Theses and Dissertations (School of Health Systems and Public Health (SHSPH))

Permanent URI for this collectionhttp://hdl.handle.net/2263/39602

Browse

Now showing 1 - 20 of 132

Prevalence of silicosis among in-service Zambian copper miners
(University of Pretoria, 2004) Smith, Ferdie; Mwansa, Connard
Objectives To determine the prevalence of silicosis among the in-service Zambian copper miners and identify which areas of operations and mines are mostly affected by silicosis. Design: Cross-sectional point prevalence. Setting: Medical records of miners from Occupational Health Research Bureau Kitwe-Zambia. Subjects and methods: 1122 miners who had worked for more than ten years in the scheduled area were randomly selected for the study from the Bureau records. Data was analyzed in Stata for descriptive summaries associations of silicosis with and independent variables. Results: The prevalence of silicosis was 8 per 1000 miners with Chibuluma mine recording the highest prevalence2.2%(1 out of 27) and with the dust producing areas recording the highest prevalence 6 out of the 9 cases (66.6% ). All the cases were from underground miners and were above 40 years of age. Silicosis was strongly associated with years of service, age of the miners and area of operations (p <0.05). Association was shown between silicosis and dust particles in chi-square test of association, but when put in binary logistic regression there was no association (p 0. 7) probably because dust particle levels was used instead of respirable free silica and the different mines contain different silica levels in their virgin rocks. If respirable free silica levels were used the association could have been there. This can be confirmed by the high prevalence in dust producing areas. Conclusion: The prevalence of silicosis among the in-service Zambia copper mines is low compared to other mining countries. There is a possibility of high prevalence among the retirees since all of them are more than 40years of age and there is strong association between years of service age, and areas of operation with silicosis.
Strengthening occupational health systems and services for health workers during the COVID-19 pandemic and beyond : the role of occupational health and safety information systems
(University of Pretoria, 2024-07-29) Voyi, Kuku; Jerry, Spiegel; Yassi, Annalee; muzimkhulu.zungu@up.ac.za; Zungu, Laszchenov Muzimkhulu
Introduction The World Health Organization’s (WHO) global plan of action on workers’ health urged member states to work towards full coverage of all workers with essential interventions and occupational health (OH) services to support primary prevention of occupational and work-related diseases and injuries. Globally, health workers (HWs) face a range of occupational hazards, especially in low- and middle-income countries (LMICs), and have limited access to OH services especially in LMICs. The calls by the WHO, International Labour Organization (ILO), International Commission on Occupational Health (ICOH) and other stakeholders, for OH systems to support needed OH services for HWs became more urgent when the COVID 19 pandemic struck. Some of the recommended OH interventions for HWs include the use of “HealthWISE”, a tool jointly developed by the WHO and ILO to support participatory action to improve workplace health, and occupational health and safety information systems, to help plan, guide and evaluate the various measures implemented. Both these interventions have been piloted and used to improve OH in South Africa. Strengthening OH systems is essential to the attainment of a healthy and safe workforce in the health sector, especially during the COVID-19 pandemic. This study aimed to assess the extent to which the introduction of HealthWISE tool along with a specific information system referred to as Occupational Health and Safety Information System (OHASIS), could strengthen OH systems in South African health settings during the COVID-19 pandemic and potentially beyond, through greater emphasis on how information was systematically collected and implemented in the workplace. Methods This was a quasi-experimental study utilising a mixed methodological approach, based on the principles of participatory action research to enable action and empower active participation in strengthening and evaluating OH. It assesses changes in OH systems (policy, leadership, and coordination; financing; human resources; infrastructure, technology and medicines; information management; and services) in health facilities, following the concurrent introduction of HealthWISE and OHASIS. Using Poisson regression models, it also assesses COVID-19 infection rates in relation to HealthWISE compliance scores in participating health facilities. Results Our baseline survey for HealthWISE, found that health facilities in all provinces had SARS-CoV-2 response routine and emergency plans for the general population but no comprehensive OH plan for HWs. Some provinces had an OH SARS-CoV-2 provincial coordinating team and a dedicated budget for OH, an occupational medical practitioner and an OH nurse, a functional health and safety committee, and health risk assessments specific to SARS-CoV-2. However, none of the assessed health facilities had an acceptable HealthWISE compliance score (≥ 75%) due to poor ventilation and inadequate administrative control measures. In addition, our analysis showed that hospitals with higher compliance scores had significantly lower infection rates (IRR 0.98; 95% CI: 0.97, 0.98). In facilities where HealthWISE had been implemented, we found that HWs adherence to administrative and personal protective equipment declined over time as the COVID-19 pandemic restrictions were relaxed, while the ventilation did not change, as it was poor from the baseline. Our Poisson regression model showed that higher HealthWISE administrative measures score in an area was associated with a reduction of the infection rate and higher HealthWISE ventilation measure score was also associated with a reduction of the infection rate. For the baseline OHASIS study, there were 71 participants comprised of hospital managers, health and safety representatives, trade unions and OH and safety professionals. At least 42% reported poor accessibility and poor timelines of OH information systems for decision-making. Only 50% had access to computers and 27% reported poor computer skills. The existing OH information systems were poorly organized and needed upgrades, with 85% reporting the need for significant reforms. Only 45% reported use of OH information systems for decision-making in their OH and safety role. Following numerous attempts to implement OHASIS, we were unable to do so within the study period. The observed barriers for the establishment, provision, maintenance and sustaining OH and IPC, particularly our OHASIS intervention for HWs during the COVID-19 pandemic was a lack of leadership or stewardship in OH; lack of interest in the use of effective intelligence systems for decision making; ineffective health and safety committees; inactive trade unions; untrained and overworked OH professionals, among others. Conclusions Despite some initial preparedness, greater effort to protect HWs is still warranted. LMICs may need to pay more attention to occupational safety and health systems and consider using tools, such as HealthWISE to protect HWs’. HealthWISE, similar to other health strengthening measures, requires commitment and motivation of stakeholders, however if implemented, there are positive implications for OH. The significant association between infection rates and HealthWISE compliance scores is notable and merits further attention. Given the persistence gaps in access to and utilization of information needed to protect the rights of HWs to a safe workplace in South Africa and beyond, more attention is warranted to the systems that support data management and information implementation in this setting, including OH information system education and training.
