Globally, particularly in Sub-Saharan Africa (SSA), Human Immunodeficiency Virus /Acquired Immune Deficiency Syndrome (HIV/AIDS) was the leading cause of morbidity and mortality among reproductive-age women [1]. In sub-Saharan Africa, it is estimated that 60% of HIV/AIDS patients were reproductive-age women [2] however, less than 30% of reproductive-age women have comprehensive knowledge of HIV/AIDS [3]. In developing countries like Ethiopia, reproductive-age women are highly infected by HIV/AIDS owing to the discriminatory cultural, social, and economic status in society [4]. Mass media exposure (television, radio, and newspapers) have become a significant source of raising awareness on HIV/AIDS to the population and an effective tool for disseminating HIV education to the group [5, 6]. Media exposure is one of the most common and cost-effective public health programs globally for improving public health issues by increasing knowledge and altering health behaviors [7]. The level of HIV/AIDS-related knowledge among reproductive-age women has varied across and within countries and strongly connected to the socio-economic contexts [8].
In the last few years, SSA countries have presented media exposure to combat HIV/AIDS to rapidly touch the community by addressing different messages [9]. However, mass media availability and accessibility have been speckled across different social groups [10, 11], and it is the most powerful tool to change the health behavior of the community [12]. To reduce the incidence of HIV/AIDS among reproductive-age women, improving comprehensive knowledge of HIV/AIDS transmission, and smearing ways of prevention methods is a crucial issue [13].
Despite mass media plays a significant role in reducing the incidence of HIV/AIDS through raising comprehensive knowledge towards HIV/AIDS, most of the reproductive-age women in developing countries including Ethiopia have poor comprehensive knowledge of HIV/AIDS. For example, in Sub-Saharan Africa Uganda, only 38% of reproductive-age women had a comprehensive knowledge of HIV/AIDS [14]. Besides, in Ethiopia especially in Afar, Gambella, Dire Dawa, and Somali regions were detected poor knowledge of HIV/AIDS among reproductive-age women [15]. However, there are few studies conducted on the magnitude and determinants of comprehensive knowledge of HIV/AIDS among reproductive-age women in Ethiopia [16, 17], these studies are unable to capture the spatial distribution of comprehensive knowledge of HIV/AIDS and the effect of media exposure on the comprehensive knowledge of HIV/AIDS. Therefore, this study aimed to investigate the effect of mass media on comprehensive knowledge of HIV/AIDS and its spatial distribution among reproductive-age women in Ethiopia using spatial and multilevel analysis. Thus, the findings of this study could help to increase media exposure in the identified hotspot areas where poor comprehensive knowledge of HIV/AIDS was clustered to reduce the incidence of HIV/AIDS among reproductive-age women in Ethiopia.
Study design, setting, and period
A community-based cross-sectional study was conducted based on the 2016 Ethiopian Demographic and Health Survey (EDHS) data. It was the fourth nationally representative survey conducted in Ethiopia employed with a 5-year interval. Ethiopia is situated in the Horn of Africa. It has 9 Regional states (Afar, Amhara, Benishangul-Gumuz, Gambela, Harari, Oromia, Somali, Southern Nations, Nationalities, and People’s Region (SNNP) and Tigray) and two Administrative Cities (Addis Ababa and Dire-Dawa). In Ethiopia, about 84% of the population are rural residents [18]. It is the13th in the world and 2nd in Africa’s most populous country [19].
Sample and source of population
All reproductive-age women within 5 years preceding the survey in Ethiopia were the source of the population whereas all reproductive-age women in the selected enumeration areas within 5 years preceding the survey were the study population. In EDHS, a stratified two-stage cluster sampling technique was employed using the 2007 Population and Housing Census (PHC) as a sampling frame. In the first stage, 645 Enumeration Areas (EAs) were selected. In the second stage, on average 28 households were systematically selected. A total weighted sample of 15,683 reproductive-age women was included in the study. The detailed sampling procedure is presented in the full EDHS 2016 report [20].
Measurement of variables
The dependent variables used for the study was “comprehensive knowledge of HIV/AIDS”, it was generated by aggregating a series of questions which were designed to evaluate knowledge of HIV/AIDS was based on the question: 1) the risk of HIV/AIDS transmission can be reduced by having sex with only one infected partner, who has no other partners, 2) a person can reduce the risk of getting HIV by using a condom every time they have sex; 3) a healthy-looking person can have HIV; 4) a person can get HIV from mosquito bites; 5) a person can get HIV by sharing food with someone infected. These questions were used by Millennium Developing Goals (MDG) to measure “comprehensive knowledge of HIV/AIDS” in this study. Then the outcome variable was coded as 0 = “No” if women didn’t have comprehensive knowledge of HIV/AIDS, and as 1 = “yes” if a woman had Comprehensive knowledge of HIV/AIDS.
