Design a coherent trial
Connect the clinical question, eligibility, two-arm design, outcome definitions, ethics, procedures and analysis plan in eight structured stages.
Clinical Trials Studio · DataStatPro
Clinical trial design & statistical analysis
DataStatPro Clinical Trials Studio is browser-based software for trial design, sample-size planning, statistical analysis and reporting. Connect your protocol, covariate-adjusted models and sensitivity analyses with clear tables, figures and exportable reports.
Start with a hypothetical study. Review its purpose and analysis plan before loading synthetic data.
What can Clinical Trials Studio do?
Move from a clinical question to an inspectable analysis, keeping decisions and assumptions connected.
Make planning assumptions explicit
Calculate sample size and recruitment allowances. Check prospective power for supported models, with Monte Carlo uncertainty and unavailable-fit counts where applicable.
Adjust for prognostic covariates
Map numeric variables and categorical reference levels for ANCOVA, adjusted logistic or multivariable Cox analysis. Review complete-case exclusions and coefficient estimates.
Understand robustness
Inspect model diagnostics and run selected population, HC3, unadjusted, missing-outcome or multiple-imputation sensitivity scenarios where supported.
Keep endpoints organized
Assign endpoint-specific variables and distinguish confirmatory secondary outcomes from exploratory results. Review adjusted family decisions alongside nominal intervals.
Produce an inspectable report
Bring together the specification, participant flow, baseline summaries, effects, figures and reporting checks. Save drafts locally or work with a named cloud project.
Clinical trial statistical analysis methods
Choose the method that matches your endpoint, randomized unit and prespecified hypothesis.
| Endpoint / design | Available analysis | Effect summary | Important distinctions |
|---|---|---|---|
| Continuous outcomes | Welch comparison or covariate-adjusted ANCOVA | Mean difference with interval and hypothesis-specific inference | Superiority, supported non-inferiority or equivalence; numeric/categorical adjustment and supplementary HC3 ANCOVA. |
| Binary outcomes | Fisher comparison or adjusted logistic regression | Unadjusted risk difference/RR, or standardized adjusted risk contrasts | Logistic coefficient ratios are conditional odds ratios. They are labelled separately from standardized risk ratios. |
| Time-to-event outcomes | Arm-only or multivariable Breslow Cox; Kaplan–Meier and log-rank | Hazard ratio with interval; adjusted Cox Wald test when selected | Kaplan–Meier curves and retained log-rank comparisons are unadjusted. The current model uses right censoring and Breslow ties. |
| Cluster-randomized continuous outcomes | Welch comparison of equally weighted cluster means | Cluster-average treatment difference | ICC and cluster-size planning are available. This path is not a mixed model or GEE analysis. |
| Baseline and one follow-up | Complete-pair change-score comparison | Between-arm difference in mean change | Paired-change planning uses baseline/follow-up correlation. Multiple-visit MMRM and GEE are outside this path. |
Adjustment with a clear estimand. Logistic standardized risk differences and risk ratios are separate from conditional odds ratios. Adjusted Cox inference is separate from unadjusted Kaplan–Meier curves and log-rank comparisons. Coefficient tests and endpoint intervals remain nominal unless a stated family procedure applies to the decision.
Clinical trial sample size and prospective power
Record assumptions first. Calculate sample size, then check power for the analysis you intend to use.
Plan evaluable and recruitment counts
Specify alpha, target power, allocation and the expected difference, response probabilities or hazard ratio. Supported paths include variance assumptions, paired-change correlation, event fraction, participant attrition, and cluster ICC/size allowances.
The planning estimate and analysis-matched power check are separate actions. Changing relevant assumptions makes saved checks outdated.
Simulate expanded adjusted models
Multivariable ANCOVA, logistic and Cox power checks use explicitly declared prospective covariate distributions and model assumptions. Choose 200, 500 or 1,000 seeded replicates and inspect Monte Carlo intervals and unavailable fits.
ANCOVA uses an externally justified multiple R-squared and equal marginal SDs. Logistic uses reference risk and conditional effects. Cox uses exponential hazards, accrual, study end and independent dropout assumptions.
Expanded simulations assume independent covariate distributions in both arms, are bounded to 1,000 evaluable participants and a computation limit, and count unavailable fits as nonrejections. They do not estimate power from observed effects, optimize recruitment automatically or establish joint endpoint-family power.
