nnt_x: Laupacis' conditional (adjusted) and harmonic mean (marginal)...

In vancak/NNTcalculator: Calculator of unadjusted, adjusted and marginal NNT

Laupacis' conditional (adjusted) and harmonic mean (marginal) NNT calculator in linear and generalized regression models.

Description

Calculates Laupacis' unadjusted, conditional (adjusted) and harmonic mean (marginal) NNT. Takes a data-set suitable for a regression analysis and returns the estimated conditional (adjusted) NNT(x), the estimated unadjusted, and the estimated harmonic mean (marginal) NNT given the explanatory variable, and a specified model.

Usage

nnt_x(model, response, x, cutoff, base, group, adj, decrease, data)

Arguments

model	specification of the regression model; anova for the one-way ANOVA model, linreg for the linear regression, and logreg for the logistic regression with the logit link-function.
response	vector of the response variable (i.e., the dependent variable).
x	vector of the explanatory variable.
cutoff	the MCID threshold. This argument is suitable for continuous response variables, namely for ANOVA and linear regression.
base	control group of the x variable in the one-way ANOVA model.
group	allocated arm variable where 1 corresponds to the treatment arm, and 0 to the control arm. Suitable for linear and logistic regression.
adj	value that the NNT need to be condtioned on (adjusted for). The default value is mean of x.
decrease	logical TRUE or FALSE. Indicates whether the MCID change is decrease in the response variable
data	data frame that contains the required variables for the computations.

Value

The estimated unadjusted, harmonic mean (marginal) and conditional (adjusted) NNT with their corresponding 95 percent confidence intervals given a specified model, and conditioned on (adjusted for) a specified value of the explanatory variables.

Examples

data(anova_data)

### SUCCESS = INCREASE
nnt_x(         model     = "anova",
               response  = anova_data$y,
               x         = anova_data$gr,
               cutoff    = 0,
               base      = 1,
               decrease  = FALSE,
               data      = anova_data)

### SUCCESS = DECREASE
nnt_x(         model     = "anova",
               response  = anova_data$y,
               x         = anova_data$gr,
               cutoff    = 2,
               base      = 4,
               decrease  = TRUE,
               data      = anova_data)

data(linreg_data)

### SUCCESS = INCREASE
nnt_x(         model    = "linreg",
               response = linreg_data$y,
               x        = linreg_data$x_var,
               cutoff   = 3,
               group    = linreg_data$gr,
               decrease = FALSE,
               adj      = 2.6,
               data     = linreg_data )

### SUCCESS = DECREASE
inv_data = data.frame( y = linreg_data$y, x_var = linreg_data$x_var, gr = 1 - linreg_data$gr )

nnt_x(  model    = "linreg",
       response  = inv_data$y,
        x        = inv_data$x_var,
        cutoff   = 3,
        group    = inv_data$gr,
        decrease = TRUE,
        adj      = 2.6,
        data     = inv_data )

data(logreg_data)

nnt_x( model    = "logreg",
       response = logreg_data$y,
       x        = logreg_data$x_var,
       group    = logreg_data$gr,
       adj      = 1.5,
       data     = logreg_data )

vancak/NNTcalculator documentation built on April 7, 2022, 3:48 a.m.