Skip to contents

Epsilon-aminocaproic acid (Stricker 2013)

Source: vignettes/articles/Stricker_2013_aminocaproicAcid.Rmd
Stricker_2013_aminocaproicAcid.Rmd

Model and source

  • Citation: Stricker PA, Zuppa AF, Fiadjoe JE, Maxwell LG, Sussman EM, Pruitt EY, Goebel TK, Gastonguay MR, Taylor JA, Bartlett SP, Schreiner MS. Population pharmacokinetics of epsilon-aminocaproic acid in infants undergoing craniofacial reconstruction surgery. Br J Anaesth 2013;112(5):788-795. doi:10.1093/bja/aes507.
  • Description: Two-compartment IV population PK model for epsilon-aminocaproic acid (EACA) in infants (2-24 months) undergoing craniofacial reconstruction surgery. Allometric scaling on body weight (reference 8.82 kg; fixed exponents 0.75 on CL and Q, 1.0 on V1 and V2), an asymptotically increasing post-natal age maturation effect on clearance (age50 = 7.36 weeks), and binary intra-operative-period multipliers on CL (0.89) and V1 (0.80) capturing the composite effect of anaesthesia, blood loss, and surgical fluid management. Parameter values from Stricker 2013 Table 4.
  • Article: https://doi.org/10.1093/bja/aes507

Population

Eighteen healthy infants aged 2-24 months undergoing elective craniofacial reconstruction surgery at the Children’s Hospital of Philadelphia were enrolled in three sequential dose-escalation cohorts of six subjects each. Median (range) postnatal age was 39 (27-107) weeks and median (range) body weight 8.8 (6.7-11.8) kg. The cohort was 10 female and 8 male, with diagnoses dominated by metopic, sagittal, unicoronal, and lambdoid synostoses (Pfeiffer and Saethre-Chotzen syndromes also represented), and procedures consisted of fronto-orbital advancement or posterior cranial vault reconstruction. Subjects with abnormal renal function, coagulation derangements, or screening haematologic abnormalities were ineligible. See Stricker 2013 Table 1 (baseline demographics) and Table 2 (intra-operative blood loss and fluid balance).

The same information is available programmatically via the model’s population metadata (readModelDb("Stricker_2013_aminocaproicAcid")()$population).

Source trace

The per-parameter origin is recorded as an in-file comment next to each ini() entry in inst/modeldb/specificDrugs/Stricker_2013_aminocaproicAcid.R. The table below collects them in one place for review.

Equation / parameter Value Source location
lcl (postoperative CL) 37.6 mL/min (= 0.0376 L/min) Table 4 row “Postoperative CL (ml min-1)”
lvc (postoperative V1) 1.27 L Table 4 row “V1 (litre)”
lq (Q) 42.3 mL/min (= 0.0423 L/min) Table 4 row “Q (ml min-1)”
lvp (V2) 2.53 L Table 4 row “V2 (litre)”
e_wt_cl_q (allometric exponent CL, Q) 0.75 (fixed) Table 4 footnote / Methods “Base model”: “fixed at a value of 0.75 for clearances”
e_wt_vc_vp (allometric exponent V1, V2) 1.0 (fixed) Table 4 footnote / Methods “Base model”: “a value of 1 for volumes”
lage50_pna (age50 for CL maturation) log(7.36 / 4.34524) = log(1.6938 months) Table 4 row “Age of 50% CL (weeks) = 7.36”; reparameterised in months via the canonical PNA in inst/references/covariate-columns.md and the Zhao 2018 precedent
lr_intraop_cl (intra-op CL multiplier) log(0.89) Table 4 row “Ratio of intra-operative CL to postoperative CL = 0.89”
lr_intraop_vc (intra-op V1 multiplier) log(0.80) Table 4 row “Ratio of intra-operative V1 to postoperative V1 = 0.8”
omega^2_CL (16.79 / 100)^2 = 0.0281904 Table 4 row “v^2 CL = 16.79” with caption “Between-subject variability = (square-root of variance) * 100”
omega^2_V1 (47.01 / 100)^2 = 0.2209940 Table 4 row “v^2 V1 = 47.01” with same caption
cov(eta_CL, eta_V1) 0.03 Table 4 caption “Covariance between CL and V1 random effects was 0.03 (102% SE)”
propSd (proportional residual SD) sqrt(0.03) = 0.1732 Table 4 row “sigma^2 proportional = 0.03” (reported as variance)
addSd (additive residual SD) sqrt(0.6) = 0.7746 mg/L Table 4 row “sigma^2 additive = 0.6” (reported as variance)
Postoperative CL equation CL = theta_CL * (WT/8.82)^0.75 * AGE/(7.36 + AGE) (AGE in weeks) Table 4 footnote, reparameterised here in months: CL = theta_CL * (WT/8.82)^0.75 * PNA/(exp(lage50_pna) + PNA)
Intra-operative CL equation CL_intra = CL_post * 0.89 Table 4 footnote
d/dt(central) and d/dt(peripheral1) ADVAN3 TRANS4 two-compartment IV Methods “Pharmacostatistical analysis - Base model” (NONMEM VI Level 2.0; FOCE-I)
Residual error structure C_obs = C_pred * (1 + eps_P) + eps_A Methods “Base model” (combined additive + proportional)