Indoor and outdoor air pollution health risks in Mabopane and Soshanguve townships in the City of Tshwane, South Africa
(University of Pretoria, 2024-05-02) Shirinde, Joyce; Wichmann, Janine; bhuda12@hotmail.com; Bhuda, Mandla Freddy
EXECUTIVE SUMMARY Background: Air pollution is a major threat to human health. Low – and middle-income countries are faced with many challenges, such as urbanisation and modernization, industrialization, traffic density, and an increase in the human population. Children are uniquely vulnerable to air pollution due to their physiologic characteristics (i.e., narrow airways and immature immune systems). Approximately 7 million people die worldwide annually due to exposure to indoor and outdoor air pollution. In South Africa, more than 28,000 people die annually due to exposure to air pollution. Therefore, air pollution is responsible for about R94.7 billion in economic costs every year. There is strong evidence that the majority of communities in the City of Tshwane Metropolitan Municipality are exposed to high levels of pollution during the winter and daily during rush hours, mostly in residential areas. Aim: To evaluate indoor and outdoor air pollution health risks at Mabopane and Soshanguve townships in the City of Tshwane, South Africa. Method: The study was conducted in Mabopane and Soshanguve Townships in the City of Tshwane Metropolitan Municipality, Gauteng Province. A cross-sectional study design was applied following the International Study on Asthma and Allergies in Childhood (ISAAC) Phase III. A total of 1844 parents and guardians of preschool children completed the modification. However, a total of 4 questionnaires were discarded due to incorrect completion by the participants. As a result, a total of 1840 questionnaires were included in the statistical analysis. Ambient PM2.5 samples were collected on the rooftop of the Mabopane fire station from 04 May 2022 to 28 February 2023, over 24 hours every six (6) days. In total, 60 PM2.5 samples were collected, including 10 duplicates. PM2.5, soot, black carbon (BC), and organic carbon (OC) levels were measured using gravimetric techniques, reflectometers, optical transmitters, energy-dispersive X-ray fluorescence, and Hybrid Single-Particle Langrangain Integrated Trajectory software. The United States Environmental Protection Agency’s (US EPA) human health risk assessment (HRA) model was followed to conduct the health risk assessment of ambient PM2.5 and trace elements. Results and discussion The study shows that the prevalence of severe asthma symptoms, asthma, rhinoconjunctivitis, eczema (EE) ever, and current eczema symptoms (current ESs) among preschool children in Mabopane and Soshanguve was 15.4%, 6.8%, 18.5%, 11.9%, and 13.3%, respectively. The use of gas for cooking or heating significantly increased the risk of current severe asthma symptoms among preschool children by 20% (CI: 2.08 – 4.91; p-value < 0.001). Furthermore, the use of open-fire sources (paraffin, wood, or coal) increased the risk of severe asthma symptoms among preschool children by 87% (CI: 0.98 – 3.55; p-value = 0.057). The use of gas and open fire sources (paraffin, wood, or coal) increased the likelihood of asthma (OR = 3.69; 95% CI: 1.22 - 11.2) and allergic rhinoconjunctivitis (OR = 2.48; 95% CI: 1.55 - 3.96). Trucks passing near homes almost the whole day during the weekdays increased the likelihood of allergic rhinoconjunctivitis (OR = 1.31; 95% CI: 0.72-1.38). Environmental tobacco smoke (ETS) exposure at preschool increased the likelihood of asthma (OR = 2.11; 95% CI: 1.49 – 2.97). ETS exposure at home increased the likelihood of allergic rhinoconjunctivitis (OR = 1.95; 95% CI: 0.52 - 7.38). The use of open fires (paraffin, wood, or coal) increased the likelihood of EE (OR = 1.63; 95% CI: 0.76 – 3.52) and current ESs (OR=1.94; 95% CI: 1.00 – 3.74). Additionally, ETS exposure at home increased the likelihood of EE (OR = 1.66; 95% CI: 1.08 – 2.55) and current ESs (OR = 1.61; 95% CI: 1.07 – 2.43). Lastly, mothers or female guardians smoking cigarettes increased the likelihood of EE (OR = 1.50; 95% CI: 0.86 – 2.62) and current ESs (OR = 1.23; 95% CI: 0.71 – 2.13). Parents should be informed about the health risks associated with exposing their children and community to smoking. The National Department of Health should speed up the process of promulgating indoor air pollution standards. South African road management policies should encourage integrated transport to decrease vehicle usage and promote lift clubs to reduce road traffic. The mean levels of PM2.5, soot, BC, and OC were 10.4 μg.m-3, 1 m−1x 10–5, 0.9 μg.m-3, and 1.0 μg.m-3 respectively. The 24-hour mean PM2.5 exceeded the WHO air quality guidelines (5 μg.m-3) but did not exceed the 24-hour annual South African National Ambient Air Quality Standards (SANAAQS) (20 μg.m-3), respectively. There was no significant seasonal variation of PM2.5, with the highest mean observed in spring (13.1 μg.m−3) (p > 0.05). However, there was a temporal variation of PM2.5 on weekdays and weekends, with the highest levels observed on weekdays (11.8 μg.m−3) (p < 0.05). Six geographic air masses were observed: Cluster 1: Southeastern Indian Ocean (23%); Cluster 2: North Limpopo (32%); Cluster 3: Northern Cape (14%); Cluster 4: Short Eastern Indian Ocean (23%); Cluster 5: Long Southeastern Indian Ocean (6%); and Cluster 6: Long Southwest Atlantic Ocean (3%). A total of 19 elements were determined: Ag, Ba, Br, Ca, Cl, Cu, Fe, K, Mn, Ni, P, S, Sb, Si, Sr, Ti, U, V, and Zn. The trace element that recorded the highest mean level for the entire sampling campaign was Fe (243 ng.m−3), followed by Si (222 ng.m−3), S (158 ng.m−3), Ca (107 ng.m−3), and K (84 ng.m−3), with no significant seasonal variation during the dry and wet seasons (p > 0.05). Fe and Si are associated with soil dust, while S and Ca are linked to industrial and construction air, such as power generation industry and construction dust. The Hazard Quotient (HQ) for total PM2.5 for adults, children, and infants was 2.1, 11.1, and 20.1, respectively, exceeding the WHO benchmark (5 μg.m−3). The HQs for trace elements S and Ti were also above 1 for infants. The Ni excess cancer was 1.2 x 10-6, and the exposure to Ni was a risk factor for cancer throughout the year. According to the HYPLIT model, air pollution in Mabopane is affected by several provinces, namely Mpumalanga, Limpopo, Northern Cape, Eastern Cape, and Free State. Additionally, the pollution is influenced by both long- and short-range factors in the Indian Ocean. Some of these provinces are known for their air pollution sources, including mining activities, coal power plants, and biomass combustion. Therefore, the South African government should strengthen air quality management to prevent transboundary air pollution. The City of Tshwane Metropolitan Municipality should enforce air quality bylaws that include vehicle emission tests to penalise road users who do not comply with the air quality legislation. Conclusion and Recommendations: The study has found that exposure to both outdoor and indoor air pollution sources can increase the risk of asthma, severe asthma symptoms, allergic rhinoconjunctivitis, and atopic eczema among preschool children aged 7 years old and below in Mabopane and Soshanguve. The sources of indoor and outdoor air pollution include environmental tobacco smoke, household fuel use (such as gas, wood, paraffin, and coal), and vehicle emissions (specifically from trucks). The study also found that although the levels of outdoor PM2.5 did not exceed the South African National Ambient Air Quality Standards (SANAAQS), they exceeded the World Health Organisation (WHO) guidelines. The health risk assessment showed that even exposure to PM2.5 levels below the SANAAQS poses a risk to adults, children, and infants in Mabopane. The study suggests that the current SANAAQS for PM2.5 needs to be reviewed to better protect human health. The study also highlighted the carcinogenic hazard of nickel (Ni), which poses a risk throughout the study period. This study contributes to the understanding of health complications associated with exposure to indoor and outdoor air pollution in the communities of Mabopane and Soshanguve. It also provides knowledge about the air distance sources contributing to air pollution in Mabopane, thus providing scientific evidence for South African policymakers to promulgate indoor and household air pollution legislation and strengthen the current air pollution legislation. This study serves as a baseline for further epidemiology studies among preschool children aged 7 years old and below. The study recommends conducting epidemiological studies to understand the burden of respiratory diseases and their association with specific air pollutants in South African townships, as well as a source apportionment study to understand the sources of air pollution in Mabopane. Keywords: Air pollution, PM2.5, Preschool, Health risk assessment, HYSPLIT, Trace elements, Mabopane, Soshanguve, South Africa.
