Replication materials for "Spatial autocorrelation of suicide mortality worldwide at three time points spanning the COVID-19 pandemic period: 2019, 2021, and 2023"

Abstract

Replication materials (data and code) for the manuscript "Spatial autocorrelation of suicide mortality worldwide at three time points spanning the COVID-19 pandemic period: 2019, 2021, and 2023", submitted to PLOS Global Public Health (PGPH-D-26-02463). The package contains: (1) the analytical dataset for 203 countries with age-standardized suicide mortality rates (Global Burden of Disease 2023, self-harm cause, both sexes) for 2019, 2021 and 2023, death-registration completeness (International Completeness of Death Registration database, 2015–2019), LISA classifications, local pseudo p-values, Benjamini-Hochberg q-values and descriptive trajectory labels; (2) the GBD 2023 data extracts; (3) a single canonical R script that reproduces every value reported in the manuscript, including global Moran's I and LISA with five-nearest-neighbour weights and 9999 permutations (seed 42), Benjamini-Hochberg false-discovery-rate correction applied separately within each year, the Friedman test, Wilcoxon signed-rank comparisons with Holm correction, Kendall's W, the exact McNemar test, the complete-registration sensitivity analysis and the robustness analyses; (4) the full log of the run from which all reported values are taken; and (5) the Python script used to draw the figures. Software environment: R 4.5.1 with sf 1.0.23, spdep 1.4.1 and rnaturalearth 1.2.0. Country boundaries: Natural Earth 1:50 m (public domain). See README.md for the codebook and reproduction instructions. Version 3: the canonical analysis script now produces every value reported in the manuscript, including the Benjamini-Hochberg FDR correction applied within each year and the exact McNemar test. The full run log and the country-level local pseudo p-values and q-values are included. All files are in English.