War-associated Civilian Vulnerability and its Predictive factors in Tigray Region, Northern Ethiopia: A Zero-Inflated Poisson regression model

Background:The political turmoil in late 2020 and throughout 2021 and 2022 in the Tigray region of Ethiopia resulted in significant conflict-induced civilian vulnerabilities. This situation represents a deeply traumatic experience that can lead to dire consequences. There is still a lack of systematic quantitative studies that provide broader evidence on the level and contribution of war to civilian vulnerability in the Tigray region. This study determines the war associated civilian vulnerability level and predictors in the community of Tigray, Ethiopia. Methods: This article was part of the integrated survey (18). In this integrated survey, a community-based cross-sectional study was conducted among systematically selected 13,915 women aged 15 years and above in the Tigray region. The sample size for this integrated survey was calculated based on Multiple Indicator Cluster Survey (MICS). A multistage sampling method was employed to interview study units. The questionnaire was administered through face-to-face interviews using open data kit (ODK) installed in mobile smart-phones. Variables that answer the Vulnerability status and its determinants were extracted for analysis. Descriptive statistics, such as frequency, percentage, mean, and standard deviation, were applied based on the nature of the variables. The vulnerability status was scored from 7 exposure variables. Different Poisson regression models were used to identify candidate variables for vulnerability using R 4.1.0, which were then included in count regression models to determine associated factors and strength of association. Akaike Information Criteria (AIC), Bayesian Information Criteria (BIC),and Log-likelihood model comparison approaches were performed. Accordingly, a Zero-inflated Poisson regression model was found to be the best-fitted model, considering the number of vulnerabilities as a dependent variable. Statistical significance was declared at a p-value less than 0.05. Result: Among the 13,915 women who participated in the study, 61.51% were vulnerable to at least one vulnerability condition. Of those, 20% were exposed to three harmful conditions. The mean and standard deviation of the count data related to vulnerability are 1.588 and 1.545, respectively. The output of the zero-inflated count model indicates that those adults who are aged (AIRR=1.115 (1.06- 1.17)), widowers (AIRR=1.21, 95% CI (1.06-1.183)), and Catholic followers (AIRR=1.257, 95% CI (1.215-1.301)). The merchant's (AIRR=1.136, 95%CI (1.06-1.215)) urban dwellers (AIRR=1.56, 95% CI (1.491-1.631)), having family members with chronic illness (AIRR=1.23, 95%CI (1.163-1.301)), food insecurity (AIRR=1.356, 95%CI (1.308-1.406)) and low household income (AIRR=1.157, 95%CI (1.113-1.202)) were more vulnerable. Conclusions:The study indicates that the vulnerability status of civilian was alarmingly high. Besides, variables such as chronic illness, perceived food insecurity, respondent age, religion, and urban residency were independent predictors of vulnerability. The study advocates for a holistic strategy to address war-related vulnerabilities, taking into account elements such as health conditions, perceptions of food security, demographic characteristics, and living environments.