Broccoli Sprout Dose Response: Bioavailability and Effects of Air Pollutants: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Broccoli Sprout Dose Response: Bioavailability and Effects of Air Pollutants (NCT02656420), a randomized phase 1, phase 2 study. The primary outcome is Sulforaphane Bioavailability Measured in Sequential 12-Hour Urine Samples.

2. Clinical Question

3. Trial Design

The registry identifies this as a Phase 1 and Phase 2 study with randomized allocation, single masking and a parallel-group model. It lists enrollment of 170 and a completed status. The recorded start was 2016-01, and primary completion was 2016-03. The sponsor was Johns Hopkins Bloomberg School of Public Health. These record details describe the trial’s registration, not the number of participants contributing to each posted endpoint result.

Four arms are described: placebo, high-dose broccoli sprout, medium-dose broccoli sprout and low-dose broccoli sprout. All beverages were 100 mL and were taken nightly for 10 days. The placebo contained pineapple juice, lime juice and water. Each active beverage contained broccoli sprout powder mixed into pineapple juice, lime juice and water, with the glucoraphanin-rich and sulforaphane-rich amounts varying by dose. The posted primary endpoint results give analyzed sample sizes for the three active arms, while the placebo arm has no analyzed primary endpoint value.

The registry’s primary endpoint is titled “Sulforaphane Bioavailability Measured in Sequential 12-Hour Urine Samples.” It reports geometric means in micromoles/24 hours, with a timeframe of 10 days. The endpoint description says metabolites were measured in sequential 12-hour urine collections and that the data for each individual were summed to provide a per-24-hour value. The registry also lists modulation of air-pollutant excretion in sequential overnight 12-hour urine samples as a secondary outcome.

The record does not report a detailed allocation procedure, how masking was maintained, or which trial personnel were masked. It also does not report a protocol-level sample-size justification or a detailed statistical analysis plan. The posted design information therefore supports describing the trial as randomized, single-masked and parallel-group, but not filling in unreported operational or planning details.

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: Placebo arm samples were not analyzed f because no sulforaphane was administered to this group. Historically, background levels from dietary sources in this population are very low compared to active intervention levels. In addition, one participant in the High Dose Broccoli Sprout arm did not complete the full protocol and was not included.

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 Sulforaphane Bioavailability Measured in Sequential 12-Hour Urine Samples, measured over 10 days.

The primary endpoint is urinary excretion of sulforaphane-mercapturic acid, identified in the record as a broccoli-derived sulforaphane metabolite. It is expressed as a geometric mean in micromoles/24 hours over the 10-day intervention. The endpoint description says that metabolite measurements from sequential 12-hour urine collections were combined within each participant to form a per-24-hour value. Such an endpoint is intended to characterize systemic availability and urinary excretion of the administered broccoli-derived compounds, rather than to measure a clinical disease outcome.

Urine measurement offers a way to quantify excreted metabolites following exposure to the active beverage. In general, a biomarker endpoint can help characterize biological uptake or processing of an intervention, but it is not automatically a direct measure of a downstream health effect. Here, the trial’s stated broader interest concerned detoxication of air pollutants excreted in urine, while the primary endpoint specifically concerns sulforaphane metabolite excretion. The registry separately lists modulation of air-pollutant excretion as a secondary outcome.

The posted primary values are 24.8 for the high-dose arm, 11.0 for the medium-dose arm and 4.3 for the low-dose arm. Their associated posted ranges are 16.8 to 33.5, 8.4 to 13.2 and 3.6 to 5.2, respectively. The registry gives no primary endpoint value for placebo because placebo samples were not analyzed. Thus, these results describe the active-arm biomarker endpoint, not a randomized active-versus-placebo contrast.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports an analysis called “Mixed Models Analysis” for the three active-dose groups. Its reported parameter is “Mean Difference (Final Values),” with an estimate of -1.73, a 95% confidence interval from -1.89 to -1.57, and p < 0.001. The analysis type is recorded as equivalence. These details identify the method label and reported result, but do not fully specify the model or the comparison being estimated.

What does a mixed model do?

In general, a mixed model can account for data that have more than one source of variation, such as repeated measurements from the same participant alongside differences between participants. The primary endpoint was collected in sequential 12-hour urine samples over the intervention period, with participant-level measurements combined into a per-24-hour value. Repeated measurements can be correlated within a person, so treating every measurement as wholly independent may understate uncertainty. A mixed model can represent within-person and between-person variation when its structure is specified appropriately.

