planned analysis.

A confidence interval represents a range of values reasonably compatible with the observed data under the statistical method used. For the primary risk difference, the interval crosses zero. Zero represents no difference on the risk-difference scale. Thus, the interval does not rule out no difference and also includes values in either direction. A p-value and confidence interval are related summaries of uncertainty, but neither describes the probability that a specific treatment effect is true.

For continuous outcomes, the reported effect measure is a mean difference. The platelet-transfusion means are 7.7 for TXA and 7.6 for placebo, producing a reported mean difference of 0.1. The standard deviations are 8.7 and 10.1, respectively. The mean difference interval runs from -1.934 to 2.134, and the p-value is 0.923. For days alive and without WHO grade 2 bleeding, the means are 28.1 and 27.7, with a reported mean difference of 0.4. The interval includes zero, and the p-value is 0.44.

As a general statistical explanation, a mean summarizes the center of a continuous outcome, while a standard deviation describes the spread of individual observations around the mean. A mean difference summarizes separation between group averages, not how much individual participants overlap or how outcomes vary within a group. Here, the standard deviations for platelet transfusions are large relative to the small difference in means. This is a descriptive observation, not a separate test of whether the groups differ.

Over-dispersion is a general concern when count outcomes vary more than a simple statistical model assumes. Greater variability can make estimates less precise and can make methods based on restrictive variance assumptions unsuitable. The posted platelet-transfusion summaries report standard deviations of 8.7 and 10.1, but the registry record does not describe a count model or a specific method for handling over-dispersion. No model choice should therefore be attributed to the trial from these summaries alone.

The mean-difference calculations are based on posted group summaries and are unadjusted recalculations. The sample-size illustrations use an assumed normal approximation and equal allocation, as described in the planning assumptions. They are not results of the original trial’s own design analysis. No discrepancy is flagged between the posted summaries and the recalculated comparisons.

The primary outcome is binary and reported as participant counts; the secondary outcomes are continuous and reported as means and standard deviations. Different outcome types call for different effect measures and modeling choices in a full analysis. For example, a binary outcome may be analyzed with a method tailored to event probabilities, while a continuous outcome may be analyzed through a mean contrast. The registry record does not report the full statistical methods used by the sponsor, so the displayed unadjusted summaries should be interpreted as descriptive comparisons rather than a substitute for an undocumented analysis plan.

The record lists three outcome measures, including the primary outcome and two secondary outcomes, but does not report analyses for the secondary outcomes. It also does not report whether covariates, stratification factors, repeated measurements, or other design features were included in the sponsor’s analysis. Statistical interpretation should consequently stay close to the posted event counts, means, intervals, and p-values, without assuming unreported adjustments or models.

11. Confidence Intervals and Precision

Precision concerns how tightly the data estimate a group contrast. The primary risk-difference interval crosses zero, so the observed results are compatible with effects on either side of no difference. Its span is substantial relative to the point estimate, indicating that the primary comparison does not precisely determine the direction or magnitude of the underlying contrast.

For the platelet-transfusion mean difference, the reported interval extends from -1.934 to 2.134. It includes zero and covers both negative and positive contrasts. The observed mean difference is 0.1, and the p-value is 0.923. For days alive and without WHO grade 2 bleeding, the mean difference is 0.4; its interval also includes zero, and the p-value is 0.44. Neither secondary interval establishes a clear separation between groups.

As a general statistical explanation, a confidence interval is not a range containing a fixed proportion of individual outcomes. It is an uncertainty range for an estimated contrast, constructed according to a statistical method. A wide interval means the data leave more uncertainty about the contrast; a narrow interval means they constrain it more tightly. The interval’s inclusion of zero is not proof of equivalence, because equivalence requires an appropriate margin and analysis, neither of which is reported in the registry record.

Precision also depends on the number of participants with outcome data and on outcome variability. The outcome-specific group counts differ, and the continuous platelet-transfusion outcome shows substantial spread. The registry record does not report reasons for the different counts or a missing-data analysis. These limits make it important to interpret the uncertainty intervals alongside the reported sample sizes and the incomplete information about analysis populations.

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 illustrative numbers of participants per arm, not total events. They are based on the observed effect and show the group size that would be needed under the stated planning assumptions to detect a chosen fraction of that effect. For each outcome, the figures are presented for planning for 100%, 75%, or 50% of the observed effect.