Independent variables included in this study were frequency of media exposure (frequency of watching television, listening radio, and reading newspaper, which was defined as “not at all”, “less than once a week” and “at least once a week”, maternal age, residence, place of delivery, covered by health insurance, occupation, maternal age, religion, wealth status, and maternal education.
Data management and analysis
STATA version 14, ArcGIS version 10.6, and SaTScan version 9.6 statistical software were used for analysis. The DHS data was hierarchal data, there might have a clustering effect hence women in one cluster might share similar characteristics than women in different clusters. Therefore we have checked the clustering effect using the Intra-class Correlation Coefficient (ICC) and Likelihood Ratio (LR) test. The ICC indicates that there was a significant clustering effect that should be considered using advanced models such as mixed-effect models. We have fitted two models such as the standard logistic regression model and the mixed-effect logistic regression model. Deviance, Akakie Information Criteria (AIC), and Bayesian Information Criteria (BIC) were used for model comparison. The mixed-effect logistic regression model was the best-fitted model for the data since it had the lowest deviance value. Both bi-variable and multivariable mixed-effect logistic regression analyses were done. Variables with < 0.2 p-values in the bi-variable analysis were considered for the multivariable mixed-effect logistic regression model. Adjusted Odds Ratio (AOR) with a 95% Confidence Interval (CI) and p-value < 0.05 in the multivariable model were used to declare significant association with comprehensive knowledge of HIV/AIDS. For the determinant factors, we used STATA version 14 statistical software using xtmelogit, xtmrho and icc packages were used.
Spatial analysis
For the spatial analysis ArcGIS version, 10.3 and SaTScan version 9.6 software were used.
Spatial autocorrelation analysis
The spatial autocorrelation (Global Moran’s I) analysis is a spatial statistics used to measure spatial autocorrelation by taking the entire data set and produce a single output value which ranges from − 1 to + 1. Moran’s I assess whether the spatial distribution of comprehensive knowledge of HIV/AIDS was dispersed, clustered, or randomly distributed in the study area [21]. Moran’s I values close to − 1 indicates there is dispersion, whereas Moran’s I close to + 1 indicate there is spatial clustering and distributed randomly if Moran’s I value is close to 0.
Hot spot analysis (Getis-OrdGi* statistic)
Getis-OrdGi* statistics were computed to measure how spatial autocorrelation varies over the study location by calculating GI* statistic for each area. Z-score is computed to determine the significant hotspot and significant cold spot areas of poor comprehensive knowledge towards HIV/AIDS. Statistical output with high GI* indicates “hotspot” whereas low GI* indicates “cold spot” [22].
Spatial interpolation
The Kriging spatial interpolation technique was used to predict the percentage of comprehensive knowledge towards HIV/AIDS among reproductive-age women on the un-sampled areas in the country based on observed measurements. There are various deterministic and geostatistical interpolation methods. Ordinary Kriging spatial interpolation method was used for predictions of comprehensive knowledge of HIV/AIDS in unobserved areas of Ethiopia since it incorporates the spatial autocorrelation and it statistically optimizes the weight [23].
Spatial scan statistical analysis
For the spatial scan statistical analysis, the Bernoulli based model was employed to test for the presence of statistically significant spatial clusters of poor comprehensive knowledge of HIV/AIDS using Kuldorff’s SaTScan version 9.6 software. The SaTScan uses a circular scanning window that moves across the study area. Women having poor comprehensive knowledge of HIV/AIDS were taken as cases and those who have good comprehensive knowledge of HIV/AIDS as controls to fit the Bernoulli model. The numbers of cases in each location had Bernoulli distribution and the model required data for cases, controls, and geographic coordinates. The default maximum spatial cluster size of < 50% of the population was used, as an upper limit, which allowed both small and large clusters to be detected and ignored clusters that contained more than the maximum limit.
For each potential cluster, a likelihood ratio test statistic and the p-value were used to determine if the number of observed comprehensive knowledge of HIV/AIDS within the potential cluster was significantly higher than expected or not. The scanning window with maximum likelihood was the most likely performing cluster, and the p-value was assigned to each cluster using Monte Carlo hypothesis testing by comparing the rank of the maximum likelihood from the real data with the maximum likelihood from the random datasets. The primary and secondary clusters were identified and assigned p-values and ranked based on their likelihood ratio test, based on 999 Monte Carlo replications [24].