The registry does not report the model formula, covariance structure, fixed or random effects, handling of time, transformation, or exact contrast. It also does not say whether the reported final-values estimate is based on a comparison among all three active groups, a particular pairwise contrast, or another model contrast. The endpoint arm summaries are geometric means, whereas the analysis parameter is named mean difference (final values). Without the model scale and estimand, the two kinds of reported numbers should not be directly combined or treated as interchangeable.

How should the interval and p-value be read?

In general, a confidence interval gives a range of values reasonably compatible with an estimate under the statistical model, its assumptions and the data. The reported interval runs from -1.89 to -1.57 around the estimate of -1.73, and the registry identifies it as 95%. The meaning of the negative scale depends on the contrast’s definition and the outcome scale, neither of which is fully described. The p-value, reported as p < 0.001, is evidence against the null hypothesis used for that model; it is not a measure of the probability that the null hypothesis is true, and does not independently establish equivalence or practical importance.

Equivalence testing generally asks whether an estimated difference is sufficiently bounded within a prespecified range of differences regarded as unimportant. The equivalence margin is central: a confidence interval must be assessed relative to that margin. The registry labels the analysis as equivalence but does not report the margin. Without it, the label and interval alone do not allow a reader to judge the equivalence criterion or verify how it was applied.

Why does the distribution summary matter?

The endpoint parameter is a geometric mean and the dispersion is named as an inter-quartile range. Geometric summaries are often used for positive, right-skewed measurements, because multiplicative variation can be more natural than additive variation. In general, the inter-quartile range describes the middle half of a distribution and is not a confidence interval for the group mean. However, the arm data contain no populated spread value, while low and high values are present. The record does not clearly identify the statistical meaning of those bounds in the posted arm results. They should therefore be read as the listed ranges without assuming they are confidence intervals or inter-quartile limits.

Over-dispersion is a general concern when observed variability exceeds what an analysis model assumes. If a model does not allow for the actual variability or dependence in the data, its standard errors and uncertainty intervals may be too narrow. The registry does not provide enough detail to assess over-dispersion or the fit of the mixed model. The endpoint’s repeated collection structure makes the modeling of within-participant dependence potentially relevant, but the analysis specification is not reported in sufficient detail to evaluate it.

Why is a recalculated comparison unavailable?

A simple unadjusted re-calculation requires a clearly defined comparison and the necessary summary statistics for its groups. The registry record has no placebo endpoint value, and the posted arm summaries do not establish the scale or contrast underlying the mixed-model result. The record also flags that values are missing or not estimable for a two-arm comparison for the other endpoint. Accordingly, there is no suitable two-arm recalculated result to place alongside the sponsor’s analysis. This absence does not negate the reported mixed-model result; it means the posted information does not support an independent numerical comparison on a matching estimand.

General sample-size illustrations, when based on an observed effect, are planning calculations rather than evidence that a trial had a particular probability of detecting an effect. The registry’s stated illustration assumptions are two-sided significance level 0.05, equal allocation between two arms and a normal approximation, with the effect to detect set as a stated fraction of the observed effect. Those assumptions do not describe the trial’s original design justification. The sample-size justification for this trial and any observed-power claim are not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry reports the sponsor’s analysis as a mixed models analysis of the three active-dose groups. It gives a mean difference (final values) of -1.73, a 95% confidence interval from -1.89 to -1.57, and p < 0.001, and classifies the analysis as equivalence. That reported analysis is the trial’s own statistical result. The registry record does not provide enough detail about the modeled contrast, scale or equivalence margin to independently interpret the estimate as a specific between-dose difference or to verify the equivalence criterion.

No matching two-arm recalculation is available from the posted endpoint summaries. Placebo samples were not analyzed for the primary endpoint, and the record does not define the model contrast needed to reconstruct the reported estimate. The active-arm geometric means and their listed ranges are descriptive summaries, not a substitute for the model’s estimate and uncertainty. Therefore, there is no recalculated numerical result to compare with the sponsor’s analysis or to use to confirm it.

The registry also states that a two-arm comparison is not estimable for the other endpoint. A sample-size illustration based on observed effects, if displayed with the results, is a planning illustration under its stated assumptions; it is not the trial’s observed power or evidence that a particular sample size was justified. The protocol’s design rationale and the complete statistical plan are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists the primary endpoint as sulforaphane bioavailability measured in sequential 12-hour urine samples. It also lists modulation of air-pollutant excretion in sequential overnight 12-hour urine samples as a secondary outcome. The record indicates two outcome measures and two analyses, but the detailed multiplicity strategy is not reported in the registry record. It does not state whether the analyses were ordered hierarchically, adjusted for multiple testing, or designated as exploratory.