Vector biting behaviour, community malaria infection risk, and field trial on repellent footwear in southern Mozambique
(University of Pretoria, 2024) Braack, LEO; Riddin, Megan; u17326339@tuks.co.za; Salome, Graca
Introduction: Effective control of malaria vectors requires targeting mosquitoes indoors and outdoors. The core interventions currently recommended by the World Health Organization are for large scale deployment of ITNs and IRS. These tools are deployed indoors and have been shown to be effective in protecting against mosquito bites and reducing malaria transmission across many settings. The emergence and spread of resistance to insecticides by mosquito vectors and the shift in vector populations towards those that feed outdoors and early in the evening when people are not protected by LLINs and or IRS, represent a substantial shortfall in protection. To address the outdoor biting and resistant vector populations, innovative interventions with new or existing tools need to be implemented. The efficacy of a repellent-impregnated footwear designed and produced as a controlled slow-release repellent system was tested for efficacy in the field and were contextualized with knowledge, attitude, and practice (KAP), and prevalence surveys, as well as a comprehensive vector survey. Methods: For the mosquito vector survey, human landing catches were conducted over a 9-month period. Morphological and molecular identification, and assays for detection of Plasmodium circumsporozoite protein were performed. A cross-sectional design was used for the KAP survey and prevalence surveys. For the KAP survey, data were collected through questionnaires, while RDTs and microscope examination of finger-prick blood samples were used for population malaria infection rates. A randomized controlled field trial was conducted to determine the efficacy of the repellent-impregnated footwear. The number of collected mosquitoes was compared between groups and the protection efficacy was determined. Results: Seventy-six Anopheles arabiensis and 117 Anopheles funestus s.s. were identified among the 1802 Anopheles mosquitoes collected. One mosquito from each species was found harbouring Plasmodium falciparum circumsporozoite protein. Anopheles arabiensis was biting equally indoors and outdoors and early in the evening (χ2 = 0.48, df = 1, P = 0.53) while Anopheles funestus s.s. was biting more indoors and late at night (χ2 = 31.84, df = 1, P < 0.001). All the infected mosquitoes were collected biting outdoors, suggesting that transmission may be occurring outdoors. The KAP and prevalence surveys revealed that the level of knowledge (89.7%) and practices (89.7%) toward malaria are good, and the attitude (93.7%) is positive among the study participants. The prevalence of malaria was 11.9%. Residing in Conhane neighbourhood (aOR = 16.01, 95% CI 1.87-137.33, P = 0.011) and having a well as the source of water (aOR = 11.82 (95% CI 1.17-119.91, P = 0.037) were identified as risk factors for malaria infection. The repellent-impregnated footwear (mean rank = 214.44) did not show efficacy in reducing Anopheles mosquito bites when compared with the non-impregnated footwear (mean rank = 228.81) (adjusted P = 1). However, the footwear either impregnated or not reduced mosquito bites when compared with a bare foot (mean rank = 287.26) (adjusted P < 0.0001), suggesting that a physical barrier effect is present. Conclusions: This study brought evidence of Plasmodium falciparum infected mosquitoes biting outdoors early in the evening and late at night. This should inform the NMCP to include interventions for malaria vector control outdoors to improve vector control in the village. The KAP and prevalence surveys contributed to reinforcing the importance of the association between living conditions and the higher risk of malaria infection. Improvement in provision of protected sources of water should be considered toward contributing to decreasing risk of malaria infection in Conhane Village. The field trial suggests that the footwear was effective in reducing mosquito bites through a physical barrier effect as opposed to that of a repellent effect. Protection of the lower legs and feet against mosquito bites can be achieved by covering these areas with footwear or clothing. Further studies with repellent-based tools to target the lower limbs should be considered as previous testing of fibres impregnated with repellents were successful.
Task sharing in mental health service provision : developing a model for clinical associates in South Africa
(University of Pretoria, 2023-12-04) Wolvaardt, Jacqueline Elizabeth (Liz); Grobler, Christoffel; saiendhra.moodley@up.ac.za; Moodley, Saiendhra Vasudevan
Background There is a shortage of the human resources needed to deliver mental health services in South Africa. Clinical associates are possibly an under-utilised resource in mental health task sharing approaches in South Africa. The study aimed to develop a model of task sharing in mental health in South Africa focussed on clinical associates. Methods and results by objective Objective 1: To describe the mental health content of the three clinical associate training programmes in South Africa A collective case study approach was utilised involving the three universities offering undergraduate clinical associate degrees. In-depth interviews using videoconferencing were conducted with individuals involved in each programme and documents such as study guides and timetables were reviewed. We found that mental health was included in the curricula and assessments of all three programmes. The approach to facility-based training was different at the three universities with one adopting a practical approach at a hospital with a mental health unit, the second a more theoretical approach with limited practical exposure, and third not have a universal approach as there was considerable variation between facilities. Objective 2: To determine knowledge, attitudes and practices of clinical associates with respect to management of mental illness A cross-sectional study of clinical associates based in South Africa was conducted. The questionnaire incorporated the 16-item Mental Illness Clinicians’ Attitudes scale version 4 (MICA-4) as well as questions related to knowledge, confidence, practices, and interest in mental health service provision and further training. Only 50.3% of participants felt ‘quite confident’ or ‘very confident’ taking a mental health history and even fewer (43.2%) in carrying out a mental health examination. The mean MICA-4 score recorded was 37.55 (SD 7.33) which is at the lower end of the scale indicating less stigmatising attitudes. There was considerable interest in mental health work (83.8%) and in a specialisation in mental health (66.5%). Objective 3. To describe the attitudes of health managers, medical doctors and nurses towards mental health task sharing involving clinical associates Focus group interviews of medical doctors and nurses involved in mental health service provision, and health managers were conducted in four districts of South Africa.The participants felt that the performance of clinical associates in other disciplines suggests their potential usefulness in mental health but there are barriers and constraints that needed to be addressed. Objective 4: To identify the key elements of a mental health task sharing model for clinical associates The Delphi method was utilised to reach consensus on the key elements of a model for mental health training and service provision. The Delphi panel consisted of family physicians and psychiatrists from the public and private sectors in South Africa. The panel reached consensus on 10 of the 21 tasks provided that could be performed based on undergraduate training and 20 of the same 21 provided tasks for those with a postgraduate qualification in mental health. Conclusion Based on our findings, clinical associates have a potentially important role to play in addressing the mental health crisis in South Africa. They could help improve access to mental health services in the public sector particularly in rural areas. The proposed model comprises training and service provision components and highlights the policy and regulatory changes that are needed to enable mental health task sharing and optimise the contribution of clinical associates to the mental health system.
