Using Bayesian Adaptive Trial Designs for Comparative Effectiveness Research: A Virtual Trial Execution.

BACKGROUND: Bayesian and adaptive clinical trial designs offer the potential for more efficient processes that result in lower sample sizes and shorter trial durations than traditional designs. OBJECTIVE: To explore the use and potential benefits of Bayesian adaptive clinical trial designs in comparative effectiveness research. DESIGN: Virtual execution of ALLHAT (Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial) as if it had been done according to a Bayesian adaptive trial design. SETTING: Comparative effectiveness trial of antihypertensive medications. PATIENTS: Patient data sampled from the more than 42 000 patients enrolled in ALLHAT with publicly available data. MEASUREMENTS: Number of patients randomly assigned between groups, trial duration, observed numbers of events, and overall trial results and conclusions. RESULTS: The Bayesian adaptive approach and original design yielded similar overall trial conclusions. The Bayesian adaptive trial randomly assigned more patients to the better-performing group and would probably have ended slightly earlier. LIMITATIONS: This virtual trial execution required limited resampling of ALLHAT patients for inclusion in RE-ADAPT (REsearch in ADAptive methods for Pragmatic Trials). Involvement of a data monitoring committee and other trial logistics were not considered. CONCLUSION: In a comparative effectiveness research trial, Bayesian adaptive trial designs are a feasible approach and potentially generate earlier results and allocate more patients to better-performing groups. PRIMARY FUNDING SOURCE: National Heart, Lung, and Blood Institute.

Not too big, not too small: a goldilocks approach to sample size selection.

We present a Bayesian adaptive design for a confirmatory trial to select a trial's sample size based on accumulating data. During accrual, frequent sample size selection analyses are made and predictive probabilities are used to determine whether the current sample size is sufficient or whether continuing accrual would be futile. The algorithm explicitly accounts for complete follow-up of all patients before the primary analysis is conducted. We refer to this as a Goldilocks trial design, as it is constantly asking the question, "Is the sample size too big, too small, or just right?" We describe the adaptive sample size algorithm, describe how the design parameters should be chosen, and show examples for dichotomous and time-to-event endpoints.

Multidisciplinary considerations for implementing Bayesian borrowing in basket trials.

Drug development has historically relied on phase I-III clinical trials including participants sharing the same disease. However, drug development has evolved as the discovery of mechanistic drivers of disease demonstrated that the same therapeutic target may provide benefits across different diseases. A basket trial condenses evaluation of one therapy among multiple related diseases into a single trial and presents an opportunity to borrow information across them rather than viewing each in isolation. Borrowing is a statistical tool but requires a foundation of clinical and therapeutic mechanistic justification. We review the Bayesian borrowing approach, including its assumptions, and provide a framework for how this approach can be evaluated for successful use in a basket trial for drug development.

Bayesian hierarchical modeling of patient subpopulations: efficient designs of Phase II oncology clinical trials.

BACKGROUND: In oncology, the treatment paradigm is shifting toward personalized medicine, where the goal is to match patients to the treatments most likely to deliver benefit. Treatment effects in various subpopulations may provide some information about treatment effects in other subpopulations. PURPOSE: We compare different approaches to Phase II trial design where a new treatment is being investigated in several groups of patients. We compare considering each group in an independent trial to a single trial with hierarchical modeling of the patient groups. METHODS: We assume four patient groups with different background response rates and simulate operating characteristics of three trial designs, Simon's Optimal Two-Stage design, a Bayesian adaptive design with frequent interim analyses, and a Bayesian adaptive design with frequent interim analyses and hierarchical modeling across patient groups. RESULTS: Simon's designs are based on 10% Type I and Type II error rates. The independent Bayesian designs are tuned to have similar error rates, but may have a slightly smaller mean sample size due to more frequent interim analyses. Under the null, the mean sample size is 2-4 patients smaller. A hierarchical model across patient groups can provide additional power and a further reduction in mean sample size. Under the null, the addition of the hierarchical model decreases the mean sample size an additional 4-7 patients in each group. Under the alternative hypothesis, power is increased to at least 98% in all groups. LIMITATIONS: Hierarchical borrowing can make finding a single group in which the treatment is promising, if there is only one, more difficult. In a scenario where the treatment is uninteresting in all but one group, power for that one group is reduced to 65%. When the drug appears promising in some groups and not in others, there is potential for borrowing to inflate the Type I error rate. CONCLUSIONS: The Bayesian hierarchical design is more likely to correctly conclude efficacy or futility than the other two designs in many scenarios. The Bayesian hierarchical design is a strong design for addressing possibly differential effects in different groups.