Diagnostics and sensitivity analyses you can inspect
Review how model assumptions, population definitions and missing observations affect interpretation.
Model diagnostics
Supported outputs include residual and Q-Q summaries, influence screening, adjustment condition/VIF checks, common covariate-effect checks, logistic fit/event screening and an adjusted Cox identity-time proportional-hazards score check.
Named robustness comparisons
Select applicable HC3 ANCOVA, matched-complete-case unadjusted analysis, randomized/per-protocol counterpart, binary missing-outcome extremes and continuous delta stress tests. Unavailable scenarios are reported explicitly.
Formal continuous-outcome MI
Supplementary normal MAR or delta multiple imputation supports individual continuous Welch and ANCOVA. Use 5–100 imputations and a declared seed, with Rubin variance pooling and Barnard–Rubin t inference.
Primary inference remains available-case. Missing covariates are not filled; MAR and delta assumptions need scientific justification. Binary/survival MI, reference-based imputation and automatic tipping-point grids are outside the current implementation.
How to plan and analyze a clinical trial in DataStatPro
A repeatable workflow for investigators, biostatisticians, research teams and teaching.
Define the clinical question
Record the population, intervention, comparator, hypothesis, primary outcome, assessment time and intended treatment effect.
Plan sample size and power
Declare external effect and variance or event assumptions, calculate evaluable and recruitment counts, then run the supported analysis-matched power check.
Prespecify the analysis
Choose the model, covariates, population, endpoint family and executable sensitivity scenarios before interpreting outcomes.
Map and validate participant data
Use one row per randomized participant. Map IDs, original arm codes, outcomes and required baseline, event or cluster fields; inspect missingness and exclusions.
Analyze and review uncertainty
Run the analysis explicitly. Review effects, confidence intervals, diagnostics, coefficient tables, sensitivity results and multiplicity decisions.
Save, collaborate and report
Save a local trial file or name a cloud project. Review reporting checks, reconcile participant flow and export the selected report.
Learn with eleven hypothetical clinical trial examples
Each scenario opens with a structured preview: clinical purpose, population, design, endpoint, planning assumptions, analysis plan, synthetic cohort and interpretation boundaries.
Blood pressure superiority
Explore baseline ANCOVA, HC3 and matched-sample comparisons.
Non-inferiority & equivalence
See how directional margins and two one-sided tests change the decision rule.
Age and site covariates
Inspect numeric and categorical adjustment, reference coding and required-measurement exclusions.
Holm, Bonferroni & fixed sequence
Compare raw evidence with declared family decisions and the consequences of gating.
Survival, clusters & change scores
Explore right censoring, equally weighted clinic means and complete-pair changes.
Response extremes
Compare available-case response estimates with explicit missing-outcome assignments.
You choose when data loads. Review the scenario, then select Load worked example to attach its trial specification and synthetic data. Canceling leaves the current workspace unchanged.
Save your trial, collaborate and export reports
Keep your study specification and results together, with explicit saving and data-sharing choices.
Save and open locally or in the cloud
Save → Local / Cloud and Open → Local / Cloud are separate controls. Cloud saving asks for a project/trial name. Local .trial.json files can include the attached dataset when you choose that option.
Work with your study team
Cloud projects support roles, invitations, section comments, review states, revision history and restoration. Participant dataset sharing is separately controlled; saving the plan does not automatically share participant records.
Export the report you need
Use Export Report for the complete report, protocol, SAP or analysis in Word/PDF. Supported outputs include baseline and outcome tables, effect/coefficient results, participant flow, figures and sensitivity summaries.
Reporting preflight: review stale results, count reconciliation, provenance, estimand labels, scenario availability, Rubin variance consistency and plan timing. Coverage panels support reporting review; they do not certify a full numbered SPIRIT/CONSORT checklist or regulatory submission.
Cloud saving requires sign-in. Review current plans and access options, privacy information and your study's data-governance requirements before choosing a data workflow.
Available now, with clear method boundaries
Assess the specific estimator and data contract before choosing a clinical trial analysis workflow.
Additional endpoint families
One primary plus up to twelve confirmatory secondary/exploratory endpoints, endpoint-specific mappings, and one declared family with Holm, Bonferroni or primary-first fixed sequence. Nominal intervals and exploratory findings remain distinct from adjusted confirmatory decisions.
