Skip to contents

Assess Baseline Balance Between Treatment Groups

Source: R/gba_utils.R
assess_baseline_balance.Rd

Performs a comprehensive baseline balance assessment calculating SMD for multiple continuous and categorical variables. Returns a BalanceAssessment object that includes SMD results, imbalance flags, summary statistics, and data formatted for Love plot visualization.

Usage

assess_baseline_balance(
  data,
  trt_var,
  continuous_vars = character(),
  categorical_vars = character(),
  ref_group = NULL,
  threshold = 0.1,
  conf_level = 0.95,
  continuous_threshold = 10,
  continuous_method = c("cohens_d", "hedges_g"),
  categorical_method = c("arcsine", "logit", "raw")
)

Arguments

data

A data frame containing the baseline data.

trt_var

Character. Name of the treatment variable.

continuous_vars

Character vector. Names of continuous variables.

categorical_vars

Character vector. Names of categorical variables.

ref_group

Character or NULL. Reference (control) group value.

threshold

Numeric. SMD threshold for imbalance (default: 0.1).

conf_level

Numeric. Confidence level for CIs (default: 0.95).

continuous_threshold

Integer. Minimum number of unique values to treat numeric variables as continuous (default: 10).

continuous_method

Character. SMD method for continuous variables. One of "cohens_d" (default) or "hedges_g".

categorical_method

Character. SMD method for categorical variables. One of "arcsine" (default), "logit", or "raw".

Value

A BalanceAssessment S7 object with:

  • smd_results: Data frame with SMD for each variable

  • imbalanced_vars: Character vector of imbalanced variable names

  • threshold: The threshold used

  • n_treatment: Sample size in treatment group

  • n_control: Sample size in control group

  • summary_stats: List with summary statistics

  • love_plot_data: Data frame formatted for Love plot

Details

A threshold of 0.1 is commonly used in clinical trials to indicate meaningful imbalance. Variables with |SMD| > threshold are flagged.

The IQWiG (German HTA agency) recommends SMD assessment for all baseline characteristics in benefit assessment dossiers.

References

Austin, P. C. (2009). Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Statistics in Medicine, 28(25), 3083-3107.

See also

create_love_plot(), add_smd_to_table()

Examples

if (FALSE) { # \dontrun{
# Assess baseline balance
balance <- assess_baseline_balance(
  data = adsl,
  trt_var = "TRT01P",
  continuous_vars = c("AGE", "WEIGHT", "HEIGHT", "BMI"),
  categorical_vars = c("SEX", "RACE", "ETHNIC"),
  ref_group = "Placebo",
  threshold = 0.1
)

# Check if balanced
balance@balanced

# View imbalanced variables
balance@imbalanced_vars

# Create Love plot
create_love_plot(balance)
} # }