This page separates reported trial information from statistical interpretation. The ClinicalTrials.gov record reports the primary endpoint and outcome measures, but do not provide a formal statistical-analysis table with effect estimates, confidence intervals, or p-values.
1. Trial at a Glance
BADGER was a randomized, quadruple-masked, phase 3 crossover trial in asthma. The study compared three step-up therapy strategies and defined its primary endpoint around whether participants demonstrated a differential response to those therapies according to fixed thresholds for three asthma-control measures.
| Feature | BADGER |
|---|---|
| Trial name | Childhood Asthma Research and Education (CARE) Network Trial - Best Add-On Therapy Giving Effective Response (BADGER) |
| NCT identifier | NCT00395304 |
| Phase | Phase 3 |
| Condition | Asthma |
| Allocation | Randomized |
| Design model | Crossover |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 182 |
| Interventions | Fluticasone propionate + montelukast; fluticasone propionate; fluticasone propionate + salmeterol |
| Primary endpoint type | Count / rate |
| Results posted | Yes |
| Outcome measures posted | 14 |
| Lead sponsor | Milton S. Hershey Medical Center |
2. Clinical Question
The BADGER trial addresses a clinically important statistical question: when a participant with asthma needs step-up therapy, do different therapeutic strategies produce different responses in different participants?
Population
The registry identifies the condition under study as asthma. The ClinicalTrials.gov record does not provide a more detailed eligibility or baseline-population description.
Interventions
Three step-up therapy strategies were evaluated: fluticasone propionate + montelukast, fluticasone propionate, and fluticasone propionate + salmeterol.
Comparator
Because the design is a crossover comparison of three therapies, each therapy is part of the within-participant treatment sequence rather than a simple permanent parallel-group comparator.
Primary question
How many participants demonstrate a differential response to the three step-up therapies according to the registry's fixed threshold criteria?
3. Trial Design
The defining feature of BADGER is its crossover design. In a conventional parallel-group trial, each participant is generally assigned to one treatment strategy and remains associated with that strategy for the principal comparison. In a crossover trial, participants can contribute observations under more than one treatment condition. That creates a different statistical structure: within-participant comparisons can become important because each participant can provide information about response to multiple therapies.
Randomization helps determine treatment sequence rather than simply assigning participants to one permanent treatment group.
The registry identifies the study as quadruple-masked. The ClinicalTrials.gov record does not specify the individual categories of people covered by the masking designation.
The same participant can provide information under different step-up therapy conditions, creating paired or repeated-measure information.
The registry classifies the primary purpose as treatment.
Why randomization is still important in a crossover trial
Randomization does not become unnecessary simply because participants receive multiple treatments. The order in which treatments are received can itself influence the observed response. Randomizing treatment sequences helps prevent systematic differences in sequence assignment from becoming confounded with treatment comparisons.
Why crossover changes the analysis
A crossover design creates observations that are not independent in the same way as observations from unrelated participants in a parallel-group trial. Measurements from the same participant can be correlated. A suitable analysis therefore needs to recognize the repeated contribution of each participant and, where relevant, account for treatment period and sequence.
4. Treatment Strategies
Fluticasone propionate + montelukast
- Drug intervention
- One of the three step-up therapies evaluated in the crossover trial
Fluticasone propionate
- Drug intervention
- One of the three step-up therapies evaluated in the crossover trial
Fluticasone propionate + salmeterol
- Drug intervention
- One of the three step-up therapies evaluated in the crossover trial
Six registered arms
- the ClinicalTrials.gov record reports 6 arms.
- The registry-reported intervention list contains the three step-up therapy strategies.
The combination of 6 registered arms and 3 named step-up therapies is consistent with a crossover study in which treatment sequences or other arm-level assignments are represented separately in the registry. The ClinicalTrials.gov record does not specify the individual six-arm sequence structure, so the page does not reconstruct it.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Primary endpoint | The Number of Participants With a Differential Response to the Three Step-up Therapies Based on Fixed Threshold Criteria for the Following Three Asthma Control Measures: Use of Oral Prednisone for Acute Asthma Exacerbations, Asthma Control Days and FEV1. | Measured during the last 12 weeks of each 16-week treatment period | Count / rate |
The registry's primary endpoint is unusual compared with a conventional single-outcome superiority endpoint. Rather than asking only whether one treatment has a larger mean response than another, the endpoint identifies participants whose responses differ across the three step-up therapies according to fixed threshold criteria.
