Study 302: BLI4900 Versus an FDA-approved Comparator in Adult Subjects Prior to Colonoscopy: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study 302: BLI4900 Versus an FDA-approved Comparator in Adult Subjects Prior to Colonoscopy (NCT04446312), a randomized phase 3 study. The primary outcome is Proportion of Subjects With Successful Bowel Preparation.

2. Clinical Question

The registry record identifies the trial as a phase 3 randomized comparison in adults before colonoscopy. Its stated purpose was to compare safety and efficacy. The primary outcome focuses on the quality of bowel preparation as judged by the endoscopist, rather than on a patient-reported measure or a clinical outcome after colonoscopy.

The central statistical question is a comparison between two proportions: the number with a successful preparation relative to the number in each outcome group. The registry also classifies the sponsor analysis as non-inferiority. That classification indicates the analysis was framed around whether the intervention could be considered sufficiently close to the control under a prespecified margin. The margin is not reported in the registry record, so the precise criterion used for that conclusion cannot be assessed from the record alone.

3. Trial Design

NCT04446312, titled Study 302, is listed as a phase 3, randomized, single-masked, parallel-group study. It compared BLI4900 with an FDA-approved bowel-preparation control in adult subjects prior to colonoscopy. The intervention is described as a drug, and both arms involved an oral bowel-preparation solution. The study purpose describes both preparations as 2-day, split-dose regimens.

The registry lists enrollment of 500 and a completed status. The recorded study start was 2020-07-13, and primary completion was 2021-01-07. These dates and the overall enrollment describe the trial record; they do not establish how many participants contributed to every analysis. For the primary outcome, the arm counts are 226 for BLI4900 and 224 for the FDA-approved control. The safety figures use different at-risk counts, 227 and 226 respectively. The registry does not explain these differences.

Single masking is recorded, but the identity of the masked party or parties is not reported in the registry record. The primary outcome description does state that the endoscopist assessing preparation quality was blinded. The registry lists one outcome measure and one analysis, and no secondary outcomes are posted.

Random allocation and a parallel-group structure support a direct comparison of the two assigned preparations. The registry does not report the allocation procedure, the concealment method, the trial’s original sample-size assumptions, or a detailed statistical analysis plan. Those omissions limit how fully the design and reported non-inferiority analysis can be reconstructed.

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

Analysis population for the primary outcome, as defined in the registry: Modified intent-to-treat population - this population includes all subjects that took any portion of study preparation and who did not withdraw for a reason other than safety, efficacy or tolerance.

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.

Population: The Baseline Analysis Population represents the Intent-to-Treat Population which includes all patients that were dispensed a study preparation kit.

6. Primary Endpoint

The registry lists the primary outcome as Proportion of Subjects With Successful Bowel Preparation, measured over 2 days.

The primary endpoint is the proportion of participants with successful bowel preparation. Success was defined as a preparation rated Excellent or Good by the blinded endoscopist on a four-point scale of Excellent, Good, Fair, or Poor. The registry gives a 2 days timeframe and identifies the outcome unit as participants.

This endpoint turns the ordered quality ratings into a binary outcome: Excellent or Good counts as success, while Fair or Poor does not meet the stated success definition. A binary outcome allows the study to compare the share of participants meeting a clinically interpretable preparation threshold in each randomized group. It does not retain distinctions between Excellent and Good, or between Fair and Poor, once the ratings are grouped this way.

Endoscopist assessment is relevant to the study question because the preparation quality is judged in the context of colonoscopy. Blinding the endoscopist, as stated in the outcome description, is intended to reduce the possibility that knowledge of the assigned preparation influences that assessment. The record does not describe additional safeguards or the rating process beyond the four-point scale and the blinded endoscopist.

The endpoint analysis population is modified intent-to-treat: participants who took any portion of study preparation and did not withdraw for a reason other than safety, efficacy, or tolerance. This definition is narrower than simply including everyone randomized, and the registry does not give further detail about participants excluded from that population.

7. Results Posted in the Registry

• BLI4900: 93.8% (212/226) versus 94.2% (211/224) on control; risk difference -0.4 percentage points (95% CI -4.8 to 4.0), risk ratio 1.00 (0.95–1.04); unadjusted two-proportion z-test p = 0.861.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is binary: each participant either met the successful-preparation definition or did not. For a binary outcome, a group proportion describes the fraction of participants in that group with the event of interest. Here, the event is a rating of Excellent or Good. The posted counts are 212 successful preparations among 226 BLI4900 participants and 211 among 224 control participants.

What does a risk difference describe?

