In thebackman/CDRalg: Create CDR (Clinical Dementia Rating) total scores from raw data

options(width = 150)

Introduction

This R package contains a function to assign a global CDR-rating based on the algorithm by [@morris1993clinical]. A implementation of algorithm can be found here. The scoring rules are also found here.

Pre usage

In order to use the function the data must be read into R. This could be done using several packages depending on your type of input data. Packages such as haven, readr or readxl are avaliable. I will show an example using data stored in an excel sheet.

There are a few prerequisites for the data.

1. The data set must have an ID-number. This is used for internal merging.
2. The data must have the variables glo5-glo9 and they must have these names.

These are the variables:

• glo5 - memory (m)
• glo6 - orientation (o)
• glo7 - judgement, problemsolving (jps)
• glo8 - community affairs (ca)
• glo9 - home and hobbies (hh)
• glo10 - personal care (pc)

In case an ID-variable is not found or any of the glo5-glo10 variables are not found the program will terminate and produce an error message. The order of the variables are not important but the names are.

Usage

The main function is code_CDR(df, id_name, deb = F) The function takes three arguments. The first argument is the data frame to be used. The second argument is a id variable found in the data set, provided as a quoted string. This is used for internal sorting and merging and must be provided. The third argument is for debugging use. This can be useful for testing purposes or in case a bug is found in the implementation of the algorithm (to the best of my knowledge the algorithm is correct).

The function will also perform a few checks on the data. All glo-variables have to be numeric and can only take the values 0, 0.5, 1, 2 and 3. In case this is not fulfilled the program will give you an error message.

A person must also have complete data in order for the algoritm to work. Therefore those that do not have complete data (on glo5-glo10) will get a missing observation on the CDR-rating (NA). The program takes care of this automatically.

Examples

I include a data set in this package. An excel file with data that I use to test the algorithm. I will use this file for my examples. After you have installed the package you can also use this code, but it is much easier to just load the data included in the package, scroll down to the comments section.

# load package to read excel files
library(readxl)

# Read data (skip the first row in the excel file as it contains comments and not data)
indata <- read_excel(system.file("extdata", "examples.xlsx", package = "CDRalg"), skip = 1)

We can now look at this file.

indata

The data file contains some blank lines (which I use to separate examples). Let´s clean it up a bit using dplyr.

# load dplyr
library(dplyr)

# filter rows that are empty
indata <- filter(indata, is.na(test) != T)

# print it
indata

The data is now ready to be processed by our function code_CDR.

# load the package
library(CDRalg)

# assign the results of the function to an object
results <- code_CDR(df = indata,
                    id_name = "id",
                    deb = F)

Print some of the results.

head(results)

We can see that we now have an extra column called CDR. This is the assigned diagnosis from the algorithm. If we want to see how the algorithm works we can use the debugging deb = TRUE option. This prints the iteration. One iteration per line in the data frame (one per person).

# subsetting the data frame so that we only show one iteration
indata2 <- slice(indata, 1:1)

# applying the function
results <- code_CDR(indata2, id_name = "id", deb = T)

Lastly we want to save the output data. There are lots of options but one way is writing it to disk as a csv-file.

# load a packade for saving csv:s
library(readr)

# write it to disk
write_csv(results, path = "CDRratings.csv")

Comments

If you want to play around with the function a data set called CDRexamples is included in the package and it is the same one that I got from the excel file, it´s just easier to load. Use:

# load data set
data("CDRexamples")

# print part of it
head(CDRexamples)

If you want to know the details of the data set, you can write

?CDRexamples

References

thebackman/CDRalg documentation built on May 23, 2019, 7:36 a.m.