Do Bayesian adaptive trials offer advantages for comparative effectiveness research? Protocol for the RE-ADAPT study.

BACKGROUND: Randomized clinical trials, particularly for comparative effectiveness research (CER), are frequently criticized for being overly restrictive or untimely for health-care decision making. PURPOSE: Our prospectively designed REsearch in ADAptive methods for Pragmatic Trials (RE-ADAPT) study is a 'proof of concept' to stimulate investment in Bayesian adaptive designs for future CER trials. METHODS: We will assess whether Bayesian adaptive designs offer potential efficiencies in CER by simulating a re-execution of the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) study using actual data from ALLHAT. RESULTS: We prospectively define seven alternate designs consisting of various combinations of arm dropping, adaptive randomization, and early stopping and describe how these designs will be compared to the original ALLHAT design. We identify the one particular design that would have been executed, which incorporates early stopping and information-based adaptive randomization. LIMITATIONS: While the simulation realistically emulates patient enrollment, interim analyses, and adaptive changes to design, it cannot incorporate key features like the involvement of data monitoring committee in making decisions about adaptive changes. CONCLUSION: This article describes our analytic approach for RE-ADAPT. The next stage of the project is to conduct the re-execution analyses using the seven prespecified designs and the original ALLHAT data.

Combining risk difference and risk ratio in noninferiority trials of safety.

Prior to marketing, the long-term safety profile of a new therapy is often uncertain. One recommendation for premarket safety studies is to compare the new therapy to an appropriate control to determine whether the 95% confidence interval of the risk ratio is entirely less than a prespecified threshold (e.g., 1.8). The restriction to the risk ratio, however, has consequences that may not be intended. Risk difference may be a more appropriate measure of risk in this setting when event rates are very low. We propose using a suitable combination of risk ratio and risk difference in demonstrating noninferiority.

Addressing the incremental benefit of histamine dihydrochloride when added to interleukin-2 in treating acute myeloid leukemia: a Bayesian meta-analysis.

IL-2 has been investigated as maintenance therapy for patients with AML in five randomized trials. None has shown a statistically significant benefit. A randomized trial of HDC + IL-2 showed a statistically significant benefit for leukemia-free survival (LFS) in comparison with standard of care. Because HDC + IL-2 has not been randomized against IL-2 alone, the question remains as to whether and to what extent HDC + IL-2 is an improvement compared to IL-2 alone. This is a literature-based meta-analysis. We employ two versions of a Bayesian hierarchical model to compare HDC + IL-2 versus IL-2 alone on the basis of LFS in patients in remission from AML.

Age and associated fibrocystic changes are prognostically significant in patients with small node-negative (T1a,bN0) invasive breast cancer.

Some patients with small (≤1.0 cm) node-negative (T1a,bN0) invasive breast cancer (IBC) who undergo only local therapy experience recurrences. There is limited information on prognostic factors in these patients. We sought to identify prognostic factors associated with disease-free survival (DFS) and overall survival (OS) in patients with T1a,bN0 IBC. Histologic sections from 273 T1a,bN0 IBC patients treated at M. D. Anderson Cancer Center (MDACC) between 1980 and 1999 were reviewed. Microscopic tumor size; multifocality; histologic type, grade of tumor; presence, type, grade of associated ductal carcinoma in situ (DCIS); presence of fibrocystic changes (FCC) with/without atypia; and lymphovascular invasion were identified. The Kaplan-Meier method was used to evaluate DFS and OS. Median patient age was 58 years, median follow-up period was 10.8 years, and median tumor size was 0.8 cm. Multifocal disease was identified in 26% of cases. At 10 years, the DFS and OS rates were 91% and 88%, respectively. Twenty-one percent of patients had extensive (>50%), and 30% had grade 3 DCIS. Nonproliferative FCC and proliferative FCC with/without atypia were present in 80%, 36%, and 38% of patients, respectively. In univariate analysis, age at diagnosis (p < 0.0001), grade (p = 0.015), and percent (p = 0.046) of DCIS were significantly associated with DFS; presence of FCC was associated with longer DFS and OS. In multivariable models, age and presence of FCC remained significantly associated with survival. Age at diagnosis and associated FCC are significant factors in predicting recurrence in patients with T1a,bN0 IBC. Adjuvant systemic therapy should be discussed with and considered for young patients with T1a,bN0 IBC.