Virtual cohort

Original individual patient data are not publicly available. The cohort below reproduces the per-subject body weight and postnatal age from Stricker 2013 Table 1 (n = 18), preserving the per-cohort dosing assignment (Cohort 1: 25 mg/kg + 10 mg/kg/h; Cohort 2: 50 mg/kg + 20 mg/kg/h; Cohort 3: 100 mg/kg + 40 mg/kg/h).

# Per-subject covariates from Stricker 2013 Table 1.
cohort <- tibble::tibble(
  id            = 1:18,
  cohort_label  = c(rep("Cohort 1", 6), rep("Cohort 2", 6), rep("Cohort 3", 6)),
  WT            = c(7.7,  9.6,  7.9, 11.4,  8.3,  7.8,
                   10.8,  6.7,  8.9,  6.8,  9.9,  7.4,
                    8.7, 11.8,  9.1, 10.9,  7.1, 10.2),
  age_weeks     = c(27.4, 38.9, 31.6, 85.9, 38.6, 34.6,
                   67.1, 69.4, 99.0, 33.0, 34.9, 30.6,
                   35.0,106.9, 42.1, 48.7, 36.4, 43.7),
  loading_mg_kg = c(rep(25, 6), rep(50, 6), rep(100, 6)),
  civi_mg_kg_h  = c(rep(10, 6), rep(20, 6), rep(40, 6))
) |>
  mutate(
    PNA       = age_weeks / 4.34524,                # canonical PNA in months
    loading_mg     = loading_mg_kg * WT,
    civi_rate_mg_min = civi_mg_kg_h * WT / 60       # mg / min
  )

knitr::kable(
  cohort |> select(id, cohort_label, WT, age_weeks, PNA,
                   loading_mg, civi_rate_mg_min) |>
    mutate(across(where(is.numeric), ~ signif(.x, 4))),
  caption = "Reconstructed virtual cohort from Stricker 2013 Table 1."
)
Reconstructed virtual cohort from Stricker 2013 Table 1.
id cohort_label WT age_weeks PNA loading_mg civi_rate_mg_min
1 Cohort 1 7.7 27.4 6.306 192.5 1.283
2 Cohort 1 9.6 38.9 8.952 240.0 1.600
3 Cohort 1 7.9 31.6 7.272 197.5 1.317
4 Cohort 1 11.4 85.9 19.770 285.0 1.900
5 Cohort 1 8.3 38.6 8.883 207.5 1.383
6 Cohort 1 7.8 34.6 7.963 195.0 1.300
7 Cohort 2 10.8 67.1 15.440 540.0 3.600
8 Cohort 2 6.7 69.4 15.970 335.0 2.233
9 Cohort 2 8.9 99.0 22.780 445.0 2.967
10 Cohort 2 6.8 33.0 7.595 340.0 2.267
11 Cohort 2 9.9 34.9 8.032 495.0 3.300
12 Cohort 2 7.4 30.6 7.042 370.0 2.467
13 Cohort 3 8.7 35.0 8.055 870.0 5.800
14 Cohort 3 11.8 106.9 24.600 1180.0 7.867
15 Cohort 3 9.1 42.1 9.689 910.0 6.067
16 Cohort 3 10.9 48.7 11.210 1090.0 7.267
17 Cohort 3 7.1 36.4 8.377 710.0 4.733
18 Cohort 3 10.2 43.7 10.060 1020.0 6.800

