Benralizumab for Eosinophilic Gastritis (BEGS): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Benralizumab for Eosinophilic Gastritis (BEGS) (NCT03473977), a randomized phase 2, phase 3 study. The primary outcome is Percent of Patients in Histological Remission (<30 Eos/Hpf).

2. Clinical Question

The study, titled Benralizumab for Eosinophilic Gastritis (BEGS), addresses a histological outcome rather than making the primary endpoint a change in symptoms. The registry also lists clinical symptoms and several other outcomes as secondary measures. The primary comparison concerns the proportion meeting the stated remission threshold in each randomized group, not the amount by which every participant’s eosinophil count changed.

The posted result records remission in 77.0% of the benralizumab group and 8.0% of the placebo group. The groups each had 13 participants. These percentages describe the recorded endpoint at the specified time; they do not by themselves establish how outcomes varied among participants or whether the result applies to populations beyond those enrolled.

3. Trial Design

The registry describes a randomized, quadruple-masked, parallel-group trial comparing benralizumab with placebo. It lists the study as Phase 2 and Phase 3 and gives an enrollment of 26 participants. The posted arm sizes are 13 for benralizumab and 13 for placebo. The record names Children's Hospital Medical Center, Cincinnati, as the sponsor.

Participants assigned to benralizumab received a subcutaneous dose of 30 mg every 4 weeks. The placebo group received placebo subcutaneously every 4 weeks. The registry identifies the condition as eosinophilic gastritis or gastroenteritis. It records a start date of 2018-04-23, a primary completion date of 2020-06-22, and a status of completed.

The primary outcome was the percentage of participants in histological remission, assessed 12 weeks after treatment started. The remission threshold was a gastric peak eosinophil count < 30 eosinophils per high powered field. The registry describes the primary analysis population as intent to treat, including participants who had at least one clinical observation after randomization.

The record lists six secondary outcomes: change in Gastric Endoscopic Score (Lanza), gastric histology score, blood eosinophil count, Eosinophilic Gastritis Diagnostic Panel, gastric peak eosinophil count, and clinical symptoms. It also reports seven outcome measures and nine analyses. Details of the original sample-size justification, the trial's planned statistical analysis, and any interim analyses are not reported in the registry record. The listed sample-size figures are planning illustrations based on the observed effect, not the original design rationale.

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: 34 participants screened, 8 excluded (8 did not meet inclusion criteria), and 26 randomized

Overall Study

Analysis population for the primary outcome, as defined in the registry: Intent to treat analysis including participants who had at least one clinical observation post randomization.

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 Percent of Patients in Histological Remission (<30 Eos/Hpf), measured over 12 weeks after start of treatment.

The primary endpoint is the percentage of participants in histological remission 12 weeks after the start of treatment. The registry defines remission as a gastric peak eosinophil count < 30 eosinophils per high powered field. It is a binary outcome: each participant either meets the remission threshold or does not. The reported percentages therefore describe the share of participants reaching that threshold in each group.

A threshold-based endpoint gives a direct way to compare the proportion meeting a specified histological criterion. It is different from comparing average changes in eosinophil counts: participants just below and well below the threshold are both counted as being in remission, while those who do not meet it are counted otherwise. The registry does not report additional details about how the gastric peak count was collected or how multiple measurements, if any, were handled.

The record also lists changes in endoscopic score, histology score, blood eosinophil count, a diagnostic panel, gastric peak eosinophil count, and clinical symptoms as secondary outcomes. Those measures can describe other aspects of the condition, but the primary result is specifically about the remission threshold. The registry record does not report whether the threshold was selected through a particular validation process or how it relates to changes in symptoms.

7. Results Posted in the Registry

• Benralizumab: 76.9% (10/13) versus 7.7% (1/13) on control; risk difference 69.2 percentage points (95% CI 42.1 to 96.3), risk ratio 10.00 (1.49–67.29); unadjusted two-proportion z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is binary because remission is defined by whether a participant's gastric peak eosinophil count is below a stated threshold. For a binary outcome, a group percentage describes the proportion of participants classified as having the event—in this case, histological remission. The registry reports 77.0% for benralizumab and 8.0% for placebo, alongside counts of 10 of 13 and 1 of 13.

What does the risk difference express?

The risk difference subtracts the outcome proportion in one group from the outcome proportion in the other group. The unadjusted recalculation gives 0.692 for benralizumab compared with placebo. A risk difference is expressed on the proportion scale and describes an absolute contrast between the two groups. It is not the same as a relative measure, and it does not describe the probability that a particular participant will experience an outcome.

How do the risk ratio and odds ratio differ?

