Trends and Spatial Variation of Female Genital Mutilation among Reproductive Age Women in Ethiopia based on 2000, 2005, and 2016 Ethiopian Demographic and Health Surveys: Spatial-temporal and Multivariate Decomposition Analysis


 Introduction: Even if FGM has been declined dramatically in the world, the rate of decline is far below the rate needed to achieve SDG in Ethiopia. FGM remains as a serious public health concern in Ethiopia and the prevalence has varied across and within countries. Therefore, this study aimed to assess the trends and geographic variation of FGM practice in Ethiopia based on EDHSs.
Methods: The study used the data from the three DHSs conducted in Ethiopia. Trend and Logistic based decomposition analysis technique was used for analyzing the change in FGM practice overtime and factors contributing to the change in FGM practice. STATA 14 was used for data management and analysis. The Bernoulli model was fitted using spatial scan statistics version 9.6 to identify hotspot areas and ArcGIS version 10.6 to explore the spatial distribution FGM.
Results : The prevalence of FGM practice was decreased from 79.9% in 2000 to 70.4% in 2016. The decomposition analysis indicated that about 95% of the overall decrease in FGM practice was due to the difference in the change in the behavior of FGM practice among urban residents, orthodox and Muslim followers, self-employed, education, and media exposure were significant factors contributed to the change in FGM over the study period. The SaTScan analysis identified 141 most likely clusters (LLR=711.9, p<0.001) in 2000, 175 (LLR=576.4, p<0.001) in 2005, and 220 (LLR= 243.6, p<0.001) in 2016 which was located in Somali, Somali, Harari, and border areas of Somali consistently over the three surveys.
Conclusions: These results showed that FGM practice has been declined in Ethiopia. The decrease in FGM practice could be attributable to the difference in the change in effects of residence, media exposure; religion, region, and educational status. The spatial analysis provides further insight into differences in FGM practice within the country and highlights primary and secondary clusters. This could enable efficient and timely spatial targeting of hotspot areas of FGM practice to achieve the goals of eliminating FGM practice in Ethiopia by 2025 and the government should scale up the public health programs to improve the rate of reduction.


Abstract
Introduction: Even if FGM has been declined dramatically in the world, the rate of decline is far below the rate needed to achieve SDG in Ethiopia. FGM remains as a serious public health concern in Ethiopia and the prevalence has varied across and within countries. Therefore, this study aimed to assess the trends and geographic variation of FGM practice in Ethiopia based on EDHSs. Methods: The study used the data from the three DHSs conducted in Ethiopia These results showed that FGM practice has been declined in Ethiopia. The decrease in FGM practice could be attributable to the difference in the change in effects of residence, media exposure; religion, region, and educational status. The spatial analysis provides further insight into differences in FGM practice within the country and highlights primary and secondary clusters. This could enable efficient and timely spatial targeting of hotspot areas of FGM practice to achieve the goals of eliminating FGM practice in Ethiopia by 2025 and the government should scale up the public health programs to improve the rate of reduction.