Simulation

The infusion duration for each subject was set to the per-cohort median CIVI duration reported in the Results (230 min for Cohort 1, 227 min for Cohort 2, 254 min for Cohort 3) so the simulated profiles span the same intra-operative window as the published data.

infusion_minutes <- function(cohort_label) {
  dplyr::case_when(
    cohort_label == "Cohort 1" ~ 230,
    cohort_label == "Cohort 2" ~ 227,
    cohort_label == "Cohort 3" ~ 254,
    TRUE                       ~ NA_real_
  )
}

make_subject_events <- function(row) {
  load_end <- 10                       # 10-min loading-dose infusion
  civi_end <- load_end + infusion_minutes(row$cohort_label)
  obs_grid <- seq(0, civi_end + 750, by = 2)

  # Dosing rows: loading-dose infusion 0 - 10 min, then CIVI 10 - civi_end.
  # `rate` is in mg/min; rxode2 treats negative rate-column entries as
  # duration-modelled infusions, but here we model them as standard rate-amt
  # pairs by computing `dur` explicitly.
  doses <- tibble::tibble(
    id   = row$id,
    time = c(0,             load_end),
    amt  = c(row$loading_mg, row$civi_rate_mg_min * infusion_minutes(row$cohort_label)),
    dur  = c(load_end,      infusion_minutes(row$cohort_label)),
    evid = 1L,
    cmt  = "central"
  )

  # Observations.
  obs <- tibble::tibble(
    id   = row$id,
    time = obs_grid,
    amt  = NA_real_,
    dur  = NA_real_,
    evid = 0L,
    cmt  = "central"
  )

  dplyr::bind_rows(doses, obs) |>
    dplyr::arrange(time, dplyr::desc(evid)) |>
    dplyr::mutate(
      WT      = row$WT,
      PNA     = row$PNA,
      # INTRAOP = 1 from just after the loading-dose sample (load_end) through
      # end of surgery (civi_end), per Stricker 2013 Methods "Full covariate
      # model".
      INTRAOP = as.integer(time >= load_end & time <= civi_end),
      cohort_label = row$cohort_label
    )
}

events <- do.call(
  dplyr::bind_rows,
  lapply(seq_len(nrow(cohort)), function(i) make_subject_events(cohort[i, ]))
)

stopifnot(!anyDuplicated(unique(events[, c("id", "time", "evid")])))
mod <- readModelDb("Stricker_2013_aminocaproicAcid")

# Typical-value simulation (BSV zeroed) to reproduce Figure 1 / Figure 4 median
# trajectories from the paper.
sim_typical <- rxode2::rxSolve(
  rxode2::zeroRe(mod),
  events = events,
  keep   = c("WT", "PNA", "INTRAOP", "cohort_label")
) |>
  as.data.frame()
#> ℹ parameter labels from comments will be replaced by 'label()'
#> ℹ omega/sigma items treated as zero: 'etalcl', 'etalvc'
#> Warning: multi-subject simulation without without 'omega'

