nomogram: Fagan's nomogram to show the relationships between the prior...

In UncertainInterval: Uncertain Interval Methods for Three-Way Cut-Point Determination in Test Results

Description

Next to plotting Fagan's nomogram, this function also calculates the minimally needed values for specificity and sensitivity to reach desired posttest probabilities (or likelihood ratios) for a grey zone (Coste et al., 2003, 2006).

Usage

nomogram(
  prob.pre.test = 0.5,
  probs.post.test = c(pos = NULL, neg = NULL),
  SeSp = c(Se = NULL, Sp = NULL),
  LR = c(PLR = NULL, NLR = NULL),
  plot = T
)

Arguments

prob.pre.test	The prior test probability, with a default value of .5. Often, (local) prevalence is used.
probs.post.test	A vector of two values that give the desired posttest probabilities of observing the event in the case of a positive test result (positive posttest probability: pos), and the posttest probability of observing the event in the case of a negative test result (negative posttest probability: neg). When not given, these probabilities are calculated using the likelihood ratios (LR).
SeSp	A vector of two values that give the desired sensitivity and specificity. When not given, the Se and Sp values are calculated from the desired posttest probabilities.
LR	A vector of two values that give the positive likelihood ratio (sensitivity / (1- specificity)): PLR of observing the event, and the negative likelihood ratio ((1 - sensitivity) / specificity): NLR of not observing the event. PLR is a value > 1, NLR is a value between 0 and 1. When not given, the LR values are calculated from the desired posttest probabilities.
plot	A Boolean that indicates whether a plot is desired.

Details

Parameter probs.post.test or SeSp or LR must be supplied, the other two values are calculated. When more than one parameter is given the other two are ignored. The basis of this function is adapted from package TeachingDemos.

Value

Vector of values:

$pre:

The given pre-test probability.

$min.LRpos:

The given or calculated minimally required positive likelihood ratio. If no value is provided, it is calculated.

$max.LRneg:

The given or calculated maximally required negative likelihood ratio. If no value is provided, it is calculated.

$post.pos:

The given or calculated positive posttest probability.

$minSp:

The minimum value for the specificity, needed to reach the desired posttest probabilities.

$minSe:

The minimum value for the sensitivity, needed to reach the desired posttest probabilities.

References

Fagan, T. J. (1975). Nomogram for Bayes theorem. The New England Journal of Medicine, 293(5), 257-257.

Coste, J., Jourdain, P., & Pouchot, J. (2006). A gray zone assigned to inconclusive results of quantitative diagnostic tests: application to the use of brain natriuretic peptide for diagnosis of heart failure in acute dyspneic patients. Clinical Chemistry, 52(12), 2229-2235.

Coste, J., & Pouchot, J. (2003). A grey zone for quantitative diagnostic and screening tests. International Journal of Epidemiology, 32(2), 304-313.

Examples

# Show calculated results (first 3 times about the same)
(nomogram(prob.pre.test = .10, probs.post.test=c(pos=.70, neg=.001), plot=FALSE))
(nomogram(prob.pre.test = .10, SeSp=c(Se=0.991416309, Sp=0.952789700), plot=FALSE))
(nomogram(prob.pre.test = .10, LR=c(pos=21, neg=0.0090090091), plot=FALSE))
(nomogram(prob.pre.test = .10, SeSp=c(Se=0.99, Sp=0.95), plot=FALSE))
# plot only
nomogram(prob.pre.test = .10, LR=c(pos=21, neg=0.0090090091))
# plot and display precise results
(nomogram(prob.pre.test = .10, probs.post.test=c(pos=.70, neg=.001)))

# check the influence of different values of prevalence
i=1
out=matrix(0,nrow = 9, ncol= 7)
for (prev in (seq(.1, .9, by=.1))) {
  out[i,]=nomogram(prob.pre.test=prev, probs.post.test=c(.95, .05), plot=FALSE)
  i=i+1
}
colnames(out) = names(nomogram(prob.pre.test=prev, probs.post.test=c(.95, .05), plot=FALSE))
out

UncertainInterval documentation built on March 3, 2021, 1:10 a.m.