The planning 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. These are planning assumptions for the illustration; the original trial’s design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the estimated participants per arm for a design with an 80% chance of detecting the specified effect if that effect were the true one and the assumptions held. The 90% power column gives the corresponding participants per arm for a design with a 90% chance. Power describes a design’s detection probability under assumptions; it is not a measure of whether the observed result is reliable.

Smaller target effects require more participants because they are harder to distinguish from random variation. That is why planning for 75% or 50% of the observed effect yields larger per-arm figures than planning for the full observed effect. These figures are not observed power.

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 = 0.542; treatment proportion p₁ = 0.503; average p̄ = 0.523.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.7064 = 1.3845.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.7059 = 0.5941.
4. n = (1.3845 + 0.5941)² / (-0.0382)² = 2,680.1, rounded up to 2,681 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 <- 0.541666666666667                          # control proportion
p1 <- 0.503448275862069                          # 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(2578901)
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 = 2,681 (80% power), 3,588 (90% power)
 75% of observed effect: n per arm = 4,762 (80% power), 6,374 (90% power)
 50% of observed effect: n per arm = 10,702 (80% power), 14,326 (90% power)
power.prop.test: 2680.1 
Simulated power at n = 2,681 per arm: 0.804 (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. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary bleeding outcome and two secondary outcomes. It does not report a multiplicity adjustment, a hierarchy for testing outcomes, or a statistical analysis plan explaining how the several outcomes were to be interpreted together. The record also lists no reported analyses for the secondary outcomes. Consequently, the posted secondary comparisons are unadjusted and should be treated as exploratory context rather than as a coordinated family of confirmatory tests.

As a general statistical principle, testing multiple outcomes can increase the chance of at least one apparently unusual result even when there is no true effect. A prespecified testing sequence or an adjustment can control that concern, but the registry record does not establish whether such a plan existed or was used. The absence of a report in the registry is not evidence that no analysis was performed; it means the approach is not reported in the registry record.

The primary and secondary outcome counts are not identical. The record does not explain the differences, specify whether the analyses followed an intention-to-treat population, or describe exclusions after randomization. It also does not report the number or handling of missing outcome observations, imputation methods, or sensitivity analyses. As a result, it is not possible to determine from the record how missing data may have affected the comparisons.

The registry record does not report whether the primary comparison was adjusted for baseline variables, stratification factors, or other covariates. Nor does it document interim analyses, stopping rules, or the trial’s original sample-size justification. These are distinct parts of a full statistical analysis plan; their absence from the registry record limits assessment of whether the posted summaries align with a prespecified analysis, but does not establish what happened outside the record.

15. 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: febrile neutropenia (24.5%), nausea (16.6%), other infection (13.5%), other immune disorder (11.0%), fatigue (10.4%).

The registry’s serious adverse event figures list 163 participants at risk in each group, with 123 affected in the TXA group and 110 in the placebo group. The other adverse event figures list 163 at risk in each group, with 161 affected in each group. The death count is 19 in each group. These are the posted group-level counts and should be read as safety summaries, not as estimates of a causal treatment effect.

The record also lists serious adverse event terms, including febrile neutropenia, nausea, other infection, other immune disorder, fatigue, and headache. The percentages associated with those terms describe reported terms in the registry record; they do not by themselves establish that an event was caused by study drug.

The safety counts use denominators of 163 per group, which differ from the primary outcome group counts. The registry record does not explain the relationship between these safety populations and the primary analysis populations, or provide further details about event ascertainment, severity grading, timing, or attribution. Without those details and an appropriate comparative analysis, the counts should not be read as evidence of a treatment effect or as a complete account of safety.

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

The posted results show that qualifying bleeding occurred in both randomized groups during the 30-day outcome period. The event count was lower in the TXA group than in the placebo group, but the unadjusted primary comparison has a p-value of 0.515 and an interval that includes no difference. The data therefore do not establish a clear difference between groups, and they do not establish that the groups are equivalent.

The secondary summaries likewise do not show clear separation in the posted unadjusted comparisons. They measure different aspects of the study period and have different outcome-specific group counts, so they should not be collapsed into a single conclusion. Their results provide additional context, but the registry lists no reported secondary analyses and does not document a multiplicity approach.

