The Spatial Distribution of HIV Prevalence Rates in Nigeria

Nigeria has witnessed a gradual decline in the national trend of HIV prevalence from the peak of 5.8% in 2001 to 3.4% in 2012. In spite of the decline nationally, there is a wide variation in the distribution of HIV at the sub-national level. This study therefore aims to explore the spatial distribution of HIV in Nigeria. The study was an ecological study of secondary data of the National HIV prevalence studies conducted between 2008 and 2012. The global Moran’s I and Local Moran’s I (LISA) test were used to measure spatial autocorrelation. A final choropleth map of local Moran’s FDR-adjusted p values was produced and a p value of ≤ 0.05 was regarded as statistically significant. The analysis of the data was carried out in R statistical package version 3.2.3. Twenty seven (73%) states showed decline in HIV while 10 (27%) states showed an increase in the HIV prevalence rate between 2008 and 2012. Global Moran I statistics for the country indicates a significant positive spatial autocorrelation of HIV in 2008 and 2010 however, there was no significant relationship in 2012. The choropleth map identified four hot spot significant clusters of HIV in 2008 which reduced to three states in 2010 and one state in 2012. The study showed a number of significant hot spot clusters for HIV in Nigeria though there was a general significant decline in the national HIV prevalence rate in the country. There is need to further understand the drivers of the epidemic in hot spot areas and target specific localized action to stem the tide of the epidemic focusing the scarce resources in the identified hotspot states.


INTRODUCTION
Globally an estimated 36. Spatial analytic techniques have been a valuable tool in the study of infectious diseases including HIV. Spatial analysis can assist in the detection of spatial variation, clustering or hot spots where targeted action can be deployed and also in the general understanding of the influence of neighbouring states on the identified hotspots. Few studies have been done using spatial analytic techniques in the understanding of HIV in Nigeria [5,6], however none of these studies have looked at the trend of HIV epidemic in the last decade. This study therefore explored the spatial distribution of HIV in Nigeria between 2008 and 2012.

Study Design
The study was an ecological study of National HIV prevalence studies conducted between 2008 and 2012. The 2008 and 2010 national HIV prevalence rates at the subnational (state) level was obtained from the 2010 Nigerian HIV sentinel survey report [7]  Nigeria is the largest country in West Africa with a projected population of about 170 million people. It has 36 independent states and the Federal Capital Territory (FCT) which are grouped into six geopolitical regions. The 36 states and the FCT were used as the unit of geographical analysis. The shape file of the country at the first administrative level (i.e 36 states and FCT of the federation) was obtained online from the global administrative areas database website (www.gadm.org).

Spatial Exploration
The global Moran's I and Local Moran's I (LISA) test were used to measure spatial autocorrelation in the data set [8]. The global autocorrelation is a single measure used to assess the global spatial autocorrelation for HIV in the country as a whole. A positive spatial autocorrelation shows clustering of similar values across geographical space than what is expected if it was a random distribution while a negative spatial autocorrelation shows that the neighbouring values are more dissimilar than expected.
The global Moran I is expressed by the formula.
Where N is the number of spatial units indexed by i and j; X is the variable of interest, is the mean of X; and W ij is an element of a matrix of spatial weights. The value of Moran's I test statistics range from -1 to +1. A Moran of +1 suggests a strong positive autocorrelation while a value of -1 suggests a strong negative spatial autocorrelation. The Local spatial autocorrelation (LISA) [9] is a measure of local spatial autocorrelation for each individual location. It is expressed as Where z i is the original variable x i in standardized form and w ij is the spatial weight.
A choropleth thematic map was used to visualise possible clustering and the states with significant spatial autocorrelations were identified by mapping the p values of the Local Moran's I statistics. The false discovery rate (FDR) method by Benjamin and Hochberg's was used for the adjustment of the local Moran's I p values [9]. A final choropleth map of local Moran's FDRadjusted p values was produced and a p value of ≤ 0.05 was regarded as statistically significant.
The analysis of the data was carried out in R statistical package version 3.2.3 [8].

