Derived Variables

Author

Tianqi Ouyang

Published

April 14, 2026

All variables computed in Tables.Rmd (and replicated in revision 0311.Rmd) from the REDCap source data.

Identifiers & Filtering

Variable	Description	Logic
`id`	Internal row ID per center	`paste0("{center}_", row_number())`
`Discont_Day`	Day terlipressin discontinued (0–13)	First day where `discont_terli_dayN == 1` via `qc_check()`
`center_id`	Center abbreviation	`str_extract(subjectid, "^[^-]+")`
`center`	Numeric center grouping (factor, 1–10)	BID/MGH/Yale=1, PEN=2, OCH=3, BSW=4, CPM/CSF/UNM=5, Mayo/MCM/NWM=7, MMC=9, CCF/INU=10

Baseline Creatinine & eGFR

Variable	Description	Logic
`baseline_scr_calculated`	Estimated SCr from eGFR=75	Inverse CKD-EPI 2021 via numerical optimization (`optimize()`)
`baseline_scr_new`	Best available baseline SCr	(1) `baseline_scr` if `days_to_admit != 0`, else (2) `baseline_scr_120to365_value`, else (3) `min(calculated, scr_admit)`
`baseline_scr_new_2`	Strict baseline (no calculated fallback)	Only options (1) and (2) above
`eGFR`	Estimated GFR	CKD-EPI 2021 from `baseline_scr_new_2`
`ckd_baseline`	Baseline CKD	`eGFR < 60` → 1

AKI Staging

Variable	Description	Logic
`ratio`	SCr ratio at terli initiation	`scr_terli_day0 / baseline_scr_new`
`delta`	SCr delta at terli initiation	`scr_terli_day0 - baseline_scr_new`
`akin`	AKIN stage at initiation	1: ratio 1.5–2 OR delta > 0.3; 2: ratio 2–3; 3: ratio ≥ 3 OR SCr ≥ 4
`discont_ratio`	SCr ratio at discontinuation	`lowest_creatinine / baseline_scr_new`
`discont_delta`	SCr delta at discontinuation	`lowest_creatinine - baseline_scr_new`
`discont_aki_stage`	AKIN stage at discontinuation	Same staging as `akin`, using `lowest_creatinine`
`difference`	AKI stage change	`akin - discont_aki_stage`

Clinical Scores

Child-Pugh Score

Variable	Description
`child_pugh_score`	Sum of 5 components (5–15)

Component scoring

Component	1 point	2 points	3 points
Bilirubin (mg/dL)	< 2	2–3	> 3
Albumin (g/dL)	> 3.5	2.8–3.5	< 2.8
INR	< 1.7	1.7–2.3	> 2.3
Ascites	Absent (1)	Slight (2)	Moderate (3)
HE grade	None (0)	Grade 1–2	Grade 3–4

CLIF-C Score

Variable	Description
`clif_c_score`	`10 * (0.33 * organ_score + 0.04 * age + 0.63 * log(WBC) - 2)`
`clif_c_score_new`	Median-imputed (missing → population median)

Organ sub-scores

Organ	1 point	2 points	3 points
Liver (T.bili mg/dL)	< 6	6–12	≥ 12
Kidney (SCr mg/dL)	< 2	2–3.5	≥ 3.5
Brain (HE grade)	1	2–3	4–5
Coagulation (INR)	< 2	2–2.5	≥ 2.5
Circulation	MAP ≥ 70, no pressors	MAP < 70, no pressors	Any pressors
Respiratory (SpO2/FiO2)	> 357	214–357	≤ 214

ACLF Grade & Organ Failures

Variable	Description	Logic
`aclf_grade`	Organ failure count (capped at 3)	Count of organ failures below
`organ_failure_sum`	Total organ failure count (uncapped)	Sum of 6 binary flags
`liver_f`	Liver failure	`bili_terli_day0 >= 6`
`brain_f`	Brain failure	`he_terli_day0 > 1`
`coagulation_f`	Coagulation failure	`inr_terli_day0 >= 2`
`circulatory_f`	Circulatory failure	`map_terli_day0_time0 < 70`
`respiratory_f`	Respiratory failure	`spo2_terli_day0 / fio2_terli_day0 <= 357`
`kidney_f`	Kidney failure	`scr_terli_day0 >= 1.5`

MELD Scores

Variable	Description	Formula
`day0_meld`	MELD at Day 0	`9.57log(Cr) + 3.78log(bili) + 11.20*log(INR) + 6.43`; Cr capped 1–4
`day0_meld_na`	MELD-Na at Day 0	MELD + sodium correction (Na capped 125–137); applied only if MELD > 11
`day0_meld_3`	MELD 3.0 at Day 0	Includes sex factor (+1.33 if female), albumin, sodium, and bilirubin-sodium / albumin-Cr interactions
`admit_meld`	MELD at admission	Same as day0_meld using admission labs
`admit_meld_na`	MELD-Na at admission
`admit_meld_3`	MELD 3.0 at admission	Median-imputed if missing
`admit_meld_3_raw`	MELD 3.0 raw (pre-imputation)