# Population simulation (with BSV) for the per-cohort VPC bands.
set.seed(20130301)  # paper acceptance date
sim_pop <- rxode2::rxSolve(
  mod,
  events = events,
  nSub   = 1L,           # 1 simulation per subject; events already span 18 IDs
  nStud  = 50L,          # 50 replicates per subject for the VPC bands
  keep   = c("WT", "PNA", "INTRAOP", "cohort_label")
) |>
  as.data.frame()
#> ℹ parameter labels from comments will be replaced by 'label()'

Replicate published figures

Figure 1 – concentration-time profiles by dose cohort

Figure 1 of Stricker 2013 plots semi-logarithmic observed plasma EACA concentrations for each of the three dose cohorts. The simulation below reproduces the typical-value time course at the per-subject covariates and overlays the per-cohort published target steady-state range (~130 mg/L target, shaded).

sim_typical |>
  dplyr::filter(time <= 1000) |>
  ggplot(aes(time, Cc, group = id)) +
  geom_hline(yintercept = 130, linetype = "dashed", colour = "grey40") +
  geom_line(alpha = 0.7) +
  facet_wrap(~cohort_label) +
  scale_y_log10(limits = c(1, 1000)) +
  labs(
    x = "Time (min)",
    y = "Plasma EACA concentration (mg/L)",
    title = "Stricker 2013 Figure 1 - typical-value simulation by cohort"
  )
#> Warning in scale_y_log10(limits = c(1, 1000)): log-10 transformation introduced
#> infinite values.
#> Warning: Removed 760 rows containing missing values or values outside the scale range
#> (`geom_line()`).
Replicates Figure 1 of Stricker 2013: simulated plasma EACA concentration vs. time, faceted by dose cohort. The dashed horizontal line marks the published therapeutic target of 130 mg/L.

Replicates Figure 1 of Stricker 2013: simulated plasma EACA concentration vs. time, faceted by dose cohort. The dashed horizontal line marks the published therapeutic target of 130 mg/L.

Figure 4 – Monte Carlo simulation for the typical Cohort 3 subject

Figure 4A of Stricker 2013 simulates the typical 8.82 kg / 38-week subject under the recommended Cohort 3 regimen (100 mg/kg loading + 40 mg/kg/h CIVI for 240 min) and overlays the 2.5th / 50th / 97.5th simulated percentiles. The plot below reproduces that simulation.

typical_subject <- tibble::tibble(
  id            = 1L,
  cohort_label  = "Cohort 3",
  WT            = 8.82,
  age_weeks     = 38,
  loading_mg_kg = 100,
  civi_mg_kg_h  = 40,
  PNA           = 38 / 4.34524,
  loading_mg    = 100 * 8.82,
  civi_rate_mg_min = 40 * 8.82 / 60
)

# Override infusion duration to exactly 240 min as in Figure 4.
load_end <- 10
civi_end <- load_end + 240
obs_grid <- seq(0, civi_end + 360, by = 2)

doses_fig4 <- tibble::tibble(
  id   = 1L,
  time = c(0, load_end),
  amt  = c(typical_subject$loading_mg,
           typical_subject$civi_rate_mg_min * 240),
  dur  = c(load_end, 240),
  evid = 1L,
  cmt  = "central"
)
obs_fig4 <- tibble::tibble(
  id   = 1L,
  time = obs_grid,
  amt  = NA_real_,
  dur  = NA_real_,
  evid = 0L,
  cmt  = "central"
)
events_fig4 <- dplyr::bind_rows(doses_fig4, obs_fig4) |>
  dplyr::arrange(time, dplyr::desc(evid)) |>
  dplyr::mutate(
    WT      = typical_subject$WT,
    PNA     = typical_subject$PNA,
    INTRAOP = as.integer(time >= load_end & time <= civi_end)
  )

set.seed(20130301)
sim_fig4 <- rxode2::rxSolve(
  mod,
  events = events_fig4,
  nSub   = 500L
) |>
  as.data.frame()
#> ℹ parameter labels from comments will be replaced by 'label()'

