← Clinical Trials
Ebola Virus Disease Phase 1 Safety & Tolerability NCT02485912

EBL06: Complete Statistical Analysis of Short-interval Ebola Vaccine Prime-boost

An independent statistical review of the randomized phase 1 EBL06 trial evaluating ChAd3-EBO Z and MVA-EBO Z in a short-interval prime-boost vaccination strategy for prevention of Ebola Virus Disease.

Study start: 2015-07  ·  Primary completion: 2016-01  ·  Status: Completed
Registry record

This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.

1. Trial at a Glance

EBL06 was a randomized, parallel, unmasked phase 1 study with 40 participants and two intervention arms. Its primary purpose was prevention, and the registered primary endpoint evaluated the safety and tolerability of administration of ChAd3-EBO Z and MVA-EBO Z 7 days later over a 26-week time frame.

40
Enrollment
2 intervention arms
2
Arms
Parallel design
1
Phase
Phase 1
0/20
Serious AEs
Group 1
FeatureEBL06
Study titleEvaluation of a New Ebola Vaccine Using a Short-interval Prime-boost Vaccination
PhasePhase 1
ConditionEbola Virus Disease
Primary purposePrevention
Enrollment40
AllocationRandomized
Design modelParallel
MaskingNone
Number of arms2
Lead sponsorUniversity of Oxford
Sponsor typeOther
Study start2015-07
Primary completion2016-01
ClinicalTrials.govNCT02485912

2. Clinical Question

The central clinical question was whether a short-interval prime-boost vaccination strategy using ChAd3-EBO Z followed by MVA-EBO Z 7 days later could be administered with an acceptable safety and tolerability profile in the context of prevention of Ebola Virus Disease.

Population

The registry identifies the condition as Ebola Virus Disease and the primary purpose of the study as prevention. Enrollment was 40 participants.

Intervention strategy

The registered primary endpoint describes administration of ChAd3-EBO Z followed by MVA-EBO Z 7 days later.

Comparator structure

The study used two randomized parallel arms without masking. The registry identifies the interventions as ChAd3-EBO Z and MVA-EBO Z.

Primary question

What was the safety and tolerability profile of the short-interval vaccination strategy over the registered 26-week assessment period?

3. Trial Design

01
Enroll40 participants
02
Randomize2 parallel arms
03
VaccinateChAd3-EBO Z and MVA-EBO Z
04
AssessSafety and tolerability
05
Follow26 weeks
INTERVENTION · BIOLOGICAL

ChAd3-EBO Z

  • Registered as a biological intervention.
  • Part of the short-interval prime-boost vaccination strategy.
  • The primary endpoint specifies ChAd3-EBO Z followed by MVA-EBO Z 7 days later.
INTERVENTION · BIOLOGICAL

MVA-EBO Z

  • Registered as a biological intervention.
  • Administered 7 days later according to the primary endpoint wording.
  • Part of the short-interval prime-boost vaccination strategy.
Allocation
Randomized allocation was used.
Design model
Parallel-group design with 2 arms.
Masking
None.
Primary purpose
Prevention.

4. Enrollment and Study Timeline

2015-07 · Study start

Study initiation

The registry records the EBL06 study start as 2015-07.

2016-01 · Primary completion

Primary completion

The registry records the primary completion date as 2016-01.

Completed · Registry status

Completed study

The registry status is listed as COMPLETED, with enrollment of 40 participants.

5. Primary Endpoint

EndpointDefinition / assessmentTime frame
Safety and Tolerability of Administration of ChAd3-EBO Z and MVA-EBO Z 7 Days Later This will be done by recording the number of participants who experience adverse events and the severity of any adverse events. The registry describes active and passive collection of adverse-event data, including:
  • Occurrence of solicited local reactogenicity signs and symptoms for 7 days following the vaccination
  • Occurrence of solicited systemic reactogenicity signs and symptoms for 7 days following the vaccination
  • Occurrence of unsolicited adverse events for 28 days following the vaccination
  • Change from baseline for safety laboratory measures
  • Occurrence of serious adverse events during the whole study duration
26 weeks

The endpoint is therefore broader than a single adverse-event count. It combines solicited local and systemic reactogenicity, unsolicited adverse events, laboratory changes, and serious adverse events across different observation windows within the overall 26-week endpoint time frame.

Solicited local reactogenicity

Specific local signs and symptoms were to be collected for 7 days following vaccination.

Solicited systemic reactogenicity

Specific systemic signs and symptoms were to be collected for 7 days following vaccination.

