Efficacy and Safety Study of Eltrombopag in Pediatric Patients With Thrombocytopenia From Chronic Idiopathic Thrombocytopenic Purpura (ITP): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety Study of Eltrombopag in Pediatric Patients With Thrombocytopenia From Chronic Idiopathic Thrombocytopenic Purpura (ITP) (NCT00908037), a randomized phase 2 study. The primary outcome is Percentage of Participants Achieving a Platelet Count >=50 Giga Cells Per Liter (Gi/L) at Least Once, Between Day 8 and Day 43 (Weeks 1 to 6) of the Randomized Period of the Study (Part 2).

2. Clinical Question

The clinical question combines a treatment comparison with an age-specific design. Randomization and placebo control were used in Part 2, while the staggered sequence allowed safety, pharmacokinetic, and platelet-count information to be reviewed before moving to younger cohorts. The record describes a double-blind randomized study overall, but also makes clear that the initial dose-finding period was open label. The main posted endpoint concerns the randomized period, so results from the open-label periods answer different questions and should not be treated as randomized comparisons.

3. Trial Design

This was a phase II, multicenter, three-part study with randomized allocation and a parallel-group model. The registry reports enrollment of 82 and a completed status. The study start is listed as 2009-09-30, and primary completion as 2014-02-01. The official description identifies the study as open label and double blind, randomized, and placebo controlled; the study summary clarifies that openness and blinding differed by part.

In Part 1, eltrombopag began in an open-label fashion in an initial group of five participants in the oldest age cohort. The study description says safety, pharmacokinetic, and platelet-count data were reviewed regularly. If the 12-week review identified no safety concerns, the plan was to randomize additional participants to placebo or eltrombopag. After seven weeks of randomized treatment, all participants were to receive eltrombopag open label. The described total duration of eltrombopag treatment was 24 weeks.

The staggered approach was also applied across the age cohorts. The next younger cohort was to begin after data from the preceding cohort had been evaluated, initially with five participants and followed by a safety review before further enrollment and randomization. The summary says a Data Safety Monitoring Board would review safety data. These details describe the planned design; whether every step occurred exactly as planned is not reported in the registry record.

The randomized primary endpoint window ran from Day 8 through Day 43 of Part 2. Posted primary results are split by cohort and treatment group. Part 2/3 also included an open-label eltrombopag period, and the registry lists further platelet and pharmacokinetic outcomes. Because the open-label periods did not retain a concurrent randomized placebo comparison for those outcomes, they should be interpreted as descriptive evidence about observed responses during eltrombopag treatment rather than as controlled treatment effects.

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: 15 par. were randomized to a 24-Week (Wk) Open-Label (OL) eltrombopag Dose-Finding period (pd) (Part 1) then did not continue. 67 par. were randomized to a 7-Wk Double-Blind placebo-controlled pd (Part 2), followed by a 24-Wk OL eltrombopag-only pd (Part 2/3) and a 4-Wk Follow-Up pd.

Only the first 6 of 12 groups are shown.

Part1 (24 Week Dose-Finding Period)

Part 2 (7 Week Randomized Period)

Reasons for not completing this period:

Part 2/3(Eltrombopag Open-Label Period)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat (ITT) Population: all enrolled participants during Part 2. The ITT Population was the primary population used for assessing efficacy. Only evaluable participants were considered for analysis where participants with a Baseline platelet count \>10Gi/L was considered as evaluable.

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.

Only the first 6 of 8 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants Achieving a Platelet Count >=50 Giga Cells Per Liter (Gi/L) at Least Once, Between Day 8 and Day 43 (Weeks 1 to 6) of the Randomized Period of the Study (Part 2), measured over From Day 8 up to Day 43 of Part 2.

The primary endpoint is the percentage of participants who achieved a platelet count of at least 50 Gi/L at least once from Day 8 through Day 43 of the randomized period, in the absence of rescue treatment. The registry describes the primary efficacy population as participants enrolled during Part 2, with evaluable participants considered for analysis. A participant was considered evaluable when the baseline platelet count was above 10 Gi/L.

This binary endpoint reduces each participant’s platelet measurements in the specified window to whether the threshold was reached at least once. That can make a clinical response easier to describe than a sequence of repeated measurements, but it does not indicate how long the count remained above the threshold, how often it was above it, or the size of any change from baseline. Those are different questions, addressed in part by other listed outcomes.

