Browsing by Author "Tanser, Frank Courteney."

Now showing 1 - 2 of 2

The application of geographical information systems to infectious diseases and health systems in Africa.
(2000) Tanser, Frank Courteney.; Solarsh, Geoffrey C.; Sharp, Brian Leslie.
The health sector has not yet begun to explore the full potential of geographical information system (GIS) technology for health research and planning. The goal of this thesis is to demonstrate this potential in Africa through the application of GIS to the most important health issues in the continent. In excess of 23,000 homesteads are mapped and interviewed throughout Hlabisa district, Kwa-Zulu Natal using differential global positioning systems (GPS). I use the GIS to analyse mode health care usage patterns. 87% of homesteads use the nearest clinic and travel an average distance of 4.72 km to do so. There is a significant logarithmic relationship between distance from clinic and usage by the homesteads (r2 = 0.774, p<0.0001). I propose the distance usage index (DUI) as a composite spatial measure of clinic usage. The index is the sum of the distances from clinic to all actual client homesteads divided by the sum of the distances from clinic to all homesteads within its distance-defined catchment. The index encompasses inclusion, exclusion and strength of patient attraction for each clinic. The DUI highlights significant disparities in clinic usage patterns across the district (mean = 110%, SD =43.7). The results of the study have important implications for health planning in Africa. I use GIS/GPS technology to quantify the spatial implications of a shift towards community-based treatment of tuberculosis using the DOTs strategy in Hlabisa. The mean distance from each homestead in the district to nearest supervision point is measured using a GIS. The shift in treatment strategy from hospital to community-based between 1991-1996 reduces the mean distance to treatment point from 29.6 km (94% of the population > 5km) to 1.5 km (entire population < 5km). GIS effectively documents and quantifies the impact of community-based tuberculosis treatment on access to treatment. I produce the first quantifiable evidence of a relationship between distance to roads and HIV prevalence using a GIS. HIV prevalence was measured through anonymous surveillance among pregnant women in Hlabisa and stratified by clinic attended. Assuming women attend the nearest clinic, the mean distance from homesteads to a primary or secondary road for each clinic catchment is strongly correlated with HIV prevalence (r = 0.66; p = 0.002). Further research is needed to better understand this relationship both at ecological and individual levels.I develop a methodology that has numerous applications to health systems provision in developing countries where limited physical access to primary health care is a major factor contributing to the poor health of populations. I use an accessibility model within a GIS to subdivide an area into units of equal workload using a range of physical and social variables. The methodology could be used to ergonomically design programmes for home-based care and tuberculosis directly observed treatment. It could also be used as a basis for more efficient distribution of community health workers. I use high-resolution long-term rainfall and temperature data to produce the first malaria seasonality (length, start and end of transmission season(s)) maps for Africa. I relate the model to population data and estimate the population exposure in a variety of transmission settings. I investigate the relationship between predicted length of transmission season and parasite ratio from 2335 geo-referenced studies of children <10 years across Africa. The research is the first to correlate actual malaria survey data with model predictions at a continental scale. The seasonality model corresponds well with historical expert opinion maps and case data. A significant logarithmic relationship is detected between predicted length of transmission season and parasite ratio (r2=0.712, p=0.001). I recompute the changes in the disease likely to occur as a result of global warming. The seasonality model constitutes an important first step towards an estimate of continental intensity of transmission.
Spatial analysis of HIV infections in high burden sub-districts in KwaZulu-Natal, South Africa.
(2017) Buthelezi, Usangiphile Evile.; Kharsany, Ayesha Bibi Mahomed.; Tanser, Frank Courteney.
Background: Substantial spatial variations in HIV prevalence and incidence at a global, national and district levels have been shown to occur. However, only a few studies have assessed variability of these infections at a highly localised level. Aim: The aim of the study was to assess the spatial variability of HIV prevalence and HIV-1 RNA viral load in two areas within the uMgungundlovu district, KwaZulu-Natal, South Africa. Methods: The data source for this study was from the HIV Incidence Provincial Surveillance System (HIPSS), a multi-stage random sampling of enumeration areas (EAs), households and individuals. From June 2014 to June 2015, HIPSS enrolled 9812 household-representative sample of men and women aged 15-49 years from 221 of the 591 randomly selected EAs. Briefly, the randomly selected households were identified through the global positioning system (GPS) co-ordinates. The head or designate of the selected household was provided with detailed study information, followed by verbal consent, collection of basic sociodemographic information and listing of household members. A single age eligible individual was randomly selected, provided with detailed study information, followed by written informed consent and or assent and enrolled. A questionnaire was administered to obtain demographic, psycho-social and behavioural information, biological samples for laboratory tests and GPS co-ordinates for the household were collected for each enrolled participant at the time of the interview. HIV prevalence, geometric mean viral load and prevalence of viraemia >1000 copies/ml were calculated and mapped per municipal ward using ArcGIS software version 10.3 (ESRI, USA). Micro-geographical cluster detection of HIV prevalence and prevalence of viraemia were performed using Kulldorff spatial scan statistic (SaTScan) at a significance level of p<0.05. Results: Based on the HIV viral load, the overall geometric mean viral load for individuals in the study area was 202 copies /ml, in men it was 735 copies/ml and in women it, was 130 copies /ml. In the south-east of the study area, two high viral load clusters were identified. The first cluster accounted for the overall population and the geometric mean viral load for this cluster was 327 copies/ml. The geometric mean viral load for individuals in the population outside of the high viral load cluster was 125 copies/ml resulting in a geometric mean viral load difference of 202 copies/ml (Log-likelihood ratio =18.95, p=0.001). The second-high viral load cluster accounted for women and the geometric mean viral load for this cluster was 237 copies/ml. The geometric mean viral load for women outside the high viral load cluster was 79 copies/ml and the geometric mean viral load difference was 158 copies/ml (Log-likelihood ratio =18.99, p=0.001). Both the high viral load clusters occurred to the south-east of the study area representing a peri-urban setting. A further analysis of the viral load at a threshold of >1000 copies/ml showed that viral load >1000 copies /ml exceeded 50% in 11 of the 30 Page | 2 municipal wards, and 10 of the 11 wards were located within urban areas. A single cluster was identified and was in the north-west of the study area and approximately over a three kilometre radius having a relative risk of 0.69 (p=0.02). A total of 309 HIV positive individuals contributed to this cluster indicating that 69% or 213 individuals had viral load of <1000 copies/ml, whilst 31% or 96 individuals had viral load >1000 copies/ml. Based on the HIV prevalence analysis, a high-prevalence cluster with a relative risk of 1.75 (p=0.02) was identified in the same north-west area and the HIV prevalence in this area was 71%. Overall men compared to women had higher log10 mean viral load (Log10 mean viral load 3.21 vs 2.62; p<0.001) and higher in the age category15-24 years compared to 25-49 years, (Log10 mean viral load 3.34 vs 2.65; p<0.001). Conclusions: The findings of this study demonstrate that applying spatial analysis to the understanding of HIV epidemiology even in a hyperendemic HIV epidemic setting is a valuable tool to monitor the epidemic. Despite the unprecedented high prevalence of HIV within geographically specific areas, the promising finding of the high prevalence of viral load <1000 copies/ml underscore the importance and impact of HIV programmes that have been rolled-out in this community. Furthermore, the high HIV viral load in young women and men in this region play a significant role in sustaining the epidemic, and there is an urgent need to prioritise interventions critical to reducing the potential for HIV transmission.