The risk ratio compares the outcome proportions between groups; the recalculated estimate is 10.0. The odds ratio compares the odds of the outcome, rather than the proportions themselves; its recalculated estimate is 40.0. When outcomes are uncommon or groups are small, these measures can differ considerably in numerical size. The registry's group counts help make clear that these relative measures arise from a small set of observed outcomes. They should be interpreted alongside the absolute group percentages and the risk difference, not as if they were equivalent expressions.

What does Fisher Exact mean?

The sponsor's reported analysis is a Fisher Exact test, with p < 0.001. As a general statistical explanation, Fisher's Exact test evaluates the association between two categorical variables using the observed counts and the possible arrangements of counts under a null hypothesis of no association. It is often useful for sparse tables, where some cells have small counts. A test result addresses compatibility with that null model; it does not estimate the size of the group contrast. The estimated differences and their intervals supply information about size and uncertainty.

The unadjusted recalculation reports p < 0.001. That is a separate presentation of the comparison, and the registry's reported Fisher Exact result remains the sponsor's analysis. A p-value below a threshold is not a measure of the probability that a result arose by chance, and it is not a measure of clinical importance. Interpretation needs to consider the endpoint, group sizes, estimates, and uncertainty together.

What is the role of the interval estimates?

The recalculated risk-difference interval has limits of 0.421 and 0.963. The risk-ratio interval has limits of 1.486 and 67.29, and the odds-ratio interval has limits of 3.579 and 447.0. As a general explanation, an uncertainty interval represents a range of estimates compatible with the observed data under the method used to construct it. A narrower interval indicates greater precision than a wider interval on the same scale. The interval method and confidence level for these recalculated estimates are not reported in the registry record, so those details should not be inferred.

The wide ranges for the relative measures emphasize that the exact magnitude of the relative contrast is uncertain, even though the posted group percentages differ markedly. Relative measures can be especially unstable when the sample is small and outcome counts are sparse. This is why the risk difference, group percentages, and counts provide useful context rather than relying on a relative measure alone.

What does the analysis population include?

The registry calls the primary population intent to treat and defines it as participants who had at least one clinical observation after randomization. That wording is the record's stated analysis population. It is important not to assume that it includes every randomized participant regardless of available outcome data: the explicit definition requires at least one post-randomization clinical observation. The registry does not report how missing primary outcome observations were handled, whether any participants lacked the endpoint measurement, or whether any additional analysis populations were used.

The trial is described as randomized and parallel-group, with quadruple masking. These design features describe the planned comparison structure and masking stated in the record; they do not replace information about the analysis itself. Further details of the statistical plan, including adjustment for other outcomes or covariates, are not reported in the registry record. Consequently, the posted primary comparison is best read as a two-group binary-outcome result with a reported Fisher Exact test and separate unadjusted recalculations.

10. Confidence Intervals and Precision

The recalculated risk difference is 0.692, and its interval limits are 0.421 and 0.963. These limits give a range around the estimated absolute contrast rather than a guarantee that the true contrast lies within it. The interval is entirely above zero, while still spanning a range of possible differences. The registry record does not report the confidence level or the method used for this recalculated interval, so those details cannot be specified.

The recalculated risk ratio is 10.0, with interval limits of 1.486 and 67.29. The odds ratio is 40.0, with interval limits of 3.579 and 447.0. These ranges are broad, indicating limited precision in the magnitude of the relative comparisons. The lower and upper limits should be read on the respective scales; an odds ratio is not a risk ratio, and their interval widths cannot be directly compared as if they measured the same quantity.

As a general statistical principle, an interval communicates both the direction and precision of an estimate. Wider intervals indicate that the data are compatible with a broader range of effect sizes. In this trial, the group sizes were 13 in each arm, and the remission counts were 10 and 1. Those small counts are important context for the breadth of the relative-effect intervals. The displayed estimates describe uncertainty in the group comparison, not variation in every individual participant's response.

The reported Fisher Exact p-value is < 0.001, while the recalculated p-value is < 0.001. Neither p-value substitutes for the interval estimates. The record's reported analysis does not include interval limits, and the recalculated intervals do not have a reported confidence level in the registry record. These distinctions should remain clear when judging how precisely the primary contrast is estimated.

11. Planning a Similar Trial: Sample Size

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

The planning figures are participants per arm, not total events. They are planning illustrations based on the observed effect and use 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 original sample-size assumptions and justification are not reported in the registry record.

For planning at 100% of the observed effect, the figures are 7 participants per arm for 80% power and 9 participants per arm for 90% power. At 75% of the observed effect, the corresponding figures are 12 and 16 participants per arm. At 50% of the observed effect, they are 24 and 31 participants per arm. The percentage describes how large an effect the hypothetical study is designed to detect relative to the observed effect; it does not describe the percentage of participants expected to have remission.