General statistical explanation: a risk difference subtracts one group’s event proportion from the other group’s event proportion. It is expressed on an absolute scale, so it describes the separation between proportions rather than their relative size. A value near zero indicates little separation in the observed proportions, but the estimate alone does not show how uncertain that comparison is. The unadjusted recalculation reports a risk difference of -0.004 and a confidence interval from -0.048 to 0.04. The registry’s sponsor analysis reports a Mantel-Haenszel risk difference of 0.15, with a 95% confidence interval from -3.97 to 4.27.

What is the Mantel-Haenszel method?

General statistical explanation: the Mantel-Haenszel method can combine comparisons across strata, producing an adjusted estimate when the analysis accounts for grouping factors. The registry names this method for the sponsor analysis but does not report the strata, adjustment variables, or other implementation details. It is therefore not possible to reproduce or fully evaluate that analysis from the registry record.

How is non-inferiority different from superiority?

General statistical explanation: a superiority analysis asks whether the data support a difference favoring one group. A non-inferiority analysis asks whether the intervention is not worse than the comparator by more than a prespecified acceptable margin. The margin must be set before interpreting the outcome; it defines how much difference the analysis permits. The registry labels the posted analysis non-inferiority, but the margin is not reported in the registry record. The analysis plan and the details of the non-inferiority decision procedure are also not reported in the registry record. Thus, the label and reported test result cannot by themselves explain the precise decision rule.

Why are estimates and p-values different?

General statistical explanation: an estimate describes the size and direction of an observed contrast, while a p-value assesses how compatible data are with a particular statistical null hypothesis under the model used. A p-value is not the probability that a treatment works, and it does not measure the size or practical importance of an effect. The sponsor analysis reports p < 0.001. The unadjusted recalculation reports p = 0.861. These results arise from analyses that do not match, as also indicated by the registry’s discrepancy flag. They should not be treated as interchangeable tests of the same fully specified procedure.

What does a confidence interval add?

General statistical explanation: a confidence interval describes uncertainty around an estimate under the method used to calculate it. It gives a range of values compatible with the observed data and the method’s assumptions; it is not a guarantee that the true value lies inside the particular interval. A wider interval indicates less precision than a narrower interval on the same scale. Comparing an interval with a non-inferiority margin can be central to a non-inferiority conclusion, which is why the missing margin matters.

The recalculation uses the posted summary counts and is explicitly unadjusted. It does not account for any unreported stratification or reproduce the sponsor’s named method. The registry does not report whether there were interim analyses, how missing outcomes were handled, or any other details needed to reconstruct the full statistical workflow. Those elements should not be inferred from the result alone.

10. Confidence Intervals and Precision

Precision concerns how tightly an estimate is determined by the data and the analysis method. A confidence interval expresses uncertainty around an estimate; a narrower interval generally indicates greater precision on the same scale, while a wider interval leaves more possible values compatible with the data and method. An interval is not a probability statement about the particular true effect after the data have been observed.

The sponsor-reported risk difference is 0.15, with a 95% confidence interval from -3.97 to 4.27. The interval includes values below and above the reported estimate. The record does not report the non-inferiority margin, so the interval cannot be evaluated against the trial’s stated decision threshold. The sponsor p-value is reported as p < 0.001, but a p-value does not replace the interval or disclose the margin.

The unadjusted recalculation gives a risk difference of -0.004 and a confidence interval from -0.048 to 0.04. It also gives a relative risk of 0.996, with an interval from 0.950 to 1.043, and an odds ratio of 0.933, with an interval from 0.428 to 2.032. These are recalculated quantities, not the sponsor’s posted analysis. Because the record flags a discrepancy, the sponsor analysis takes precedence as the sponsor-reported result; the recalculated intervals should be read as a separate summary-count check.

The estimates and intervals use different measures, and their scales should not be conflated. The registry does not explain the difference between the sponsor-reported values and the unadjusted recalculation. That unresolved mismatch limits any attempt to give a single precision statement for the trial’s primary comparison.

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 event counts. They are illustrations based on the observed effect, rather than a report of the trial’s original design assumptions. The registry record does not report the original sample-size justification or assumptions.

The calculation assumptions listed for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set at the stated fraction of the observed effect. The 80% power column means the number of participants per arm in a hypothetical design that would have an 80% chance of detecting the specified effect under those assumptions. The 90% power column describes the corresponding hypothetical number per arm for a 90% chance.

The rows labelled 100%, 75%, and 50% of the observed effect show planning for progressively smaller target effects. The listed participant counts increase as the target effect becomes smaller. This is a general feature of sample-size planning: smaller differences are harder to distinguish from random variation, so more participants are typically needed under otherwise comparable assumptions.