Towards development of a novel approach for enhancement of TB diagnostic services during the pandemic : a case of primary health care clinics in eThekwini District KwaZulu-Natal
(University of Pretoria, 2023) Mashamba-Thompson, Tivani; Musekiwa, Alfred; thobekadlangalala@gmail.com; Dlangalala, Thobeka Nomzamo
Tuberculosis (TB) is an important public health issue in South Africa that has burdened health systems for decades. The past decade has seen progress in the management of this disease. However, the advent of the coronavirus disease 2019 (COVID-19) has disrupted the provision of essential TB services, resulting in TB detection dropping in the early phase of the pandemic. This has subsequently affected TB incidence and mortality. Therefore, in addition to recovery plans, TB services require strengthening to withstand future health crises. As such, this study aimed to generate evidence to inform a novel approach for improving TB diagnostic services in high-burden settings using the eThekwini district in KwaZulu Natal (KZN) Province, South Africa, as a study setting. Methods The study employed a multiphase mixed methods study design consisting of four phases. Initially, a scoping review was conducted to gather the available evidence on TB services during the COVID-19 pandemic. The findings were used to inform the objectives for the rest of the study. During the first phase, a geospatial analysis was conducted to calculate the geographic accessibility of TB diagnostic services at primary healthcare clinics in eThekwini district. The second phase consisted of a quasi-experimental study that determined the impact of COVID-19 on TB diagnostic services. The barriers and facilitators to providing quality diagnostic services were explored for the third phase through a facility audit and patient interviews. In the final phase, a Nominal Group Technique (NGT) was conducted with relevant TB stakeholders to develop an approach for enhancing TB diagnostic services during the pandemic. Together, the study’s findings were synthesized and used to inform a framework to improve and strengthen the quality of TB diagnostic services. Results The scoping review revealed that the COVID-19 pandemic severely impacted TB detection due to various factors, including limited access to facilities. This prompted recommendations that could facilitate better service provision amid the pandemic. The study’s geographic access evaluation determined that diagnostic services were highly accessible to most (92.6%) of the vii eThekwini population. The areas of poor accessibility mainly consisted of the rural population. Moreover, the analysis determined that many TB cases were in urban and sub-urban regions. The study also found that the impact of COVID-19 on TB detection was severe during the lockdown, showing 45% and 40% decreases in TB investigations and confirmed cases, respectively. These indicators recovered when lockdown measures were lifted. However, the peaks of SARS-CoV-2 variant-driven infection resulted in overall decreases in confirmed cases of TB. The assessment of the quality of TB diagnostic services revealed that many IPC aspects and continuous TB training were lacking at facilities, in addition to long turnaround times for GeneXpert results. Patients perceived long wait times, staff attitudes, and drug stockouts as barriers to quality services. Lastly, stakeholders identified key barriers to diagnostic services during the pandemic and developed an approach to overcome them. They suggested integrating TB/COVID-19 activities, continuous training among staff, strengthening IPC, decentralizing TB testing, using Point-of-Care tests (POC) and raising public awareness through social media platforms to enhance diagnostic services. Conclusion The present study has successfully developed a novel approach for enhancing tuberculosis (TB) diagnostic services at Primary Health Clinics (PHCs) in high-burden regions using the eThekwini district as a study setting. The approach involved devising a consolidated framework for providing high-quality diagnostic services, which is informed by the evidence generated in the thesis. The framework provides guidance on improving structural factors such as accessibility, infection prevention and control (IPC), and care processes, including continuous training and service integration. This comprehensive approach has the potential to improve service delivery and boost public confidence in the health system, ultimately leading to better health outcomes.
Association between temperature variability, cause-specific mortality, hospital admissions, and synoptic weather types in South Africa
(University of Pretoria, 2023-09-30) Wichmann, Janine; Rautenbach, Hannes; malebomakunyane@gmail.com; Makunyane, Malebo Sephule
Non-communicable diseases (NCDs), such as cardiovascular and respiratory diseases are weather-sensitive diseases that are regarded as a cause of premature death globally. As much as 85% of NCDs occur in low- and middle-income countries, such as South Africa. Approximately 57% of all the deaths that occurred in South Africa in 2017 were attributable to NCDs, specifically, deaths that were caused by diseases of the circulatory system (18.4%) and deaths caused by diseases of the respiratory system (9.5%). However, epidemiological understanding of the health effects of climate indicators and variability in South Africa is scarce and sparse. Evaluating the relationship between key climate indicators or meteorological variables and cause-specific health outcomes can assist in identifying the potential health risks associated with climate change, developing strategies to mitigate the associated risks, and informing policies that will reduce the burden of diseases and environmental sustainability. Climate change does not only cause an increase in average temperatures, but also the variability of temperatures. To the author's knowledge, there are no studies conducted in Africa that assessed the combined effects of inter- and intraday temperature variability (an important meteorological indicator of climate variability) on morbidity and mortality. Correct exposure-response relationships are required to optimally assess the burden of diseases and health outcomes due to temperature variability (TV). This thesis aims to address the following research questions: • What is the current state of knowledge or literature on the association between climate variability indicators and cause-specific health outcomes? • Are there any associations between TV and hospital admissions and mortality? • Which subpopulations are most likely to be hospitalised or die from exposure to TV? • Do season and spatial synoptic classification (SSC) weather types (i.e., an entire suite of meteorological indicators) modify the association between TV and health outcomes? The primary research objectives of this study were answered through a series of papers (manuscripts), of which five manuscripts were identified to address these research questions. The objective of Manuscript 1 was to systematically review the current state of knowledge of time series and case-crossover epidemiological studies on the impact of TV on cardiovascular disease (CVD) and respiratory disease (RD) mortality as well as hospital admissions at various time scales. Manuscript 2 investigated the association between TV, RD, and CVD hospital admissions in Cape Town during 2011–2016. Manuscript 3 investigated whether season and SSC weather types modified the association between TV and CVD and RD hospital admissions in the City of Cape Town from 2011 to 2016. Manuscript 4 examined the association between TV, RD, and CVD mortality in five South African cities, namely Bloemfontein, Cape Town, Durban, Johannesburg, and Gqeberha (formerly known as Port Elizabeth) located in different Kӧppen-Geiger regions in South Africa during the study period 2006–2016. Manuscript 5 investigated the association between TV and cardiorespiratory (total CVD and RD) mortality in the afore-mentioned five cities from 2006–2016 and assessed whether season and SSC weather types modified the association between TV and mortality. Manuscript 1 was achieved by following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to systematically review the current literature on the health effects of temperature variation. A literature search was conducted using PubMed, MEDLINE/ Web of Science, and Scopus. The review focused on time series and case-crossover epidemiological studies that investigated the association between TV and CVD and/or RD morbidity and mortality and reported the empirical results. The review was restricted to English studies published before the 30th of March 2023. Studies that only reported counts of hospital visitations or mortality and those that focused on outpatients or excess risks were excluded from the review. The exposure variable of interest for this thesis was defined as a composite of intra- and interday variability using the minimum and maximum temperatures. For example, the TV for the three preceding days’ exposure before the outcome was calculated as TV0-2 = standard deviation (minimum and maximum temperature on the same day as exposure, minimum and maximum temperature a day after exposure, minimum and maximum temperature two days after exposure). To achieve the aim of Manuscripts 2, 3, 4, and 5, time series analysis using quasi-Poisson generalised linear models combined with a distributed lag nonlinear modelling framework was used. The models were adjusted for the potential confounding effects of the day of the week as a categorical variable, public holidays as a binary variable, and seasonal and long-term trends as a nonlinear spline. Furthermore, a meta-analysis was used for Manuscript 4 to combine city-specific effect estimates to obtain the national effects of TV on mortality. In addition to achieving the aims of Manuscripts 3 and 5, the SSC framework was used to classify each day of the study period into one of the seven distinct categories, namely Dry moderate [DM], Dry polar [DP], Dry tropical [DT], Moist moderate [MM], Moist polar [MP], Moist tropical [MT], and Transition [TR]. Furthermore, modification of the TV and health outcomes association was investigated either by adding an interaction term of TV and SSC weather types to the regression models (Manuscript 3) or through stratification (Manuscript 5). In addition, stratification by age, gender, and season was performed for all the Manuscripts to identify vulnerable subgroups and determine whether hospital admissions or mortality risks differed by season. All regression analyses were performed with R software version 4.2.3 (2023-03-15 ucrt). For Manuscript 1, of the 1 465 identified studies, only 45 met the inclusion criteria of which 23 studies investigated the association between TV indexes and CVD and RD hospital admission and the remaining 22 studies focused on CVD and RD mortality risks associated with TV. Most of these studies were conducted in Asia (particularly China), the United States, and Europe, with only two conducted in Africa. The studies investigated various TV indexes, including diurnal temperature range (DTR), temperature change between neighbouring days (TCN), and standard deviation between the minimum and maximum temperature. In most of the studies, children, people ≥ 65-years of age and males had higher risks of hospitalisation or dying from CVD and RD diseases due to exposure to TV. For Manuscript 2, in general, a positive and statistically significant association between TV and CVD and RD hospital