The 80% power column gives the estimated number of participants per arm under which the stated planning assumptions would yield an 80% chance of detecting the selected effect at the stated significance level, if that effect and assumptions were correct. The 90% column expresses the same planning idea with a higher target chance of detection. Smaller effects require more participants under these assumptions, which is why the listed requirements rise as the assumed fraction of the observed effect decreases.

These calculations are planning illustrations, not observed power. They do not show that the completed trial had a particular probability of detecting its true effect, and they do not establish that the original trial was designed using these assumptions. The displayed curve should be read in the same planning context as the listed figures.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.077; treatment proportion p₁ = 0.769; average p̄ = 0.423.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6987 = 1.3694.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.4985 = 0.4196.
4. n = (1.3694 + 0.4196)² / (0.6923)² = 6.7, rounded up to 7 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 <- 0.0769230769230769                          # control proportion
p1 <- 0.769230769230769                          # 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(3473977)
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 = 7 (80% power), 9 (90% power)
 75% of observed effect: n per arm = 12 (80% power), 16 (90% power)
 50% of observed effect: n per arm = 24 (80% power), 31 (90% power)
power.prop.test: 6.7 
Simulated power at n = 7 per arm: 0.851 (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.

13. Analyses Reported by the Sponsor

The registry reports the sponsor's own primary comparison as a Fisher Exact analysis with p < 0.001. The unadjusted recalculation gives a risk difference of 0.692, with interval limits of 0.421 and 0.963, and p < 0.001. It also provides recalculated risk-ratio and odds-ratio estimates. These recalculations use the posted group counts and are not adjusted analyses.

The two p-value formats are not identical, but both are reported as small values. The registry record does not provide further detail about the sponsor's test implementation, such as a particular tail convention, or an interval estimate for the sponsor analysis. Accordingly, the sponsor result should be described as the registry-reported Fisher Exact finding, while the other estimates are unadjusted recalculations that help express the magnitude and precision of the contrast. The record does not flag a discrepancy between the sponsor analysis and recalculation.

The recalculated estimates do not replace the sponsor's reported analysis or supply missing details of the original statistical plan. The registry record does not report the planned method for the primary comparison beyond the posted Fisher Exact analysis, nor does it describe whether other analyses were prespecified.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists six secondary outcomes and reports seven outcome measures and nine analyses. The secondary outcomes include endoscopic and histology scores, blood eosinophil count, a diagnostic panel, gastric peak eosinophil count, and clinical symptoms. The primary endpoint is the percentage in histological remission at the stated follow-up time. The record does not explain how the analyses correspond to every outcome measure or analysis count.

When a study examines multiple outcomes or performs multiple statistical tests, the chance of obtaining at least one small p-value can increase. As a general statistical concept, multiplicity procedures can address that issue by controlling an error rate or by defining a hierarchy for interpretation. Whether any such procedure was used for this trial is not reported in the registry record. The posted primary p-value should therefore not be assumed to be adjusted for the secondary outcomes, and no multiplicity adjustment should be inferred.

The primary analysis population is described as intent to treat and includes participants who had at least one clinical observation after randomization. The registry record does not report the number of participants with missing primary endpoint data, how missing values were handled, or whether participants with no post-randomization observation were included in any other analysis. It also does not report sensitivity analyses for missing data.

Other analysis populations, such as a per-protocol or safety population definition, are not reported in the registry record. Details of interim analyses, covariate adjustment, subgroup analysis, and the complete statistical analysis plan are also not reported. These absences limit assessment of how the primary and secondary results were selected, whether additional analyses affected interpretation, and how robust the result is to alternative assumptions about missing outcomes.

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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry lists zero serious adverse events in the benralizumab group and zero in the placebo group. Each group had 13 participants at risk for serious adverse events. It also records other adverse events in 11 of 13 participants in the benralizumab group and 8 of 13 participants in the placebo group. Deaths are recorded as zero in both groups.

These are counts for the categories as labeled in the record. The registry does not list serious adverse event terms in the posted summary. It does not define the category of other adverse events in the information, or provide event-level details, severity, timing, or attribution. The figures therefore describe how many participants were recorded in each category, not the nature or cause of individual events.

With 13 participants per group, the safety counts are descriptive and do not establish whether the groups differ reliably in safety outcomes. A zero count means that no event in that category was recorded among the listed participants; it does not demonstrate that an event cannot occur. The registry data do not support conclusions about rare events or safety beyond the recorded follow-up and categories.

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

The posted primary result compares the proportion reaching the registry's histological remission threshold 12 weeks after treatment began. Remission was recorded for 77.0% of the benralizumab group and 8.0% of the placebo group. The unadjusted risk difference is 0.692, with interval limits of 0.421 and 0.963. These statistics describe a difference in the recorded outcome proportions in this trial.

