Determinants of partner violence in low and middle-income countries : exploring variation in individual and population-level risk

摘要: Background: Lack of understanding factors that contribute to an individual woman's risk partner violence as well collectively account for the distribution violence across settings, continues compromise efforts design effective prevention programs. Likewise, key methodological questions remain unanswered, most notably how best conceptualize, capture, and measure purposes research. This thesis attempts bridge these gaps by analyzing and protective a variety low middle-income with emphasis on Brazil Peru. Methods: The analysis herein draws data from WHO Multi-Country Study Domestic Violence Women's Health, population-based survey interviewed over 24,000 reproductive age women, in 15 sites about their experiences violence. examines patterning Peru explores relative utility using Latent Class analysis (LCA) compared traditional case definitions, identify classify cases partner It then uses generalized estimating equations develop explanatory model the predict experiencing severe violence, identified LCA. Later chapters present two ecological analyses: one identifies cluster- level emerge predictive cluster-level prevalences domestic violence; second full set, 18 Demographic Health Surveys, United Nations independent bases test various theories on macro-level work influence country's overall level Results: LCA categorizes differently than definition, although both tend similar factors. approach, however, seriously underestimates effect size serious Without further research it remains unclear whether categories identified through represent fundamentally different "types" suggested some high income countries, or merely differential groupings serverity. At level, partner-related lifetime risk including exposure child, controlling behavior, frequency drunkenness, history fights other men having outside sexual partners. Marital conflict, more children, living together versus being married, not completing secondary school, poor communication between couple are also strongly associated cluster proportion women secondary norms around male dominance, households which routinely comes home drunk among strongest variables predicting mean At macro level, range related women's status, gender inequality, social socio-economic development prevalence multivariate analysis, acceptability wife beating control female behavior, access formal wage employment appear linked past year A drinking binge 2 does levels abuse, illustrating can be those population-level risk. Conclusion: next generation should focus longitudinal mixed method studies to help clarify temporal associations among why certain emerge markers