admissions was observed in Cape Town, even after controlling for the confounding variables. For the entire study group, TV showed the greatest effect at short exposure days, at 0-2 days for CVD and 0-1 days for RD hospitalisations. The 15-64 age group had the highest risk for CVD hospitalisation, with the effects reaching a peak at 0-3 days of exposure (3.02%, 95% CI:0.86%-5.23%). For RD hospitalisations, the highest risks were observed for the elderly, with the effects reaching a peak at 0-4 days of TV exposure (5.18%, 95% CI:2.03%-8.43%). In general people ≥65-years had higher risks of RD hospitalisation and the 15-64-years age group was more likely to be hospitalised due to CVD-related diseases. Males had higher risks of hospitalisations compared to females. For Manuscript 3, season and SSC weather types modified the association between TV and hospital admissions in Cape Town from 2011 to 2016. Specifically, according to the seven SSC types, a stronger TV-CVD association occurred during MT weather conditions, with the most significant risks observed after four days of exposure (6.72% [95% CI:1.95%, 11.71%]). Immediate risks of TV on RD appeared with DM conditions (3.35% [95% CI:0.92%, 5.83%]). Grouping SSC categories by similar temperature and humidity characteristics (e.g., both tropical types), a modified association was observed, especially for the 15–64-year age group during tropical conditions for the CVD admissions. The 0-14- and 15-64-year age groups for RD hospital admissions appeared more vulnerable during dry and transitional weather conditions, respectively. For Manuscript 4, a positive association between TV and mortality was observed in the five cities. Similar to Manuscript 3, the effect estimates were higher for RD health outcomes (mortality) as compared to CVD mortality. The pooled estimates showed the highest RD mortality risks were associated with a decrease of RR=1.21 (95% CI:1.04; 1.38) from the 5th to the 50th percentile in TV at 0-2 days for all ages combined. The elderly appeared more vulnerable to RD mortality than <65 years age group, with significant mortality risks per increase in TV at 0-2 days (RR=1.18, 95% CI:1.04;1.32), 0-3 days (RR=1.16, 95% CI:1.04;1.28) and at 0-7 days (RR=1.12, 95% CI: 1.02;1.22) from the 50th to the 75th percentile. No significant results were observed for CVD mortality. For Manuscript 5, The pooled estimates showed the highest and significant increase in RD mortality of 1.21(95% CI:1.04;1.38) per an increase in TV at 0-2 days from the 25th to the 50th percentile for all ages combined. The elderly appeared more vulnerable to RD mortality than <65 years age group, with significant mortality risks per increase in TV at 0-2 days (RR=1.18, 95% CI:1.04;1.32), 0-3 days (RR=1.16, 95% CI:1.04;1.28) and at 0-7 days (RR=1.12, 95% CI: 1.02;1.22) from the 50th to the 75th percentile. A stratified analysis showed the elderly and women as more vulnerable. For the entire study population, greater association was observed during MT weather conditions in Bloemfontein (1.87 95% 0.96-3.63), DT weather conditions in Cape Town (1.27; 95% CI0.83-1.95), MM weather conditions in Johannesburg (1.35; 95% CI:0.51-3.56) and during TR weather conditions in Durban (1.17; 95% CI0.63-2.19) and Gqeberha (1.35; 95% CI:0.51-3.56). This thesis demonstrated that climate change and variability are a huge public health threat in South Africa. Specifically, the results of this thesis showed that TV is an important risk factor for cardiorespiratory health outcomes in South Africa. Furthermore, it demonstrated that season, synoptic weather types, and age modify the association between TV and health outcomes. Application of the SSC scheme can provide insights into how atmospheric circulation patterns and synoptic weather types affect temperature and health outcomes. The occurrence of individual air masses can be forecasted, providing helpful information for planning public health resources. The results of the present study imply that the synergistic effects of other environmental and meteorological factors should be considered when conducting risk assessments for health and climate-related research. This thesis advances the knowledge of mortality risk factors in South Africa.
An Afrocentric approach to peer mentoring in a decolonised curriculum for first-year Clinical Associate students at the University of Pretoria
(University of Pretoria, 2023-12-14) Wolvaardt, Jacqueline Elizabeth (Liz); Du Toit, Pieter Hertzog; corne.nel@up.ac.za; Nel, Corne
Introduction: First-year Clinical Associate (ClinA) students face more challenges than the average first-year student as they are educated in decentralised healthcare facilities predominantly in rural areas, where institutional support is not always available. Another concern is throughput rate – only 63% of BCMP students graduate in the minimum time of three years, while in comparison the throughput rate for medical students is 89 %. Their educational strategy is different from traditional strategies, incorporating authentic learning, self-directed learning, Whole Brain® learning and collaborative learning. The absence of existing literature on mentoring programmes tailored to students who navigate much of their academic experience without substantial access to university student support services underscored the necessity for a tailor-made approach. Peer mentoring was identified as an option, especially for ClinA students who do not have access to academic staff or campus-based support structures. This research study unfolded in multiple phases, with the overarching aim being the development, implementation, and evaluation of a peer mentoring programme expressly tailored for first-year undergraduate ClinA students. This aim was achieved through the formulation of a Short Learning Course (SLC) designed for the training of peer mentors within an Afrocentric, decolonised context. Methods: A concurrent, embedded mixed methods study design was used for the multi-phased study. In Phase 1, the exploration of ClinA students' needs and challenges utilized a phenomenological research approach. Two qualitative methodologies, appreciative inquiry interviews and focus group discussions, were employed for this purpose. Phase 2 was done using a modified nominal group technique (mNGT) to establish group consensus on the learning outcomes and activities suitable for inclusion in the short learning course designed for peer mentor training. Following the mNGT, the curriculum was developed using Biggs’ constructive alignment as the design principle. Phase 3 of the study adopted an action research methodology. Action research is characterized by the inclusion of participants' feelings, perspectives, and patterns without external control or manipulation by the researcher. The procedural steps associated with action research were adhered to, including planning (identifying the need), acting (developing and implementing the SLC), observing, reflecting, and evaluating the SLC. Results: The study revealed themes across various aspects of ClinA students’ environment. Challenges faced by ClinA students encompassed concerns about the teaching and learning strategies employed by programme facilitators, the decentralised nature of learning platforms, and a lack of institutional support at clinical learning centres. Benefits of peer mentoring were identified, including mentors serving as proxy representatives of institutional support, offering all-encompassing support, and functioning as advocates and role models. The perceived enablers of a peer mentoring programme's success included identifying individuals with suitable characteristics for peer mentoring, comprehensive training for peer mentors, and crucial support from the faculty. Conversely, perceived constraints involved the shortage of human and financial resources, time limitations, and concerns about the commitment levels of peer mentors. The efficacy of the modified Nominal Group Technique process was shown, emphasizing its ease of facilitation and cost-effectiveness. Learning outcomes and activities for a SLC for training of peer mentors, were carefully formulated, with a focus on academic support, guidance, and psychosocial assistance. The SLC curriculum emerged as a comprehensive framework aligning learning outcomes with objectives for peer mentors, emphasizing personal and academic development, effective communication, teamwork, and professional role modelling. The unintended alignment of identified outcomes with institutional graduate attributes reflects the programme's holistic nature, aiming to cultivate graduates with academic, clinical, and societal impact skills. The curriculum's commitment to decolonisation principles showcased practical applications, promoting cooperation, diverse representation, and shared knowledge benefits. Additionally, Whole Brain® learning was integrated, encouraging diverse thinking approaches among peer mentors. Participant perspectives, both quantitative and qualitative, provided valuable insights. While there was a shared understanding of programme goals, a consensus emerged that regular meetings were lacking. Qualitative themes encompassed the impact of strong relationships, challenges faced by peer mentor programme participants, and the advantages mentioned. Afrocentricity served as the philosophical paradigm for this study with incorporating the African canons, ujamaa (familyhood/community), utulivu (justice/peacefulness), kujitoa (dedication/commitment), uhaki (harmony/justice), and ukweli (truth) as standards of judgement. Conclusion: This study embarked on a comprehensive journey to address the challenges of low throughput rates within the BCMP programme, specifically focusing on first-year ClinA students. Through an exploration of the unique needs and challenges of ClinA students in South Africa and an examination of the enablers and constraints of an effective peer mentoring programme, the study laid the groundwork for a transformative initiative. The next step was the development and implementation of a tailor-made online short learning course (SLC) for training peer mentors, rooted in decolonised curriculum principles and an Afrocentric lens as conceptual framework. The SLC achieved its objectives of cultivating personal and academic development, effective communication, teamwork, and professional role modelling. The structured outline of preparation, application, and consolidation phases contributed to the formation of well-rounded peer mentors capable of fostering a supportive environment for their mentees. Moreover, the study showcased the practical application of Le Grange and Makhele's 4Rs in evaluating educational programmes, emphasizing relational accountability, respectful representation, and reciprocal appropriation. The SLC not only contributed to individual growth but also instilled a culture of inclusivity and adaptability within the educational landscape. Overall, this research provides valuable insights and practical applications for enhancing peer mentoring programmes, specifically tailored for the unique context of ClinA students in the BCMP programme, setting a foundation for a more interconnected, diverse, and mutually beneficial educational environment. Embracing ubuntu and the African Canons in our ethical framework challenges existing perspectives, fostering an inclusive and compassionate society. This shift, coupled with an Afrocentric research approach, contributes to a more just and equitable future through heightened awareness of interconnectedness.