The result does not, by itself, establish what happened to symptoms or other clinical outcomes. The registry lists six secondary outcomes, but their results are not reported in the registry record. Nor does a single follow-up assessment establish whether remission persisted at later times. The endpoint is a threshold classification, so it does not reveal the full distribution of eosinophil counts or how far participants were from the threshold.

The result also does not determine how broadly it applies beyond this trial. The record reports 26 enrolled participants, with 13 in each posted arm. The observed counts are 10 and 1, and the relative-effect intervals are broad. The registry record does not report enough detail about participant characteristics, missing outcomes, or other aspects of the statistical plan to assess all sources of generalizability or analytic uncertainty.

These results should not be read as medical advice or as a complete evaluation of benralizumab. They describe one randomized, masked, parallel-group comparison, its primary histological outcome, and the posted safety counts. The distinction between a statistically notable group contrast and a conclusion about treatment in other settings is important. Additional outcomes and details not reported in the registry record cannot be inferred from the primary endpoint alone.

17. Why This Trial Matters Statistically

This trial is statistically informative because its primary endpoint turns a histological measurement into a clearly defined binary comparison: remission or no remission under the stated threshold. That makes the group percentages and the absolute risk difference interpretable in direct terms. The endpoint also illustrates an important distinction: a threshold outcome answers whether participants crossed a specified boundary, not how much their measurements changed.

The trial also shows why estimates and their precision should be read together. The posted proportions differ, and the recalculated risk difference is 0.692, but the groups each contain 13 participants. The interval estimates for the relative measures cover broad ranges. The observed contrast is therefore not the same thing as a highly precise estimate of its magnitude. The small number of participants is especially relevant to interpreting the relative estimates derived from sparse counts.

The sponsor's Fisher Exact analysis offers a method for testing the categorical comparison, while the recalculated risk difference and relative measures express its size on different scales. Keeping those roles separate helps avoid treating a p-value as an effect-size estimate. The record also makes visible the boundaries of what can be concluded: missing-data procedures, multiplicity handling, and original design assumptions are not reported.

Finally, the listed planning illustrations connect observed effects with hypothetical future study sizes under explicit assumptions. They are not evidence of the completed trial's power and do not reveal its original design rationale. Taken together, the trial is a useful example of interpreting a binary endpoint, sparse event counts, multiple effect measures, and uncertainty without extending the posted result beyond its scope.

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

The trial enrolled 26 participants, with 13 in each posted arm. This small sample means that each participant's outcome has a substantial influence on the group proportions, and the relative estimates are based on sparse counts. The recalculated risk-ratio and odds-ratio intervals are broad, limiting precision about the size of those relative contrasts.

The primary outcome is a threshold-based histological measure at 12 weeks after treatment began. The registry record does not report the distribution of gastric peak eosinophil counts, measurement details, or results at other time points. The threshold outcome therefore cannot describe the full range of histological changes or the duration of any remission.

The primary population definition includes participants with at least one clinical observation after randomization, but the record does not report whether any primary outcome data were missing or how missing data were handled. The registry also does not report multiplicity procedures for the listed secondary outcomes, details of interim analyses, or the complete statistical analysis plan. These omissions make it difficult to evaluate potential analytic choices beyond the reported result.

The registry lists six secondary outcomes but their results are not reported in the information. The record also does not provide the original sample-size justification or design assumptions. The planning illustrations use the observed effect and should not be mistaken for the original rationale. These limitations constrain interpretation of the posted primary comparison; they do not change the counts and estimates that are reported.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Benralizumab for Eosinophilic Gastritis (BEGS) — NCT03473977.
• Registry record last updated 2022-02-03; status last verified 2022-01.
• Results first posted 2021-08-17.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

NCT03473977, Benralizumab for Eosinophilic Gastritis (BEGS), is recorded as a completed randomized, quadruple-masked, parallel-group study listed as Phase 2 and Phase 3. The sponsor is Children's Hospital Medical Center, Cincinnati. The registry gives an enrollment of 26 and lists eosinophilic gastritis or gastroenteritis as the condition.

The groups received subcutaneous benralizumab 30 mg every 4 weeks or placebo every 4 weeks. The primary endpoint was the percentage in histological remission 12 weeks after treatment began, defined as a gastric peak eosinophil count < 30 eosinophils per high powered field. The posted percentages were 77.0% and 8.0%, with 13 participants in each group.

The registry reports a Fisher Exact p-value of < 0.001. Recalculated unadjusted estimates include a risk difference of 0.692 and interval limits of 0.421 and 0.963. The record lists six secondary outcomes; additional analysis and missing-data details are not reported in the registry record.