Package {cprd.search}

Search the Clinical Practice Research Datalink Aurum Dictionary

Description

Load a user-supplied dictionary with load_cprd_dictionary(), search with cprd_search(), cprd_search_or(), cprd_search_exclude(), or cprd_search_fuzzy(), combine with cprd_combine(), and save with cprd_export(). cprd_browse() provides an interactive console.

Details

Only synthetic example data is included. Real CPRD dictionary files must be obtained separately. Search results are candidates for review, not validated study phenotypes.

Author(s)

Maintainer: Ramin Rezaeianzadeh ramin.rezaeianzadeh@outlook.com

See Also

Useful links:

• https://github.com/raminrzn/cprd_search

• Report bugs at https://github.com/raminrzn/cprd_search/issues

Interactive Dictionary Browser

Description

Repeatedly search the loaded dictionary at a console prompt. Space-separated keywords use AND logic. Use ⁠OR: asthma, diabetes⁠ for OR logic, or diabetes -gestational for exclusions. Prefixes are case-insensitive. Type q to quit. Keywords are literal, case-insensitive substrings.

Usage

cprd_browse(min_obs = 0)

Arguments

Value

NULL, invisibly, for its printing and prompting side effects.

Examples

if (interactive()) {
    load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                     package = "cprd.search"))
    cprd_browse()
}

Combine Multiple Search Results

Description

Combine search result tables, keeping the first occurrence of each MedCodeId. Columns absent from an input are filled with missing values.

Usage

cprd_combine(..., select = NULL)

Arguments

Value

A tibble deduplicated by MedCodeId, sorted by decreasing Observations when available (missing counts last). Otherwise input order is preserved. Conflicting duplicate rows use the first row, without merging its values with later rows.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
a <- cprd_search("diabetes", select = c("MedCodeId", "Term"))
b <- cprd_search_or("diabetes", "asthma", select = c("MedCodeId", "Term"))
cprd_combine(a, b)

Export Search Results to CSV

Description

Save a tibble, data.frame, or data.table of search results to a UTF-8 CSV. Internal term_lower and fuzzy dist columns are omitted.

Usage

cprd_export(results, filename, open = FALSE)

Arguments

Details

Character identifiers are written without numeric conversion. When importing the CSV, specify character types for identifier columns (for example, colClasses = "character" in utils::read.csv()) to avoid another application removing leading zeroes or rounding long identifiers.

Value

The supplied file path, invisibly.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
results <- cprd_search("diabetes")
file <- tempfile(fileext = ".csv")
cprd_export(results, file)
utils::read.csv(file, colClasses = "character")
unlink(file)

Search the CPRD Dictionary with AND Logic

Description

Find rows containing every supplied keyword. Matching is case-insensitive and uses literal substrings unless regex = TRUE.

Usage

cprd_search(..., min_obs = 0, column = "Term", select = NULL, regex = FALSE)

Arguments

Value

A tibble, sorted by decreasing Observations, with missing counts last. No matches returns a zero-row tibble with the same columns and types.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
cprd_search("family history", "lung cancer")
cprd_search("diabetes", min_obs = 50, select = c("MedCodeId", "Term"))
cprd_search("H33..", column = "CleansedReadCode")
cprd_search("^asthma$", regex = TRUE)

Search with Exclusions

Description

Find rows matching inclusion keywords but none of the exclusion keywords. Matching is case-insensitive and literal unless regex = TRUE.

Usage

cprd_search_exclude(
  include,
  exclude = NULL,
  min_obs = 0,
  or_logic = FALSE,
  select = NULL,
  column = "Term",
  regex = FALSE
)

Arguments

Value

A tibble, sorted by decreasing Observations, with missing counts last. No matches returns a zero-row tibble with the same columns and types.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
cprd_search_exclude("diabetes", exclude = "gestational")

Fuzzy Search Complete Dictionary Terms

Description

Compare a query with complete, case-insensitive term descriptions using string distances. Every eligible term is compared; no prefix filter is used.

Usage

cprd_search_fuzzy(
  term,
  max_dist = NULL,
  method = "osa",
  min_obs = 0,
  max_results = 50,
  select = NULL
)

Arguments

Details

Distances compare the whole query with the whole description, not individual words or substrings. Thus "diabeties" can match "diabetes" at an edit distance of 1, but a longer description such as "gestational diabetes" requires a larger edit threshold. Use cprd_search() for literal substring searches. Search candidates still require review when building a study code list.

Value

A tibble containing dictionary columns and numeric dist, ordered by increasing distance, then decreasing observations. select can include dist. Missing or empty descriptions do not match.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
cprd_search_fuzzy("diabeties", select = c("MedCodeId", "Term", "dist"))
cprd_search_fuzzy("xiabetes", max_dist = 1)
cprd_search_fuzzy("diabeties", method = "jw", max_dist = 0.1)

Search the CPRD Dictionary with OR Logic

Description

Find rows containing any supplied keyword, using case-insensitive matching.

Usage

cprd_search_or(..., min_obs = 0, column = "Term", select = NULL, regex = FALSE)

Arguments

Value

A tibble, sorted by decreasing Observations, with missing counts last. No matches returns a zero-row tibble with the same columns and types.

Examples

load_cprd_dictionary(system.file("extdata", "synthetic-dictionary.tsv",
                                 package = "cprd.search"))
cprd_search_or("asthma", "lung cancer", select = c("MedCodeId", "Term"))

Load the CPRD Aurum Medical Dictionary

Description

Read a user-supplied tab-delimited dictionary and cache it for searching. The package includes only a small, entirely synthetic example dictionary.

Usage

load_cprd_dictionary(filepath = NULL, force = FALSE)

Arguments

Details

Required columns are MedCodeId, Term, and Observations. Standard additional columns are OriginalReadCode, CleansedReadCode, SnomedCTConceptId, SnomedCTDescriptionId, Release, and EmisCodeCategoryId. All columns except Observations are read as character to preserve identifiers. Observation counts must be non-negative whole numbers; blank or NA counts are treated as missing. Missing terms do not match searches. Empty or missing medical code identifiers are rejected.

Supplying a different path loads a new dictionary. Failed loads leave the previous cache intact. Use force = TRUE after changing a cached file. Real dictionaries must be obtained separately through the Clinical Practice Research Datalink; see https://www.cprd.com/ for access information.

Value

An independent copy of the loaded data.table, invisibly. Identifier columns are character, Observations is numeric, and term_lower is an internal lowercase copy of Term.

Examples

path <- system.file("extdata", "synthetic-dictionary.tsv", package = "cprd.search")
dict <- load_cprd_dictionary(path)
nrow(dict)