High correlation warnings while executing Generalizability

[parkdongsu]

Hello, @ameliaaveritt

Recently, I'm working for executing generalizability study.

ps <- createPs(cohortMethodData = cohortMethodData, population = z, stopOnError = FALSE, prior = createPrior("laplace", exclude = c(0), useCrossValidation = T), control = createControl(cvType = "auto", startingVariance = 0.01, noiseLevel = "quiet", tolerance = 2e-07, cvRepetitions = 10, threads = 24))

While executing the code above, high correlation warnings occurred as below:
"
High correlation between covariate(s) and treatment detected:
covariateId covariateName analysisId conceptId
19011773414 drug_era group during day -365 through 0 days relative to index: Ascorbic Acid 414 19011773
997276412 drug_era group during day -30 through 0 days relative to index: Cimetidine 412 997276
1750500415 drug_era group during day -180 through 0 days relative to index: Clarithromycin 415 1750500
1112921412 drug_era group during day -30 through 0 days relative to index: Ipratropium 412 1112921
19069425406 drug_era during day -365 through 0 days relative to index: nimesulide 406 19069425
19011773406 drug_era during day -365 through 0 days relative to index: Ascorbic Acid 406 19011773
21603274412 drug_era group during day -30 through 0 days relative to index: Adrenergics in combination with corticosteroids or other drugs, excl. anticholinergics 412 21603274
21604035416 drug_era group during day -30 through 0 days relative to index: TOPICAL PRODUCTS FOR JOINT AND MUSCULAR PAIN 416 21604035
1112921404 drug_era during day -30 through 0 days relative to index: Ipratropium 404 1112921
21600109416 drug_era group during day -30 through 0 days relative to index: Other drugs for peptic ulcer and gastro-oesophageal reflux disease (GORD) 416 21600109
19069425410 drug_era group during day -365 through 0 days relative to index: nimesulide 410 19069425
21490587701 measurement any time prior through 0 days relative to index: Arterial blood temperature 701 21490587
19069425414 drug_era group during day -365 through 0 days relative to index: nimesulide 414 19069425
21490587702 measurement during day -365 through 0 days relative to index: Arterial blood temperature 702 21490587
21604034416 drug_era group during day -30 through 0 days relative to index: TOPICAL PRODUCTS FOR JOINT AND MUSCULAR PAIN 416 21604034
21604036416 drug_era group during day -30 through 0 days relative to index: Antiinflammatory preparations, non-steroids for topical use 416 21604036
997276404 drug_era during day -30 through 0 days relative to index: Cimetidine 404 997276
19069425402 drug_era during day -365 through 0 days relative to index: nimesulide 402 19069425
1112921413 drug_era group during day 0 through 0 days relative to index: Ipratropium 413 1112921
21603274413 drug_era group during day 0 through 0 days relative to index: Adrenergics in combination with corticosteroids or other drugs, excl. anticholinergics 413 21603274
1750500407 drug_era during day -180 through 0 days relative to index: Clarithromycin 407 1750500
3017569701 measurement any time prior through 0 days relative to index: Oxygen content in Arterial blood 701 3017569
1112921405 drug_era during day 0 through 0 days relative to index: Ipratropium 405 1112921
19134101304 drug_exposure during day -30 through 0 days relative to index: Ipratropium Bromide 0.25 MG/ML Inhalant Solution 304 19134101
1750569303 drug_exposure during day -180 through 0 days relative to index: Clarithromycin 500 MG Extended Release Oral Tablet 303 1750569
3017569702 measurement during day -365 through 0 days relative to index: Oxygen content in Arterial blood 702 3017569
19018873304 drug_exposure during day -30 through 0 days relative to index: Cimetidine 200 MG Oral Tablet 304 19018873
"

[ameliaaveritt]

Hi @parkdongsu

Re: the high correlation, I have a few things to note…

1. It is important that you run the line that reads “excludedConcepts <- c(1597756, 1580747)” AND in the FeatureExtraction::createCovariateSettings function, you include the argument, "excludedCovariateConceptIds = excludedConcepts, addDescendantsToExclude = TRUE”. Both of these are already in the code, you do not have to add them.

2. You have included the createPs argument, "stopOnError = FALSE” did your code break (ie, does the ps object not exist?) Or was it just a warning?

Amelia