Distill EPHT Data Down to XML • distiller

Motivation

As a newbie who has to submit data to the CDC’s EPHT program, I was dismayed to find out that the documentation is buried under many layers inside their SharePoint. It is also highly fragmented, convoluted and in many cases, conflicts with itself.

My goal is to make this process easier and reproducible for myself, and others.

So who is this highly specific package for?

Do you submit data to the CDC’s EPHT program?
Do you use R? Or are interested in incorporating R into your workflow?
Do you struggle with the CDC’s EPHT documenation and/or tooling?
Do you want to make your submission process more reproducible?

If you answered yes to the first question and any of the others, then this package might be for you.

What does this package do?

I think it’s important to state up front what this package doesn’t do - and that is, it will not wrangle your data for you. There are a few helpers, and and a whole slew of checks distiller will run on your data and metadata to ensure that everything is reasonably close to the correct format for submission to the CDC’s EPHT program.

distiller still expects your data to have specific variable names, and to have the required variables for each type of data. However, if you’ve ever wondered why the epht requires different variable names in a different order for the same types of data, even for the same disease you’ll be pleased to know that distiller takes care of the facility-type-specific naming conventions and the ordering for you. Users just need to bring the data and now they can spend less time worrying about XML semantics and more time polishing their data products.

disitller is no replacement for the CDC EPHPT Test Submission portal, however, creating the XML, and shuffling files around and then dropping them into the portal and waiting an indeterminate amount of time for feedback eats up time and is a pain. distiller aims to provide feedback on your data and metadata before you send it off to the CDC. This way, you can fix any obvious issues before you sink 20+ minutes waiting to find out you forgot to replace your NA’s with “U”.

What’s in the box?

distiller contains the following core functions:

check_submission() - a function that checks your data and metadata and provides quick feedback
make_xml_document() - a function that creates an xml document for submission based on your data and the metadata your provide it

distiller also contains functions for:

collapsing race and ethnicity values into the CDC’s required format
converting month integers to 0-padded character strings
return the proper health outcome identifier for a given content group identifier
Starting from scratch? Most of the mini-functions that make up the two core ones are exposed to the user, so you can check your work in pieces as you make progress with your data wrangling

`distiller` expectations and scope

distiller works for the following content group identifiers:

AS-HOSP
AS-ED
CO-HOSP
CO-ED
MI-HOSP
HEAT-HOSP
HEAT-ED
COPD-HOSP
COPD-ED

distiller expects the following variables in your data:

For every content group identifier:

agegroup
county
sex
ethnicity
race
health_outcome_id,
monthly_count
month
year

For content group identifiers CO-HOSP and CO-ED, the above plus the following:

fire_count
nonfire_count
unknown_count

Installation

You can install the development version of distiller from GitHub with:

# install.packages("pak")
pak::pak("asenetcky/distiller")

Example

Here is a basic example of how to use it:

library(distiller)

# Take you already-wrangled data
# note the specific variable names
data <-
  mtcars |>
  dplyr::rename(
    month = mpg,
    agegroup = cyl,
    county = disp,
    ethnicity = hp,
    health_outcome_id = drat,
    monthly_count = wt,
    race = qsec,
    sex = vs,
    year = am
  ) |>
  dplyr::select(-c(gear, carb))

# And your metadata
content_group_id <- "AS-HOSP"
mcn <- "1234-1234-1234-1234-1234"
jurisdiction_code <- "two_letter_code"
state_fips_code <- "1234"
submitter_email <- "submitter@email.com"
submitter_name <- "Submitter Name"
submitter_title <- "Submitter Title"

# Optionally check your submission data structure and metadata
check_submission(
  data,
  content_group_id,
  mcn,
  jurisdiction_code,
  state_fips_code,
  submitter_email,
  submitter_name,
  submitter_title
)
#> ℹ Checking submission metadata
#> ✔ Success: content_group_id
#> ! Warning: mcn may not have correct format
#> Troublemakers: length, format
#> ! Warning: jurisdiction_code may not have correct format
#> Troublemakers: length, format
#> ! Warning: state_fips_code may not have correct format
#> Troublemakers: length, format
#> ✔ Success: submitter_email
#> ✔ Success: submitter_name
#> ✔ Success: submitter_title
#> ℹ Checking data structure and content
#> ✔ Success: dataframe_structure
#> ✖ Danger: month does not have allowable value/s
#> Troublemakers: allowed_values
#> ✔ Success: agegroup
#> ✖ Danger: county does not have allowable value/s
#> Troublemakers: length
#> ✖ Danger: ethnicity does not have allowable value/s
#> Troublemakers: allowed_values
#> ✖ Danger: health_outcome_id does not have allowable value/s
#> Troublemakers: allowed_values
#> ✖ Danger: sex does not have allowable value/s
#> Troublemakers: allowed_values
#> ✖ Danger: year does not have allowable value/s
#> Troublemakers: allowed_values
#> ✖ Danger: race does not have allowable value/s
#> Troublemakers: allowed_values
#> ✔ Success: monthly_count

# This can also be checked with `check_first = TRUE` in `make_xml_document()`


# And then make your xml document
make_xml_document(
  data,
  content_group_id,
  mcn,
  jurisdiction_code,
  state_fips_code,
  submitter_email,
  submitter_name,
  submitter_title
)
#> {xml_document}
#> <HospitalizationData schemaLocation="http://www.ephtn.org/NCDM/PH/HospitalizationData ephtn-ph-HospitalizationData.xsd" xmlns="http://www.ephtn.org/NCDM/PH/HospitalizationData" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
#> [1] <Header>\n  <MCN>1234-1234-1234-1234-1234</MCN>\n  <JurisdictionCode>two_ ...
#> [2] <Dataset>\n  <Row>\n    <RowIdentifier>1</RowIdentifier>\n    <AdmissionM ...

distiller