Description Usage Arguments Value Warning Background History/development log Author(s) See Also Examples

View source: R/classifyScoresVectorByRCI.R

Function to classify changes using the RCI paradigm

1 2 3 4 5 6 7 8 9 10 11 12 13 14 |

`scoreChange` |
numeric vector of score changes |

`score1` |
numeric vector of baseline scores (if not giving scoreChange) |

`score2` |
numeric vector of final scores (if not giving scoreChange) |

`RCI` |
numeric value of the RCI |

`cueing` |
whether measure is cued positively or negatively |

`verbose` |
logical: suppresses messages if FALSE |

`returnTable` |
logical: whether summary table returned (TRUE) or a tibble of classified scores (FALSE) |

`dp` |
numeric: number of decimal places in percentages (if returnTable TRUE) |

`addCI` |
logical: whether to add confidence intervals around observed percentages e.g. "69.6% to 87.5%" |

`returnNumeric` |
logical: whether to return values as numeric or as nicely formatted strings (the default) |

`CLsSeparate` |
logical: returns CI as two separate variables instead of that string |

`conf` |
numeric, gives the width of the confidence interval (if addCI is TRUE) |

a tibble, either of the summary breakdown with n and % by classification, or a tibble of the data with classified change

Function is still under development. Currently handles vector input. I will be modifying it to handle data frame and tibble input but not sure how best to do that yet. Currently heading for making that a separate function: classifyScoresInDataByRCI()

The RCI is part of Jacobson, Follette and Revenstorf's "RCSC": Reliable and Clinically Significant Change paradigm.

Started before 5.iv.21 Tweaked 11.iv.21 to fix error in examples.

Chris Evans

`getCSC`

for more general information about the paradigm and

`getRCIfromSDandAlpha`

for more detail about the RCI.

Note that by the logic of the criterion, change has to exceed the RCI, not just equal it, for change to be deemed reliable.

Splitting change into three levels: reliable deterioration, no reliable change, and reliable improvement is pretty trivial. However, as I worked on this function it did seem to grow into something not so trivial that probably will save people time. It can be used in two main ways:

to return a tibble of the scoreChange or the score1, score2 and computed scoreChange with the RCI categories added (returnTable = FALSE).

to return a tabulation of the categories (returnTable = TRUE).

I have given examples below but there are some aspects to option 2) wthat can be adjusted by:

using dp to determine the decimal places on the percentage breakdown (defaults to 1 decimal place)

using addCI = TRUE to add a column giving the confidence interval (CI) around the observed percentage and perhaps ...

using conf to choose something other than the default .95, i.e. a 95% interval ... though I can't really think of a good reason to do this!

Adding the CI is probably mainly useful when you want to compare the breakdown from a particular set of data with some other data or some published figure or even, though I hope not, some managerial or political target.

Bear in mind that the three percentages must sum to 100% so they are not independent of one another and the three CIs are therfore also not independent.

The CI is calculated using `binconf`

function from Frank Harrell's Hmisc package of miscellaneous
functions. That uses Wilson's method to compute the CIs. See
the documentation for `binconf`

for a brief
discussion of that choice and references.

Other RCSC functions:
`getBootCICSC()`

,
`getCSC()`

,
`getRCIfromSDandAlpha()`

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 | ```
## Not run:
### start with some very simple change values:
scoreChanges <- -5:5
scoreChanges # is:
# [1] -5 -4 -3 -2 -1 0 1 2 3 4 5
## so now apply an RCI value of 2 to that:
classifyScoresVectorByRCI(scoreChange = scoreChanges, RCI = 2)
## produces:
# You input 11 values for scoreChange. There were no missing values.
## A tibble: 3 x 4
# RCIclass n percent valid_percent
# <ord> <int> <chr> <chr>
# 1 Reliable deterioration 3 27.3% 27.3%
# 2 No reliable change 5 45.5% 45.5%
# 3 Reliable improvement 3 27.3% 27.3%
### you could pipe that to your R/tidyverse table formatting
### tool of choice, \code{pander}, \code{huxtable}, whatever.
### you can add the 95% CI:
### create some spurious scores
n <- 75
score1 <- rnorm(n) # random Gaussian scores
score2 <- score1 - rnorm(n, mean = .2, sd = .2) # and some random change
scoreChange <- score1 - score2 # get the change
classifyScoresVectorByRCI(scoreChange, RCI = .3)
classifyScoresVectorByRCI(score1 = score1, score2 = score2, RCI = .3)
## End(Not run)
``` |

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.