Multiplicity matters in general when several outcomes, dose comparisons, time points or model contrasts are tested. Looking across multiple tests can increase the chance of at least one apparently notable result, depending on the analysis plan and the tests’ relationship. The registry excerpt does not describe the number of contrasts examined, whether all planned analyses are posted, or whether an adjustment was used. The reported p-value should therefore be understood as belonging to the specific analysis as recorded, rather than assumed to account for every outcome or comparison in the study.

Missing-data handling and analysis populations are also not fully described. The primary endpoint description says placebo samples were not analyzed because no sulforaphane was administered to that group. It also begins to note that one participant in the high-dose arm did not complete something, but the description ends before explaining what was not completed or how the participant was handled. The reason for any missing endpoint data, the handling of incomplete urine collections, and the analysis population definition are not reported in the registry record.

The primary endpoint results include 24 high-dose, 35 medium-dose and 55 low-dose participants, with no placebo result. These arm-specific endpoint counts do not by themselves establish whether analyses used an intention-to-treat, per-protocol or other population. The overall enrollment is listed as 170, but the record does not reconcile enrollment with endpoint-specific analytic samples. No assumptions should be made about the participants not represented in the posted endpoint summaries.

In general, incomplete outcome data can affect an estimate if the likelihood of missingness is related to treatment or outcome. A complete assessment would need the amount and reasons for missing data, prespecified handling rules, and sensitivity analyses if applicable. Those details are not reported in the registry record. The absence of those details is a limitation in interpreting the endpoint estimates, not evidence that a particular missing-data method was or was not used.

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.

The adverse-event record reports zero serious adverse events and zero other adverse events in the listed groups: placebo, high-dose broccoli sprout, medium-dose broccoli sprout and low-dose broccoli sprout. The corresponding at-risk counts are 55, 25, 35 and 55, respectively. These are the registry’s posted counts; they should be read as event counts during the reported study period, not as proof that no events could occur or that risks are identical across groups.

Deaths are recorded as null for each group. A null entry is not the same as a reported zero, so deaths should be described as not reported in the registry record rather than as none. The serious-event terms list is empty. The record does not provide additional detail about event definitions, collection procedures, severity grading for non-serious events, timing, or follow-up beyond these counts.

In general, a zero event count in a group does not establish that the underlying event risk is zero. The ability to characterize uncommon harms depends on the amount and duration of observation and on how events were collected. The posted safety counts offer a narrow description of what the registry records, but do not support a broad safety conclusion or a reliable comparison of risks between arms. They also should not be confused with the primary endpoint sample sizes, which are different and do not include an analyzed placebo value.

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

The result means that the registry reports different geometric mean urinary sulforaphane-metabolite values across the three active-dose arms: 24.8 in the high-dose arm, 11.0 in the medium-dose arm and 4.3 in the low-dose arm. The endpoint is expressed in micromoles/24 hours and reflects urinary excretion of sulforaphane-mercapturic acid during the 10-day intervention. The reported pattern is descriptive of the posted active-arm summaries.

The result does not, on its own, establish a causal dose-response relationship. Although the trial is recorded as randomized, the endpoint summaries do not include placebo, and the mixed-model contrast is not fully described. The registry’s equivalence label cannot be evaluated without the equivalence margin and more detail about the estimand. The model’s reported estimate and p-value should be interpreted within those limits, rather than mapped directly onto the arm geometric means.

The endpoint is a biomarker measure, not a clinical outcome. It does not directly show that air-pollutant excretion changed, even though the study’s stated aim concerns detoxication of air pollutants excreted in urine. The registry lists air-pollutant excretion as a secondary outcome, but the detailed results for that outcome are not reported in the record. Nor do the metabolite values establish a health benefit or a change in disease risk.

The reported ranges are not sufficiently defined in the arm data to assign them a confidence-interval interpretation. The record names inter-quartile range as the dispersion measure, but does not populate a separate spread statistic. The truncated description of the high-dose group’s participant completion status and the absence of a complete missing-data account further constrain interpretation.

