openFDA API • openFDA

openFDA makes querying the openFDA API from R a breeze. The API itself serves publicly available data from the FDA about foods, drugs, devices, and more. This data includes data such as recall enforcement reports, adverse events, manufacturer details, and – again – even more! Note that the data on openFDA has not been validated for clinical or production use.

Installation

The easiest way to install openFDA is to get it from CRAN. However, given a spate of recent improvements which are not yet on CRAN, you may have a better time using the development version. Nonetheless, to install from CRAN:

install.packages("openFDA")

Development version

# install.packages("pak")
pak::pkg_install("simpar1471/openFDA")

Using openFDA

library(openFDA)

The full documentation for the API is online, so look at the openFDA website to get a full feel for the API itself.

Before using the package, you will need to generate an API key on the openFDA website. With an API key you can make 240 requests per minute, and are limited to 120,000 requests per day (the package will handle this request rate throttling for you). Without a key, you can only make 1000 requests a day. For this package, if you try to run openFDA() without setting an API key, it will fail:

search <- openFDA(
  search = "openfda.generic_name:furosemide",
  limit = 5
)
#> Error:
#> ! No openFDA API key was found by `keyring::key_get()`.
#> ℹ To use openFDA, you must set an openFDA API key. Go to
#>   <https://open.fda.gov/apis/authentication/> to get an openFDA API key, then
#>   use `set_api_key()` to cache it for use in this and future sessions.

That’s not good. To fix this error, get an API key from the webpage linked above. Once you have your API key, use set_api_key() to store your API key. Once you run this function, openFDA will ‘know’ your API key, even for future sessions:

# n.b. this API key is a fake, just for demonstration
set_api_key(api_key = "iHXeqlbXtlYEwX9MYLwyCBuiZxfn98Sj3oX2FNtSx")
#> ! OpenFDA API key set.
#> ℹ You can retrieve the key with the following code:
#>   `keyring::key_get(service = "OPENFDA_KEY", user = "openFDA")`

Let’s try using openFDA() again, now that you have an API key:

search <- openFDA(
  search = "openfda.generic_name:furosemide",
  limit = 5
)

search
#> <httr2_response>
#> GET
#> https://api.fda.gov/drug/drugsfda.json?search=openfda.generic_name:furosemide&limit=5
#> Status: 200 OK
#> Content-Type: application/json
#> Body: In memory (29294 bytes)

The returned httr2 response object contains JSON data from the API - you can use httr2::resp_body_json() to extract the JSON data as a nested list, then extract the fields you want.

json <- httr2::resp_body_json(search)

json$results[[1]]$openfda$brand_name
#> [[1]]
#> [1] "FUROSEMIDE"
json$results[[1]]$openfda$pharm_class_epc
#> [[1]]
#> [1] "Loop Diuretic [EPC]"

I’ve found purrr to be very useful for parsing this data quickly.

purrr::map_chr(
  .x = json$results, 
  .f = \(result) purrr::pluck(result, "openfda", "manufacturer_name", 1)
)
#> [1] "Ipca Laboratories Limited"      "Hikma Pharmaceuticals USA Inc."
#> [3] "Phlow Corporation"              "Amneal Pharmaceuticals LLC"    
#> [5] "Accord Healthcare, Inc."

Other R packages for openFDA

The openfda package from rOpenHealth also wraps the openFDA API from R, and is available on GitHub. It’s got a pretty neat structure whereby you build up a query using individual functions for each parameter - though it’s my personal preference to keep these as parameters to a single function. It also makes results available in data frames, which is nice, but I think working with the response object and parsing the underlying JSON yourself permits more powerful interactions with the API.

There is also FDAopenR, which is all object-oriented. I couldn’t quite wrap my head around it, but it all appears to be in working order!