Background
According to World Health Organization (WHO), Female Genital Mutilation (FGM) is defined as partial or total removal of the external female genitalia or other injuries to the female genital organs for non-medical reasons that have no health benefits for girls and women (1). FGM is widely practiced in Africa, Asia, Middle East and some countries in South America and it has strong cultural beliefs that are linked to women's sexuality and their reproductive role in the society and is recognized internationally as a violation of the human rights of girls and women (2). Globally, more than 200 million girls and women undergo female genital mutilation, of these 125 million were occurred in Africa (3). Despite efforts to eradicate the FGM practice, an estimated 3 million girls are at risk of undergoing female genital mutilation every year and exposed to the health consequences of FGM and the parents allow their daughters to conform to the social tradition (4).
Female genital mutilation remains a major public health problem worldwide (2,5). It is highly practiced in Africa, mostly in the north-eastern region of Africa: Djibouti, Eritrea, Ethiopia, and Somalia (6). The trend of girls undergoing female genital mutilation has reduced dramatically in Africa. According to the systematic analysis based on DHS data from 29 African countries and 2 west Asian countries, the prevalence of female genital mutilation has varied greatly between countries and regions and also varied within countries over time. The rates of FGM have been decreased in North Africa from 57.7% in 1990 to 14.1% in 2015. In West Africa, the trends of FGM have decreased from 73.6% in 1996 to 25.4% in 2017. Whereas, the trends of FGM in East Africa has been reduced from 71.4% in 1995 to 8% in 2016 (7). Countries with the highest prevalence of FGM among women aged 15 to 49 are Somalia at 98 percent, Guinea at 97 percent, Djibouti 93 percent and Egypt at 87 percent (8).
Eliminating FGM is the focus of governments and NGOs all over the world. The Ethiopian government has prioritized sustainable development goal 5.3 as one of the national development targets and it outlined strategic measures to reduce FGM practice that can contribute to the elimination of FGM in Ethiopia (9,10). In Ethiopia, FGM has been decreased from 79.9 in 2000 to 65% in 2016 (11)(12)(13).
Despite this progress, FGM remains a serious public health problem in Ethiopia and has affected 23.8 million women and girls, making Ethiopia the second-highest FGM practice in Africa (14). Despite a significant reduction in FGM nationally over the past decades, the UNICEF report from the 2011 Welfare and Monitoring Survey one in every four girls are subjected to the practice and the risk varies across regions with the highest prevalence in Somali and Afar regions (15). FGM has serious physical and psychological consequences to women, risking their lives at the time of circumcision, at marriage, during birth and decrease their attendance and performance in schools (16). The practice has been associated with significant health consequences such as complications during pregnancy and childbirth, menstrual abnormality, post-traumatic stress disorder, recurrent pelvic and vaginal infection (17).
Although there are studies conducted on the prevalence of female genital mutilation in different parts of Ethiopia (18)(19)(20)(21)(22)(23)(24) and one study done on Geographic variation of FGM based on nationally representative EDHS data (21) and the result of these studies showed that there is geographic variability of the prevalence of FGM practice within the country but this study only explore the spatial distribution of prevalence of FGM in Ethiopia, they did not conducted Spatial autocorrelation, Hotspot analysis, interpolation, SaTScan analysis and trend analysis since this is important to identify significant hotspot areas and change in FGM practice across areas which is important for designing effective public health interventions and to identify the factors contributing for the decline in FGM practice over time in Ethiopia which is important to evaluate already implemented policies and programmers and to work on it.
Even if the prevalence of FGM has been decreased overtime in Ethiopia (11)(12)(13), still there are areas with a high prevalence of FGM in different parts of the countries. Therefore having information on the trend of FGM among women is necessary for the development of public health policies for the prevention of FGM practice in the identified hotspot areas of FGM, and would help policymakers to work on factors contributing to the change in FGM overtime and design public health interventions.
The ultimate aim of this study is to inform decision and policymakers to design effective public intervention across geographic regions. To fill this research gap, we therefore used the three consecutive Ethiopian Demographic and Health Survey (EDHS) 2000, 2005 and 2016 data sets to assess the trends and spatial variation of FGM in Ethiopia using a multivariate decomposition and spatial analysis to identify the determinants factors contributing for the decline in FGM from 2000 to 2016. Thus, the government gives more attention to the most significant contributing factors and design and implement appropriate interventions for further decline of FGM in Ethiopia. Understanding the trends and geographic variations of FGM also would help public health planners, policymakers, programmers, and partners to design effective strategies and interventions to reduce FGM practice in Ethiopia.

Study design and Study settings
Repeated cross-sectional study design was employed using Ethiopian Demographic and Health surveys (EDHSs) in 2000, 2005, and 2016 to explore the trend and spatial variation of FGM. Since EDHS 2011 FGM data was not collected, EDHS 2011 was not used for this study. Ethiopia is located in the horn of Africa and it has been divided into nine regions and two city administrations ( Figure 1).
Ethiopia is an agrarian country and agriculture accounts for 43 percent of the gross domestic product (GDP) and 84% of the population lives in rural areas. More than 80 percent of the country's total population lives in the regional states of Amhara, Oromia, and SNNPRs (25). In 2016, the population was 102 million of which 43.47% the population ages less than 14 years with birth rate of 36.5 births per 1000 population and fertility rate of 4.46, Ethiopia is the13th in the world and 2nd in Africa most populous country Ethiopia has 3 tiers health systems, Primary health care unit (Primary hospital, health center, health post, primary clinic, and medium clinic); Secondary health care (General hospital, specialty clinics and specialty centers); and Tertiary health care (Specialized hospital). The number of hospitals varies from region to region in response to differences in population size. The most populous region, Oromia has 30 hospitals. The other two predominant regions Amhara and SNNPR have 19 and 20 respectively with Tigray in fourth place with 16 hospitals, Gambela has only one hospital and Benishangul-Gumuz two had two hospitals (26).