Where the current module stops
Stratified/Efron/exact Cox, competing risks, RMST, delayed entry, repeated-visit mixed models, generalized cluster models, co-primary joint planning and adaptive designs need separate methods. Synthetic allocation previews do not provide concealed operational randomization, enrollment or eCRF management.
Feature scope reviewed 2026-10-05 against the Clinical Trials Studio implementation. Explore DataStatPro's statistical validation approach and limitations.
Clinical Trials Studio: frequently asked questions
Direct answers about design, models, power, missing data, examples and reports.
What is DataStatPro Clinical Trials Studio?
DataStatPro Clinical Trials Studio is a browser-based workspace for clinical trial planning, sample-size calculation, statistical analysis, sensitivity checks and reporting. It connects a structured trial specification with participant data, worked examples and local or cloud projects.
Which clinical trial designs are supported?
The studio supports two-arm parallel individual trials and selected continuous-outcome cluster trials. Superiority is available across supported endpoint methods; non-inferiority and equivalence use supported individual continuous Welch or ANCOVA analyses. Change-score analysis covers baseline and one follow-up, rather than a repeated-visit mixed model.
Can I adjust for multiple covariates?
Yes. Select up to ten additional numeric or categorical covariates with explicit reference levels. Continuous ANCOVA can also include a dedicated baseline measure. Adjusted logistic regression reports standardized risk contrasts separately from conditional odds ratios, while multivariable Breslow Cox regression reports adjusted hazard ratios.
How do I calculate clinical trial sample size and power?
Specify the hypothesis, endpoint, allocation, alpha, target power and relevant effect, variance or event assumptions, then calculate sample size. Run analysis-matched power separately at the calculated evaluable counts. Expanded ANCOVA, logistic and Cox simulations require explicit prospective covariate and model assumptions; they do not use observed treatment effects or automatically optimize sample size.
How does the studio handle missing outcomes?
The primary analysis uses available required observations in the selected population. Supplementary options include population comparisons, binary missing-outcome extremes, continuous delta stress tests and normal-outcome multiple imputation for supported individual Welch or ANCOVA analyses. MAR and delta MI use Rubin pooling and Barnard–Rubin degrees of freedom. Missing covariates remain excluded; binary and survival multiple imputation are not implemented.
Can I analyze multiple clinical trial endpoints?
Yes. Define one primary endpoint and up to twelve additional confirmatory secondary or exploratory endpoints with their own mappings. One declared confirmatory family supports Holm, Bonferroni or primary-first fixed-sequence testing. Confidence intervals remain nominal; exploratory results and primary-only power do not establish joint family or co-primary success.
Are the worked examples based on real patients?
No. All eleven worked examples describe hypothetical studies and use synthetic participant records. A read-only scenario preview explains the purpose, design, outcomes, assumptions and analysis plan. Data loads only when you select Load worked example, and canceling preserves the current workspace.
What reports and files can I export?
Export a complete report, protocol, statistical analysis plan or analysis report as Word or PDF. Supported tables have CSV downloads and figures have SVG or PNG export options. Save an editable local .trial.json file with optional attached data, or save a named cloud project when signed in.
How do cloud saving and trial collaboration work?
Choose Save, then Cloud, and confirm a project or trial name. Cloud projects support roles, invitations, section comments, review states and revision history. Open, then Cloud, lists existing and shared projects. Participant dataset sharing is a separate explicit action; signing in is required for cloud saving.
Is this an operational clinical trial management system?
Clinical Trials Studio focuses on study planning, statistical analysis and reporting. It does not provide concealed operational allocation, participant enrollment, eCRFs, live safety-event management or regulatory signatures. Allocation previews are synthetic planning demonstrations, and reporting preflight is not a complete numbered SPIRIT or CONSORT checklist.
Related clinical research and statistics resources
Explore study planning and statistical reporting alongside the trial workspace.
Sample-size calculators
Explore focused planning tools and understand the inputs behind sample-size estimates.
Clinical trial design guide
Review design considerations and connect the clinical question with an appropriate study plan.
Publication-oriented reporting
Explore DataStatPro's broader tables, figures and reporting resources.
Bring your clinical question into a structured workspace
Explore a hypothetical example, specify the study, review analysis assumptions and prepare an inspectable clinical trial report.
Open Clinical Trials Studio