survey_sim(): manually specifying new survey designs results in errors

[luccoffeng]

If a user-defined survey design is created, survey_sim() will complain that it does not know the name (e.g., "NS_51") and terminate. Likewise, simulating an arbitrary number of pre-treatment days of sampling will result in an error. I'm not sure whether or why this level of error control is necessary.

Minimal working example:

# Prep session ----
rm(list = ls())

library(eggSim)
library(data.table)


# Set basic simulation parameters ----
species <- "hookworm"  # just a placeholder
T_E <- 0.98   # expected treatment efficacy
delta <- .05  # non-inferiority margin

scen <- as.data.table(survey_scenario(parasite = species)[1, ])
scen[, true_efficacy := T_E]

pars <- as.data.table(survey_parameters(design = c("NS_11"),
                                        parasite = species,
                                        method = "kk"))
pars <- pars[1]
pars[, min_positive_pre := 0]
pars[, min_positive_screen := 0]
pars[, efficacy_expected := T_E]
pars[, efficacy_lower_target := delta]
pars[, n_day_pre := 5]  # will trigger error about Invalid N_day_pre (does not match template)
# pars[, design := "NS_51"]  # uncomment to trigger error about unrecognised design


# Run simulations for non-inferiority trial----
test <- survey_sim(
  scenario = scen,
  parameters = pars,
  n_individ = 1e3,
  iterations = 100,
  cl = 10,
  output = "full",
  analysis = "delta",
  check_memory = TRUE
)