Threshold-based response classification
The registry definition specifies that one treatment period was ranked as better than another if the total amount of prednisone received during the period was at least 180 mg less, if the number of annualized asthma-control days during the final 12 weeks of the period was increased by at least 31 days, or if FEV1 at the end of the period was at least 5% higher.
6. Planned Analysis
ClinicalTrials.gov reports that results are posted and that 14 outcome measures are available, but the ClinicalTrials.gov record contains no formal statistical analyses. Consequently, there are no registry-provided hazard ratios, mean differences, confidence intervals, p-values, or other formal comparative estimates to reproduce here.
Primary endpoint status
The registry identifies the primary endpoint and its measurement window, but the ClinicalTrials.gov recordset contains no formal analysis record for that endpoint.
What would typically be analysed?
For a crossover endpoint based on participant-level differential response, the analysis would ordinarily begin with the prespecified response classification for each participant and treatment period. The primary comparison would then need to respect the crossover structure rather than treating every treatment-period observation as an independent observation.
Depending on the prespecified estimand and endpoint construction, an appropriate analysis could involve a comparison of paired treatment-period outcomes, a model for repeated binary or categorical response classifications, or another crossover-specific model. If treatment-period measurements are analysed directly, treatment, period, sequence, and within-participant dependence may need consideration.
Unit of analysis
The primary endpoint is expressed as a number of participants with differential response. That makes participant-level classification central to the analysis.
Repeated observations
Because the design is crossover, a participant can contribute information under multiple therapies. Those observations are correlated within participant.
Thresholds
The fixed thresholds define what counts as a clinically meaningful ranking between treatment periods for the primary endpoint.
Formal inference
The ClinicalTrials.gov record does not identify the inferential model, standard error calculation, confidence intervals, or p-value for the primary endpoint.
7. Statistical Methodology
Crossover analysis
The statistical structure of BADGER is fundamentally different from that of a simple parallel-group trial. In a crossover study, each participant can act as a partial internal comparator because the participant receives more than one treatment condition. This can reduce between-participant variability when the analysis focuses on within-participant contrasts.
The important feature is that both responses can come from the same participant. The two observations are therefore not treated as if they came from unrelated people.
Period effects
A crossover analysis generally needs to consider the possibility that outcomes differ because of when treatment was received rather than because of treatment itself. A period effect could arise if asthma control changes over calendar time, treatment period, season, or another time-related factor.
The ClinicalTrials.gov record does not state whether a period effect was included in the formal analysis. Accordingly, period adjustment is discussed here as an analytical issue rather than as a documented BADGER analysis.
Sequence effects
In a crossover study, treatment order can matter. If participants who receive Therapy A first differ systematically in their subsequent response from participants who receive Therapy B first, the treatment comparison can become entangled with sequence. Randomization of treatment sequence is designed to protect against this type of systematic imbalance.
Carryover
Carryover occurs when the effect of a treatment persists into a later treatment period and influences the response attributed to the next treatment. Whether carryover is plausible, how it is assessed, and how it is handled are important design and analysis questions in crossover studies.
The registry-reported BADGER data do not report a carryover assessment or statistical adjustment. The existence of a crossover design alone therefore does not justify claiming that carryover was absent or successfully controlled.
Participant-level differential response
The primary endpoint is particularly suited to a responder-classification framework. Instead of estimating only an average difference, the trial asks whether participants meet prespecified thresholds indicating that one treatment period is better than another.
The primary endpoint therefore turns continuous or count-based treatment-period measurements into a participant-level classification according to fixed criteria.
Count / rate endpoints
The registry classifies the primary endpoint type as Count / rate. For a count-based endpoint, analysis can involve comparing the number or proportion of participants meeting the specified response definition. The correct inferential method depends on the exact endpoint construction, whether participants can contribute to more than one category, and how the crossover structure is represented.
8. Statistical Methods Explained
Why was a crossover design used?
A crossover design is useful when investigators want participants to experience multiple treatment conditions and then compare their responses across those conditions. This can make each participant informative about more than one therapy and can reduce some sources of between-person variability. The trade-off is that treatment-period, sequence, and carryover effects become part of the statistical problem.
Why does randomization still matter?
Randomization helps determine who receives particular treatment sequences and reduces systematic allocation differences. In a crossover trial, randomization can therefore protect the treatment comparison from being confounded by the order in which therapies are administered.
What does "differential response" mean?
It means that response is not assumed to be identical across treatments for every participant. The BADGER primary endpoint uses fixed threshold criteria to classify whether one treatment period ranks better than another based on prednisone use, asthma-control days, or FEV1.
Why use fixed thresholds?