Discrete choice experiments in medical education : the role of management, health economics and research
(University of Pretoria, 2023-12-15) Wolvaardt, Jacqueline Elizabeth (Liz); Ryan, Mandy; astrid.turner@up.ac.za; Turner, Astrid Chrisilda
In this thesis, Discrete choice experiments in medical education: the role of management, health economics and research, the candidate utilised the stated preference technique—discrete choice experiment — to elicit the preferences of medical doctors regarding the allocation of their time between clinical training and training in management, health economics, and research in the medical curriculum. The findings suggest a preference among doctors to undergo formal training in self-management, the management of others, and the utilization of research skills, demonstrating their willingness to give up clinical training time to acquire these skills. Specifically, participants expressed a readiness to allocate 4.46 hours per week for management training and 4.38 hours per week for research training. By leveraging the insights derived from doctors' preferences with an attributes-based method, medical training can align with the expressed needs and priorities of doctors in a real-world context, ensuring a more effective and responsive educational framework.
The value of house screening as an addition to long-lasting insecticidal nets in protecting against malaria in Zambia
(University of Pretoria, 2024-01-30) Mutero, Clifford Maina; De Jager, Christiaan; kochelani.saili@gmail.com; Saili, Kochelani
The aim of this study was to evaluate the impact of adding house screening to long-lasting insecticide-treated net (LLINs) use on malaria vector densities and malaria transmission potential in rural southeast Zambia. The study was conducted in Nyimba district in four phases. First, baseline information on malaria vector species composition, relative abundance, sporozoite infectivity and entomological inoculation rates (EIRs) was collected. Second, the impact of combining house screening with LLINs on indoor mosquito densities and EIRs were evaluated in a randomized controlled trial. Intervention houses received LLINs plus house screening whilst the control arm households received LLINs only. Third, the durability of the window and door screens were assessed a year after screening. Fourth, community acceptability of the house screening intervention by the participants was assessed. Anopheles rufipes, Anopheles funestus and Anopheles arabiensis were the main vectors in the study area. Closing eaves and screening doors and windows reduced indoor densities by an average 65%. EIR in unscreened houses was 2.91 infectious bites/person/six months (ib/p), higher than that in screened houses (1.88 ib/p/six months). After a year, window screens were intact. However, the wire mesh on most door screens was damaged on the bottom half. Participants accepted this intervention and linked house screening to reduced malaria in their households and cited sleeping peacefully due to reduced mosquito biting. House screening has the potential to reduce malaria incidence, offer prevention against diseases, and provide additional benefits against nuisance biting and must therefore be promoted as a public health intervention.
Endocrine disruptive activity and occurrence of pharmaceuticals and viral content in selected water sources in Melusi, Pretoria
(University of Pretoria, 2022) Patrick, Sean Mark; Shirinde, Joyce; miane.swanepoel@gmail.com; Swanepoel, Hermina Johanna
The quality of drinking water is a global concern. The increase in ineffective Waste Water Treatment Plants (WWTP) and increased human activity contribute to water pollution, resulting in adverse health effects in various populations dependent on these water sources. Water sources are often contaminated with chemicals and pollutants, some of which have endocrine-disrupting chemicals (EDCs) properties. The p study site Melusi is a non-sewage informal settlement in Pretoria North, home to over 3000 inhabitants, dependent on external water sources for daily water use and consumption. Thus it is imperative to screen water contaminants, particularly EDCs, to protect the drinking water source and decontamination to ensure drinking water supply security and resilience. This study aimed to determine the occurrence of endocrine-disrupting chemicals, pharmaceuticals and viral content in drinking water sources and WWTP in Melusi, Pretoria. The study is novel in that although these contaminants have been noted in an array of literature; however, not much has been reported in a South African context. Thus the results are critical to investigating the extent of the problem from these contaminants.
