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
Beyond was a phase 3 clinical trial in relapsing-remitting multiple sclerosis evaluating three interferon beta-1b groups in a parallel design. The registry describes the allocation as non-randomized and the masking as double, with treatment as the primary purpose. The enrollment was 1420 participants.
| Feature | Beyond |
|---|---|
| Trial name | Beyond |
| Brief title | BEYOND Follow-up: Betaferon®/Betaseron® Efficacy Yielding Outcomes of a New Dose |
| Phase | Phase 3 |
| Condition | Multiple Sclerosis, Relapsing-Remitting |
| Therapeutic area | Neurology |
| Allocation | Non-randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 1420 |
| Number of arms | 3 |
| Start | May 2007 |
| Primary completion | March 2008 |
| Status | Completed |
| Lead sponsor | Bayer |
2. Clinical Question
The registry describes a phase 3 treatment study of interferon beta-1b in patients with relapsing-remitting multiple sclerosis. The primary registered questions concern the incidence of four safety-related clinical event categories over 309 days: flu-like syndrome, injection-site reactions, liver enzyme elevations, and hematological abnormalities.
Population
Participants with multiple sclerosis, relapsing-remitting, enrolled in the Beyond study.
Intervention
Interferon beta-1b (Betaseron, BAY86-5046), evaluated in three study groups.
Comparator structure
The registry describes three parallel interferon beta-1b groups rather than a randomized treatment-versus-placebo comparison.
Primary questions
How frequently were the prespecified flu-like, injection-site, liver-enzyme, and hematological event categories observed during 309 days?
3. Trial Design
IFNB-1b 500 mcg
- Interferon beta-1b (Betaseron, BAY86-5046)
- Serious adverse events: 11/586
IFNB-1b 250 mcg
- Interferon beta-1b (Betaseron, BAY86-5046)
- Serious adverse events: 16/645
IFNB-1b 250 mcg*
- Interferon beta-1b (Betaseron, BAY86-5046)
- Serious adverse events: 6/180
4. Registered Primary Endpoints
The registry lists four primary endpoints, each evaluated over 309 days. All four are safety-related event categories constructed from specified MedDRA terms.
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Flu-like-syndrome | 309 days | The variable consists of a combination of MedDRA terms, including Preferred Terms and Lower Level Terms, indicative for this condition. |
| Injection-site Reactions | 309 days | The variable consists of a combination of MedDRA Preferred Terms indicative for this condition. |
| Liver Enzyme Elevations | 309 days | The variable consists of a combination of MedDRA Preferred Terms indicative for this condition. |
| Hematological Abnormalities | 309 days | The variable consists of a combination of MedDRA Preferred Terms indicative for this condition. |
These endpoint definitions illustrate an important feature of adverse-event analysis: a clinically meaningful syndrome may be defined as a prespecified grouping of coded medical terms rather than as one individual adverse-event term.
5. Statistical Methodology
Safety analysis population
The registry states that the analysis set used for safety analysis comprised 1411 patients. Adverse events were defined as events with an onset date after the first application of study drug.
For the posted primary analyses, the reported estimate is the percentage of participants with the specified event category during the 309-day time frame. The registry describes the corresponding 95% confidence interval as being calculated for the number of patients with the event.
Binary endpoint structure
Each posted primary analysis is classified as a binary endpoint. For a binary safety outcome, the fundamental quantity is whether an eligible participant experienced the prespecified event category during the analysis window. The resulting incidence proportion can be expressed as the number of participants with the event divided by the relevant analysis population.
The registry reports percentage estimates and confidence intervals rather than a relative treatment-effect measure such as a risk ratio or odds ratio.
Confidence intervals
A 95% confidence interval accompanies each posted estimate for flu-like syndrome and injection-site reactions. The interval describes statistical uncertainty around the estimated incidence under the applicable sampling framework. A narrower interval generally indicates greater precision, whereas a wider interval indicates greater uncertainty.
No reported analysis method
The registry does not identify a specific statistical method or method category for the six posted analyses. The reported analyses are described as incidence rates with 95% confidence intervals. No p-values are reported for these analyses.
6. Results
The registry posts formal statistical analyses for the flu-like-syndrome and injection-site-reaction primary endpoints. Each reported estimate is accompanied by a 95% confidence interval. The posted analyses do not report p-values.