Excluding responses after rescue treatment helps distinguish the measured platelet response from one occurring after rescue intervention. The endpoint nevertheless reflects a specific threshold, time window, and analysis population. It should not be read as a complete summary of a participant’s course or as a measure of every outcome important in ITP. The registry description says a 95% confidence interval was calculated by the exact binomial method. Numerical confidence limits for the primary groups are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What a percentage represents. For a binary endpoint, each participant is classified according to whether the defined event occurred. The reported percentage is the proportion meeting that definition in the analyzed group. Here, the event for the primary outcome is reaching a platelet count of at least 50 Gi/L at least once during the stated window without rescue treatment. A percentage is straightforward to read, but its precision depends on the number of observations and the underlying event frequency. With small groups, one participant can materially change the proportion.

Confidence intervals. A confidence interval describes uncertainty around an estimate under the statistical method used. A wider interval indicates less precision; an interval for a treatment contrast that includes no difference is compatible with a range of effects that includes no difference. The primary endpoint description says a 95% confidence interval was calculated using the exact binomial method. However, numerical primary confidence limits are not reported in the registry record, and the posted primary results do not give a confidence interval for a between-group contrast.

The exact binomial approach is designed for estimating a proportion when outcomes are binary, including settings where large-sample approximations may be unreliable. An interval around a single group’s proportion is not the same as an interval for the difference between treatment groups. Keeping those estimands distinct matters: the former describes uncertainty about a group response rate, while the latter describes uncertainty about a contrast.

Randomized comparison and interpretation. Random assignment in Part 2 is intended to balance, on average, factors that could affect outcomes, making a between-group comparison more informative about treatment than a comparison of nonrandomized groups. Blinding can also reduce some sources of bias. These protections apply to the randomized period and do not automatically extend to open-label periods. They also do not remove uncertainty caused by small group sizes, missing outcomes, departures from the analysis plan, or selective reporting.

The primary result is presented as cohort-specific percentages rather than as a single model-based or pooled treatment contrast. The registry does not report a primary contrast estimate, its standard error, a primary p-value, or an analysis model. Those details are not reported in the registry record. Accordingly, the percentages can be described as posted, but a formal inferential conclusion for the primary comparison cannot be reconstructed from the record’s primary results alone.

Unadjusted secondary comparisons. The posted secondary comparisons include risk differences and confidence intervals, along with unadjusted p-values. A risk difference expresses the difference between two group proportions on the absolute scale. A p-value measures how incompatible the observed data are with a specified null model, assuming that model and the analysis conditions apply. It does not give the probability that the null hypothesis is true, nor does it measure the size or importance of an effect. An unadjusted p-value does not account for the fact that multiple outcomes or comparisons may have been examined.

One secondary randomized-period comparison reports an unadjusted p-value of 0.001. That result concerns the outcome defined by reaching the platelet threshold in at least 60% of assessments during its specified window; it is not the primary endpoint’s “at least once” definition. The distinction between endpoints is important because changing the response rule changes the clinical and statistical question. A result for one definition should not be transferred to another.

For the open-label secondary comparisons, the record reports risk differences and intervals, and also reports risk ratios and odds ratios for the cohort contrasts. These measures describe relative as well as absolute associations, but their uncertainty is substantial in small groups. Intervals that span the null value do not establish a difference. Since these are comparisons across open-label cohorts rather than randomized treatment groups, they also do not isolate a causal treatment effect.

Planning illustrations. The sample-size illustrations are based on observed secondary effects and are not the trial’s design justification. The stated assumptions are a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the effect to detect set as a stated fraction of the observed effect. Such calculations illustrate how a future study’s required participants per arm could vary with the effect targeted. They are not observed power, do not show that the current study had a particular probability of detecting an effect, and should not be treated as a recommendation without a prospective design and clinically justified target effect.

