epiR: rsu.sep.rb – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

rsu.sep.rb

Surveillance system sensitivity assuming risk-based sampling and varying unit sensitivity

Description

Calculates surveillance system (population-level) sensitivity assuming one-stage, risk-based sampling and varying unit sensitivity using either the binomial or hypergeometric methods.

Usage

rsu.sep.rb(N, rr, ppr, df, pstar, method = "binomial")

Arguments

`N`	vector of the same length as `rr`, population size estimates for each risk group.
`rr`	vector of length equal to the number of risk strata, the relative risk values.
`ppr`	vector of the same length as `rr`, population proportions for each risk group.
`df`	a dataframe of values for each combination of risk stratum and sensitivity level. Column 1 = risk group index, column 2 = unit sensitivities, column 3 = the sample size for risk group and unit sensitivity).
`pstar`	scalar, the design prevalence.
`method`	character string indicating the method to be used. Options are `binomial` or `hypergeometric`. See details, below.

Details

If method = binomial N is ignored and values for ppr need to be entered. Conversely, if method = hypergeometric, ppr is ignored and calculated from N.

Value

A list comprised of five elements:

`sep`	scalar, the population-level sensitivity estimate.
`epi`	vector, effective probability of infection estimates.
`adj.risk`	vector, adjusted risks.
`n`	vector, sample size by risk group
`se.u`	a vector of the mean sensitivity for each risk group.

Examples

## EXAMPLE 1:
## Calculate the surveillance system sensitivity assuming one-stage risk-
## based sampling assuming a population comprised of high risk (n = 200 
## clusters) and low risk (n = 1800 clusters) where the probability of 
## disease in the high risk group is 5 times that of the low risk group.

## Four clusters will be sampled with n = 80, 30, 20 and 30 surveillance
## units within each cluster tested using a test with diagnostic sensitivity
## at the surveillance unit level of 0.92, 0.85, 0.92 and 0.85, respectively.

## Assume a design prevalence of 0.01.

rg <- c(1,1,2,2)
se.u <- c(0.92,0.85,0.92,0.85)
n <- c(80,30,20,30)
df <- data.frame(rg, se.u, n)

rsu.sep.rb(N = c(200,1800), rr = c(5,1), ppr = NA,  df = df, pstar = 0.01, 
   method = "hypergeometric")

## The expected surveillance system sensitivity is 0.993.

 
## EXAMPLE 2:
## Recalculate, assuming that we don't know the size of the cluster population
## at risk.

## When the size of the cluster population at risk is unknown we set N = NA 
## and enter values for ppr (the proportion of the population in each risk
## group). Assume (from above) that 0.10 of the cluster population are in the
## high risk group and 0.90 are in the low risk group.

rsu.sep.rb(N = NA, rr = c(5,1), ppr = c(0.10,0.90), df = df, pstar = 0.01, 
   method = "binomial")

## The expected surveillance system sensitivity is 0.980.

epiR

Tools for the Analysis of Epidemiological Data

v2.0.19

GPL (>= 2)

Authors

Mark Stevenson <mark.stevenson1@unimelb.edu.au> and Evan Sergeant <evansergeant@gmail.com> with contributions from Telmo Nunes, Cord Heuer, Jonathon Marshall, Javier Sanchez, Ron Thornton, Jeno Reiczigel, Jim Robison-Cox, Paola Sebastiani, Peter Solymos, Kazuki Yoshida, Geoff Jones, Sarah Pirikahu, Simon Firestone, Ryan Kyle, Johann Popp, Mathew Jay and Charles Reynard.

Initial release

2021-01-12