Unsolicited adverse events

Unsolicited adverse events were to be assessed for 28 days following vaccination.

Serious adverse events

Serious adverse events were to be assessed during the whole study duration.

6. Planned Analysis

The registry reports that results have been posted, but it does not post formal statistical analyses for the primary endpoint. The available endpoint is fundamentally a safety and tolerability endpoint, so the principal statistical task is to describe the number and severity of adverse events and other safety findings within each randomized arm and over the specified observation periods.

For adverse-event outcomes, a typical analysis would summarize participants experiencing at least one event, as well as event severity and clinically relevant event categories. Because participants can experience more than one adverse event, participant-level incidence and event-level counts answer different questions and should not be treated as interchangeable.

For binary safety outcomes, proportions within each arm are natural descriptive measures. With only 40 enrolled participants, exact or other small-sample confidence intervals can be useful for showing the uncertainty around those proportions. A confidence interval around an adverse-event proportion describes uncertainty in the estimated rate; it does not describe the probability that an individual participant will experience an event.

For safety laboratory measures, the registered endpoint specifies change from baseline. A longitudinal analysis would normally preserve the repeated-measure structure and distinguish within-participant change from between-arm differences. The registry, however, does not report a formal statistical model for these laboratory measurements.

Registry reporting: ClinicalTrials.gov reports that results are posted for the primary endpoint, but no formal statistical analyses are posted in the record information available here. The reported serious-adverse-event data are therefore presented descriptively rather than as the result of a reported hypothesis test.

7. Reported Safety Results

The registry reports serious adverse events by arm as 0/20 in Group 1 and 0/20 in Group 2. The available arm-level result therefore indicates that no participants in either group were reported as affected by a serious adverse event in the registry's posted result.

Safety measureGroup 1Group 2
Participants affected by serious adverse events0/200/20
Participants affected by serious adverse events
Group 1
0/20
Group 2
0/20
Clinical Biostats interpretation

The reported result is 0 serious adverse events among 20 participants in Group 1 and 0 among 20 participants in Group 2. This describes the observed serious-adverse-event experience in the enrolled groups.

It does not establish that the true serious-adverse-event rate is zero, because an observed count of zero in a small study still leaves statistical uncertainty about the underlying event rate.

There is also no reported effect estimate, confidence interval, or p-value for this comparison. A formal between-group test would generally be of limited interpretive value in a small phase 1 safety study, particularly when no events are observed. Confidence intervals or other exact small-sample methods are often more informative for communicating the uncertainty around an observed event proportion.

The result should also be distinguished from the broader primary endpoint. The registry endpoint includes solicited local and systemic reactogenicity, unsolicited adverse events, safety laboratory changes, and serious adverse events; the posted arm-level ClinicalTrials.gov record concerns serious adverse events specifically.

8. Statistical Methodology

Descriptive safety analysis

The central statistical role of the primary endpoint is descriptive. Safety data are commonly summarized by the number and proportion of participants experiencing an adverse event, together with severity and event categories. This preserves the distinction between how many participants were affected and how many total events occurred.

Participant-level incidence
Incidence proportion = participants with ≥1 event ÷ participants evaluated

For the posted serious-adverse-event result, the numerator is 0 in each group and the denominator is 20 in each group.

Severity is part of the endpoint

The registry explicitly defines the primary endpoint in terms of both the number of participants who experience adverse events and the severity of any adverse events. Consequently, a simple ever/never adverse-event indicator would not fully capture the registered endpoint.

Different observation windows

The safety endpoint contains multiple follow-up windows. Solicited local and systemic reactogenicity were collected for 7 days following vaccination, unsolicited adverse events for 28 days following vaccination, and serious adverse events during the whole study duration. These windows should be kept separate when interpreting event frequencies because they define different periods of observation.

Safety laboratory changes

The registry specifies change from baseline for safety laboratory measures. Statistically, change scores can be described within each treatment group and compared between groups when the underlying measurement and analysis plan support such a comparison. Repeated laboratory measurements may also require methods that account for correlation within participants.

Small-sample inference

With an enrollment of 40 participants and 20 participants reported in each serious-adverse-event group, safety estimates can be imprecise. In particular, zero observed events do not imply a zero underlying probability. Exact binomial confidence intervals are one established way to quantify uncertainty for a proportion when sample sizes are small.

9. Statistical Methods Explained

Why is safety primarily descriptive in a phase 1 study?

A phase 1 safety endpoint is generally intended to characterize tolerability and identify adverse events rather than to establish a definitive comparative efficacy claim. Counts, proportions, severity categories, and laboratory changes can therefore be more informative than reducing the safety experience to a single p-value.