As a general statistical explanation, failing to find clear evidence of a difference is not the same as demonstrating that any difference is too small to matter. A conclusion of equivalence or non-inferiority would require a specified margin and an analysis designed for that purpose. Neither is reported in the registry record. The same caution applies to interpreting a nonsignificant p-value as proof of no effect.

The results are limited to the population and outcomes described in the record. The registry does not report sufficient detail about missing data, analysis populations, outcome assessment, or the sponsor’s full statistical analysis to resolve those uncertainties. These results should therefore be understood as registry-posted summaries rather than a complete account of every aspect of the trial.

No treatment recommendation follows from these statistical summaries. They describe the estimated comparisons and their uncertainty, not what an individual patient should do or whether an intervention should be used in a particular clinical setting.

17. Why This Trial Matters Statistically

This trial is statistically instructive because its randomized, masked, parallel-group design is paired with a primary binary outcome and two continuous secondary outcomes. That combination illustrates why effect measures must match outcome type: the primary comparison is expressed as a risk difference, while the secondary comparisons use mean differences.

The primary event counts alone might invite a simple directional reading, but the uncertainty interval and p-value show why counts are not enough. The interval crosses no difference, and the unadjusted p-value is 0.515. Statistical interpretation therefore requires attention to both the observed contrast and the range of contrasts compatible with the data.

The secondary outcomes also demonstrate that similar group averages do not automatically mean precise estimates. Their intervals include zero, and the platelet-transfusion outcome has notable variability in both groups. Sample-size planning illustrations further show a general design principle: smaller effects require larger samples to detect under the same assumptions. The illustrations are distinct from the trial’s own design rationale, which is not reported in the registry record.

Finally, the record highlights the importance of transparent reporting. Participant counts vary by outcome, no secondary analyses are reported, and details about missing data, analysis populations, multiplicity, and the sponsor’s analysis plan are absent. Those omissions do not invalidate the posted results, but they constrain the strength and scope of conclusions that can be drawn from the registry record.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several trial-specific limitations arise from the information available in the registry record. The primary endpoint group counts are 145 for TXA and 144 for placebo, while the secondary outcome counts vary, including 165 in each group for platelet transfusions and 136 and 132 for days alive and without bleeding. The record does not explain these differences or state how missing outcomes and analysis populations were handled.

The registry gives a primary outcome definition and timeframe, but does not report further detail about bleeding assessment or adjudication. It also does not report the sponsor’s full statistical analysis, including any covariate adjustment, multiplicity approach, or sensitivity analyses. There are no reported analyses listed for the secondary outcomes. Consequently, the posted unadjusted summaries cannot answer whether other prespecified analyses would yield different estimates or conclusions.

The record does not provide the trial’s original sample-size justification or statistical design assumptions. The sample-size figures associated with the posted results are planning illustrations based on observed effects and stated assumptions, not evidence of the trial’s planned detection capability.

Safety counts are reported with 163 participants at risk per group, which differs from the primary outcome group counts. The registry does not explain this population difference or provide further context on attribution and event assessment. Finally, the results address the described trial population and outcomes; the registry record alone does not establish how well they generalize beyond that setting.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: American Trial Using Tranexamic Acid in Thrombocytopenia — NCT02578901.
• Registry record last updated 2021-03-24; status last verified 2021-02.
• Results first posted 2021-03-24.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

A-TREAT (NCT02578901) is recorded as a completed, phase 3, randomized, quadruple-masked, parallel-group trial sponsored by the University of Washington. It compared tranexamic acid with placebo in patients with thrombocytopenia related to primary bone marrow disorders or cancer treatment, with a registry enrollment of 330.

The primary outcome was WHO grade 2 or above bleeding within 30 days after activation of study drug. The posted event counts were 73 of 145 for TXA and 78 of 144 for placebo; the unadjusted p-value was 0.515, and the confidence interval for the risk difference includes zero. The two secondary outcomes also have unadjusted intervals that include no difference.

The registry record does not report the full statistical analysis plan, missing-data handling, analysis populations, or multiplicity approach. The posted results therefore support a cautious description of the observed comparisons, not a claim of equivalence or a treatment recommendation.