PEPFAR's annual planning process is done either at the country (COP) or regional level (ROP).
PEPFAR's programs are implemented through implementing partners who apply for funding based on PEPFAR's published Requests for Applications.
Since 2010, PEPFAR COPs have grouped implementing partners according to an organizational type. We have retroactively applied these classifications to earlier years in the database as well.
Also called "Strategic Areas", these are general areas of HIV programming. Each program area has several corresponding budget codes.
Specific areas of HIV programming. Budget Codes are the lowest level of spending data available.
Expenditure Program Areas track general areas of PEPFAR expenditure.
Expenditure Sub-Program Areas track more specific PEPFAR expenditures.
Object classes provide highly specific ways that implementing partners are spending PEPFAR funds on programming.
Cross-cutting attributions are areas of PEPFAR programming that contribute across several program areas. They contain limited indicative information related to aspects such as human resources, health infrastructure, or key populations programming. However, they represent only a small proportion of the total funds that PEPFAR allocates through the COP process. Additionally, they have changed significantly over the years. As such, analysis and interpretation of these data should be approached carefully. Learn more
Beneficiary Expenditure data identify how PEPFAR programming is targeted at reaching different populations.
Sub-Beneficiary Expenditure data highlight more specific populations targeted for HIV prevention and treatment interventions.
PEPFAR sets targets using the Monitoring, Evaluation, and Reporting (MER) System - documentation for which can be found on PEPFAR's website at https://www.pepfar.gov/reports/guidance/. As with most data on this website, the targets here have been extracted from the COP documents. Targets are for the fiscal year following each COP year, such that selecting 2016 will access targets for FY2017. This feature is currently experimental and should be used for exploratory purposes only at present.
The purpose of this activity is to build a ‘geodatabase' that stores geographic data for Mozambique using visual and spatial techniques to discern patterns and to identify relationships between the location of features (such as location of testing sites) and health outcomes related to HIV/AIDS. In addition, it will build capacity at both the USG staff and the local Eduardo Mondlane University in Geographic Information Systems (GIS) skills. This will allow for an improved monitoring and evaluation ability from the in-country PEPFAR team.
ArcGIS will be the main GIS software package used to manipulate, process, and map the geographic data available at the Global AIDS Program, the National Statistics Institute (INE), and the National Health Institute (INS) in Mozambique. We will also closely collaborate with Health Information Systems Program-Mozambique (HISP-Mozambique), an established group of researchers from the local Eduardo Mondlane University.
Examples of geographic data (e.g. shapefiles) to be included in the ‘geodatabase' include boundary maps of provinces and health districts in Mozambique and morbidity/mortality data from around the country. Centralizing these types of data will allow us to generate maps that show the distribution of selected outcomes of interest. We will use the various layers (e.g. shapefiles from the ‘geodatabase') to provide multiple options for visualizing patterns and identify whether values cluster or are evenly distributed by province or district. Possible outcomes of this portion of the work are kernel density maps, bivariate choropleth maps, and spatial cluster maps.
Two types of density maps will be created, by area and by creating a density surface. To create a density map by area, we will summarize the data by area such as by district or province. We will also create a density surface as a raster layer, which is an approach that provides the most detailed information. We will create a density surface from individual locations, or linear features such as roads or rivers to show where point or line features are concentrated. Various issues to consider in the creation of such density maps include calculating density values for defined areas, choosing parameters for their display and looking at the results to identify patterns on the maps. For example, we can use kernel density mapping to portray variations in overall point locations and create a smoothly curved surface around each point. Kernel density maps help illustrate density of geographic phenomena in a raster/grid cell method that eliminates the exaggeration effect of polygon maps.
Bivariate choropleth maps are another type of map that we can use to identify associations between characteristics of provinces or districts and locations of facilities or aggregate health statistics. Bivariate maps allow for the display of areas showing combinations of two different variables. With this type of map it is possible to see, for example, where areas of high migration and higher HIV prevalence coincide. An additional strength of this type of mapping is that these maps can also be used to describe the same variable at two different points in time, to see whether it has a stationary or migratory spatial pattern.
Finally, we will use the warehouse of data that we assembled to create spatial cluster maps. These maps illustrate any statistically significant clusters of high or low values for variables of interest, suggesting hot spots or cool spots. The same variables can be observed at different time periods, illustrating whether clusters were stationary or migratory over time. In addition, since the data to be analyzed was collected over a significant period of time, the temporal aspects of the data will also be explored. Time-series maps will be created that illustrate how the distribution of health statistics and location of health facilities have changed over time—in comparison to other districts or provinces- both for PEPFAR and for the National Health System as a whole.
In addition to producing different types of maps to visualize patterns in the geospatial data, we can further employ techniques of spatial autocorrelation and spatial regression. Exploratory spatial data analysis is a collection of methods that will allow us to explore the correlation between values in an area and neighboring values. We can conduct tests to identify evidence of spatial dependency/correlation or if patterns of selected values are randomly scattered. To conduct these types of spatial analyses, we will use GeoDa (http://www.csiss.org/clearinghouse/GeoDa/) and GeoVista Studio (http://www.geovistastudio.psu.edu/jsp/index.jsp) software to perform ESDA and any
other spatial statistical analyses.
Producing these types of maps and conducting spatial analyses will allow us to find priority areas that require action, examine specific research questions, and monitor changing conditions over the long term.
