Skip to contents

Methodology & conventions

Source: vignettes/methodology.Rmd
methodology.Rmd

This article is the reference layer: it explains how indicators are specified, where to look up variable mappings, the function naming scheme, and the two kinds of dictionaries the package provides.

Function naming conventions

Pattern Purpose Example
calc_*_dhs() Survey-weighted DHS estimates (long format) calc_itn_dhs()
calc_*_mbg() Run MBG model for an indicator family calc_itn_mbg()
prep_*_mbg() Prepare cluster-level data for MBG prep_itn_mbg()
fit_mbg_indicator() All-in-one MBG fit + admin aggregation fit_mbg_indicator()
run_mbg_pipeline() Full MBG pipeline (all indicators) run_mbg_pipeline()
dhs_dictionary() Unified dictionary across all DHS domains dhs_dictionary()
calc_incidence() Routine-data incidence (N0-N5 cascade) calc_incidence()
calc_tpr() Test positivity rate with fallbacks calc_tpr()

Two kinds of dictionaries

Indicator dictionaries describe the outputs - each indicator’s numerator, denominator, codes, and metadata. Use dhs_dictionary() for the unified view, or per-domain helpers (itn_dictionary(), act_dictionary(), pfpr_dictionary(), …):

Raw variable dictionaries describe the inputs - every variable present in a specific DHS recode, with its label. Build these with make_dhs_raw_dictionary() (or spot-check with list_dhs_var_labels()) before computing indicators, to confirm a survey actually carries the variables an indicator needs:

kr <- dhs_read(path = path_dhs_parquet, file_type = "KR",
               country_code = "TG", survey_year = 2017)
make_dhs_raw_dictionary(kr)     # full variable list for this recode

See DHS survey analysis → inspect the variables for the recommended “Step 0” workflow.

Methodology specifications (inst/methods/)

Detailed methodology for each indicator lives in machine-readable YAML at inst/methods/. Each file documents DHS variable mappings, inclusion criteria, calculation logic, and references to WHO / World Malaria Report standards.

File Domain
pfpr_dhs.yml Parasite prevalence
itn_dhs.yml ITN ownership/access/use
irs_dhs.yml Indoor residual spraying
fever_dhs.yml Fever prevalence
csb_dhs.yml Care-seeking behaviour
malaria_dx_dhs.yml Malaria diagnostic testing
antimalarial_dhs.yml Antimalarial treatment
act_dhs.yml ACT treatment
anc_dhs.yml Antenatal care
iptp_dhs.yml IPTp dosing
epi_dhs.yml EPI vaccination
u5mr_dhs.yml Under-5 mortality
anemia_dhs.yml Anemia prevalence
smc_dhs.yml SMC coverage
wealth_dhs.yml Wealth index
incidence.yml Incidence cascade (N0-N5)
tpr.yml Test positivity rate
reporting_rate.yml Reporting-rate calculations
outlier_detection.yml Outlier detection methods
active_status.yml Facility activity classification

Methodological notes

  • Direct survey estimates stay at adm0/adm1. DHS/MIS weights are representative at region level; sub-region estimates come from the MBG pipeline, not direct aggregation.
  • ACT is a drug class across multiple variables - only artemisinin-based combination therapies count; monotherapies are excluded.
  • U5MR is reported per 1,000 live births; intervention-coverage indicators are percentages (0-100).
  • ITN use_if_access (use among those with access) is computed by default with standard age groups (u5, 5_14, ov15).

Country configuration

Country-specific survey settings (variable overrides, eligibility notes) live in inst/countries/.

See also