A threshold provides a prespecified definition of what constitutes a sufficiently large difference to count as a better treatment period. This differs from simply asking whether two group means are statistically different. The threshold therefore has a clinical-interpretation role in addition to its statistical role.
Why can the same participant be an advantage statistically?
When two treatment responses come from the same participant, stable characteristics of that participant are shared by both observations. A within-participant comparison can therefore remove some person-to-person variability that would remain in an unrelated parallel-group comparison.
Why can the same participant also create a statistical complication?
Repeated measurements from one participant are correlated. Standard methods that assume every observation is independent can underestimate uncertainty if that correlation is ignored. Crossover analysis therefore needs a method that reflects the repeated-measures structure.
What cannot be concluded from the ClinicalTrials.gov record?
The ClinicalTrials.gov record does not provide a formal statistical-analysis record for the primary endpoint.
9. Endpoint Measurement Window
16-week treatment period
The registry specifies a 16-week treatment period for the primary endpoint.
Last 12 weeks
The primary endpoint was measured during the last 12 weeks of each 16-week treatment period.
The distinction between the full treatment period and the endpoint assessment window is statistically meaningful. It indicates that the primary endpoint was not simply calculated from every observation during the entire treatment period. The registry specifically anchors the primary measurement to the last 12 weeks.
10. Interpreting the Three Primary Response Measures
Oral prednisone use
The registry uses total prednisone received during a treatment period as one criterion for ranking treatments, with a threshold of at least 180 mg less.
Asthma-control days
The registry uses annualized asthma-control days during the final 12 weeks, with an increase of at least 31 days serving as one ranking criterion.
FEV1
The registry uses FEV1 at the end of the treatment period, with a difference of at least 5% higher serving as one ranking criterion.
Combined decision rule
The primary endpoint combines these treatment-ranking criteria into a participant-level differential-response assessment.
These three measures operate on different scales and capture different dimensions of asthma control. Prednisone use reflects treatment required for acute exacerbations; asthma-control days reflect control over time; and FEV1 provides a pulmonary-function measure.
Because the measures are heterogeneous, the fixed threshold rule is especially important. Without a prespecified decision rule, investigators could face substantial flexibility in deciding which measure should determine whether one treatment is considered better for an individual participant.
11. Safety
The ClinicalTrials.gov record reports serious adverse events as 7/182 among all participants. The dataset does not provide a treatment-arm breakdown of those serious adverse events.
| Safety measure | Reported value |
|---|---|
| Serious adverse events | 7/182 among all participants |
| Arm-specific serious adverse events | Not provided in the ClinicalTrials.gov record |
This distinction matters. An aggregate safety count cannot be converted into treatment-specific rates without knowing the number of affected participants and the corresponding number at risk within each treatment arm or sequence.
12. Results Reporting Status
ClinicalTrials.gov identifies BADGER as having results posted and reports 14 outcome measures. However, the registry-reported statistical-analysis section is empty: there are no formal statistical analyses in the ClinicalTrials.gov record.
| Reporting element | Status in the ClinicalTrials.gov record |
|---|---|
| Results posted | Yes |
| Outcome measures posted | 14 |
| Statistical analyses reported | None |
| Formal primary-endpoint effect estimate | Not provided |
| Confidence interval for primary endpoint | Not provided |
| P-value for primary endpoint | Not provided |
This is an important distinction for evidence interpretation. A registry can contain outcome data without presenting the inferential analysis needed to quantify uncertainty around a treatment comparison.
13. What a Formal Crossover Analysis Would Need to Establish
A rigorous analysis of BADGER would need to make several analytical choices explicit. These choices are not reported in the ClinicalTrials.gov record, so they are presented here as the questions an analyst would need to resolve rather than as claims about what the investigators actually did.
| Analytical issue | Question to resolve |
|---|---|
| Treatment effect | How is differential response attributed to each of the three therapies? |
| Participant dependence | How is correlation among repeated treatment-period observations handled? |
| Period effect | Does the analysis account for differences between treatment periods? |
| Sequence effect | Does treatment order contribute systematically to the observed response? |
| Carryover | Could effects from one treatment persist into a subsequent treatment period? |
| Missing data | How are incomplete treatment periods or unavailable endpoint measurements handled? |
| Multiplicity | How are comparisons among three therapies controlled when more than one treatment contrast is examined? |
These questions are not merely technical details. Each can change the estimated uncertainty or the interpretation of a treatment comparison. A crossover analysis that ignores within-participant correlation, for example, can produce standard errors that do not reflect the actual information structure.
