Skip to contents

Generate a Bayesian meta-analysis report

Source: R/bayesma_report.R
bayesma_report.Rd

Renders a parameterised Quarto template into a publication-style statistical report. Section structure follows the bayesma workflow guide; prose style follows a typical systematic review (Introduction, Methods, Results, Overall Interpretation). Most sections are toggleable via run_* arguments.

Usage

bayesma_report(
  data,
  studyvar,
  yearvar,
  estimand = c("OR", "RR", "HR", "IRR", "MD", "SMD", "RD", "ARR", "ATE", "ATT", "CATE"),
  cate_covariate = NULL,
  baseline_risk = NULL,
  event_ctrl = NULL,
  event_int = NULL,
  mean_ctrl = NULL,
  mean_int = NULL,
  sd_ctrl = NULL,
  sd_int = NULL,
  n_ctrl = NULL,
  n_int = NULL,
  outcome_name = "the primary outcome",
  null_range,
  title = "Bayesian meta-analysis report",
  subtitle = "Generated with bayesma",
  author = "",
  prospero_id = "",
  background = "",
  hypothesis = "",
  inclusion_criteria = character(),
  exclusion_criteria = character(),
  quality_tool = "Cochrane RoB 2",
  software_note = "",
  rob_data = NULL,
  priors = NULL,
  run_inspect_sr = FALSE,
  run_rob = FALSE,
  run_egger = TRUE,
  run_robma = TRUE,
  run_metareg = FALSE,
  run_subgroup = FALSE,
  run_sensitivity_priors = TRUE,
  moderators = character(),
  subgroup_var = character(),
  primary_model = "auto",
  re_min_k = NULL,
  output_file,
  render = FALSE
)

Arguments

data

A data frame with one row per study (or per arm).

studyvar

Character. Column name with study identifiers.

yearvar

Character. Column name with publication year.

estimand

Character. Target effect: "OR", "RR", "HR", "IRR", "MD", "SMD", "RD" / "ARR", "ATE", "ATT", or "CATE". The marginal estimands (RD/ATE/ATT/CATE) are computed via bayesma_marginal() on the natural scale.

cate_covariate

Character. Study-level covariate name. Required when estimand = "CATE".

baseline_risk

Numeric in (0, 1) or "study_mean". Reference baseline used to back-transform two-stage binomial fits to RD/ATE/ATT.

event_ctrl, event_int, n_ctrl, n_int

Character. Column names for event counts and arm sizes (binary measures).

mean_ctrl, mean_int, sd_ctrl, sd_int

Character. Column names for arm means and SDs (continuous measures).

outcome_name

Character. Human-readable outcome label used in titles and prose.

null_range

Numeric length-2 vector. Null/ROPE range on the natural scale (e.g. c(0.9, 1.1) for an OR).

title, subtitle, author

Character. Report metadata.

prospero_id

Character. Optional PROSPERO registration identifier.

background, hypothesis

Character (scalar or vector). Prose for the Introduction and Methods sections. Each element becomes a paragraph.

inclusion_criteria, exclusion_criteria

Character vectors. Each element becomes a bullet point.

quality_tool

Character. Risk-of-bias tool name (default Cochrane RoB 2).

software_note

Character. Optional override for the Software paragraph.

rob_data

Optional pre-scored risk-of-bias data frame for rob_plot(). Used when run_rob = TRUE.

priors

Optional named list of prior settings for the sensitivity panel. Each element is a list with mu_prior and tau_prior. Defaults to vague / weak_reg / informative.

run_inspect_sr, run_rob, run_egger, run_robma, run_metareg, run_subgroup, run_sensitivity_priors

Logical. Section toggles.

moderators

Character vector. Moderator columns for meta-regression. Required when run_metareg = TRUE.

subgroup_var

Character vector. Subgroup column(s). Required when run_subgroup = TRUE.

primary_model

Character. "auto" (default; selected via compare_models()) or one of the candidate model labels (e.g. "two_stage_re", "two_stage_re_hksj").

re_min_k

Optional numeric. Forwarded to bayesma() (and used per subgroup level): downgrade random-effects to common-effect when the number of studies in scope is below this threshold. Default NULL.

output_file

Character. Path for the generated .qmd.

render

Logical. If TRUE, also render the .qmd to HTML.

Value

Invisibly, the path to the generated .qmd file.

Details

The function writes a .qmd file plus a companion .rds (containing data, rob_data, and priors) to the directory of output_file. With render = TRUE it then renders the .qmd to HTML via the Quarto CLI.

Examples

if (FALSE) { # \dontrun{
bayesma_report(
  data         = eeg_data,
  studyvar     = "Author",
  yearvar      = "Year",
  estimand     = "OR",
  event_ctrl   = "c_event", event_int = "i_event",
  n_ctrl       = "c_n",     n_int     = "i_n",
  outcome_name = "Postoperative Delirium",
  null_range   = c(0.9, 1.1),
  title        = "EEG-guided anaesthesia and POD",
  author       = "Dr Benjamin Moran",
  output_file  = "eeg_pod_report.qmd",
  render       = TRUE
)
} # }