These figures are not observed power and do not say how much power the completed trial actually had. They are planning illustrations derived from the observed effect, which itself is uncertain and subject to the discrepancy between the sponsor result and the unadjusted recalculation. The record does not report whether the original trial design used these assumptions or 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 = 0.942; treatment proportion p₁ = 0.938; average p̄ = 0.940.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.3358 = 0.6582.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.3358 = 0.2826.
4. n = (0.6582 + 0.2826)² / (-0.0039)² = 57,867.1, rounded up to 57,868 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 <- 0.941964285714286                          # control proportion
p1 <- 0.938053097345133                          # 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(4446312)
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 = 57,868 (80% power), 77,468 (90% power)
 75% of observed effect: n per arm = 102,091 (80% power), 136,670 (90% power)
 50% of observed effect: n per arm = 227,936 (80% power), 305,142 (90% power)
power.prop.test: 57867.1 
Simulated power at n = 57,868 per arm: 0.801 (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 a sponsor analysis using the Mantel-Haenszel method, with a risk difference of 0.15, a 95% confidence interval from -3.97 to 4.27, and p < 0.001. The analysis is identified as non-inferiority. Separately, the recalculation from posted group counts is unadjusted and gives a risk difference of -0.004, with a confidence interval from -0.048 to 0.04 and p = 0.861.

The registry flags a discrepancy between the sponsor analysis and the recalculation. The differences should not be resolved by assuming that one analysis used a particular adjustment, scale, or population; those explanations are not reported in the registry record. The sponsor’s own analysis takes precedence when describing the sponsor-reported finding. The recalculation remains useful as a check of the posted summary counts, but it is not a substitute for the sponsor’s analysis or a reproduction of its method.

The non-inferiority interpretation also depends on the margin and analysis details, which are not reported in the registry record. Consequently, the available record supports reporting what the sponsor analysis states, while leaving the exact decision criterion and the source of the discrepancy unresolved.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure and one analysis, and it reports no secondary outcomes. That is the extent of the outcome information in the record; it does not establish whether the trial protocol included other outcomes or analyses that were not posted.

Multiplicity arises when several outcomes, time points, subgroups, or statistical tests are considered. General statistical explanation: evaluating many comparisons can increase the chance that at least one appears statistically notable by chance, so a prespecified analysis plan may describe how such testing is handled. The registry does not report multiplicity adjustments, a hierarchy of outcomes, or a complete analysis plan. It is therefore not possible to determine from the record whether multiplicity was relevant to analyses beyond the single posted outcome and analysis.

The primary endpoint population is described as modified intent-to-treat. It includes participants who took any portion of study preparation and did not withdraw for a reason other than safety, efficacy, or tolerance. This differs from a definition that automatically includes every randomized participant. The registry does not report how many randomized participants were excluded under this definition, why any particular participant was excluded, or how the analysis handled outcome data that were unavailable.

Missing-data handling matters because the result can depend on whether participants without an observed endpoint are excluded, classified in a particular way, or handled using a model or imputation procedure. General statistical explanation: different reasonable approaches can yield different estimates when outcomes are missing, especially if missingness is related to treatment or outcome. The registry record does not report the amount or pattern of missing primary-outcome data, the method used to address missingness, or sensitivity analyses.

The record also does not report interim analyses, stopping rules, prespecified subgroup analyses, or further details of the analysis population. Those items should not be inferred from the randomized design or the single posted result. Their absence limits evaluation of how completely the posted analysis represents the planned statistical work.

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 1.5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: concussion (0.4%), covid-19 (0.4%), rectal hemorrhage (0.4%).

The registry reports serious adverse events for 0 of 227 participants at risk in the BLI4900 group and 2 of 226 at risk in the FDA-approved control group. It reports other adverse events for 23 of 227 at risk with BLI4900 and 26 of 226 at risk with the control. Deaths are listed as 0 in both groups.

The at-risk counts for these safety summaries differ from the primary-outcome group counts, which are 226 for BLI4900 and 224 for the control. The registry does not explain why the populations differ, so the safety and efficacy counts should be read as their separately posted denominators rather than assumed to use identical analysis populations.

The listed serious-event terms are concussion, COVID-19, and rectal hemorrhage, each shown as 0.4%. The record does not identify an arm alongside these terms in the listed summary, so they should not be attributed to a specific group based on this information alone.

These are descriptive event counts. They do not establish comparative risk or explain the severity, timing, relatedness, or clinical course of each event. The registry does not report those details in the safety summary.

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

The results describe a randomized comparison of two bowel-preparation solutions before colonoscopy, using the proportion rated Excellent or Good as the primary measure. The posted counts show 212 successful preparations among 226 participants in the BLI4900 group and 211 among 224 in the FDA-approved control group. The sponsor reports a Mantel-Haenszel non-inferiority analysis, while an unadjusted recalculation from the posted counts produces different estimates and a different p-value.