Multiplicity and missingness. When many endpoints and comparisons are examined, the chance of at least one apparently unusual result can increase. Adjustment methods or a prespecified hierarchy can address that issue, but multiplicity procedures are not reported in the registry record. The record also does not report the handling of missing endpoint data, imputation, sensitivity analyses, or a detailed analysis plan. These omissions limit evaluation of how robust the posted results are to analytic choices and incomplete observations.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Part 2 (Randomized Period) Cohort 1-Placebo", "Part 2 (Randomized Period) Cohort 1- Eltrombopag", "Part 2 (Randomized Period) Cohort 2-Placebo", "Part 2 (Randomized Period) Cohort 2-Eltrombopag", "Part 2 (Randomized Period) Cohort 3-Placebo", "Part 2 (Randomized Period) Cohort 3- Eltrombopag")
n      <- c(8, 16, 9, 17, 5, 10)
events <- c(0, 10, 3, 11, 4, 6)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Part 2 (Randomized Period) Cohort 1-Placebo  8  0  0.0  0.0  36.9  0.0  32.4  0.3  33.6
ARM  Part 2 (Randomized Period) Cohort 1- Eltrombopag  16  10  62.5  35.4  84.8  38.6  81.5  38.3  81.6
ARM  Part 2 (Randomized Period) Cohort 2-Placebo  9  3  33.3  7.5  70.1  12.1  64.6  12.2  65.2
ARM  Part 2 (Randomized Period) Cohort 2-Eltrombopag  17  11  64.7  38.3  85.8  41.3  82.7  41.0  82.7
ARM  Part 2 (Randomized Period) Cohort 3-Placebo  5  4  80.0  28.4  99.5  37.6  96.4  35.9  95.7
ARM  Part 2 (Randomized Period) Cohort 3- Eltrombopag  10  6  60.0  26.2  87.8  31.3  83.2  30.8  83.3

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 40 outcome measures and reports 39 secondary outcomes. Examining many outcomes creates a multiplicity concern: even when no true effect exists for any one outcome, repeated testing can produce an apparently compelling result by chance. A prespecified primary endpoint, a defined hierarchy of secondary endpoints, or a multiplicity adjustment can help control this issue. The registry record does not report such a hierarchy or an adjustment procedure, so unadjusted secondary p-values should be interpreted cautiously.

The record identifies an Intent-to-Treat population for the primary efficacy assessment as all enrolled participants during Part 2, but also states that only evaluable participants were considered for analysis. Evaluable participants were those with a baseline platelet count above 10 Gi/L. The relationship between this stated population and the analysis denominators shown for each cohort is not further explained. The exact handling of participants who were not evaluable, had missing platelet measurements, or received rescue treatment is not reported in the registry record beyond the endpoint’s exclusion of responses in the presence of rescue treatment.

Methods for handling missing data, sensitivity analyses, protocol deviations, and analysis-set exclusions are not reported in the registry record. These details matter because a binary response outcome can be affected by whether participants without an observed measurement are excluded, treated as nonresponders, or handled through another method. The record does not provide enough information to determine how alternative missing-data assumptions would change the estimates.

The registry describes an exact binomial confidence interval for the primary endpoint, but numerical primary intervals and a primary between-group inferential analysis are not posted. Formal analysis plans, interim analyses, and sample-size justification are not reported in the registry record. The described safety reviews are part of the staggered study plan, but they do not establish that interim efficacy testing occurred. These reporting gaps limit assessment of prespecification and the robustness of the efficacy conclusions.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: neutropenia (6.7%), epistaxis (6.7%), impetigo (4.8%), varicella (4.8%), conjunctivitis (4.8%).

The adverse-event results are reported by study period, not as one simple randomized comparison. In Part 1, 2 participants had serious adverse events among 15 at risk, and 15 had other adverse events. In the randomized Part 2 placebo group, 2 had serious adverse events among 21 at risk, and 20 had other adverse events. In the randomized Part 2 eltrombopag group, 4 had serious adverse events among 44 at risk, and 36 had other adverse events. During the Part 2/3 open-label eltrombopag period, 8 had serious adverse events among 65 at risk, and 62 had other adverse events.

These counts should be read in light of the differing periods, exposure, and group sizes. The open-label group is not a concurrent randomized comparator, and participants may have had different treatment histories across study parts. The counts alone do not establish that treatment caused an event or that one period was safer than another. Death counts are not reported in the registry record. The listed serious-event terms and percentages are descriptive; the record does not provide enough detail to assign them to a particular treatment group or to infer comparative risk.

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

The posted primary results describe how often participants in each randomized cohort reached the platelet threshold at least once during the defined window without rescue treatment. The observed percentages favor eltrombopag in two cohorts and placebo in the youngest cohort. This is a mixed cohort-specific pattern, not a uniform result across all ages. The group sizes are limited, particularly in the youngest cohort, and numerical primary confidence intervals and a formal primary contrast are not reported in the registry record.

These results do not establish that eltrombopag works or does not work overall. The record does not report a pooled primary effect, a treatment-by-age analysis, or a primary p-value. The endpoint is a specific platelet-count response measure and does not capture duration of response, all dimensions of ITP, or all potential benefits and harms. It should not be read as an individual prediction.