Unsupervised machine learning in air pollution epidemiology in South Africa : artificial intelligence subset application
(University of Pretoria, 2023) Wichmann, Janine; Junger, Washington; u17242496@tuks.co.za; Mwase, Nandi Sisasenkosi
Clean air is a human right and a condition for healthy living, but air pollution remains a global concern. The World Health Organization (WHO) has stated the detrimental health effects of air pollution, equating the effects to other health risks including an unhealthy diet and smoking tobacco. Air pollution is a complex mixture of droplets, solid particles, and gases, such as particulate matter (PM), nitrogen dioxide (NO2), ground-level ozone (O3), and sulphur dioxide (SO2). Air pollution is globally recognised as the most significant environmental threat to human health. Exposure to air pollution is associated with increased risk of respiratory diseases, cardiovascular diseases, and cancers, as well as increased risk of mortality. The global estimation of the number of deaths from air pollution ranges from 6.7 to 7 million deaths. Low- and middle-income countries (LMIC) are reported to account for a substantial proportion of these fatalities, with Africa accounting for approximately one-million deaths. Long-term exposure to household air pollution has also contributed 4% of global deaths. There are a number of pollutants that have been associated with negative health effects. As of 2019, in South Africa, the State of Global Air estimated 24 800 premature deaths due to exposure to PM2.5. However, this may be an underestimation as there are only a few studies in South Africa sampling PM2.5 and associating the pollutant with mortality. Ground-level ozone has contributed to approximately 365 000 deaths, equating to 11% of chronic obstructive pulmonary disease (COPD) deaths globally. However, all air pollutant estimations and the associated number of deaths are reliant on exposure-response functions derived from epidemiological studies that are predominantly conducted in developed countries. Currently, there are limited studies conducted in LMIC, like South Africa that provide a comprehensive understanding of the impact of air pollution. Hence, it is critical for more epidemiological studies on air pollution to be conducted in countries such as South Africa. The epidemiological evidence on the health effects of air pollution mixtures is lacking globally. This could indicate a current underestimation of the health risks from merely adding air pollutants together in statistical models. There are various traditional statistical methods that have been proposed to investigate the health effects of air pollution mixtures, such as multi-linear regression, classification and regression tree analysis (CART), cox proportional hazards regression, etc. Recently researchers have also applied Machine Learning (ML) methods, which is a subset of Artificial Intelligence (AI), to address this topic. The majority of studies have applied unsupervised ML, such as k-means clustering, however, such studies are lacking in Africa. Additionally, there are multiple sources, both man-made and natural, that can lead to different mixtures of air pollutants, such as PM10 and PM2.5. While many epidemiological studies mainly focus on the mass of PM10 and PM2.5, few studies investigate the chemical composition and identification of their sources. Positive Matrix Factorization (PMF) is a well-regarded method for source apportionment. Similar to other research areas, ML methods such as k-means and spectral clustering are being used as alternative source apportionment methods. Even fewer studies in South Africa are investigating the use of ML as a source apportionment method. Therefore, the aim of this PhD thesis was to address some of the research gaps identified above, namely, the lack of studies in Africa on the health effects of air pollution mixtures and PM2.5 source apportionment, whilst also assessing the applicability of AI methods, such as unsupervised ML, in air pollution epidemiology in South Africa. The thesis objectives were to: • Assess the perceptions and attitudes regarding AI in public health among postgraduate students registered for the online Postgraduate Diploma in Public Health at the School of Health Systems and Public Health (SHSPH), University of Pretoria (UP). • Determine the joint effects of SO2, NO2, O3, PM2.5, and PM10 on hospital admissions for respiratory disease (RD) and cardiovascular disease (CVD) in Vereeniging and Vanderbijlpark, Gauteng, using traditional statistical analysis, specifically, classification and regression trees. Thereafter, unsupervised Machine Learning methods are utilised to determine the joint effects of the air pollutants on RD and CVD hospital admissions. • Compare two methods of source apportionment of PM2.5 in Pretoria – a traditional method such as Positive Matrix Factorization (PMF) and unsupervised Machine Learning clustering methods. Method: The PhD project was divided into three parts. The first was a cross-sectional survey among students enrolled in the Postgraduate Diploma in Public Health at UP to assess perceptions and attitudes regarding AI in public health. The second part of the project was to determine the joint effects of SO2, NO2, O3, PM2.5, and PM10 on RD and CVD hospital admissions in Vereeniging and Vanderbijlpark, in the Vaal Triangle Airshed Priority Area (VTAPA), South Africa. There was a total of 3 346 observations from 2 January 2011 to 29 February 2020 (before the first recorded COVID-19 case in South Africa). The statistical CART analysis was used to assess the joint effects. Seven air pollution mixtures were created in the analyses, i.e. (mixture 1) PM10, NO2, and SO2, (mixture 2) PM2.5, NO2, and SO2, (mixture 3) PM10, NO2, and O3, (mixture 4) PM2.5, NO2, and O3, (mixture 5) PM10, SO2, and O3, (mixture 6) PM2.5, SO2, and O3, and (mixture 7) O3, NO2, and SO2. Thereafter, unsupervised ML clustering methods – k-means, spectral clustering, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN) – were applied to the air pollution data to determine their joint effects on RD and CVD hospital admissions. Lastly, source apportionment for PM2.5 in Pretoria was performed using PMF analysis and unsupervised ML clustering methods, i.e. k-means, spectral clustering and principle component analysis (PCA). There was a total of 428 observations collected from 18 April 2017 to 12 February 2021. Gravimetric analysis was used to calculate the concentration levels and species identification was done through X-ray Fluorescence (XRF). The following fifteen identified species were used in the PMF model: PM2.5, BC, UV-PM, S, Cl, K, Ca, Ti, Fe, Ni, Cu, Zn, Br, U, and Si. Results: 618 respondents completed an online survey (81.5% response rate). Generally, respondents thought AI would be capable of performing various tasks that did not provide direct care to individuals. Most (69%) agreed that the introduction of AI could reduce job availability in public health fields. Respondents agreed that AI in public health could raise ethical (84%), social (77%), and health equity (77%) challenges. Relatively few respondents (52%) thought they were being adequately trained to work alongside AI tools and the majority (76%) felt training of AI competencies should begin at an undergraduate level. The air pollution (SO2, NO2, O3, PM2.5, and PM10) and meteorological data (relative humidity and temperature) used was from 1 January 2011 to 29 February 2020 (before the first recorded COVID-19 case in South Africa). Due to the missing air pollution and meteorological data for the VTAPA area, data was imputed using the multiple imputation by chain equations (mice) method. There were 54 822 respiratory disease (RD) hospital admissions in VTAPA from 2 January 2011 to 29 February 2020 (before the first recorded COVID-19 case). Generally, the risk of RD hospital admissions increased by 1.04 (95% CI 1.01, 1.08) when exposed to mixtures with high levels of NO2 and varying levels of SO2, O3, PM2.5, and PM10. There were 22 205 cardiovascular disease (CVD) hospital admissions in VTAPA during the study period. The RRs of CVD hospital admissions increased among those exposed to air pollution mixtures numbered (2), (3), (4), (6), and (7) by 1.11 (95% CI 1.02, 1.20), 1.15 (95% CI 1.04, 1.29), 1.13 (95% CI 1.05, 1.21), 1.11 (95% CI 1.02, 1.20), and 1.14 (95% CI 1.06, 1.22), respectively. Similar to findings for RD, the highest risk for CVD hospitalisation was found when exposed to high levels of NO2 and varying levels of SO2, O3, PM2.5, and PM10. The unsupervised ML clustering methods used – k-means clustering and spectral clustering – showed that the air pollution data SO2, NO2, O3, PM2.5, and PM10 were best grouped into two clusters. However a three-cluster spectral clustering model using the normalised Laplacian matrix, showed that the risk of RD hospital admission increased when exposed to SO2, NO2, PM2.5, and PM10 in higher concentration levels, and lower levels of O3 by 1.04 (95% CI 1.01-1.08). None of the formed cluster mixtures were found to increase the risk of CVD hospital admission. The DBSCAN clustering method did not prove to be an appropriate clustering method, as it greatly reduced the dataset and produced ill-distributed observations within formed clusters. A seven-factor PMF model was assigned to PM2.5 data collected over a 46-month period in Pretoria, South Africa. The seven contributing sources identified included mining (43.2%), biomass/coal burning (14.2%), secondary sulphur (12.1%), road traffic (11.3%), industry/base metal (8.7%), resuspended dust (8.5%), and general exhaust emissions (2.0%). PMF analysis was relatively easy to conduct and analyse, however, the process proved to be computationally taxing for medium to large datasets. Additionally, the modelled PM2.5 concentration levels was lower than the actual PM2.5 concentration levels; the correlation between modelled PM2.5 and actual PM2.5 data was R2 = 0.6. The seven-cluster spectral clustering model, using the normalized Laplacian matrix, showed feasible sources for the PM2.5 data during the 46-month period in Pretoria, South Africa. The possible identified sources of PM2.5 were coal burning (42.89%), industry (22.0%), resuspended dust (10.4%), base metal (6.7%), road traffic (6.8%), general exhaust