Daily oral sodium bicarbonate preserves glomerular filtration rate by slowing its decline in early hypertensive nephropathy.

In most patients with hypertensive nephropathy and low glomerular filtration rate (GFR), the kidney function progressively declines despite the adequate control of the hypertension with angiotensin-converting enzyme inhibition. Previously we found that 2 years of oral sodium citrate slowed GFR decline in patients whose estimated GFR (eGFR) was very low (mean 33 ml/min). This treatment also slowed GFR decline in an animal model of surgically reduced nephron mass. Here, we tested if daily oral sodium bicarbonate slowed GFR decline in patients with hypertensive nephropathy with reduced but relatively preserved eGFR (mean 75 ml/min) in a 5-year, prospective, randomized, placebo-controlled, and blinded interventional study. Patients matched for age, ethnicity, albuminuria, and eGFR received daily placebo or equimolar sodium chloride or bicarbonate while maintaining antihypertensive regimens (including angiotensin-converting enzyme inhibition) aiming for their recommended blood pressure targets. After 5 years, the rate of eGFR decline, estimated using plasma cystatin C, was slower and eGFR was higher in patients given sodium bicarbonate than in those given placebo or sodium chloride. Thus, our study shows that in hypertensive nephropathy, daily sodium bicarbonate is an effective kidney protective adjunct to blood pressure control along with angiotensin-converting enzyme inhibition.

Neoadjuvant therapy with paclitaxel followed by 5-fluorouracil, epirubicin, and cyclophosphamide chemotherapy and concurrent trastuzumab in human epidermal growth factor receptor 2-positive operable breast cancer: an update of the initial randomized study population and data of additional patients treated with the same regimen.

PURPOSE: Findings from our previously published phase III randomized trial showed a high pathologic complete remission (CR) rate in patients with human epidermal growth factor receptor 2-positive breast cancer after the concurrent administration of trastuzumab and paclitaxel, followed by concurrent trastuzumab and 5-fluorouracil, epirubicin, and cyclophosphamide (FEC) preoperative chemotherapy. The safety and efficacy data of initial population were updated, with inclusion of additional experience with the same therapy. STUDY DESIGN: The initial randomized study population of 42 patients were randomly assigned to either four cycles of paclitaxel followed by four cycles of FEC or to the same chemotherapy with simultaneous weekly trastuzumab for 24 weeks. All data were updated through November 2005. RESULTS: Pretreatment characteristics of the initial patients and of the second cohort were similar. In the second cohort, pathologic CR rate was 54.5% (95% confidence interval, 32.2-75.6%) and the pathologic CR rate among all patients treated with chemotherapy plus trastuzumab was 60% (95% confidence interval, 44.3-74.3%). Three patients in the chemotherapy only group have recurred, and one has died. There has been no recurrences in the patients randomized to chemotherapy plus trastuzumab, and the estimated disease-free survival at 1 and 3 years was 100% (P = 0.041). In additional cohort treated with chemotherapy and trastuzumab at the median follow-up of 16.3 months, no patients had recurred. No new safety concerns were observed in this study. CONCLUSION: Our expanded cardiac safety data and the updated efficacy data showed that the natural history of this subset of breast cancer patients can be substantially modified by this treatment approach.

Circulating tumor cells in metastatic breast cancer: biologic staging beyond tumor burden.

BACKGROUND: The detection of circulating tumor cells (CTCs) predicts overall survival in patients with metastatic breast cancer (MBC). However, it is unknown whether CTCs have superior value compared with other standard prognostic factors. We compared the prognostic significance of CTCs with clinical and laboratory measures of tumor burden and phenotypic subtype of disease. PATIENTS AND METHODS: One hundred fifty-one patients with MBC evaluated between 2000 and 2006 were included in this retrospective analysis. Circulating tumor cells were isolated and enumerated in whole blood using an immunomagnetic bead system (CellSearch System). Overall survival was evaluated according to the level of CTCs (negative: <5 CTCs per 7.5 mL of blood; positive: >or=5 CTCs per 7.5 mL of blood), Swenerton score, cancer antigen 27-29 level, age (<50 years vs. >or=50 years), hormone-receptor status and HER2 status, metastatic site, and type and line of therapy. RESULTS: The median age of patients was 53 years (range, 24-88 years), and 44% of the patients had >5 CTCs. The median overall survival for negative versus positive CTCs were 29.3 months and 13.5 months, respectively (P<0.0001). In the multivariable Cox model, the detection of>or=5 CTCs demonstrated the highest hazard ratio with 2.2 times the risk of death (P=0.003). The prognostic value was independent of measure of tumor burden and type and line of therapy, and phenotypic subtype of the disease. CONCLUSION: Circulating tumor cells have superior and independent prognostic value of tumor burden and disease phenotype and might represent an important marker of tumor biology in MBC. Detection of CTCs should be considered for new staging stratification of patients with MBC.

