Changing contribution of a shared neighbourhood pattern to reported sexual and non-sexual violence against women and girls in Medellín, Colombia, 2018--2022
Abstract
During the COVID-19 restrictions, notifications of violence against women and girls may have changed unevenly across neighbourhoods. We asked whether the contribution of a shared neighbourhood pattern changed over this period, and whether any change was common to sexual and non-sexual violence or specific to one. We analysed 46,390 suspected cases notified to the public-health surveillance system of Medellín, Colombia, across 285 neighbourhoods in 2018-2022. A Bayesian shared-component model with age-standardised expected counts allowed the loading of a common spatial field on each form of violence to differ between the pre-pandemic (2018-2019), restriction (2020-2021) and post-restriction (2022) periods. Across the urban communes, the shared field was generally higher where living conditions were poorer. During the restrictions, its loading increased by 0.26 (95% credible interval 0.16-0.37) for non-sexual violence and by 0.23 (0.09-0.37) for sexual violence, without clear evidence of a change in their ratio. Between neighbourhoods at the 90th and 10th percentiles of the shared field, the shared component's contribution to their risk ratio widened by about 36% for non-sexual and 31% for sexual violence. In 2022, the non-sexual loading remained 0.19 (0.07-0.31) above its pre-pandemic value, whereas the sexual-violence loading was close to its pre-pandemic level. The direction of these changes was retained among notifications from health institutions only and under alternative specifications. Twenty-five neighbourhoods had joint probabilities above 0.95 of both relative risks exceeding one in all five years. The contribution of the shared neighbourhood pattern increased for both outcomes during the restrictions and remained elevated in 2022 for non-sexual violence.
Related
- Changing Geographies of Violence in North and West Africa
- Neighbourhood complexity and identification problems for graphs of bounded treewidth and pathwidth
- A duality principle for the neighbourhood of manifolds and applications to counting rational points near curves
- Four colours are needed for closed-neighbourhood conflict-free colouring of all planar graphs
- Neighbourhood-Based Generalized Dynamic Principal Components for Spatial Functional Data
BibTeX
@misc{rodrigueztaborda2026changing,
title = {{Changing contribution of a shared neighbourhood pattern to reported sexual and non-sexual violence against women and girls in Medellín, Colombia, 2018--2022}},
author = {Tomás Rodríguez-Taborda and Alejandra-Estefanía Patiño-Hoyos and Simón Ruiz-Martínez and Johnatan Cardona-Jiménez},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.31741},
eprint = {2610.31741},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.31741},
}
RIS
TY - UNPB
TI - Changing contribution of a shared neighbourhood pattern to reported sexual and non-sexual violence against women and girls in Medellín, Colombia, 2018--2022
AU - Tomás Rodríguez-Taborda
AU - Alejandra-Estefanía Patiño-Hoyos
AU - Simón Ruiz-Martínez
AU - Johnatan Cardona-Jiménez
PY - 2026
DA - 2026/10/22
PB - arXiv
DO - 10.48550/arxiv.2610.31741
UR - https://arxiv.org/abs/2610.31741
AB - During the COVID-19 restrictions, notifications of violence against women and girls may have changed unevenly across neighbourhoods. We asked whether the contribution of a shared neighbourhood pattern changed over this period, and whether any change was common to sexual and non-sexual violence or specific to one. We analysed 46,390 suspected cases notified to the public-health surveillance system of Medellín, Colombia, across 285 neighbourhoods in 2018-2022. A Bayesian shared-component model with age-standardised expected counts allowed the loading of a common spatial field on each form of violence to differ between the pre-pandemic (2018-2019), restriction (2020-2021) and post-restriction (2022) periods. Across the urban communes, the shared field was generally higher where living conditions were poorer. During the restrictions, its loading increased by 0.26 (95% credible interval 0.16-0.37) for non-sexual violence and by 0.23 (0.09-0.37) for sexual violence, without clear evidence of a change in their ratio. Between neighbourhoods at the 90th and 10th percentiles of the shared field, the shared component's contribution to their risk ratio widened by about 36% for non-sexual and 31% for sexual violence. In 2022, the non-sexual loading remained 0.19 (0.07-0.31) above its pre-pandemic value, whereas the sexual-violence loading was close to its pre-pandemic level. The direction of these changes was retained among notifications from health institutions only and under alternative specifications. Twenty-five neighbourhoods had joint probabilities above 0.95 of both relative risks exceeding one in all five years. The contribution of the shared neighbourhood pattern increased for both outcomes during the restrictions and remained elevated in 2022 for non-sexual violence.
ER -