Flu-like-syndrome
The registry reports the percentage of participants with flu-like syndrome during the 309-day time frame for each of the three interferon beta-1b groups.
| Group | Estimate | 95% CI | Time frame |
|---|---|---|---|
| IFNB-1b 500 mcg | 17.4% | 14.4%–20.7% | 309 days |
| IFNB-1b 250 mcg | 15.8% | 13.1%–18.9% | 309 days |
| IFNB-1b 250 mcg* | 31.1% | 24.4%–38.4% | 309 days |
The reported incidence estimates were 17.4%, 15.8%, and 31.1% across the three listed groups. The estimates describe the proportion of participants meeting the registry's flu-like-syndrome definition during 309 days; they do not establish that one dose caused a higher or lower incidence than another.
The 95% confidence intervals quantify uncertainty around each incidence estimate: 14.4%–20.7%, 13.1%–18.9%, and 24.4%–38.4%, respectively. The interval is not a range containing individual patients' risks, nor does it imply that future study populations must fall inside the interval.
No p-value is reported. A p-value would address a prespecified hypothesis about a comparison or parameter under a statistical model; it would not measure the size or clinical importance of the observed incidence itself.
Because allocation was non-randomized, differences between the groups should not be interpreted as a clean causal comparison without accounting for potential differences between participants entering the groups.
Injection-site Reactions
The registry also reports the percentage of participants with injection-site reactions during 309 days.
| Group | Estimate | 95% CI | Time frame |
|---|---|---|---|
| IFNB-1b 500 mcg | 31.7% | 28.0%–35.7% | 309 days |
| IFNB-1b 250 mcg | 26.2% | 22.8%–29.8% | 309 days |
| IFNB-1b 250 mcg* | 31.7% | 24.9%–39.0% | 309 days |
The reported incidence estimates for injection-site reactions were 31.7%, 26.2%, and 31.7% across the three listed groups. These are observed percentages over 309 days using the registry's prespecified MedDRA-based endpoint definition.
The associated 95% confidence intervals were 28.0%–35.7%, 22.8%–29.8%, and 24.9%–39.0%. The wider interval for the third listed group is consistent with its smaller reported affected/at-risk safety count relative to the other two groups, although the confidence interval itself should be interpreted as uncertainty around the estimated incidence rather than as a prediction interval.
The estimates do not establish a treatment effect, and the confidence intervals do not constitute formal evidence that two groups are statistically different from one another. No comparative p-value is reported.
For a binary safety endpoint, an effect measure such as a risk difference or risk ratio could provide a direct comparison between groups, but neither is reported in the posted analyses.
Other registered primary endpoints
Liver enzyme elevations and hematological abnormalities are also registered primary endpoints with a 309-day time frame. The registry record identifies their MedDRA-based definitions, but the posted statistical analyses summarized here do not contain an estimate and 95% confidence interval for either endpoint.
| Registered endpoint | Time frame | Posted estimate in the statistical analyses |
|---|---|---|
| Liver Enzyme Elevations | 309 days | Not reported in the posted statistical analyses |
| Hematological Abnormalities | 309 days | Not reported in the posted statistical analyses |
7. Safety Results
The registry reports serious adverse events by study group using affected participants over participants at risk. These counts provide a separate safety perspective from the four registered primary endpoint categories.
| Group | Serious adverse events | At risk |
|---|---|---|
| IFNB-1b 500 mcg | 11 | 586 |
| IFNB-1b 250 mcg | 16 | 645 |
| IFNB-1b 250 mcg* | 6 | 180 |
These serious-adverse-event counts should be interpreted separately from the posted flu-like-syndrome and injection-site-reaction incidence estimates. The registry does not provide a comparative effect measure or confidence interval for these serious-adverse-event counts in the statistical analyses summarized here.
8. Statistical Methods Explained
Why are the primary endpoints treated as binary outcomes?
The posted analyses classify flu-like syndrome and injection-site reactions as binary endpoints. Each participant is classified according to whether the prespecified event category occurred during the 309-day assessment period. This produces an incidence percentage that is straightforward to describe clinically.
What does a 17.4% incidence mean?
For the IFNB-1b 500 mcg group, the registry reports that 17.4% of participants in the safety analysis experienced the prespecified flu-like-syndrome event category during 309 days. It does not mean that the event occurred continuously in 17.4% of patients, nor does it describe event severity or duration.
What does the 95% confidence interval add?
The estimate is a sample-based quantity. The 95% confidence interval provides a measure of statistical precision around that estimate. For example, the reported flu-like-syndrome estimate of 17.4% for the IFNB-1b 500 mcg group has a 95% confidence interval of 14.4%–20.7%. The interval therefore communicates more information than the point estimate alone.
Why is a p-value different from a confidence interval?
A confidence interval describes uncertainty around an estimated quantity, while a p-value evaluates the compatibility of observed data with a specified null hypothesis. Neither quantity is itself a measure of clinical importance. The posted analyses provide confidence intervals but do not provide p-values.