fig4_bands <- sim_fig4 |>
  dplyr::group_by(time) |>
  dplyr::summarise(
    Q025 = stats::quantile(Cc, 0.025, na.rm = TRUE),
    Q50  = stats::quantile(Cc, 0.50,  na.rm = TRUE),
    Q975 = stats::quantile(Cc, 0.975, na.rm = TRUE),
    .groups = "drop"
  )

ggplot(fig4_bands, aes(time, Q50)) +
  geom_ribbon(aes(ymin = Q025, ymax = Q975), alpha = 0.2) +
  geom_line() +
  geom_hline(yintercept = 130, linetype = "dashed", colour = "grey40") +
  labs(
    x = "Time (min)",
    y = "Plasma EACA concentration (mg/L)",
    title = "Stricker 2013 Figure 4A - 500-replicate simulation, typical 8.82 kg 38-wk subject",
    subtitle = "100 mg/kg loading over 10 min, then 40 mg/kg/h CIVI for 240 min"
  )

PKNCA validation

Per-cohort steady-state NCA over the median intra-operative infusion duration of each cohort, using the typical-value simulation so the NCA targets the model’s central tendency.

# Concentrations -- one row per id x time during the infusion + post-CIVI tail.
sim_nca <- sim_typical |>
  dplyr::filter(!is.na(Cc), Cc > 0) |>
  dplyr::select(id, time, Cc, cohort_label) |>
  as.data.frame()

# Doses -- collapse the loading dose + CIVI total into a single combined-amount
# row per subject so PKNCA computes AUC-relevant clearance against the total
# administered.
dose_df <- cohort |>
  dplyr::mutate(
    civi_total_mg = civi_rate_mg_min *
      infusion_minutes(cohort_label),
    total_amt     = loading_mg + civi_total_mg,
    time          = 0
  ) |>
  dplyr::select(id, time, amt = total_amt, cohort_label) |>
  as.data.frame()

conc_obj <- PKNCA::PKNCAconc(
  sim_nca,
  formula = Cc ~ time | cohort_label + id,
  concu   = "mg/L",
  timeu   = "min"
)

dose_obj <- PKNCA::PKNCAdose(
  dose_df,
  formula = amt ~ time | cohort_label + id,
  doseu   = "mg"
)

intervals <- data.frame(
  start      = 0,
  end        = Inf,
  cmax       = TRUE,
  tmax       = TRUE,
  aucinf.obs = TRUE,
  auclast    = TRUE,
  half.life  = TRUE
)

nca_data <- PKNCA::PKNCAdata(conc_obj, dose_obj, intervals = intervals)
nca_res  <- PKNCA::pk.nca(nca_data)
#> Warning: Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed
#> Requesting an AUC range starting (0) before the first measurement (2) is not allowed

knitr::kable(
  summary(nca_res),
  caption = "PKNCA summary of the typical-value simulation by dose cohort."
)
PKNCA summary of the typical-value simulation by dose cohort.
Interval Start Interval End cohort_label N AUClast (min*mg/L) Cmax (mg/L) Tmax (min) Half-life (min) AUCinf,obs (min*mg/L)
0 Inf Cohort 1 6 NC 166 [0.457] 10.0 [10.0, 10.0] 114 [1.69] NC
0 Inf Cohort 2 6 NC 330 [1.28] 10.0 [10.0, 10.0] 111 [6.21] NC
0 Inf Cohort 3 6 NC 665 [0.770] 10.0 [10.0, 10.0] 115 [3.07] NC

Comparison against published clearance summaries

Stricker 2013 Discussion states the typical 8.82 kg, 38-week subject has a pre-/postoperative clearance of 32 mL/min. Table 5 lists the model-implied typical pre-/postoperative clearance at a range of body weights (without the age adjustment). The check below recomputes those typical values from the packaged model and compares them to Table 5.

table5 <- tibble::tibble(
  WT        = c(1, 5, 10, 15, 25, 50, 70),
  CL_paper  = c(7.35, 24.56, 41.31, 56.00, 82.14, 138.14, 177.79)
)