The randomized secondary result offers additional evidence for a different platelet-response definition, but it is unadjusted and one among many listed outcomes. The open-label outcomes add descriptive information during eltrombopag exposure but do not provide a controlled comparison. Safety counts likewise describe events by period and cannot, on their own, establish comparative causation.

The registry record is useful for understanding the study’s structure, endpoint definitions, group-level results, and reported adverse events. Its limitations include incomplete reporting of primary uncertainty and inferential testing, unclear details on missing data, and no reported multiplicity strategy. The appropriate conclusion is therefore bounded: the posted findings describe observed response and safety events in this study, while important uncertainties remain about precision, analysis choices, and how broadly the results can be generalized.

14. Why This Trial Matters Statistically

This trial raises several statistical issues that matter in pediatric studies with age-defined cohorts. The staggered design allowed safety and pharmacokinetic information to be reviewed before moving to younger groups, while randomization and placebo control were used in Part 2. That structure is important to understand because results from open-label and randomized periods answer different questions. Combining them without distinction would obscure whether a comparison is controlled.

The primary outcome is binary and measured over a defined interval. That makes the result readily interpretable as a proportion, but the information is limited when groups are small and event rates vary across cohorts. Cohort-specific reporting preserves age-group detail, yet it also means that any overall claim requires a suitable pooled analysis and consideration of whether treatment effects differ by age. Neither is reported in the registry record.

The study also illustrates why endpoint definitions and analysis populations need to be explicit. “At least once” is not the same as response at repeated assessments, and a response without rescue treatment is not equivalent to any platelet-threshold crossing. Likewise, the stated Intent-to-Treat population is qualified by an evaluability condition. Without complete detail on missing observations and exclusions, readers cannot fully assess the consequences of those choices.

Finally, the secondary findings and sample-size illustrations demonstrate the distinction between evidence generation and planning. Unadjusted secondary results can be informative, but multiplicity weakens the strength of isolated signals. Sample-size calculations based on observed effects are planning exercises, not observed power. Clear separation of these concepts helps prevent descriptive or exploratory results from being overstated.

15. 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 primary results are presented as cohort-specific percentages without numerical confidence intervals or a reported between-group effect estimate. The registry describes an exact binomial interval calculation, but the primary interval values are not reported in the registry record. A primary p-value and formal comparison method are also not reported, which limits assessment of the precision and inferential strength of the main result.

The cohort-specific estimates are based on small treatment groups, most notably in the youngest cohort. Percentages from small groups are sensitive to individual outcomes, and the posted results vary in direction across cohorts. A pooled estimate or formal assessment of variation by age is not reported. The trial’s staggered enrollment also means that age cohorts entered after prior cohort data had been evaluated; the consequences of that sequencing for comparisons are not quantified in the registry record.

The primary analysis description combines an Intent-to-Treat population with an evaluability requirement based on baseline platelet count. Further information about exclusions, missing data, rescue treatment handling, and sensitivity analyses is not reported. The record also does not report multiplicity control despite many listed outcomes, nor does it provide a detailed statistical analysis plan or sample-size justification.

Several longer-term secondary outcomes are measured during open-label treatment. These are descriptive and cannot establish comparative effects against placebo. The safety figures are organized by period, with different exposure and treatment histories, so direct comparisons across those periods may be misleading. These limitations are specific to interpreting the posted record and do not, by themselves, determine what additional analyses or data might show.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Efficacy and Safety Study of Eltrombopag in Pediatric Patients With Thrombocytopenia From Chronic Idiopathic Thrombocytopenic Purpura (ITP) — NCT00908037.
• Registry record last updated 2018-10-12; status last verified 2015-03.
• Results first posted 2014-10-31.
• 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.

19. Record Summary

The registry record describes a completed phase II randomized, placebo-controlled study of eltrombopag in previously treated pediatric patients with chronic ITP. It enrolled 82 participants and used three age-determined cohorts, with an open-label dose-finding period, a randomized placebo-controlled period, and a later open-label period. The primary endpoint was platelet-count threshold attainment at least once during the randomized window without rescue treatment.

Primary percentages are reported separately by cohort and treatment group, with a mixed pattern across ages. The registry does not report numerical primary confidence intervals or a formal primary treatment contrast. A secondary randomized-period response outcome has an unadjusted comparison, while longer-window outcomes are described during open-label treatment. Safety events are reported by study period. Missing-data methods, multiplicity procedures, and a detailed primary inferential analysis are not reported in the registry record.