Why does non-randomized allocation matter?
Randomization helps balance measured and unmeasured prognostic factors across treatment groups in expectation. The registry classifies Beyond as non-randomized. Consequently, a difference in observed incidence between groups may reflect differences in the groups themselves as well as differences associated with the treatment regimens.
Why define adverse-event categories using MedDRA terms?
Clinical trials collect adverse events using standardized medical terminology. A syndrome such as flu-like syndrome may encompass multiple related Preferred Terms and Lower Level Terms. A prespecified grouping can therefore capture a clinically defined category more consistently than relying on one individual coded term.
Why should the 250 mcg* group be kept separate?
The registry reports the third group as IFNB-1b 250 mcg* and provides a separate at-risk count for it. The asterisk is part of the reported group label. Combining it with the other 250 mcg group would remove a distinction made in the registry and would change the denominator structure of the reported safety summaries.
9. Interpreting Incidence Rates and Confidence Intervals
The most informative reading of the posted safety results combines the percentage, its 95% confidence interval, the 309-day time frame, the safety-analysis population, and the non-randomized study design.
The point estimate is the most compact description of the observed incidence. The confidence interval adds precision. The time frame tells the reader over what period the event classification was made. The analysis population establishes the denominator context. Finally, the allocation method determines how confidently differences between groups can be interpreted as treatment effects.
Point estimate
The percentage summarizes the observed frequency of the prespecified event category during the 309-day period.
Confidence interval
The 95% interval communicates uncertainty around the incidence estimate rather than individual-level variability.
Analysis population
The safety analysis set comprised 1411 patients, and adverse events were defined by onset after the first application of study drug.
Study design
Because allocation was non-randomized, observed between-group differences require caution in causal interpretation.
10. Comparing the Posted Safety Estimates
The two primary endpoints with posted statistical analyses show different incidence patterns across the three listed groups.
| Endpoint | IFNB-1b 500 mcg | IFNB-1b 250 mcg | IFNB-1b 250 mcg* |
|---|---|---|---|
| Flu-like syndrome | 17.4% 95% CI 14.4%–20.7% | 15.8% 95% CI 13.1%–18.9% | 31.1% 95% CI 24.4%–38.4% |
| Injection-site reactions | 31.7% 95% CI 28.0%–35.7% | 26.2% 95% CI 22.8%–29.8% | 31.7% 95% CI 24.9%–39.0% |
Within this descriptive comparison, injection-site reactions were reported more frequently than flu-like syndrome in each listed group. The third listed group has a wider confidence interval for both endpoints than the other groups, consistent with its smaller reported number at risk.
However, a visual difference in percentages should not automatically be translated into a statistically significant treatment difference. Formal comparison would require a prespecified inferential framework, including an appropriate null hypothesis and treatment of the non-randomized design. Those comparative analyses are not reported here.
11. Analysis Population and Adverse-Event Timing
The safety analysis set comprised 1411 patients. The registry defines an adverse event for these analyses as an event with an onset date after the first application of study drug.
This timing definition is important because an adverse event analysis depends on when follow-up begins. Restricting the definition to events with onset after first study-drug application establishes the temporal relationship used for the safety summaries.
12. Registry Design Features That Affect Interpretation
| Design feature | Registry description | Statistical implication |
|---|---|---|
| Allocation | Non-randomized | Between-group differences may reflect participant or enrollment differences in addition to treatment. |
| Design model | Parallel | Participants are evaluated within separate study groups rather than through a crossover comparison. |
| Masking | Double | Masking can reduce some sources of observational and assessment bias, although it does not replace randomization. |
| Primary purpose | Treatment | The trial is designed around treatment evaluation. |
| Endpoint time frame | 309 days | Reported incidence estimates are tied to this specified assessment period. |
The combination of double masking and non-randomized allocation is particularly important to distinguish. Masking concerns awareness of treatment assignment; randomization concerns how participants are assigned to treatment groups. These are separate design properties with different statistical implications.
13. Study Termination and Missing Functional Assessments
The registry reports that the study was terminated before the start of Phase B according to protocol-defined stopping rules. As a result, functional assessments of multiple sclerosis using FAMS and EuroQoL using EQ-5D were not performed.
This matters statistically because a planned endpoint that is not collected cannot contribute an observed treatment comparison. The available registry results therefore center on the safety-related endpoints for which analyses were posted.
14. What the Confidence Intervals Tell Us
The reported 95% confidence intervals are 14.4%–20.7% for IFNB-1b 500 mcg, 13.1%–18.9% for IFNB-1b 250 mcg, and 24.4%–38.4% for IFNB-1b 250 mcg*. These intervals provide a statistical measure of precision around the corresponding incidence estimates of 17.4%, 15.8%, and 31.1%.