What does 0/20 mean?

A result of 0/20 means that 0 participants among 20 evaluated participants in that group were reported as affected by the specified serious adverse event outcome. It is an observed proportion of zero, not proof that the underlying risk is zero.

Why does zero events still have uncertainty?

Statistical inference concerns the underlying event probability, not only the observed sample. Even if no events occur among 20 participants, another sample from the same underlying population could contain an event. Small samples therefore produce substantial uncertainty around apparently simple safety proportions.

Why separate solicited and unsolicited adverse events?

Solicited events are actively sought according to predefined symptoms or signs, while unsolicited events arise without the same prespecified solicitation framework. The EBL06 endpoint also assigns different observation windows to these categories, so combining them without preserving their definitions could obscure the intended safety assessment.

Why does severity matter?

Two treatment groups could have the same number of participants experiencing at least one adverse event while having very different distributions of severity. The registered endpoint therefore includes severity as an explicit component of the safety and tolerability assessment.

What would a confidence interval add to 0/20?

A confidence interval would communicate the uncertainty surrounding the observed proportion. This is especially important when the observed event count is zero: the estimate itself is simple, but the plausible range for the underlying event probability can still be materially wider than zero.

10. Analysis Populations and Interpretation

The registry reports enrollment of 40 participants and a randomized, parallel design with two arms. It does not provide a separate detailed definition of an intention-to-treat population, a safety population, or another analysis population in the information summarized here.

For randomized trials, the distinction between the randomized population and the participants actually exposed to an intervention can be important for safety interpretation. A safety analysis generally relates adverse events to actual exposure, whereas the randomized population is the natural reference population for preserving the treatment assignment created by randomization.

Statistical conceptWhat can be established from the registry
RandomizationThe study is registered as randomized.
Parallel allocationThe design model is parallel with 2 arms.
MaskingThe study is registered as having no masking.
Enrollment40 participants.
Arm-level serious adverse events0/20 in Group 1 and 0/20 in Group 2.
Formal statistical modelNo formal statistical analyses are posted in the registry information available here.

11. Interpreting the Serious-Adverse-Event Result

Observed serious adverse events

0/20 vs 0/20

Group 1: 0 affected of 20 at risk  ·  Group 2: 0 affected of 20 at risk

The most direct interpretation is that the registry reports no serious adverse events among the 20 participants at risk in either group. Because both observed counts are zero, there is no observed difference in the number of affected participants between the two groups.

What the result does mean

Within the participants represented by the posted result, the observed number of participants affected by serious adverse events was zero in both groups.

What the result does not mean

The result does not establish that the probability of a serious adverse event is exactly zero in the broader population. It also does not quantify the frequency or severity of non-serious adverse events, because those data are not represented by the reported 0/20 serious-adverse-event result.

Why the sample size matters

Each group contains 20 participants in the reported serious-adverse-event comparison. With this group size, an absence of observed serious adverse events provides limited information about very uncommon risks. Rare events can remain statistically difficult to characterize even when none are observed.

12. Endpoint Structure and Follow-up

Safety componentObservation periodStatistical interpretation
Solicited local reactogenicity signs and symptoms7 days following vaccinationTime-limited participant-level incidence and severity assessment.
Solicited systemic reactogenicity signs and symptoms7 days following vaccinationTime-limited participant-level incidence and severity assessment.
Unsolicited adverse events28 days following vaccinationParticipant-level incidence and event characterization over a longer post-vaccination window.
Safety laboratory measuresBaseline change within the registered endpointQuantitative change-from-baseline assessment.
Serious adverse eventsWhole study durationParticipant-level serious-event surveillance throughout study participation.

This structure illustrates an important statistical principle: the denominator and observation window are part of the meaning of a safety rate. A 7-day solicited reactogenicity rate and a whole-study serious-adverse-event rate should not be interpreted as though they were measurements of the same exposure period.

13. Randomization and Comparative Interpretation

Randomization creates the basic framework for comparing the two study arms. In a randomized parallel trial, treatment assignment is determined before subsequent outcomes are observed, reducing systematic differences in treatment assignment that could otherwise confound the comparison.

The absence of masking is also important to recognize. An unmasked design means participants and investigators were not blinded to treatment assignment. For objective outcomes, this may have less influence than for subjective outcomes, but safety reporting can include participant-reported symptoms and investigator-assessed events. The registry's inclusion of solicited reactogenicity therefore makes the unmasked design a relevant interpretive feature.

