Calculate Reference Interval and Corresponding Confidence Interval

This function is used to establish the reference interval for target population with parametric, non-parametric and robust methods that follows the CLSI-EP28A3 and NMPA guideline. In additional, it also provides the corresponding confidence interval for lower/upper reference limit if needed. Given that outliers should be identified beforehand, Tukey and Dixon methods can be applied depending on distribution of the data.

Usage

refInterval(
  x,
  out_method = c("doxin", "tukey"),
  out_rm = FALSE,
  RI_method = c("parametric", "nonparametric", "robust"),
  CI_method = c("parametric", "nonparametric", "boot"),
  refLevel = 0.95,
  bootCI = c("perc", "norm", "basic", "stud", "bca"),
  confLevel = 0.9,
  rng.seed = NULL,
  tol = 1e-06,
  R = 10000
)

Arguments

x

(numeric)
numeric measurements from target population.

out_method

(string)
string specifying the which outlier detection to use.

out_rm

(logical)
whether the outliers is removed or not.

RI_method

(string)
string specifying the which method for computing reference interval to use. Default is parametric, options can be nonparametric and robust.

CI_method

(string)
string specifying the which method for computing confidence interval of reference limit(lower or upper) to use. Default is parametric, options can be nonparametric and boot.

refLevel

(numeric)
reference range/interval, usual is 0.95.

bootCI

(string)
string specifying the which bootstrap confidence interval from boot.ci() function in boot package. Default is perc(bootstrap percentile), options can be norm(normal approximation), boot(basic bootstrap), stud(studentized bootstrap) and bca(adjusted bootstrap percentile).

confLevel

(numeric)
significance level for the confidence interval of reference limit.

rng.seed

(integer)
number of the random number generator seed for bootstrap sampling. If set to NULL currently in the R session used RNG setting will be used.

tol

(numeric)
tolerance for when the iterative process can be stopped in robust method.

R

(integer)
number of bootstrap replicates, is used in boot() function.

Value

A RefInt object contains relevant results in establishing of reference interval.

Note

There are some conditions of use to be aware of:

• If parametric method is used to calculate reference interval, confidence interval should be the same method as well.

• If non-parametric method is used to calculate the reference interval and the sample size is up to 120 observations, the non-parametric is suggested for confidence interval. Otherwise if the sample size is below to 120, the bootstrap method is the better choice. Beside the non-parametric method for confidence interval only allows the refLevel=0.95 and confLevel=0.9 arguments, if not the bootstrap methods will be used automatically.

• If robust method is used to calculate the reference interval, the method for confidence interval must be bootstrap.

Examples

data("calcium")
x <- calcium$Value
refInterval(x, RI_method = "parametric", CI_method = "parametric")
#> 
#>  Reference Interval Method: parametric, Confidence Interval Method: parametric 
#> 
#>  Call: refInterval(x = x, RI_method = "parametric", CI_method = "parametric")
#> 
#>   N = 240
#>   Outliers: NULL
#>   Reference Interval: 9.05, 10.32
#>   RefLower Confidence Interval: 8.9926, 9.1100
#>   Refupper Confidence Interval: 10.2584, 10.3757
refInterval(x, RI_method = "nonparametric", CI_method = "nonparametric")
#> 
#>  Reference Interval Method: nonparametric, Confidence Interval Method: nonparametric 
#> 
#>  Call: refInterval(x = x, RI_method = "nonparametric", CI_method = "nonparametric")
#> 
#>   N = 240
#>   Outliers: NULL
#>   Reference Interval: 9.10, 10.30
#>   RefLower Confidence Interval: 8.9000, 9.2000
#>   Refupper Confidence Interval: 10.3000, 10.4000
refInterval(x, RI_method = "robust", CI_method = "boot", R = 1000)
#> Bootstrape process could take a short while.
#> 
#>  Reference Interval Method: robust, Confidence Interval Method: boot 
#> 
#>  Call: refInterval(x = x, RI_method = "robust", CI_method = "boot", 
#>     R = 1000)
#> 
#>   N = 240
#>   Outliers: NULL
#>   Reference Interval: 9.04, 10.32
#>   RefLower Confidence Interval: 8.9777, 9.0979
#>   Refupper Confidence Interval: 10.2568, 10.3751