Sample and Source population
The source population was all reproductive-age women within five years before the survey in Ethiopia and all reproductive-age women in the selected enumeration areas within five years before the survey were the study population. The EDHS used a stratified two-stage cluster sampling technique were used for analysis. The detailed sampling procedure was presented in the full EDHS report (11, 12, and 27).

Outcome variables
The EDHS asked women to answer to the question "having experiencing Female genital mutilation?".
The response variable of this study was whether circumcised or not. The response variable for the ith mother is represented by a random variable Yi with two possible values coded as "yes" and "No". So, the response variable of the ith mother Yi was measured as a dichotomous variable with possible values Yi = Yes if ith mother had experienced circumcision and Yi = No if mother did not experience circumcision.

Independent variables
Independent factors that were considered to affect the practice of female genital mutilation like the place of residence, religion, geographic region, age of women, maternal education, women occupation, media exposure and wealth status were included for the study.

Data collection procedure
The study was conducted based on EDHS data by accessing from the DHS program official database (www.measuredhs.com) after permission was granted through an online request by explaining the objective of our study. The raw data was collected from all parts of the country on childbearing aged women using a structured and pre-tested questionnaire (11)(12)(13). We used the women data set and extracted the outcome and independent variables. Geographic coordinate data (longitude and latitude coordinates) was taken at the cluster level/ enumeration area level.

Data management and analysis
Before the actual data collection, the Pre-test was conducted which consisted of in-class training, biomarker training, and field practice days. The questionnaires were pretested in all three local languages (Amharic, Oromia, and Tigrigna) to make sure that the questions were clear and could be understood by the respondents. The data were weighted using sampling weight, primary sampling unit, and strata before any statistical analysis to restore the representativeness of the survey and to tell the STATA to take into account the sampling design when calculating standard errors to get reliable statistical estimates. Cross tabulations and summary statistics were conducted to describe the study population. Descriptive and summary statistics were done using STATA version 14 software. The changes of FGM over time can be attributed to compositional changes between surveys and changes in the effects of the selected explanatory. Hence, the observed difference in FGM between surveys is additively decomposed into a characteristics (or endowments) component and a coefficient (or effects of characteristics) component.
For logistic regression, the Logit or log-odd of FGM is taken as: The E component refers to the part of the differential owing to differences in endowments or characteristics. The C component refers to that part of the differential attributable to differences in coefficients or effects.
The equation can be presented as: The recently developed multivariate decomposition for the non-linear model was used for the decomposition analysis of stillbirth using mvdcmp STATA command (28).

Spatial analysis
For the spatial analysis ArcGIS version 10.6 software and SaTScan version 9.6 software.

Spatial autocorrelation analysis
The spatial autocorrelation (Global Moran's I) statistic measures whether FGM patterns were dispersed, clustered or randomly distributed in the study area (29). Moran's I 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 Values close to −1 indicate FGM dispersed, whereas Moran's I close to +1 indicate FGM clustered and disease distributed randomly if I value is zero. A statistically significant Moran's I (p < 0.05) leads to rejection of the null hypothesis (FGM is randomly distributed) and indicates the presence of spatial autocorrelation.
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 statistical significance of clustering, and the p-value computed for the significance. Statistical output with high GI* indicates "hotspot" whereas low GI* means a "cold spot"(30).

Spatial interpolation
It is very expensive and laborious to collect reliable data in all areas of the country to know the burden of a certain event. Therefore, part of a certain area can be predicted by using observed data using a method called interpolation. The spatial interpolation technique is used to predict stillbirth on the un-sampled areas in the country based on sampled EAs. There are various deterministic and geostatistical interpolation methods. Among all of the methods, ordinary Kriging and empirical Bayesian Kriging are considered the best method since it incorporates the spatial autocorrelation and it statistically optimizes the weight (31). Ordinary Kriging spatial interpolation method was used for this study for predictions of FGM in unobserved areas of Ethiopia.

Spatial scan statistical analysis
Spatial scan statistical analysis Bernoulli based model was employed to test for the presence of statistically significant spatial clusters of FGM using Kuldorff's SaTScan version 9.6 software. The spatial scan statistic uses a circular scanning window that moves across the study area. Women are circumcised were taken as cases and women who are not circumcised were taken 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 FGM cases 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 (32).

