Citation

Clark CJ, Bergenfeld I, Cheong YF, Najera H, Sardinha LM, Garcia-Moreno C, Heise L. SSM Popul. Health 2023; 24: e101500.

Copyright

(Copyright © 2023, Elsevier Publishing)

DOI

10.1016/j.ssmph.2023.101500

PMID

unavailable

Abstract

Introduction
Under Sustainable Development Goal 5, prevalence of intimate partner violence (IPV) is a globally reportable indicator. There is a lack of consensus on how to measure and report psychological IPV, affecting prevalence estimates and cross-country comparability. We examine similarities and differences in the patterning of women's experiences of psychological abuse in low- and middle-income countries (LMICs) to inform common cut points.
Methods
Data include 13,452 ever-partnered women from six LMICs participating in the WHO multi-country study on women's health and domestic violence against women and 306,101 from 47 LMICs participating in the Demographic and Health Surveys. A confirmatory latent class analysis (LCA) approach was applied to identify the optimal class structure using the 3 DHS and 4 WHO psychological IPV items, assessed the impact of physical and sexual IPV on class structure, and tested class generalizability across countries. We validated the three-class solution by regressing the classes on physical IPV, sexual IPV, controlling behaviors, and injury due to domestic violence. We used item response theory (IRT) methods to assess item-level characteristics of the items.
Results
Analysis confirmed the three-class structure in most countries. Addition of physical and sexual IPV did not change overall class structure or improve discrimination or homogeneity of the items. The three-class structure was invariant within most WHO-classified regions. Operationalized classes informed by the LCA resulted in prevalences of roughly 90% low-to-no class, 7% moderate-intensity class, and 3% high-intensity class. Classes showed convergent validity with all outcomes tested. IRT analysis revealed good discriminations but substantial information overlaps over a narrow range of the latent psychological violence construct.
Conclusions
This study confirms the three-class pattern but suggests some differences across countries. and regions. We suggest cut points distinguishing violent from non-violent acts and demarcating levels of severity for future study.

FINDINGS offer evidence-based guidance to rectify challenges.

Keywords

Intimate partner violence; Latent class analysis; Latent patterns; Low- and middle-income countries; Profile similarity; Psychological violence