Randomization is not the same as large-sample precision. Randomization addresses the structure of the comparison; it does not by itself guarantee precise estimates. With 40 enrolled participants, uncertainty around uncommon safety outcomes remains important even in a randomized design.

14. Why This Trial Matters Statistically

EBL06 is a useful teaching example because its statistical problem is fundamentally different from a large confirmatory efficacy trial. The central issue is how to describe safety information accurately when the sample is small, the endpoint contains several event definitions and follow-up windows, and the available arm-level serious-adverse-event result contains no observed events.

ConceptHow it appears in EBL06
RandomizationThe study uses randomized allocation.
Parallel-group designTwo intervention arms are registered under a parallel design model.
Phase 1 safety assessmentThe primary purpose is prevention and the registered primary endpoint is safety and tolerability.
Participant-level incidenceThe endpoint records the number of participants experiencing adverse events.
Severity assessmentThe primary endpoint explicitly includes severity of adverse events.
Time-window definitionSolicited events, unsolicited events, laboratory changes, and serious adverse events have different assessment structures.
Small-sample inference40 participants were enrolled, with 20 at risk in each posted serious-adverse-event group.
Zero-event interpretation0/20 in each group requires careful distinction between an observed zero and a zero underlying risk.
Unmasked designThe registry identifies masking as none, which is relevant when interpreting subjective safety reporting.

15. Important Limitations and Interpretation Issues

16. What a Statistical Safety Analysis Should Separate

Participants affected

How many participants experienced at least one event? This is the natural basis for a participant-level incidence proportion.

Number of events

How many events occurred in total? This can exceed the number of affected participants when participants experience multiple events.

Severity

How severe were the events? The registered endpoint explicitly includes severity.

Observation period

Over what period was the event sought or observed? EBL06 uses 7-day, 28-day, and whole-study safety windows.

Keeping these dimensions separate prevents a common analytical error: treating a single adverse-event percentage as though it completely describes safety. In this trial, the endpoint definition itself signals that safety is multidimensional.

17. Clinical Interpretation vs Statistical Interpretation

Statistical interpretation

The posted serious-adverse-event result is 0/20 in Group 1 and 0/20 in Group 2. With no observed events in either group, there is no observed between-group difference for this particular outcome, while the small sample leaves uncertainty about the underlying event rate.

Clinical interpretation

The registry reports no serious adverse events among the 20 participants at risk in either group. The broader safety endpoint also includes local and systemic reactogenicity, unsolicited adverse events, and safety laboratory changes, so serious adverse events represent only one component of the overall safety assessment.

18. Statistical Questions Raised by the Design

Several features of EBL06 illustrate why trial design and statistical interpretation cannot be separated. Randomization provides a structured comparison, but the small enrollment limits precision. Lack of masking can matter for subjective safety assessments. The primary endpoint spans several event categories and time frames, so the analysis needs to preserve those definitions.

The short interval between the two registered vaccine components also makes the temporal definition of adverse events especially important. A safety event occurring during the 7-day solicited-reactogenicity window is not analytically identical to an unsolicited event observed through 28 days or a serious adverse event recorded during the whole study duration.

For a small safety study, descriptive estimates accompanied by appropriate measures of uncertainty can often communicate more information than a single significance test. The purpose is to show what was observed, how much information contributed to the estimate, and how much uncertainty remains.

19. Sources

Continue through the Clinical Biostats trial library

Explore additional clinical-trial analyses and statistical explanations across the Clinical Biostats library.

20. Record Summary

EBL06 is a randomized phase 1, parallel, unmasked study of 40 participants evaluating a short-interval prime-boost vaccination strategy involving ChAd3-EBO Z and MVA-EBO Z 7 days later. The registered primary endpoint is safety and tolerability over 26 weeks, incorporating solicited local and systemic reactogenicity, unsolicited adverse events, safety laboratory changes, and serious adverse events. The posted serious-adverse-event result is 0/20 in Group 1 and 0/20 in Group 2.

Statistically, the most important lesson is that a zero observed event count is not equivalent to zero underlying risk. With 20 participants at risk in each reported group, the observed result provides a direct description of the recorded serious-adverse-event experience but leaves uncertainty about uncommon events. Interpretation also requires attention to the endpoint's distinct observation windows, the study's unmasked design, and the difference between serious adverse events and the broader safety and tolerability endpoint.

Clinical Biostats interpretation: The value of a phase 1 safety analysis lies in disciplined description of what occurred, careful definition of denominators and observation periods, explicit treatment of severity, and appropriate communication of uncertainty—especially when the observed number of serious events is zero.