High rate of durable remissions after treatment of newly diagnosed aggressive mantle-cell lymphoma with rituximab plus hyper-CVAD alternating with rituximab plus high-dose methotrexate and cytarabine.

PURPOSE: To determine the response, failure-free survival (FFS), and overall survival rates and toxicity of rituximab plus an intense chemotherapy regimen in patients with previously untreated aggressive mantle-cell lymphoma (MCL). PATIENTS AND METHODS: This was a prospective phase II trial of rituximab plus fractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone (hyper-CVAD; considered one cycle) alternating every 21 days with rituximab plus high-dose methotrexate-cytarabine (considered one cycle) for a total of six to eight cycles. RESULTS: Of 97 assessable patients, 97% responded, and 87% achieved a complete response (CR) or unconfirmed CR. With a median follow-up time of 40 months, the 3-year FFS and overall survival rates were 64% and 82%, respectively, without a plateau in the curves. For the subgroup of patients < or = 65 years of age, the 3-year FFS rate was 73%. The principal toxicity was hematologic. Five patients died from acute toxicity. Four patients developed treatment-related myelodysplasia/acute myelogenous leukemia, and three patients died while in remission from MCL. A total of eight treatment-related deaths (8%) occurred. CONCLUSION: Rituximab plus hyper-CVAD alternating with rituximab plus high-dose methotrexate and cytarabine is effective in untreated aggressive MCL. Toxicity is significant but expected. Because of the shorter FFS concurrent with significant toxicity in patients more than 65 years of age, this regimen is not recommended as standard therapy for this age subgroup. Larger prospective randomized studies are needed to define the role of this regimen in the treatment of MCL patients compared with existing and new treatment modalities.

CCR7 and CXCR4 as novel biomarkers predicting axillary lymph node metastasis in T1 breast cancer.

PURPOSE: The chemokine receptors CCR7 and CXCR4 have been shown to play an important role in cancer metastasis. We therefore studied the differential expression of CCR7 and CXCR4, along with that of the biomarker HER2-neu, to evaluate whether these biomarkers could predict axillary lymph node metastasis in breast cancer. EXPERIMENTAL DESIGN: Biomarker expression levels were evaluated using paraffin-embedded tissue sections of lymph node-negative (n = 99) and lymph node-positive (n = 98) T1 breast cancer by immunohistochemical staining. RESULTS: Lymph node-positive tumors showed higher rates of high cytoplasmic CCR7 staining (21.5% versus 8.5%, P = 0.013) and HER2-neu overexpression (21.5% versus 9.3%, P = 0.019) than did lymph node-negative tumors. Similarly, high cytoplasmic CXCR4 expression occurred more commonly in lymph node-positive tumors (11.2% versus 5.1%, P = 0.113). In contrast, predominantly nuclear CXCR4 staining was more likely to be found in lymph node-negative tumors (54.5% versus 37.8%, P = 0.018). Furthermore, cytoplasmic CXCR4 coexpressed with HER2-neu was the only factor associated with involvement of four or more lymph nodes (16.7% versus 1.2%, P = 0.04) among lymph node-positive tumors. When all three biomarkers (CCR7, CXCR4, HER2-neu) were utilized together, 50.0% of lymph node-positive tumors highly expressed one of these biomarkers compared with 18.8% of the lymph node-negative tumors (P < 0.0001). CONCLUSIONS: Our results suggest that the chemokine receptor CCR7 is a novel biomarker that can predict lymph node metastases in breast cancer. Utilization of additional markers, such as CXCR4 and HER2-neu, further improves the prediction of the presence and extent of lymph node involvement.