Ethical approval and consent to participate
The study was based on secondary analysis of existing survey data with all identifying information removed. Permission for data access was obtained from measure demographic and health survey through an online request from http://www.measuredhsprogram.com. Prior to the actual interview, each woman was asked if she agreed to participate in the study. The GIS data were obtained through explaining the purpose of using GPS data and approval from Measure DHS. Informed consent was obtained from the participants, their guardian or household heads.  Table 2).

Decomposition analysis
The decomposition analysis revealed that about 95% of the overall percentage change in FGM practice among reproductive-age women in Ethiopia was due to difference in coefficient (effects of characteristics) and the remaining 5% was due to the difference in characteristics (compositional factors or endowment) but the difference was not significant (Table 3) Table 6). It showed that women within the spatial window had 1.42 times higher risk of being circumcised as compared to women outside the spatial window and the secondary clusters were located in Benishangul, Amhara, Oromia, and Gambella regions (Figure 7).
In EDHS 2016, a spatial scan statistics identified a total of 581 significant primary and secondary clusters. Of these, 220 clusters were the most likely clusters (primary clusters) and the spatial window was located in the entire Somali, Afar and border areas of Somali, centered at 7.717178 N, 46.991580 E with a radius of 900.49 km, a Relative Risk (RR) of 1.42, and Log-likelihood (LLR=243.6, p-value<0.0001) ( Table 7). It showed that women within the spatial window had 1.42 times higher risk of circumcision as compared to women outside the spatial window (Figure 7). The rate of FGM practice in Ethiopia has been declined from 79.9% in 2000 to 70.4% in 2016 with the annual rate of reduction of 0.9%. Even though the prevalence of FGM has been dropped, the rate of decline is far below to achieve the sustainable development goal 5. which is slower than systematic study in, east Africa which was reduced from 71% to 8% within 20 years with annual rate of reduction of 3.2%, west Africa which was reduced from 73.6% to 25.4% within 21 years with annual rate of reduction of 2.3% and North Africa from 57.7% to 14.1% within 23 years with annual rate of reduction of 1.9% (38). This could be due to that FGM practice in Ethiopia is deeply rooted in traditions and persisted as a social norm for a long period and lack of resource allocation to provide health education to the communities (39). Therefore much progress is needed to achieve the SDG goals.

Kriging interpolation of FGM practice
The multivariate decomposition analysis identified the determinants contributing to the change in FGM practice over the last 16 years. Hence, understanding the factors contributing to the decline in FGM practice in Ethiopia has public health importance. The decomposition analysis revealed that about 95% of the decrease in FGM practice in Ethiopia over the past 16 years is due to the difference in effects of characteristics (coefficient) but the change due to the difference in endowment Therefore, women's education, media exposure and creating job opportunities were significantly contributing to the decline in FGM practice.
GIS-based spatial and SaTScan analysis showed that FGM practice was non-random. This was supported by a spatial study done in Nigeria (43), and Kenya (44). The hotspot and SaTScan analysis identified significant hotspot areas across the three EDHS surveys which were consistently located in the entire Somali, Afar, Harari and border areas of Somali regions. Even though there is a coordinated effort by engaging faith and community organizations to accelerate the abandonment of FGM practice in Ethiopia, Afar, and Harari and Somali regions were significant hot spot areas of FGM practice. This could be due to deeply rooted strong belief, attitudes and behaviors towards FGM practice in these regions and the belief that if a girl not get circumcised, the daughter may not get married and this makes the community female circumcision to get social acceptance(45). There is also evidence that hence FGM is deeply rooted in their tradition, there is resistance to abandoning FGM practice in Somali and afar region. This implies that the government needs to strengthen efforts to reduce the FGM practice in these regions by highly engaging the community leaders, civil society, school, media, faith leaders and females.

Strengths And Limitations
This study had several strengths. First, the study was based on nationally representative large datasets, and thus it had adequate statistical power. Second, the estimates of the study were done after the data were weighted for the probability sampling and non-response, to make it representativeness at national and regional levels: therefore, it can be generalized to all women in The study was based on secondary analysis of existing survey data with all identifying information removed. Permission for data access was obtained from measure demographic and health survey through an online request from http://www.measuredhsprogram.com. Prior to the actual interview, each woman was asked if she agreed to participate in the study. The GIS data were obtained through explaining the purpose of using GPS data and approval from Measure DHS. Informed consent was obtained from the participants, their guardian or household heads

Consent for publication
Not applicable since the study was a secondary data analysis already collected by CSA

Availability of data and materials
Data is available online and you can access it from www.measuredhs.com.

Competing Interests
Authors declare that they have no conflict of interest