14. Missing Data and Incomplete Treatment Periods
The ClinicalTrials.gov record does not specify an imputation strategy, missing-data model, censoring rule, or handling of participants who do not complete a treatment period.
That omission is especially relevant for a crossover endpoint because the primary endpoint is measured separately for treatment periods. Missing information from one period can prevent a complete within-participant comparison even when information from another period remains available.
Complete crossover information
A participant with usable endpoint information across the relevant treatment periods can contribute directly to within-participant treatment comparisons.
Incomplete information
A participant missing one or more treatment-period measurements may require a prespecified rule for inclusion in the differential-response endpoint.
For this reason, the statistical analysis should clearly distinguish the number randomized from the number contributing evaluable treatment-period data. The ClinicalTrials.gov record provides the randomized enrollment total of 182 but do not provide those additional analysis-population counts.
15. Multiplicity in a Three-Treatment Crossover Trial
Three treatment strategies create multiple possible pairwise comparisons. Even when the primary endpoint is expressed as a single differential-response measure, the underlying treatment-ranking process can involve more than one contrast.
These are the three possible pairwise treatment contrasts among three therapies. Whether all three are formally tested is a separate question that must be determined from the prespecified statistical analysis.
Multiplicity becomes important when several hypotheses are tested and the investigator wants to maintain a specified familywise type I error rate. The registry-reported BADGER data do not report an alpha-adjustment procedure, hierarchical testing strategy, or multiplicity plan, so none is attributed to the trial here.
16. Crossover, Carryover, and Period Effects
Crossover is not the same as repeated parallel trials
It is tempting to treat each treatment period as if it were an independent mini-trial. That is not generally valid. The same participant contributes observations across periods, so the responses share participant-level characteristics.
Carryover can complicate causal interpretation
If an earlier therapy changes the response observed during a later treatment period, the later measurement may reflect both the current treatment and residual effects from the previous treatment. The statistical and clinical relevance of carryover depends on the therapies, treatment-period length, washout structure, and endpoint.
Period effects can work in the opposite direction
Even in the absence of carryover, participants may change over time. Seasonal variation, natural disease variation, or other temporal influences can make outcomes in one treatment period systematically different from outcomes in another period.
17. Blinding and Randomization
The trial was randomized and quadruple-masked. Both features strengthen the design in different ways.
Randomization
Random assignment reduces systematic differences in treatment-sequence allocation and supports a causal treatment comparison when the trial is analysed according to its randomized structure.
Quadruple masking
Masking can reduce the influence of treatment knowledge on participant behavior, assessment, treatment delivery, or other study processes. The ClinicalTrials.gov record does not specify the four masked categories.
Masking is particularly relevant when outcomes can be influenced by behavior, reporting, clinical management, or assessment. It does not, however, eliminate statistical dependence caused by repeated measurements within the same participant.
18. Trial Timeline
Study start
The registry lists March 2007 as the study start.
Randomized crossover study
The trial was conducted as a randomized, quadruple-masked, phase 3 crossover study in asthma.
Primary completion
The registry lists December 2011 as the primary completion date.
Registry status
the ClinicalTrials.gov record lists the study status as COMPLETED.
19. Limitations
- Incomplete formal analysis information: the ClinicalTrials.gov recordset contains no formal comparative estimates, confidence intervals, or p-values for the primary endpoint.
- Crossover complexity: repeated treatment exposure creates dependence among observations and requires analysis that respects the within-participant structure.
- Period and sequence effects: treatment order and treatment period can influence observed responses in crossover trials and need appropriate prespecified handling.
- Carryover: residual treatment effects can complicate interpretation when one therapy is followed by another. The ClinicalTrials.gov record does not report how this issue was addressed.
- Missing data: the ClinicalTrials.gov record does not specify an imputation strategy or the handling of incomplete treatment periods.
- Multiplicity: three therapies create multiple possible contrasts, but the ClinicalTrials.gov record does not state how multiplicity was handled.
- Safety duration and sample size: the registry specifically cautions that the study was not designed or powered to evaluate the long-term safety of long-acting beta-agonists in children. The duration of the trial and its sample size preclude statements regarding long-term risks.
- Safety by treatment: serious adverse events are reported only as 7/182 among all participants; arm-specific counts are not reported.