RESULTS
Nigeria is divided into 36 states and the Federal Capital territory (FCT) but for the purpose of this study the country is divided into 37 spatial units as shown in Fig. 1. Fig. 2 showed the trend of the national HIV prevalence rate from 2008 to 2012. The national HIV prevalence rate has been on the decline from 4.6% in 2008, 4.1 in 2010 and 3.4% in 2012. However, there has been wide geographical variation within the country. Table 1 showed        [6]. Nasarawa state in the North central has remained consistently a hot spot for HIV infection since 1999 when all the 36 states and FCT were involved in the national HIV sentinel survey. This may suggest that Nasarawa state is the epicenter of HIV infection in Nigeria. A cursory look at some studies conducted in Nasarawa state on HIV showed common traditional practices that predispose to HIV infection such as polygamy, levirate marriage (a form of marriage where the brother of a deceased man is obliged to marry his brother's widow), female genital mutilation, early and child marriage [11].

DISCUSSION
There is however need for future studies to better understand the drivers of the HIV epidemic in the state and therefore target the state with specific localized action to stem the tide of the epidemic and avoid diffusion of the HIV epidemic to contiguous states.
Though this study was not aimed at exploring the factors that may be responsible for the high HIV prevalence rate because of lack of data to compare across states, other studies have suggested other possible explanations why there is variation of HIV prevalence at the states. Samuels et al. [12] argued that though the country is still experiencing a generalized epidemic, there are concentrated pockets of areas in the country with a high rate of HIV infection especially where there is a large concentration of individuals that constitute high risk groups for HIV such as female sex workers (FSW), men who have sex with men and injection drug users who may be reservoir of infection to the general population. Other drivers of the epidemic identified includes low awareness of HIV/AIDS, stigma and discrimination, multiple sexual partners and low condom use, religion and cultural practices such as polygamy and ''spouse-sharing'' sharing practices with relatives among "Okun" people in North central geopolitical zone of Nigeria [13].
Another study suggests that poor and inequitable distribution of health infrastructure and gender/economic inequality that predispose individuals to transactional and intergenerational sexual practices may be responsible for the unequal distribution of HIV in the country [14]. The complex interactions between these factors may be responsible for the geographical distribution of the disease in the country. There is need for further research to explore the contribution of these factors to the HIV epidemic in the country.
The reasons for the decline in the national HIV prevalence and the number of significant hot spot states in the country cannot be immediately ascertained. However, it may be attributed to the impact of interventions to reduce HIV prevalence in the country [10]. There has been considerable financial investment from both national and international donors in the fight against HIV in the country. The total funding to combat the HIV epidemic in Nigeria rose from USD 415,287,430 in 2009 to USD577, 432,903 in 2012 [10]. This funding was used to expand prevention services such as awareness creation, improving access to HIV counseling and testing services, prevention of mother to child transmission services and treatment and support services for people living with HIV/AIDS [15]. This may be responsible for the decline in the number of states with significant clustering from 2008 to 2012. Other studies have suggested that the decline in HIV prevalence may reflect the natural dynamics in HIV transmission in the general population from the epidemic threshold of sustainability [16,17], the variability on the host susceptibility to HIV infection [18], the changes in sexual risk behavior [19][20][21], the patterns of migration of high risk groups and the increased HIV associated mortality [22,23]. Further research will be required to assess the contribution of these factors to the decline in the national HIV prevalence rate in Nigeria.

LIMITATION OF STUDY
Disaggregation of the analysis of this study was at the state levels which are not homogeneous entities because of the large population. This is largely due to unavailability of HIV prevalence data at lower levels like the local government area (district). In addition, the government needs to make geographic and epidemiological data on HIV risk factors disaggregated by states accessible to aid comparison among states. In addition, there is need for government to strengthen epidemiological and geographical data generation and reporting at lower levels in the country especially at the local government areas to allow for more localized spatial analysis that can provide detailed information to policy makers for effective decision making and strategic planning. The study suggests that subsequent studies should explore the determinants of HIV driving the epidemic in the identified hotspots/states

CONCLUSION
This study had built on the existing knowledge of the application of spatial analytic techniques in the study of the spatial distribution of HIV at the subnational level in Nigeria. The study has shown a decline in HIV prevalence rate at the national level and in some states. In addition, a decline in the number of significant clusters was observed between 2008 and 2012. There is therefore the need for policy makers to plan targeted action to the identified hot spots for HIV transmission in the country in order to effectively stem the tide of the epidemic in Nigeria.

CONSENT
Patient consent was not required for this study. The study utilised is publicly available data on HIV prevalence rate in Nigeria with no personal identifiers.

ETHICAL APPROVAL
Ethical approval was not obtained because the data utilised for this study is publicly available with no personal identifiers.