redcap_metadata_coltypes: Suggests a col_type for each field in a REDCap project

In OuhscBbmc/REDCapR: Interaction Between R and REDCap

Suggests a col_type for each field in a REDCap project

Description

This function inspects a REDCap project to determine a readr::cols() object that is compatible with the the project's current definition. It can be copied and pasted into the R code so future calls to the server will produce a tibble::tibble() with an equivalent set of data types.

Usage

redcap_metadata_coltypes(
  redcap_uri,
  token,
  print_col_types_to_console = TRUE,
  http_response_encoding = "UTF-8",
  locale = readr::default_locale(),
  verbose = FALSE,
  config_options = NULL,
  handle_httr = NULL
)

Arguments

redcap_uri	The uri/url of the REDCap server typically formatted as "https://server.org/apps/redcap/api/". Required.
token	The user-specific string that serves as the password for a project. Required.
print_col_types_to_console	Should the readr::cols() object be printed to the console?
http_response_encoding	The encoding value passed to httr::content(). Defaults to 'UTF-8'.
locale	a readr::locale() object to specify preferences like number, date, and time formats. This object is passed to readr::read_csv(). Defaults to readr::default_locale().
verbose	A boolean value indicating if messages should be printed to the R console during the operation. The verbose output might contain sensitive information (e.g. PHI), so turn this off if the output might be visible somewhere public. Optional.
config_options	A list of options passed to httr::POST(). See details at httr::httr_options(). Optional.
handle_httr	The value passed to the handle parameter of httr::POST(). This is useful for only unconventional authentication approaches. It should be NULL for most institutions. Optional.

Details

redcap_metadata_coltypes() returns a readr::cols() object in two ways. First, a literal object is returned that can be passed to redcap_read() or redcap_read_oneshot().

Second, the function acts as a code generator. It prints text to the console so that it can be copied and pasted into an R file. This is useful to (a) document what fields and data types are expected, and (b) adjust those fields and data types if the defaults can be customized for your needs. For instance, you may choose to exclude some variables or tweak a data type (e.g., changing a patient's height from an integer to a double).

When printing to the console, each data type decision is accompanied by an explanation on the far right. See the output from the examples below. Please file an issue if you think something is too restrictive or can be improved.

The overall heuristic is assign a data type down a waterfall of decisions:

1. Is the field built into REDCap? This includes an autonumber record_id, redcap_event_name, redcap_repeat_instrument, redcap_repeat_instance, and an instrument's completion status.

2. What is the field's type? For example, sliders should be an integer, while check marks should be logical(https://stat.ethz.ch/R-manual/R-devel/library/base/html/logical.html.

3. If the field type is "text", what is the validation type? For instance, a postal code should be a character (even though it looks like a number), a "mdy" should be cast to a date, and a "number_2dp" should be cast to a floating point

4. If the field type or validation type is not recognized, the field will be cast to character. This will happen when REDCap develops & releases a new type. If you see something like, "# validation doesn't have an associated col_type. Tell us in a new REDCapR issue", please make sure REDCapR is running the newest GitHub release and file a new issue if it's still not recognized.

For details of the current implementation, the decision logic starts about half-way down in the function's source code

*Validation does NOT Guarantee Conformity

If you're coming to REDCap from a database world, this will be unexpected. A validation type does NOT guarantee that all retrieved values will conform to complementary the data type. The validation setting affects only the values entered after the validation was set.

For example, if values like "abcd" where entered in a field for a few months, then the project manager selected the "integer" validation option, all those "abcd" values remain untouched.

This is one reason redcap_metadata_coltypes() prints it suggestions to the console. It allows the developer to adjust the specifications to match the values returned by the API. The the "abcd" scenario, consider (a) changing the type from col_integer to col_character, (b) excluding the trash values, then (c) in a dplyr::mutate() statement, use readr::parse_integer() to cast it to the desired type.

Value

A readr::cols() object is returned, which can be passed to redcap_read() or redcap_read_oneshot().

Additionally objected is printed to the console, see the Details below.

Author(s)

Will Beasley, Philip Chase

References

The official documentation can be found on the 'API Help Page' and 'API Examples' pages on the REDCap wiki (i.e., https://community.projectredcap.org/articles/456/api-documentation.html and https://community.projectredcap.org/articles/462/api-examples.html). If you do not have an account for the wiki, please ask your campus REDCap administrator to send you the static material.

Examples

## Not run: 
uri      <- "https://bbmc.ouhsc.edu/redcap/api/"

# A simple project
token      <- "9A81268476645C4E5F03428B8AC3AA7B" # 153
col_types  <- redcap_metadata_coltypes(uri, token)
redcap_read_oneshot(uri, token, col_types = col_types)$data

# A longitudinal project
token      <- "0434F0E9CF53ED0587847AB6E51DE762" # 212
col_types  <- redcap_metadata_coltypes(uri, token)
redcap_read_oneshot(uri, token, col_types = col_types)$data

# A repeating instruments project
token      <- "56F43A10D01D6578A46393394D76D88F" # 2603
col_types  <- redcap_metadata_coltypes(uri, token)
redcap_read_oneshot(uri, token, col_types = col_types)$data

# A project with every field type and validation type.
#   Notice it throws a warning that some fields use a comma for a decimal,
#   while other fields use a period/dot as a decimal
token      <- "8F5313CAA266789F560D79EFCEE2E2F1" # 2634 - Validation Types
col_types  <- redcap_metadata_coltypes(uri, token)
redcap_read_oneshot(uri, token, col_types = col_types)$data

## End(Not run)

OuhscBbmc/REDCapR documentation built on Jan. 31, 2024, 8:30 p.m.