Outcome Variables

HRS Response

Variable	Description	Logic
`lowest_creatinine`	Lowest SCr around discontinuation	`min(last_scr_2, last_scr, scr_day1_post_terli)`
`last_scr`	SCr on discontinuation day	`scr_terli_day{Discont_Day}`
`last_scr_2`	SCr day before discontinuation	`scr_terli_day{Discont_Day - 1}`
`HRS_reversal_new`	Primary outcome	`lowest_creatinine <= 1.5 AND (scr_day0 - lowest_creatinine) >= 0.3`
`hrs_responders`	Response category	0 = No stage improvement, 1 = Partial (improved but SCr > 0.3 above baseline), 2 = Complete (≤ 0.3 above baseline)
`hrs_responders_cat_2`	Binary responder	0 = Non-responder, 1 = Any response
`hrs_rebound`	SCr rebound post-terli	`scr_day1_post_terli - last_scr >= 0.3`

Survival / Competing Risk

Intermediate Variables

Variable	Description	Logic
`dc_death_status`	In-hospital death	1 if `disposition == 5`, else 0
`after_discharge_90days_death_status`	Died within 90 days post-discharge	1 if `dc_death_90days == 1`, else 0
`time_to_death`	Days from terlipressin initiation to death or censoring	In-hospital death: `days_to_discharge`; post-discharge death: `days_to_discharge + dc_days_to_death`; lost to follow-up: `days_to_discharge + days_to_lost_to_followup`; alive at 90 days: `days_to_discharge + 90` (intentional censoring). All minus `days_to_terli_initiation_day0`
`time_to_death_status`	Overall death indicator	1 = died (in-hospital or post-discharge within 90 days), 0 = alive or lost to follow-up

90-Day Death Outcomes

Variable	Description	Values	Logic
`time_to_death_90days`	Follow-up time for 90-day death analysis (days)	0–90	`min(time_to_death, 90)` – all follow-up capped at 90 days
`time_to_death_status_90days`	90-day death event indicator	1 = died within 90 days; 0 = alive at 90 days, OR died after 90 days (censored at 90), OR lost to follow-up	1 only if `time_to_death_status == 1` AND `time_to_death <= 90`; all other patients get 0

90-Day Transplant Outcomes

Variable	Description	Values	Logic
`time_to_transplant_90days`	Follow-up time for 90-day transplant analysis (days)	0–90	`min(time_to_transplant_cmp, 90)`
`time_to_transplant_status_90days`	90-day transplant event indicator	2 = transplanted within 90 days; 0 = not transplanted within 90 days, OR transplanted after 90 days (censored at 90)	2 only if `transplant_lt_status == 2` AND `time_to_transplant_cmp <= 90`; all other patients get 0

90-Day Competing Risk (Death vs. Transplant)

Variable	Description	Values	Logic
`time_to_death_90days_cmp`	Competing risk follow-up time (days)	0–90	`min(time_to_death_90days, time_to_transplant_90days)` – whichever event came first
`time_to_death_status_90days_cmp`	Competing risk event indicator	0 = censored (alive, no transplant within 90 days); 1 = death within 90 days (death came first or tied); 2 = transplant within 90 days (transplant came first)	If death time ≤ transplant time: use death status; if transplant time < death time: use transplant status

How the competing risk status works

Patient dies day 30, no transplant: time = 30, status = 1 (death)
Patient transplanted day 20, alive at 90 days: time = 20, status = 2 (transplant)
Patient alive at 90 days, no transplant: time = 90, status = 0 (censored)
Patient dies day 50, transplanted day 40: time = 40, status = 2 (transplant came first)
Patient died after 90 days: time = 90, status = 0 (censored – death beyond window)
Patient lost to follow-up day 60: time = 60, status = 0 (censored)

Transplant

Variable	Description	Logic
`transplant_lt`	LT timing category	1 = During hospitalization (LT or SLKT), 2 = Within 90 days post-discharge, 3 = After 90 days (within 1 year), 0 = No transplant
`transplant_slkt`	Simultaneous liver-kidney transplant indicator	1 if `slkt_during_hospital == 1` OR `dc_lt_90days == 2` OR `lt_type_1yr == 2`, else 0
`transplant_lt_status`	Transplant status for competing risk	0 = no transplant, 2 = transplanted (coded as 2 for competing risk with death = 1)
`time_to_transplant`	Days from terlipressin initiation to transplant	NA if no transplant
`time_to_transplant_cmp`	Transplant time for competing risk (999 if no transplant)	999 serves as a large censoring time so `pmin` with death time selects death
`transplant_outcome`	Transplanted within 90 days (binary)	1 if `time_to_transplant_status_90days == 2`, else 0
`listed_transplant`	Listed for liver transplant	1 if `lt_eval == 2` OR `lteval_listingstatus == 1`, else 0

