OHDSI Home | Forums | Wiki | Github

Review, Evaluate, and Validate the frailty score


(Seng Chan You) #1

Colleagues,

Since when I worked in the hospital, I’ve been interested in the frailty. Recently, a Hospital Frailty Risk Score (HFRS) based on ICD-10 code system was published in Lancet (Gilbert et al., Development and Validation of a Hospital Frailty Risk Score Focusing on Older People in Acute Care Settings Using Electronic Hospital Records: An Observational Study.” The Lancet 391, no. 10132 (May 5, 2018)).

I implemented this score in FeatureExtraction package, and I encourage you to review, evaluate, and validate this score in various real-world database.

The code below describes how to extract this score for your own cohort

devtools::install_github('ABMI/FeatureExtraction', ref= 'hfrs')
library(FeatureExtraction)

connectionDetails <- DatabaseConnector::createConnectionDetails(dbms = dbms,
                                                                server = server,
                                                                user = user,
                                                                password = pw,
                                                                port = port)
cdmDatabaseSchema <- ""
cohortDatabaseSchema <- ""
cohortTable <- "cohort"
oracleTempSchema <- NULL
cdmVersion <- "5"

cohortId = cohort definition id of yours (INT)

settings<-createCovariateSettings(useHfrs = TRUE)
covs <- getDbCovariateData(connectionDetails = connectionDetails,
                           oracleTempSchema = oracleTempSchema,
                           cdmDatabaseSchema = cdmDatabaseSchema,
                           cohortDatabaseSchema = cohortDatabaseSchema,
                           cohortTable = cohortTable,
                           cohortId = cohortId,
                           rowIdField = "subject_id",
                           covariateSettings = settings,
                           aggregated = FALSE)

cov.d<-as.data.frame(covs$covariates)
hist(cov.d$covariateValue)

t