recode_meds_cycles1to2() and recode_meds_cycles3to6() recode medication data and merge medication columns into the main cycle data frame in one call. Both share the same call signature, so the workflow is identical regardless of cycle:
cycle1 <- recode_meds_cycles1to2(cycle1, cycle1_meds, c("any_htn_med", "diab_med"))
cycle3 <- recode_meds_cycles3to6(cycle3, cycle3_meds, c("any_htn_med", "diab_med"))
aggregate_meds_by_person() collapses long-format medication data to one row per respondent, preserving tagged_na("a") (valid skip) and tagged_na("b") (missing) distinctions across the aggregation.
recode_after_meds() recodes derived variables (e.g. htn_status, diab_status) whose inputs include pre-computed medication variables, automatically excluding meds-database rows from variable_details so medication columns are passed through rather than re-derived from raw ATC/MHR columns.
16 ATC-classification functions (is_ace_inhibitor(), is_beta_blocker(), is_calcium_channel_blocker(), is_diuretic(), is_diabetes_med(), is_nsaid(), is_any_antihtn_med(), is_other_antihtn_med(), plus their is_*_cycles1to2() variants) are exported for custom workflows.
New vignette recoding_medications.qmd documents the medication-recoding workflow for both cycle groupings, with a section on calling the underlying is_* classifiers directly.
any_htn_med2 argument of derive_hypertension(), derive_hypertension_adj(), derive_hypertension_control(), and derive_hypertension_control_adj() has been renamed to any_htn_med. The diab_med2 argument of derive_diabetes_status() has been renamed to diab_med. Calls that pass the old argument by name will fail with an "unused argument" error.any_htn_med2 and diab_med2 entries in inst/extdata/variable-details.csv have been removed. New any_htn_med and diab_med entries cover all six cycles via three database tiers: DerivedVar rows for cycles 1-2 (atc_*/mhr_* columns), DerivedVar rows for cycles 3-6 (meucatc, npi_25b), and copy passthrough rows for the main cycle databases once medications have been merged in.