Association of Pathologic Complete Response to Neoadjuvant Therapy in HER2-Positive Breast Cancer With Long-Term Outcomes: A Meta-Analysis.

IMPORTANCE: The expense and lengthy follow-up periods for randomized clinical trials (RCTs) of adjuvant systemic therapy in breast cancer make them impractical and even impossible to conduct. Randomized clinical trials of neoadjuvant systemic therapy for breast cancer may help resolve this dilemma. OBJECTIVE: To assess the utility of pathologic complete response (pCR) for neoadjuvant drug development in human epidermal growth factor receptor 2 (HER2 [also referred to as ERBB2])-positive breast cancer. DATA SOURCES: We searched MEDLINE (Ovid), Embase (Ovid), CENTRAL (Wiley), and Northern Light Life Sciences Conference Abstracts (Ovid) in December 2014. Searches combined terms for "breast cancer" and "neoadjuvant therapy," with no limit on publication date. STUDY SELECTION: Cohort studies and RCTs were selected that met following criteria: stages I to III HER2-positive breast cancer, neoadjuvant therapy, and reports of both pCR and an event-free survival (EFS)-type outcome. The initial search identified 2614 publications, of which 38 studies met the selection criteria. DATA EXTRACTION AND SYNTHESIS: Two authors independently screened each study for inclusion and extracted the data. Data were analyzed using Bayesian hierarchical models. MAIN OUTCOMES AND MEASURES: Event-free survival and overall survival (OS) hazard ratios (HRs) for pCR vs non-pCR. For RCTs, main outcome measures were treatment benefits in pCR and the corresponding treatment HRs for EFS and OS. RESULTS: A total of 36 studies with EFS by pCR status representing 5768 patients with HER2-positive breast cancer were included in the patient-level analysis. Overall, the improvement in EFS for pCR vs non-pCR was substantial: HR, 0.37 (95% probability interval [PI], 0.32-0.43). This association was greater for patients with hormone receptor-negative disease (HR, 0.29 [95% PI, 0.24-0.36]) than hormone receptor-positive disease (HR, 0.52 [95% PI, 0.40-0.66]). In RCTs, the R2 correlations between odds ratios for pCR and HRs were 0.63 for EFS and 0.29 for OS. Based on absolute treatment improvements in pCR rate, predicted HRs for EFS for RCTs were concordant with observed HRs. CONCLUSIONS AND RELEVANCE: Pathologic complete response in HER2-positive breast cancer is associated with substantially longer times to recurrence and death. This relationship is maintained in RCTs. For any particular new therapy the relationship between pCR and survival may differ. Quantifying the importance of pCR is necessary for designing efficient clinical trials, which should adapt to the relationship between pCR and survival for the therapy under investigation.

Bayesian dose escalation in oncology with sharing of information between patient populations.

We present a Phase I dose escalation trial design based on a modified continual reassessment method that allows for sharing of information between populations. We describe our approach in the context of a trial for patients with acute lymphoblastic leukemia (ALL) that is currently being conducted. The ALL trial enrolls both adult and pediatric patient populations. Dose escalation and the determination of the maximum tolerated dose (MTD) are performed separately for each population, but to increase efficiency, information about the dose-toxicity curve is shared. Dose escalation rules allow pediatric patients to skip dose levels provided safety has been shown in adults and the dose level is estimated to be safe for pediatric patients. Trial objectives are to efficiently determine the MTD for each population and to minimize the number of pediatric patients required for dose escalation.

The Stroke Hyperglycemia Insulin Network Effort (SHINE) trial: an adaptive trial design case study.

