rm_mvsum: Format a regression model nicely for 'Rmarkdown'

In reportRmd: Tidy Presentation of Clinical Reporting

Format a regression model nicely for 'Rmarkdown'

Description

Multivariable (or univariate) regression models are re-formatted for reporting and a global p-value is added for the evaluation of factor variables.

Usage

rm_mvsum(
  model,
  data,
  digits = getOption("reportRmd.digits", 2),
  covTitle = "",
  showN = TRUE,
  showEvent = TRUE,
  CIwidth = 0.95,
  vif = TRUE,
  whichp = c("levels", "global", "both"),
  caption = NULL,
  tableOnly = FALSE,
  p.adjust = "none",
  unformattedp = FALSE,
  nicenames = TRUE,
  include_unadjusted = FALSE,
  chunk_label,
  fontsize
)

Arguments

model	model fit
data	data is no longer required as it is extracted from the model. This argument will be removed in the future
digits	number of digits to round estimates to, does not affect p-values
covTitle	character with the names of the covariate (predictor) column. The default is to leave this empty for output or, for table only output to use the column name 'Covariate'.
showN	boolean indicating sample sizes should be shown for each comparison, can be useful for interactions
showEvent	boolean indicating if number of events should be shown. Only available for logistic.
CIwidth	width for confidence intervals, defaults to 0.95
vif	boolean indicating if the variance inflation factor should be included. See details
whichp	string indicating whether you want to display p-values for levels within categorical data ("levels"), global p values ("global"), or both ("both"). Irrelevant for continuous predictors.
caption	table caption
tableOnly	boolean indicating if unformatted table should be returned
p.adjust	p-adjustments to be performed. Uses the p.adjust function from base R
unformattedp	boolean indicating if you would like the p-value to be returned unformatted (ie not rounded or prefixed with '<'). Should be used in conjunction with the digits argument.
nicenames	boolean indicating if you want to replace . and _ in strings with a space
include_unadjusted	Logical. If TRUE, includes univariate estimates alongside multivariable estimates. Default is FALSE.
chunk_label	Deprecated, previously used in Word to allow cross-referencing, this should now be done at the chunk level.
fontsize	PDF/HTML output only, manually set the table fontsize

Details

Global p-values are likelihood ratio tests for lm, glm and polr models. For lme models an attempt is made to re-fit the model using ML and if successful LRT is used to obtain a global p-value. For lmer models (lme4), if the lmerTest package is installed, Satterthwaite-based p-values and F-test global p-values are used; otherwise Wald z-based p-values and chi-squared LRT global p-values are returned. For glmer models (lme4), Wald z-based p-values are used with chi-squared LRT global p-values. Estimates are exponentiated for binomial (OR) and poisson/negative binomial (RR) families. For coxph models the model is re-run without robust variances with and without each variable and a LRT is presented. If unsuccessful a Wald p-value is returned. For GEE and CRR models Wald global p-values are returned. For negative binomial models a deviance test is used.

If the variance inflation factor is requested (VIF=TRUE, default) then a generalised VIF will be calculated in the same manner as the car package.

As of version 0.1.1 if global p-values are requested they will be included in the p-value column.

As of R 4.4.0 profile likelihood confidence intervals will be calculated automatically and there is no longer an option to force Wald tests.

The number of decimals places to display the statistics can be changed with digits, but this will not change the display of p-values. If more significant digits are required for p-values then use tableOnly=TRUE and format as desired.

Value

A character vector of the table source code, unless tableOnly=TRUE in which case a data frame is returned

References

John Fox & Georges Monette (1992) Generalized Collinearity Diagnostics, Journal of the American Statistical Association, 87:417, 178-183, \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1080/01621459.1992.10475190")}

John Fox and Sanford Weisberg (2019). An R Companion to Applied Regression, Third Edition. Thousand Oaks CA: Sage.

Examples

data("pembrolizumab")
glm_fit = glm(change_ctdna_group~sex:age+baseline_ctdna+l_size,
data=pembrolizumab,family = 'binomial')
rm_mvsum(glm_fit)

#linear model with p-value adjustment
lm_fit=lm(baseline_ctdna~age+sex+l_size+tmb,data=pembrolizumab)
rm_mvsum(lm_fit,p.adjust = "bonferroni")
#Coxph
require(survival)
res.cox <- coxph(Surv(os_time, os_status) ~ sex+age+l_size+tmb, data = pembrolizumab)
rm_mvsum(res.cox, vif=TRUE)

# lmer (lme4 mixed effects model) - single random intercept

if (require(lme4)){
lmer_fit <- lme4::lmer(age ~ sex + pdl1 + (1|cohort), data = pembrolizumab)
rm_mvsum(lmer_fit)
}


# lmer with multiple random effects and global p-values

if (require(lme4) && require(geepack)){
data(dietox, package = "geepack")
dietox$Cu <- as.factor(dietox$Cu)
lmer_fit2 <- lme4::lmer(Weight ~ Cu + Time + (1|Pig) + (1|Litter), data = dietox)
rm_mvsum(lmer_fit2, whichp = "both")
}


# glmer (binomial mixed effects model) - odds ratios

if (require(lme4)){
data(cbpp, package = "lme4")
glmer_fit <- lme4::glmer(cbind(incidence, size - incidence) ~ period + (1|herd),
  data = cbpp, family = binomial)
rm_mvsum(glmer_fit)
}


# glmer.nb (negative binomial mixed effects model) - rate ratios

if (require(lme4) && require(geepack)){
data(dietox, package = "geepack")
dietox$Cu <- as.factor(dietox$Cu)
nb_fit <- lme4::glmer.nb(Weight ~ Cu + Time + (1|Pig), data = dietox)
rm_mvsum(nb_fit, whichp = "both")
}

reportRmd documentation built on March 20, 2026, 5:06 p.m.