The discrepancy is important. The sponsor’s analysis takes precedence as the sponsor-reported result, but the registry does not provide enough detail to reproduce that analysis or explain why it differs from the summary-count recalculation. The non-inferiority margin is also not reported. Without that margin, readers cannot independently assess the specific threshold against which the sponsor’s interval and test were evaluated.

These data do not establish that the preparations are identical, nor do the posted counts alone establish that one is better or worse. A failure to show a conventional difference would not by itself demonstrate equivalence or non-inferiority; those conclusions require the relevant design and decision criteria. Likewise, a reported p-value does not measure the probability that either preparation is effective.

The outcome represents endoscopist-rated preparation quality under the trial’s stated definition. It does not provide a complete account of all aspects of safety, tolerability, or participant experience. The safety summaries list event counts, but do not provide enough detail to determine causality or compare the full safety profiles. The findings should therefore be understood as the posted statistical results for this trial, with the limitations of the registry record kept in view.

No treatment recommendation follows from these statistics. They describe group-level observations and analyses, not a prediction for any individual participant.

17. Why This Trial Matters Statistically

This trial illustrates why a clinical result cannot be reduced to a single p-value. The registry’s sponsor analysis is labelled non-inferiority, but judging non-inferiority depends on a prespecified margin and a clearly described analysis. The margin is not reported in the registry record, so the result cannot be independently evaluated against the intended threshold.

It also demonstrates the value of checking whether posted counts and reported estimates tell a consistent story. The raw outcome counts support an unadjusted recalculation, but that calculation differs from the sponsor’s Mantel-Haenszel result. The record flags this discrepancy, and the sponsor’s own analysis takes precedence. At the same time, the mismatch makes it important to separate the sponsor-reported analysis from the simple recalculation rather than presenting them as if they were the same result.

The endpoint itself offers a useful statistical example: an ordinal assessment of preparation quality is converted to a binary success measure by grouping Excellent and Good together. This makes the comparison straightforward to express as proportions, but it also discards distinctions within the original four-point scale. The endpoint definition and the analysis population therefore shape what the result means.

Finally, the difference between outcome and safety denominators, limited information about missing data, and absence of original sample-size assumptions show how registry detail affects interpretation. Statistical conclusions depend not only on observed counts but also on the estimand, analysis choices, precision, and decision rules that connect the data to the claim.

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.

The registry record does not report the non-inferiority margin, original sample-size justification, or a detailed statistical analysis plan. These omissions are especially relevant because the sponsor analysis is classified as non-inferiority, and its decision criterion cannot be reconstructed without the margin and associated analysis details.

The record flags a discrepancy between the sponsor analysis and the unadjusted recalculation from the posted group counts. It names the Mantel-Haenszel method for the sponsor result but does not describe strata, adjustment variables, or implementation details. The difference therefore cannot be explained from the registry information, and the recalculation should not be treated as a reproduction of the sponsor analysis.

The primary analysis population is modified intent-to-treat, with eligibility based on taking any portion of preparation and not withdrawing for reasons other than safety, efficacy, or tolerance. The record does not report the number excluded, the number with missing outcomes, or the approach to missing data. It also does not explain why the safety denominators differ from the primary-outcome denominators.

Single masking is recorded, and the outcome description identifies a blinded endoscopist, but the registry does not state who else was masked or provide details of the assessment process. No secondary outcomes are posted. Interim analyses, multiplicity procedures, subgroup analyses, and the original randomization and allocation-concealment procedures are not reported in the registry record.

These limitations restrict independent assessment of the design and analysis. They do not invalidate the posted counts, but they narrow the conclusions that can be drawn from the registry summary alone.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Study 302: BLI4900 Versus an FDA-approved Comparator in Adult Subjects Prior to Colonoscopy — NCT04446312.
• Registry record last updated 2023-10-27; status last verified 2023-10.
• Results first posted 2023-10-27.
• 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

NCT04446312 is a completed phase 3 randomized, single-masked, parallel-group study comparing BLI4900 with an FDA-approved bowel-preparation control in adults prior to colonoscopy. The registry lists enrollment of 500 and records the study start as 2020-07-13 and primary completion as 2021-01-07.

The primary endpoint was successful preparation, defined as an endoscopist rating of Excellent or Good. The posted outcome counts are 212 of 226 for BLI4900 and 211 of 224 for the control. The sponsor reports a Mantel-Haenszel non-inferiority analysis; a separate unadjusted recalculation differs, and the record flags the discrepancy. The sponsor analysis takes precedence as the sponsor-reported result.

The non-inferiority margin, original sample-size assumptions, detailed analysis plan, and missing-data methods are not reported in the registry record. Safety summaries list serious and other adverse-event counts, with no deaths reported in either group.