In general, a trial result is strongest when the outcome, analysis population, comparison, uncertainty measure and analysis plan are clearly specified. Here, the registry supports a description of the primary active-arm biomarker results and a summary of the sponsor’s mixed-model output, but not a complete independent reconstruction of the analysis or a placebo-controlled estimate for that endpoint.

14. Why This Trial Matters Statistically

This trial illustrates a useful distinction between a biomarker endpoint and the broader biological or environmental question that motivates a study. Urinary sulforaphane-mercapturic acid provides a measure of excretion of a broccoli-derived metabolite. The trial’s broader interest is detoxication of air pollutants excreted in urine, which is listed separately as a secondary outcome. A change in a metabolite endpoint should not be treated as equivalent to evidence about pollutant excretion or a health outcome.

It also demonstrates why an assigned comparator and an analyzed comparator are not always the same. The registry describes a placebo arm, but says placebo samples were not analyzed for the primary endpoint. That means the primary endpoint’s active-arm values cannot be used to estimate an active-versus-placebo contrast. Readers need to distinguish the trial’s randomized design from the subset of arms contributing to a particular outcome analysis.

The record’s mixed-model analysis and its “equivalence” label raise a further statistical point: a method name and p-value do not fully explain an estimand. Interpretation of equivalence depends on a prespecified margin and the relationship between the interval and that margin. Since the registry does not report the margin or fully identify the modeled contrast, the result cannot be independently assessed as an equivalence finding from the posted detail alone.

Finally, this record shows how missing analysis details matter even when numerical summaries are available. Arm values, ranges and an analysis estimate each answer different questions. Without a clear account of the outcome scale, contrast, analysis population, missing-data handling and multiplicity strategy, those pieces cannot be combined into a stronger conclusion than the record supports. Transparent reporting of such details is important for reproducibility and for separating descriptive patterns from confirmatory evidence.

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.

Several limitations are specific to the registry information for this trial. The primary endpoint has no analyzed placebo value, because the registry says placebo samples were not analyzed when no sulforaphane was administered. As a result, the posted primary endpoint data do not provide an active-versus-placebo comparison. The registry’s note about historically low dietary background levels does not supply a trial-specific placebo estimate.

The mixed-model result is difficult to interpret fully because the record does not identify the exact contrast, outcome scale, model specification or equivalence margin. The estimate of -1.73 and its 95% confidence interval from -1.89 to -1.57 cannot be confidently translated into a comparison among the posted geometric means without those details. The “equivalence” classification is not enough by itself to determine whether a prespecified equivalence criterion was met.

The primary endpoint description is truncated after mentioning that one participant in the high-dose group did not complete something. The record therefore does not fully explain the issue or its effect on the analysis. It does not report the analysis-population definition, the approach to missing urine samples, or sensitivity analyses. It also does not detail the handling of repeated measurements in the model.

The arm data identify inter-quartile range as the dispersion measure, but do not populate a separate spread value; the posted low and high values are not explicitly characterized as confidence limits. The record does not report a multiplicity approach for the listed outcomes and analyses. Detailed results for the listed secondary outcome are not included in the registry record, and a two-arm comparison for the other endpoint is marked not estimable.

Finally, the safety record gives event counts but leaves deaths null and supplies little information about event ascertainment or follow-up. These reporting limitations do not establish that a specific analysis or procedure was absent; they mean that it is not reported in the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Broccoli Sprout Dose Response: Bioavailability and Effects of Air Pollutants — NCT02656420.
• Registry record last updated 2019-05-01; status last verified 2019-04.
• Results first posted 2019-05-01.
• 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 identifies NCT02656420 as a completed, randomized, single-masked, parallel Phase 1 and Phase 2 study sponsored by Johns Hopkins Bloomberg School of Public Health. Enrollment is listed as 170. The study compared placebo with high-, medium- and low-dose broccoli sprout-derived beverages taken nightly for 10 days.

The primary endpoint was urinary sulforaphane-mercapturic acid, expressed as geometric mean micromoles/24 hours. Posted active-arm values were 24.8 for high dose, 11.0 for medium dose and 4.3 for low dose. Placebo samples were not analyzed for this endpoint. The registry also reports a mixed-model analysis classified as equivalence, but does not provide the full contrast, model specification or equivalence margin needed for complete independent interpretation.

The registry lists modulation of air-pollutant excretion as a secondary outcome. Detailed outcome results, multiplicity procedures, full missing-data handling and the analysis-population definition are not reported in the registry record. Safety entries record no serious or other adverse events in the listed groups; deaths are not reported.