Use Cases › Overview Group Comparison & ANOVA ↳ Two-Group Comparison ↳ One-Way ANOVA ↳ Two-Way ANOVA ↳ Repeated Measures Regression & Curve Fitting ↳ Linear Regression ↳ Logistic Regression & AUC ↳ Dose-Response & IC50 ↳ Mixed Models (GLMM) Survival Analysis ↳ Kaplan-Meier Workflow ↳ Cox Regression Workflow Normality & Power ↳ Pre-Study Power Calculation ↳ Normality Testing Categorical & Association ↳ Chi-Square Test ↳ Fisher's Exact Test ↳ Odds Ratio & Risk Life Sciences Applications ↳ Flow Cytometry ↳ Gene Expression / Omics ↳ Animal Study Workflow ↳ Clinical Trial Endpoints

Clinical Trial Endpoints

Clinical trial analysis requires strict adherence to the pre-specified statistical analysis plan (SAP): the primary endpoint determines the primary test, secondary endpoints are analyzed with multiplicity awareness, and all populations (ITT, PP, safety) are clearly defined. Licklider enforces these disclosures and generates a CONSORT-aligned draft analysis package.

STEP 1 - The Scenario

Trial Profile

A Phase III randomized controlled trial comparing a new antihypertensive agent vs. placebo in patients with uncontrolled hypertension. The trial has one primary endpoint and three secondary endpoints, all pre-specified in the registered SAP.

Research Question

"Does the antihypertensive agent significantly reduce systolic blood pressure (SBP) at 12 weeks compared to placebo (primary endpoint)? Does it also reduce cardiovascular events, improve diastolic BP, and improve patient-reported outcomes (secondary endpoints)?"

Dataset

  • n = 240 randomized: Treatment (n = 120), Placebo (n = 120)
  • Primary endpoint: Change from baseline in SBP at 12 weeks (continuous)
  • Secondary endpoints: (1) Composite cardiovascular events (binary, time-to-event), (2) Change in diastolic BP at 12 weeks (continuous), (3) Patient-reported quality of life score (continuous)
  • Analysis populations: ITT (all randomized), PP (per-protocol compliant), Safety (at least one dose)

STEP 2 - Licklider Analysis Path

1. SAP Import & Endpoint Declaration

The researcher declares: primary endpoint = change in SBP at 12 weeks; analysis population = ITT; test = two-sample t-test or ANCOVA with baseline SBP as covariate. Secondary endpoints are listed with their planned tests. The Outcome Type Lock confirms each endpoint's type and the analysis purpose (confirmatory for primary, supporting evidence for secondary).

2. Population Definition & CONSORT-Aligned Counts

Licklider constructs the CONSORT-aligned population summary: randomized N = 240, excluded pre-treatment (n = 0), received treatment (n = 118 treated, n = 119 placebo; 3 withdrawals noted), analyzed in ITT (all 240 by modified ITT including withdrawals with last observation carried forward or multiple imputation). The Methods draft and export-facing disclosures retain these counts, but the researcher remains responsible for the final CONSORT flow diagram.

3. Primary Endpoint Analysis

ANCOVA: SBP change ~ Treatment + Baseline_SBP. Treatment effect: adjusted mean difference (Treatment vs. Placebo), 95% CI, t-statistic, p-value. This is the pre-specified primary test. Licklider notes that this is the confirmatory test, powered at alpha = 0.05, two-sided, and that no multiplicity adjustment applies to the primary endpoint.

4. Secondary Endpoint Analysis

Each secondary endpoint is analyzed per SAP:

  • (1) Composite CV events: Kaplan-Meier + log-rank test; hazard ratio from Cox regression
  • (2) Change in DBP: ANCOVA with baseline DBP covariate
  • (3) QoL score: SAP-style example (e.g. Wilcoxon rank-sum for ordinal-like scales) — narrative only; verify rank-test execution in product against Non-Parametric Alternatives

Licklider applies Holm or Hochberg correction across the three secondary endpoint p-values. The corrected and uncorrected p-values are both reported.

5. Figure Generation

Forest plot of treatment effects for all endpoints (primary + secondary) with 95% CIs. Each row is labeled with the endpoint name, analysis method, and alpha-level (primary: alpha = 0.05; secondary: multiplicity-adjusted). CONSORT population counts are carried into the draft report and can be reused in a separately prepared flow diagram.

STEP 3 - Guards & Disclosures Activated

Analysis Family Ledger (1.2.2)

The three secondary endpoints constitute one confirmatory analysis family. Any exploratory analyses conducted post-hoc are labeled as exploratory and tracked in a separate family. The family boundary, pre-specified vs. post-hoc, is disclosed in the statistical report.

Multiple Comparison Compliance (4.1.5)

Holm correction is applied across the three secondary endpoints. The guard requires that multiplicity-adjusted and unadjusted p-values are both reported for secondary endpoints. The primary endpoint p-value is not subject to multiplicity adjustment because it is the sole confirmatory test.

One-Sided / Two-Sided Disclosure (3.1.3)

The primary test is two-sided as pre-specified in the SAP. If the researcher attempts a one-sided primary test not specified in the SAP, the guard blocks the change and requires a protocol-amendment disclosure.

Effect Size + CI + N Mandatory (3.1.2)

For every endpoint, the point estimate, 95% CI, test statistic, and p-value are all required. ITT n for the primary analysis and PP n for the sensitivity analysis are both disclosed. Power for the primary endpoint is disclosed.

Missing Data Disclosure (4.1.7)

For the ITT analysis, the missing endpoint data handling strategy is declared and disclosed. The number of patients with missing 12-week BP data per arm is reported, together with the missingness summary carried into the draft Methods text.

STEP 4 - Export Package

  • Figures: Multi-endpoint forest plot; CONSORT population counts exported in the report and draft text for reuse in a final flow diagram
  • Statistical Report: Primary ANCOVA output (adjusted mean difference, 95% CI, t, p); secondary analyses (all endpoints, with unadjusted and Holm-adjusted p); CONSORT-aligned population counts; missing data summary per arm
  • Methods Text: "The primary analysis was a two-sided ANCOVA comparing change in SBP from baseline at 12 weeks between treatment and placebo, adjusting for baseline SBP, in the ITT population (alpha = 0.05). Secondary endpoints were analyzed as pre-specified in the SAP; Holm correction was applied across secondary endpoints. Missing primary endpoint data were imputed using multiple imputation. n = 240 randomized (Treatment: n = 120, Placebo: n = 120)."
  • Figure Legend (Forest Plot): Endpoint names, effect sizes with units, CI method, alpha-levels, correction method for secondary endpoints, ITT n per arm
  • CONSORT Flow Support: Allocation, follow-up, and analysis numbers per arm surfaced in the draft report and Methods text
  • Preprocessing Audit Log: ITT/PP/Safety population construction; missing data count per arm; imputation method; SAP endpoint declaration dates