# Typical-value postoperative CL with PNA -> infinity (Mat_CL = 1).
# Reference: CL = 37.6 * (WT/8.82)^0.75.
table5 <- table5 |>
  dplyr::mutate(CL_model = 37.6 * (WT / 8.82)^0.75) |>
  dplyr::mutate(rel_diff_pct = 100 * (CL_model - CL_paper) / CL_paper)

knitr::kable(
  table5 |> mutate(across(where(is.numeric), ~ signif(.x, 4))),
  caption = "Pre-/postoperative typical CL (mL/min) from Stricker 2013 Table 5 vs. the packaged model (no age adjustment)."
)
Pre-/postoperative typical CL (mL/min) from Stricker 2013 Table 5 vs. the packaged model (no age adjustment).
WT CL_paper CL_model rel_diff_pct
1 7.35 7.347 -0.0462600
5 24.56 24.560 0.0197000
10 41.31 41.310 0.0071870
15 56.00 56.000 -0.0076390
25 82.14 82.140 -0.0030590
50 138.10 138.100 -0.0012760
70 177.80 177.800 0.0006001 

stopifnot(max(abs(table5$rel_diff_pct)) < 1.0)

The maximum relative difference is well below 1%, confirming the packaged allometric scaling reproduces Stricker 2013 Table 5 verbatim.

Assumptions and deviations

  • Postnatal age unit reparameterisation. Stricker 2013 reports postnatal age in weeks and parameterises the maturation AGE / (7.36 + AGE) in weeks. The canonical PNA covariate carries months. The packaged model stores age50 on the log scale in months (log(7.36 / 4.34524) = log(1.6938)) and applies the maturation formula in months (Mat_CL = PNA / (exp(lage50_pna) + PNA)); because the Emax-style ratio is dimensionless, the numerical maturation factor is identical to the paper’s AGE-in-weeks form when both inputs are converted using 1 month = 4.34524 weeks. The Zhao 2018 omeprazole vignette established this PNA-unit-reparameterisation precedent.
  • Intra-operative time window definition. Stricker 2013 Methods “Full covariate model” defines the intra-operative period as “the time immediately after the post-loading dose PK sample through the end of the surgery”. In the virtual cohort here, INTRAOP = 1 runs from the end of the 10-min loading infusion (load_end) through the per-subject end of CIVI (load_end + infusion_minutes(cohort_label)). The paper’s exact window endpoints for each subject are not tabulated, so the simulation uses each cohort’s median CIVI duration (230 / 227 / 254 min for Cohorts 1, 2, 3).
  • Race / ethnicity distribution not used. The source publication does not tabulate the race / ethnicity of the 18 enrolled infants and the final model carries no race effect; the virtual cohort therefore omits race entirely.
  • Sex distribution not used. The 10F / 8M split (Stricker 2013 Table 1) is recorded in population$sex_female_pct but does not enter the final model, so it is not carried through the simulation either.
  • Covariance between eta_CL and eta_V1. The published covariance is 0.03 (Table 4 caption, 102% SE). The high relative standard error indicates the estimate is poorly identified; the packaged model retains it for fidelity to the published parameter vector even though a user fitting fresh data may prefer to drop or refit it. The implied correlation (0.382) is plausible given the known coupling of clearance and central volume in this short infusion window.
  • Intra-operative blood loss not extracted. Stricker 2013 Discussion notes that the intra-operative period effect “may be simply reflecting the net effect of a possible increased CL due to blood loss and decreased CL due to other confounding intra-operative factors”. The binary INTRAOP covariate is the only published mechanism for that composite effect; the continuous per-subject blood-loss column in Table 2 is not used as a covariate in the final model and is not added to the virtual cohort here.
  • Dose-record encoding. rxode2 dosing is encoded with amt (total mg) and dur (infusion duration in minutes) for both the 10-min loading bolus and the CIVI; this is equivalent to the NONMEM RATE = amt / dur parameterisation in the paper’s NMTRAN dataset.