20. Why This Trial Matters Statistically
BADGER is a useful teaching case because its central statistical question is different from the usual "Which treatment has the larger average?" framework. It asks whether participants differ in their response to alternative step-up therapies and uses a fixed-threshold algorithm to define differential response.
| Concept | How it appears in BADGER |
|---|---|
| Randomization | Participants were randomized in a phase 3 trial. |
| Blinding | The study was quadruple-masked. |
| Crossover design | Participants were evaluated across different step-up therapy conditions. |
| Within-participant comparison | The crossover structure allows treatment-period responses to be compared within participants. |
| Fixed thresholds | Prednisone, asthma-control days, and FEV1 thresholds define treatment-period ranking. |
| Count / rate endpoint | The primary endpoint is classified by the registry as Count / rate. |
| Multiple outcomes | The primary differential-response definition incorporates three asthma-control measures. |
| Period effects | Potential time-period differences are an important crossover-analysis consideration. |
| Carryover | Residual effects from an earlier treatment can affect later-period interpretation. |
| Multiplicity | Three treatment strategies create multiple possible treatment contrasts. |
| Missing data | Incomplete treatment periods can affect participant-level crossover comparisons. |
| Safety interpretation | Overall serious adverse events are reported as 7/182, without arm-specific counts. |
21. Clinical Biostats Interpretation of the Primary Endpoint
The primary endpoint is designed to identify the number of participants with a differential response to the three step-up therapies under a prespecified set of treatment-ranking thresholds. It therefore focuses on heterogeneity of response rather than only an average treatment effect.
The existence of a differential-response endpoint does not by itself establish that one therapy is superior for the entire study population. A participant-level responder classification and a population-average treatment effect are different estimands.
The thresholds of 180 mg less prednisone, 31 additional annualized asthma-control days, and 5% higher FEV1 determine what constitutes a sufficiently large treatment-period difference for the endpoint. These are decision thresholds, not p-values.
A p-value requires a specified statistical model or test statistic, its sampling distribution, and the observed data. The ClinicalTrials.gov record does not provide a statistical-analysis record or sufficient participant-level results from which a valid p-value can be reconstructed.
Because treatment-period responses come from the same participants, statistical uncertainty must reflect within-participant dependence. Treating repeated observations as unrelated would change the effective information content of the data and could distort inference.
22. Long-Term Safety Caveat
This caveat is statistically important because absence of evidence for a long-term safety difference is not equivalent to evidence that long-term risk is absent. Detecting uncommon or delayed safety events generally requires sufficient exposure, follow-up, and sample size. The registry explicitly limits what can be concluded from this study about long-term risks.
23. Data Interpretation: What Can and Cannot Be Said
| Question | Supported by the ClinicalTrials.gov record? | Reason |
|---|---|---|
| Was the trial randomized? | Yes | The registry identifies allocation as RANDOMIZED. |
| Was it a crossover trial? | Yes | The registry identifies the design model as CROSSOVER. |
| Was it quadruple-masked? | Yes | The registry identifies masking as QUADRUPLE. |
| Were three step-up therapies evaluated? | Yes | The intervention list identifies three named therapies. |
| Was the primary endpoint a count/rate endpoint? | Yes | the ClinicalTrials.gov record identifies the primary endpoint type as Count / rate. |
| Was the primary endpoint measured during the last 12 weeks? | Yes | The registry gives this time frame. |
| Was a formal p-value reported? | No formal value reported | No formal statistical analyses were posted. |
| Was a confidence interval reported? | No formal value reported | No formal statistical analyses were posted. |
| Can treatment-specific serious-AE rates be calculated? | No | Only 7/182 among all participants is reported. |
| Can long-term LABA safety be established? | No | The registry explicitly states the study was not designed or powered for that purpose. |
24. Related Tutorials
Learn more about the methods used in this trial:
25. Related Statistical Calculators
26. Sources
- ClinicalTrials.gov: NCT00395304, the official registry record for the BADGER trial.
- Linked publication: PubMed record for PMID 20197425.
Continue through the Clinical Biostats statistical pathway
Use the related tutorials and calculators to explore the crossover, randomization, and masking concepts that define the BADGER study design.
27. Record Summary
BADGER is a phase 3, randomized, quadruple-masked crossover trial evaluating three step-up therapies for asthma: fluticasone propionate + montelukast, fluticasone propionate, and fluticasone propionate + salmeterol. Its primary endpoint is a participant-level count of differential responders based on fixed thresholds for oral prednisone use, asthma-control days, and FEV1 during the last 12 weeks of each 16-week treatment period.
Statistically, the study is most useful as an example of how a crossover design changes the analysis problem. Treatment-period observations are linked within participant, while treatment sequence, period effects, carryover, missing treatment periods, and multiplicity can all influence inference. The ClinicalTrials.gov record identifies the design and endpoint clearly but does not provide the formal statistical-analysis records needed to report a treatment-effect estimate, confidence interval, or p-value.