The reported 95% confidence intervals are 28.0%–35.7%, 22.8%–29.8%, and 24.9%–39.0% for the three groups, respectively. The intervals should be considered alongside their group-specific denominators and the non-randomized allocation.
A confidence interval around an incidence estimate does not establish whether one treatment group is superior to another. To make that inference, a comparative analysis with a specified null hypothesis and appropriate handling of the study design would be required.
15. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The posted analyses provide incidence estimates and 95% confidence intervals for flu-like syndrome and injection-site reactions over 309 days. The statistical-analysis record does not provide comparative p-values or a relative effect measure.
Clinical interpretation
The reported percentages describe how frequently the registry-defined event categories occurred in each listed study group. Their clinical meaning depends on the nature of the events, the study population, follow-up period, and the broader safety context.
The distinction is useful because statistical precision and clinical importance are different concepts. A precise incidence estimate may still describe an outcome whose clinical implications require additional context, while a clinically notable difference may require a formal comparative analysis before a causal interpretation is justified.
16. Important Limitations and Interpretation Issues
- Non-randomized allocation: the registry classifies allocation as non-randomized, so observed differences between study groups cannot automatically be attributed to treatment.
- No reported comparative effect measure: the posted analyses report incidence percentages with 95% confidence intervals rather than risk ratios, risk differences, odds ratios, or hazard ratios.
- No reported p-values: the statistical analyses summarized here do not provide p-values for the reported endpoint estimates.
- Incomplete primary-endpoint results: liver enzyme elevations and hematological abnormalities are registered primary endpoints, but the posted statistical analyses summarized here do not contain corresponding estimates and confidence intervals.
- Endpoint construction: the clinical event categories are constructed from MedDRA terms, so the definition depends on the prespecified coding framework.
- Fixed time frame: the primary endpoint analyses use a 309-day time frame, so the estimates should not be interpreted as lifetime or unrestricted follow-up incidence.
- Study termination: termination before the start of Phase B according to protocol-defined stopping rules meant that FAMS and EQ-5D assessments were not performed.
- Denominator differences: serious adverse-event summaries use group-specific numbers at risk, so raw event counts are not directly comparable without considering the denominators.
17. Why This Trial Matters Statistically
Beyond is a useful teaching case because it demonstrates how safety endpoints can be analyzed and interpreted in a multicohort clinical trial without automatically treating descriptive incidence differences as causal treatment effects.
| Concept | How it appears in Beyond |
|---|---|
| Binary endpoint | Flu-like syndrome and injection-site reactions are analyzed as event/non-event outcomes. |
| Incidence proportion | The posted analyses report the percentage of participants experiencing each specified event category. |
| Confidence interval | Each posted incidence estimate is accompanied by a 95% confidence interval. |
| Safety analysis population | The safety analysis set comprised 1411 patients. |
| Adverse-event timing | Events were defined by onset after the first application of study drug. |
| MedDRA grouping | Primary safety categories were constructed from specified MedDRA terms. |
| Non-randomized allocation | Between-group differences require caution because treatment allocation was not randomized. |
| Denominator interpretation | Serious adverse-event summaries report affected participants and participants at risk separately for each group. |
| Study termination | Termination before Phase B prevented performance of FAMS and EQ-5D assessments. |
18. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
19. Related Statistical Calculators
20. Sources
- ClinicalTrials.gov: Beyond, NCT00459667.
- PubMed: PMID 21952094.
Continue through Clinical Biostats
Explore statistical tutorials and calculators that provide deeper explanations of the methods used to interpret clinical-trial evidence.
21. Record Summary
Beyond was a completed phase 3, double-masked, parallel clinical trial with non-randomized allocation and 1420 enrolled participants. Its registered primary endpoints focused on four safety-related event categories over 309 days. The posted statistical analyses provide incidence estimates with 95% confidence intervals for flu-like syndrome and injection-site reactions using a 1411-patient safety analysis set. The reported estimates were 17.4%, 15.8%, and 31.1% for flu-like syndrome and 31.7%, 26.2%, and 31.7% for injection-site reactions across the three listed groups.
The most important statistical distinction is between describing observed incidence and estimating a causal treatment effect. Confidence intervals add information about precision around the reported percentages, but the absence of a reported comparative effect measure or p-value and the non-randomized allocation mean that the group-specific incidence estimates should be interpreted descriptively. The registry also reports that the study was terminated before Phase B according to protocol-defined stopping rules, with FAMS and EQ-5D assessments therefore not performed.