BACKGROUND: The 'Adaptive Designs Accelerating Promising Trials into Treatments (ADAPT-IT)' project is a collaborative effort supported by the National Institutes of Health (NIH) and United States Food & Drug Administration (FDA) to explore how adaptive clinical trial design might improve the evaluation of drugs and medical devices. ADAPT-IT uses the National Institute of Neurologic Disorders & Stroke-supported Neurological Emergencies Treatment Trials (NETT) network as a 'laboratory' in which to study the development of adaptive clinical trial designs in the confirmatory setting. The Stroke Hyperglycemia Insulin Network Effort (SHINE) trial was selected for funding by the NIH-NINDS at the start of ADAPT-IT and is currently an ongoing phase III trial of tight glucose control in hyperglycemic acute ischemic stroke patients. Within ADAPT-IT, a Bayesian adaptive Goldilocks trial design alternative was developed. METHODS: The SHINE design includes response adaptive randomization, a sample size re-estimation, and monitoring for early efficacy and futility according to a group sequential design. The Goldilocks design includes more frequent monitoring for predicted success or futility and a longitudinal model of the primary endpoint. Both trial designs were simulated and compared in terms of their mean sample size and power across a range of treatment effects and success rates for the control group. RESULTS: As simulated, the SHINE design tends to have slightly higher power and the Goldilocks design has a lower mean sample size. Both designs were tuned to have approximately 80% power to detect a difference of 25% versus 32% between control and treatment, respectively. In this scenario, mean sample sizes are 1,114 and 979 for the SHINE and Goldilocks designs, respectively. CONCLUSIONS: Two designs were brought forward, and both were evaluated, revised, and improved based on the input of all parties involved in the ADAPT-IT process. However, the SHINE investigators were tasked with choosing only a single design to implement and ultimately elected not to implement the Goldilocks design. The Goldilocks design will be retrospectively executed upon completion of SHINE to later compare the designs based on their use of patient resources, time, and conclusions in a real world setting. TRIAL REGISTRATION: ClinicalTrials.gov NCT01369069 June 2011.

A meta-analysis platform methodology for determining the comparative effectiveness of antihepatitis C virus regimens.

AIM: Many hepatitis C virus regimens are unlikely to be compared head to head. In more difficult to treat populations where there is no standard of care, trials are single arm. We describe a flexible meta-analysis platform in this setting. METHODS: Our meta-analysis is literature based. We illustrate our methodology and show how inference can be extended to single-arm trials. RESULTS: As an example, in the single arm setting, a regimen with response rates of 84, 72 and 54% in genotype 1a across treatment naive, previous partial responders and previous null responders, respectively, would have 95% probability of superiority to IFN-α + RBV + TPV. CONCLUSION: This is a rigorous approach to comparative effectiveness that accounts for varying patient populations and plans for the incorporation of emerging treatments.

Predicting clinical trial results based on announcements of interim analyses.

BACKGROUND: Announcements of interim analyses of a clinical trial convey information about the results beyond the trial's Data Safety Monitoring Board (DSMB). The amount of information conveyed may be minimal, but the fact that none of the trial's stopping boundaries has been crossed implies that the experimental therapy is neither extremely effective nor hopeless. Predicting success of the ongoing trial is of interest to the trial's sponsor, the medical community, pharmaceutical companies, and investors. We determine the probability of trial success by quantifying only the publicly available information from interim analyses of an ongoing trial. We illustrate our method in the context of the National Surgical Adjuvant Breast and Bowel (NSABP) trial, C-08. METHODS: We simulated trials based on the specifics of the NSABP C-08 protocol that were publicly available. We quantified the uncertainty around the treatment effect using prior weights for the various possibilities in light of other colon cancer studies and other studies of the investigational agent, bevacizumab. We considered alternative prior distributions. RESULTS: Subsequent to the trial's third interim analysis, our predictive probabilities were: that the trial would eventually be successful, 48.0%; would stop for futility, 7.4%; and would continue to completion without statistical significance, 44.5%. The actual trial continued to completion without statistical significance. CONCLUSIONS: Announcements of interim analyses provide information outside the DSMB's sphere of confidentiality. This information is potentially helpful to clinical trial prognosticators. 'Information leakage' from standard interim analyses such as in NSABP C-08 is conventionally viewed as acceptable even though it may be quite revealing. Whether leakage from more aggressive types of adaptations is acceptable should be assessed at the design stage.

Bayesian adaptive trials offer advantages in comparative effectiveness trials: an example in status epilepticus.

OBJECTIVE: We present a novel Bayesian adaptive comparative effectiveness trial comparing three treatments for status epilepticus that uses adaptive randomization with potential early stopping. STUDY DESIGN AND SETTING: The trial will enroll 720 unique patients in emergency departments and uses a Bayesian adaptive design. RESULTS: The trial design is compared to a trial without adaptive randomization and produces an efficient trial in which a higher proportion of patients are likely to be randomized to the most effective treatment arm while generally using fewer total patients and offers higher power than an analogous trial with fixed randomization when identifying a superior treatment. CONCLUSION: When one treatment is superior to the other two, the trial design provides better patient care, higher power, and a lower expected sample size.