Treatment Course Variables

Variable	Description	Logic
`discont_reason_terli`	Reason at discontinuation day	`discont_reason_terli_day{Discont_Day}`
`terli_ae_abdpain`	Abdominal pain AE	From discontinuation day
`terli_ae_nausea`	Nausea AE	From discontinuation day
`terli_ae_diarrhea`	Diarrhea AE	From discontinuation day
`terli_ae_volover`	Volume overload AE	From discontinuation day
`terli_ae_brady`	Bradycardia AE	From discontinuation day
`terli_7`	Respiratory failure reason	`discont_reason_terli == 7`
`terli_8`	Ischemic event reason	`discont_reason_terli == 8`
`terli_ae_respfail`	Respiratory failure composite	`terli_7 == 1 OR terli_ae_volover == 1`
`any_ae`	Any adverse event	Any AE flag == 1
`mean_map_day0`	Mean MAP on Day 0	Average of 4 daily MAP time-point readings
`dosechange_day`	First day of dose adjustment	First day (2–13) with `dosechange_terli_dayN == 1`
`dosechange_status`	Any dose change	Binary: any change days 2–12
`time_to_dosechange`	Day of first dose change
`total_albumin`	Total albumin across treatment	`rowSums(albumintotal_terli_day0:day13)`

Criteria / Exclusion Flags

Variable	Description	Logic
`criteria_1`	Discontinued day 1 with no dose	`total_terli_day1 == 0 AND discont_terli_day1 == 1`
`criteria_2`	ACLF grade ≥ 3	`aclf_grade >= 3`
`criteria_3`	SCr ≥ 5	`scr_terli_day0 >= 5`
`criteria_4`	Respiratory compromise	`SpO2/FiO2 < 357`
`criteria_5`	High MELD 3.0	`admit_meld_3 > 35`
`HRS_Criteria`	Any criteria 1–4 met
`HRS_criteria_3`	Criteria 2, 3, or 4 (without 1)
`HRS_criteria_meld`	Criteria 2–5

Dose-Change Subgroups (revision 0311)

Variable	Description	Logic
`discont_before_day3`	Discontinued before day 3	`Discont_Day < 3`
`pc_change_before_day3`	Dose changed days 1–3, continued > day 3
`pc_nochange_after_day3`	No change days 1–3, continued > day 3
`pc_day3`	Discontinued exactly day 3	`Discont_Day == 3`

Other

Variable	Description	Logic
`fena`	Fractional excretion of sodium	`100 * (scr_admit * ua_na) / (sodium_admit * ua_cr)`
`urine_pc`	Urine protein:creatinine ratio	`uaprotein_hrs / ua_cr_value_hrs`
`race_2`	Binary race	White or missing → 1, else 0
`rrt_2`	RRT binary (need-not-candidate = 0)	`rrt == 2` recoded to 0
`bili_cal`	Bilirubin (coalesced)	`coalesce(bili_terli_day0, tbili_admit, mean)`

Longitudinal Lab Tracking

Lab	Admission	AKI	Day 0–13	Post-Terli	Follow-up
SCr	`scr_admit`	`scr_aki`	`scr_terli_day0`–`13`	`scr_day1`–`14_post_terli`	`dc_scr`, `dc_scr30`, `60`, `90`, `scr_value_1yr`
T.Bili	`tbili_admit`	`tbili_aki`	`bili_terli_day0`–`13`	–	`dc_tbili`
INR	`inr_admit`	`inr_aki`	`inr_terli_day0`–`13`	–	`dc_inr`
Albumin	`albumin_admit`	`albumin_aki`	`alb_terli_day0`–`13`	–	`dc_albumin`
Sodium	`sodium_admit`	`sodium_aki`	`na_terli_day0`–`13`	–	`dc_sna`
Cystatin C	`cystatinc_admit`	`cystatinc_aki`	`cystatinc_terli_day0`–`13`	–	–

Multiple Imputation (revision 0311)

Predictive mean matching (PMM), m=10 imputations, 50 iterations, seed=12345.

Imputed predictors: sex_male, scr_admit, tbili_admit, sodium_admit, inr_admit, albumin_admit, bili_terli_day0, he_terli_day0, inr_terli_day0, map_terli_day0_time0, fio2_terli_day0, spo2_terli_day0, scr_terli_day0, wbc_terli_day0

Post-imputation: Clinical scores (CLIF-C, ACLF grade, Child-Pugh, MELD) are recomputed on each imputed dataset.

Models fit: Fine-Gray competing risk models pooled with Rubin’s rules via run_pooled_fg(), stratified by HRS criteria, transplant listing status, and other subgroups.