ctrdata (version 1.22.3)

f.likelyPlatformTrial: Calculate if study is likely a platform trial or not

Description

Trial concept calculated: platform trial, research platform. As operational definition, at least one of these criteria is true: a. trial has "platform", "basket", "umbrella", "multi.?arm", "multi.?stage" or "master protocol" in its title or description (for ISRCTN, this is the only criterion; some trials in EUCTR lack data in English), b. trial has more than 2 active arms with different investigational medicines, after excluding comparator, auxiliary and placebo medicines (calculated with f.numTestArmsSubstances; not used for ISRCTN because it cannot be calculated precisely), c. trial has more than 2 periods, after excluding safety run-in, screening, enrolling, extension and follow-up periods (for CTGOV and CTGOV2, this criterion requires results-related data).

Usage

f.likelyPlatformTrial(df = NULL)

Value

data frame with columns `_id` and `.likelyPlatformTrial`, a logical, and two complementary columns, each with lists of identifiers: `.likelyRelatedTrials` (based on other identifiers provided in the trial record, including `associatedClinicalTrials` from CTIS; listing identifiers whether or not the trial with the other identifier is in the database collection) and `.maybeRelatedTrials` (based on similar short terms in the first set of brackets or before a colon in the trial title; only listing identifiers from the database collection).

Arguments

df

data frame such as from dbGetFieldsIntoDf. If `NULL`, prints fields needed in `df` for calculating this trial concept, which can be used with dbGetFieldsIntoDf.

Details

For EUCTR, requires that results have been included in the collection, using `ctrLoadQueryIntoDb(queryterm = ..., euctrresults = TRUE, con = ...)`. Requires packages dplyr and stringdist to be installed; stringdist is used for evaluating terms in brackets in the trial title, where trials may be related if the term similarity is 0.77 or higher.

Publication references considered: EU-PEARL WP2 2020 and Williams RJ et al. 2022, tools:::Rd_expr_doi("10.1136/bmj-2021-067745")

Examples

# fields needed
f.likelyPlatformTrial()

# apply trial concept when creating data frame
dbc <- nodbi::src_sqlite(
  dbname = system.file("extdata", "demo.sqlite", package = "ctrdata"),
  collection = "my_trials", flags = RSQLite::SQLITE_RO)
trialsDf <- dbGetFieldsIntoDf(
  calculate = "f.likelyPlatformTrial",
  con = dbc)
trialsDf