Demo for basic stratification

In LTASR: Functions to Replicate the Center for Disease Control and Prevention's 'LTAS' Software in R

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(dplyr)
library(readr)
library(knitr)
library(LTASR)
per_form <- tibble(Variable = c('id',
'gender',
'race',
'dob',
'pybegin',
'dlo',
'vs',
'code',
'rev'
),
Description = c('Unique identifier for each person',
'Sex of person ("M" = male / "F" = female)',
'Race of person ("W" = white / "N" = nonwhite)',
'date of birth',
'date to begin follow-up',
'date last observed. Minimum of end of study, date of death, date lost to follow-up',
'an indicator of ‘D’ for those who are deceased. All other values will be treated as alive or censored as of dlo.',
'ICD code of death; missing if person is censored or alive at study end date',
'revision of ICD code (7-10); missing if person is censored or alive at study end date'
),
Format = c('',
'character',
'character',
'character',
'character',
'character',
'character',
'character',
'numeric'
)
)

• It is highly recommended to use R as well as RStudio. RStudio is a separate piece of software used to write and run R code.

• It is also highly recommended to use LTASR in conjunction with the tidyverse package. If this has not been installed yet, it will need to be installed for your first use:

install.packages('tidyverse')
install.packages('LTASR')

This will only need to be run once, ever. It will install the necessary packages and functions to your computer.

1. The next step is to create a data file containing person level information. Each row of the file represents one unique person. The data file must contain the following variables:

kable(per_form)

Note: R variable names are case sensitive! LTASR assumes all variable names are lower-case. This can be done in any software. Below is an example person file format:

options(knitr.kable.NA = '')
person_example %>%
  kable()

1. Once the person file has been created, there are several tools that can be used to read it into R. RStudio has a button in the top-right labeled 'Import Dataset' which provides tools for reading in data from various sources. This example will use the read_csv() function to read in a csv file into R (alternatively, other options are excel and SAS datasets from the 'Import Dataset' drop-down). The read_csv() function takes the file path of the person csv file as the first argument.

NOTE: Paths in R use forward slashes (/) instead of the standard back slashes (\) used by windows.

The date columns will need to be converted to R date values. This can be done using the as.Date() function. This function requires the user to specify the format of the dates being converted. Below gives a key in how to describe the date format:

The above date columns in the above example person file have the general format of “%m/%d/%Y”.
Below gives example code to read in the person file as well as code to convert the date columns to date values:

library(LTASR)
library(tidyverse)

person <- read_csv('C:/person.csv') %>%
  mutate(dob = as.Date(dob, format='%m/%d/%Y'),
         pybegin = as.Date(pybegin, format='%m/%d/%Y'),
         dlo = as.Date(dlo, format='%m/%d/%Y'))

The LTASR package comes with an example person file, called person_example, that can be used for testing. For this example, it will be read in, its dates will be converted and it will be saved as 'person':

person <- person_example %>%
  mutate(dob = as.Date(dob, format='%m/%d/%Y'),
         pybegin = as.Date(pybegin, format='%m/%d/%Y'),
         dlo = as.Date(dlo, format='%m/%d/%Y'))

1. Next, a rate file object must be created. This can be done using the parseRate() function while reads in an LTAS xml rate file. These xml files are available upon request and contain national population death rates for various outcomes. Below gives an example of reading in this file:

rateobj <- parseRate('C:/UCOD_WashingtonState012018.xml')

Alternatively, the LTASR package comes pre-installed with a rate object for the 119 LTAS minors of the US population from 1960-2022. This can be used in any analyses:

rateobj <- us_119ucod_recent

1. Once a person file and a rate file have been read in, the get_table() function can be used to stratify the data from the person file according to the strata of the rate file (which is generally gender, race, 5-year age and 5-year calendar period categories):

py_table <- get_table(person, rateobj)

1. Once the py_table has been created, it can be saved using the write_csv() function. This file contains person day counts (pdays) and outcome counts for each minor (_o1, _o2, …..).

2. With your stratified data, you can quickly create a table with all SMRs for each minor found in the rate file using the smr_minor() function as below:

smr_minor_table <- smr_minor(py_table, rateobj)

1. This table can be viewed with View() and saved with write_csv():

View(smr_minor_table)
write_csv(smr_minor_table, "C:/SMR_Minors.csv")

1. Next, SMRs for various minor groupings can be calculated from this table. One particular grouping are the Majors (as was done in the original LTAS software) defined within the rate file. The smr_major() function can be used for this, which reads in the smr_minor table created in the previous step:

smr_major_table <- smr_major(smr_minor_table, rateobj)
View(smr_major_table)
write_csv(smr_major_table, "C:/SMR_Majors")

Additionally, any custom grouping can be calculated as well using the smr_custom() function where groupings are defined as a vector of numbers. Below defines groupings for all deaths and all cancers:

minor_grouping <- 1:119
all_causes <- smr_custom(smr_minor_table, minor_grouping)
View(all_causes)

minor_grouping <- 4:40
all_cancers <- smr_custom(smr_minor_table, minor_grouping)
View(all_cancers)

LTASR documentation built on Sept. 11, 2024, 6:51 p.m.