Synthetic demonstration projectSource dataset preserved and checksummed

Analysis plan

Review the analysis plan

Plan version 1 · awaiting explicit approval

Needs review

Primary analysis

Design
Cross-sectional
Outcome
Hypertension (1 = yes, 0 = no)
Effect measure
Adjusted prevalence ratios
Model
Modified Poisson regression with robust standard errors
Diagnostics
Convergence, sparse cells, collinearity, influence, N reconciliation

Why this model?

Prevalence ratios are directly interpretable; odds ratios can exaggerate association when outcomes are common.

Covariates are prespecified and confounder-informed; no automated p-value selection.

Alternative: binary logistic regression
Available when an odds ratio is the intended estimand.

Approval record

Approval seals this analysis-spec version. Later changes create a new version and preserve this plan.

Approve immutable plan