Skip to contents

Run a conditional poisson model for across many geographic units

condPois_2stage.Rd

MixMeta nope its this: https://www.sciencedirect.com/science/article/pii/S0140673614621140?via%3Dihub avgtmax and range and random effect for county unclear, Gasp does both here https://github.com/gasparrini/RiskExtrapolation/blob/3caa5c46a748d6a789021554f435e53f9960d4c1/24_MetaRegression.R and it goes counrty/city https://github.com/gasparrini/MCC-SO2/blob/1a158b31bf83f3bbca3449b553a6570280306299/Rcode/06.nonlin.R#L47 nlmeta <- mixmeta(nlcoefall, nlvcovall, random=~1|country/city, data=cities, control=list(showiter=T, igls.inititer=10)) basically, did Gasp 2015 (temp_mean, temp_iqr, random effect for county) and included the hierarchicacy structure for ids within countries https://github.com/gasparrini/2015_gasparrini_EHP_Rcodedata https://github.com/gasparrini/2019_sera_StatMed_Rcode/blob/master/03.MultilevelMA.R

Usage

condPois_2stage(
  exposure_matrix,
  outcomes_tbl,
  global_cen = NULL,
  argvar = NULL,
  arglag = NULL,
  maxlag = NULL,
  min_n = NULL,
  rf = NULL,
  strata_min = 0,
  verbose = 0
)

Arguments

exposure_matrix

a matrix of exposures, with columns for lag, usually created by make_exposure_matrix

outcomes_tbl

a data.table of outcomes, created by make_outcome_table

global_cen

a global centering point

argvar

a list containing the argvar components for the crossbasis

arglag

a list containing the arglag components for the crossbasis

maxlag

an integer of the maximum lag

min_n

an integer describing the minimum number of cases for a single region

rf

optional character string to use as the random formula in mixmeta

strata_min

minimum number of cases per strata

verbose

whether to print geo_units as they are run

Details

update as of 2025.12.21, its this https://github.com/gasparrini/Extended2stage/blob/cf82fd2f0d616210e8aa6911d2fb49a732f65795/02.multilevel.R#L68

Examples