# Clinical Trials

> Clinical trial design and analysis methods in R, including randomization, estimands, multiplicity, and reporting.

- **Type:** Skill
- **Install:** `agentstack add skill-choxos-biostatagent-clinical-trials`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [choxos](https://agentstack.voostack.com/s/choxos)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [choxos](https://github.com/choxos)
- **Source:** https://github.com/choxos/BiostatAgent/tree/main/plugins/r-tidy-modeling/skills/clinical-trials

## Install

```sh
agentstack add skill-choxos-biostatagent-clinical-trials
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Clinical Trials Statistical Methods

## Overview

Comprehensive clinical trial design and analysis methods in R covering sample size calculation, randomization, interim analyses, multiplicity adjustment, and regulatory-compliant statistical methods.

## Sample Size Calculation

### Two-Group Comparisons

```r
library(pwr)

# Two-sample t-test
pwr.t.test(
  d = 0.5,           # Cohen's d effect size
  sig.level = 0.05,
  power = 0.80,
  type = "two.sample",
  alternative = "two.sided"
)

# Proportions (chi-square)
pwr.2p.test(
  h = ES.h(p1 = 0.6, p2 = 0.4),  # Cohen's h
  sig.level = 0.05,
  power = 0.80
)

# Two proportions (unequal groups)
pwr.2p2n.test(
  h = ES.h(p1 = 0.6, p2 = 0.4),
  n1 = 100,
  sig.level = 0.05
)
```

### Survival Endpoints

```r
library(gsDesign)

# Log-rank test sample size
nSurv(
  lambda1 = log(2)/12,  # Control median = 12 months
  lambda2 = log(2)/18,  # Treatment median = 18 months (HR = 0.67)
  Ts = 24,              # Study duration
  Tr = 12,              # Accrual duration
  alpha = 0.025,        # One-sided

  beta = 0.20,          # 80% power
  ratio = 1             # 1:1 randomization
)

# Using rpact
library(rpact)
getSampleSizeSurvival(
  hazardRatio = 0.67,
  lambda1 = log(2)/12,
  accrualTime = 12,
  followUpTime = 12,
  alpha = 0.025,
  beta = 0.20,
  allocationRatioPlanned = 1
)
```

### Non-Inferiority Trials

```r
library(TrialSize)

# Non-inferiority for proportions
TwoSampleProportion.NIS(
  p = 0.80,           # Expected proportion in both groups
  delta = 0.10,       # Non-inferiority margin
  alpha = 0.025,      # One-sided
  power = 0.80
)

# Non-inferiority for means
TwoSampleMean.NIS(
  sigma = 10,         # SD
  delta = 5,          # Non-inferiority margin
  alpha = 0.025,
  power = 0.80
)
```

## Randomization

### Simple Randomization

```r
# Base R simple randomization
set.seed(123)
n =65")
)

rand_lists 
  dplyr::filter(ITTFL == "Y") |>
  dplyr::select(TRT01P, AGE, SEX, BMI, BASELINE_SCORE) |>
  tbl_summary(
    by = TRT01P,
    statistic = list(
      all_continuous() ~ "{mean} ({sd})",
      all_categorical() ~ "{n} ({p}%)"
    )
  ) |>
  modify_header(label = "**Characteristic**")
```

If the SAP requires standardized mean differences, report them as descriptive diagnostics rather than randomization tests.

## Group Sequential Designs

### gsDesign Package

```r
library(gsDesign)

# O'Brien-Fleming boundaries
gs_design 
  group_by(subgroup) |>
  summarise(
    n = n(),
    effect = mean(outcome[trt == 1]) - mean(outcome[trt == 0]),
    se = sqrt(var(outcome[trt == 1])/sum(trt == 1) +
              var(outcome[trt == 0])/sum(trt == 0)),
    lower = effect - 1.96 * se,
    upper = effect + 1.96 * se
  )

# Create forest plot
forestplot(
  labeltext = subgroup_effects$subgroup,
  mean = subgroup_effects$effect,
  lower = subgroup_effects$lower,
  upper = subgroup_effects$upper,
  zero = 0,
  xlab = "Treatment Effect (95% CI)"
)
```

### Interaction Tests

```r
# Test for treatment-by-subgroup interaction
interaction_model <- lm(outcome ~ treatment * subgroup, data = df)
anova(interaction_model)

# Quantitative interaction test
library(QI)
qi_test(outcome ~ treatment | subgroup, data = df)
```

## Regulatory Considerations

### ICH E9 Estimands Framework

```r
# Define estimand components:
# 1. Treatment condition
# 2. Population
# 3. Variable (endpoint)
# 4. Intercurrent events and strategies
# 5. Population-level summary measure

# Example: Treatment policy estimand with MMRM
mmrm_fit <- mmrm(
  change ~ treatment * visit + baseline + us(visit | subject),
  data = data_all_randomized  # Include all randomized (ITT)
)
```

Missing-data methods should follow the estimand and the plausible missingness mechanism. Prespecify primary handling and sensitivity analyses rather than defaulting to complete-case analysis.

### CONSORT Diagram

```r
library(consort)

# Create CONSORT diagram
consort_plot(
  data = trial_data,
  orders = c(
    Assessed = "Assessed for eligibility",
    Randomized = "Randomized",
    Arm_A = "Allocated to Arm A",
    Arm_B = "Allocated to Arm B",
    Lost_A = "Lost to follow-up (Arm A)",
    Lost_B = "Lost to follow-up (Arm B)",
    Analyzed_A = "Analyzed (Arm A)",
    Analyzed_B = "Analyzed (Arm B)"
  ),
  side_box = c("Excluded", "Discontinued_A", "Discontinued_B"),
  cex = 0.8
)
```

## Key Packages Summary

| Package | Purpose |
|---------|---------|
| pwr | Power analysis |
| gsDesign | Group sequential designs |
| rpact | Adaptive designs |
| blockrand | Randomization |
| gMCP | Graphical multiplicity |
| mmrm | MMRM analysis |
| mice | Multiple imputation |
| rbmi | Reference-based imputation |
| emmeans | Least squares means |
| consort | CONSORT diagrams |

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [choxos](https://github.com/choxos)
- **Source:** [choxos/BiostatAgent](https://github.com/choxos/BiostatAgent)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-choxos-biostatagent-clinical-trials
- Seller: https://agentstack.voostack.com/s/choxos
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.