emissions (5.8%), and secondary sulphur (5.5%). Spectral clustering was easy to run, not computationally taxing, and utilised the complete dataset within the clustering. This suggests that it was a good dimension reduction tool that can produce plausible results for source apportionment. However, there was an issue of overlapping clusters and a lack of external validation for the formed clusters. This is a reason of concern when using spectral clustering for source apportionment. Conclusion: The study contributes to the limited, but growing, knowledge and application of ML and AI in public health and air pollution epidemiology. The survey yielded a variety of views. There was a general assumption that AI in public health could assist in performing particular tasks at different health levels that did not involve direct care. There was also a general consensus that AI had the potential to raise unemployment and ethical challenges in the public health field in South Africa. SO2, NO2, O3, PM2.5, and PM10 mixtures proved to be associated with RD and CVD hospital admission. The mixtures showed that a higher concentration of NO2 in combination with varying concentrations of SO2, O3, PM2.5, and PM10 can lead to increased risk of both RD and CVD hospitalisation. This result contributes epidemiological evidence that can help policy makers to introduce stricter policies for improving the air quality of national priority areas, such as VTAPA in South Africa. Unsupervised ML could be useful in determining joint effects of air pollutants on hospital admission and other health outcomes. K-means and spectral clustering were both relatively easy to run and analyse; they were also less time consuming in comparison to the CART analyses. The process also showed promise for analysing more than three air pollutants, in spite of the different interactions. However, it is evident that further study is needed before unsupervised ML can be considered a reliable and definite tool to study the joint effects of air pollution on different health outcomes. PMF modelling suggested that mining and industry were the main contributing factors to PM2.5 in Pretoria. However, there is a great need for more studies that sample PM2.5 in Africa. Source apportionment studies are vital in the evaluation of policies intended to protect communities from the detrimental health effects of PM2.5. The PMF software was relatively easy to use and the data produced was relatively easy to analyse for possible sources of PM2.5. However, the three model runs only showed 0.4 to 0.6 correlation with the original data. Unsupervised ML for source apportionment is still a relatively new concept and needs to be further explored. In comparison with PMF, spectral clustering showed potential as a dimension reducing tool for source apportionment. Although the sources identified in the spectral clustering model showed similar sources identified in the PMF model, there were some noticeable limitations. Extensive studies are needed to continue exploring the potential of clustering for source apportionment studies. Furthermore, there is a need to increase air pollution epidemiology and source apportionment studies in South Africa. This will increase African-based evidence of the detrimental effects of air pollution. Air pollution studies using unsupervised ML has the potential to be used in air pollution and public health studies. This project produces a baseline in the current perceptions of AI in public health and could lead to more in-depth studies on the topic. With hopes to initiate conversation around including AI in public health, this project shows epidemiological evidence that can be used to advocate for stricter, more effectively enforced air quality standards and management plans in VTAPA. Lastly, the project also produces a baseline framework for including the application of ML in epidemiological and source apportionment studies. Spectral clustering provided plausible results in comparison to the results obtained using statistical and traditional models. Although the study used a limited number of unsupervised ML methods, it is highly recommended that other unsupervised ML methods be used in further public health studies to continue investigating the practical implementation of AI in public health.
Developing a novel approach for improving supply chain management for SARS-CoV-2 point-of-care diagnostic services in resource-limited settings : a case study of Mopani district Municipality in Limpopo Province, South Africa
(University of Pretoria, 2023-06-20) Mashamba-Thompson, Tivani; Musekiwa, Alfred; u15266304@tuks.co.za; Maluleke, Kuhlula
Introduction: In settings with limited access to laboratory diagnostic services, point-of-care (POC) testing offers a suitable alternative for diagnosing COVID-19. We conducted a scoping review to guide the objectives of this study. The scoping review highlighted the importance of equitable access to diagnostic tests at POC through well-coordinated supply chain management (SCM) systems, particularly for settings with limited access to diagnostic laboratory services. It also revealed a research gap on SCM of POC diagnostic services in low- and middle-income countries (LMICs). Guided by the scoping review results, the overarching aim of this thesis is to contribute knowledge to inform development of a novel approach for improving SCM for COVID-19 POC diagnostic services in resource-limited settings with poor access to laboratory diagnostic services, using Mopani District in Limpopo Province, South Africa, as a study setting. Methods: This multiphase mixed methods study consisted of four phases, starting with the scoping review that guided the thesis objectives. Phase 2 involved a geospatial analysis to assess the spatial distribution of COVID-19 POC testing services in the Mopani District. Phase 3 involved an audit of primary healthcare (PHC) clinics providing COVID-19 POC diagnostic services to evaluate the impact of SCM on accessibility and identify barriers and enablers. Based on the findings from the initial phases, Phase 4 employed a nominal group technique (NGT) to collaborate with key stakeholders in co-creating a novel approach for improving SCM systems for COVID-19 POC diagnostic services. Finally, we synthesised results from the above phases to inform development of an evidence-informed context specific framework for improving POC diagnostics services SCM. Results: The geospatial analysis indicated that the majority of the population (78.2%) had adequate accessibility to COVID-19 diagnostic services, assuming they utilized the nearest healthcare facility. However, an uneven distribution of services within the region was identified. The audit revealed non-compliance with SCM practices in PHC clinics in the Mopani District, particularly in inventory management, distribution, and human resource capacity. However, compliance was observed in procurement, redistribution, and quality assurance. Through collaboration with key stakeholders we were able to identify key priority areas that needed to be addressed ix to improve SCM systems for POC diagnostic services in the Mopani District. The following areas were identified to be a priority: availability of testing kits, monitoring of stock levels, unknown demand, information on SCM during a pandemic, demand planning and standardisation of procurement policies. Informed by the above results, we proposed an intersectoral POC diagnostics SCM framework for resource-limited settings. Conclusion: This thesis has successfully guided the development of a novel approach to improving SCM of SARS-CoV-2 POC diagnostic services in resource-limited settings, using Mopani District in Limpopo Province, South Africa, as a study setting. This thesis offers guidance on achieving increased accessibility, responsiveness, optimal inventory management, quality assurance, standardization, and data-driven decision-making. These advantages can contribute to more effective healthcare delivery, improved patient outcomes, and enhanced management of public health. Additionally, this thesis proposes implementing an intersectoral framework for improvement of SCM for POC diagnostics in resource-limited settings. The findings of this thesis have significant implications for policymakers and implementers involved in POC diagnostics in resource-constrained settings. Further research is necessary to determine the feasibility of implementing the intersectoral framework for improving SCM for POC diagnostics in resource-limited settings.
Framework for mitigating the risk of waterborne diarrheal diseases in peri-urban areas of Lusaka district Zambia
(University of Pretoria, 2023) Ncube, Esper Jacobeth; Voyi, Kuku V.V.; cdmeki@gmail.com; Meki. Chisala Deborah
Waterborne diarrheal diseases are a public health problem in developing countries including Zambia. Despite implementing various interventions, the diseases have persisted in Zambia. This study aimed to develop a framework for identifying appropriate interventions for mitigating the risk of waterborne diarrheal diseases in peri-urban areas of Lusaka district Zambia. The study employed a sequential mixed methods design. The first step of the study involved a systematic review to determine interventions for mitigating risk of waterborne diarrheal diseases. This was followed by a longitudinal study to investigate trends of diarrheal diseases over a 10 year period (2010 to 2019) using secondary data from the Health Management Information System in 15 health care facilities of Lusaka district. A scoping review was then conducted to identify frameworks for mitigating risk of waterborne diarrheal diseases. These frameworks were analyzed using Strength, Weakness, Opportunity, and Threat analysis to identify gaps and used as a basis for drafting the framework. Finally, the draft framework was validated by health workers and other WASH experts for correctness of information and acceptability, after which the refined framework was developed. Under the systematic review, the study found 56 studies that met the inclusion criteria reporting several interventions including: vaccines for rotavirus disease (Monovalent, Pentavalent and Lanzhou lamb vaccine); enhanced water filtration for preventing Cryptosporidiosis, Vi polysaccharide for typhoid; cholera 2 dose vaccines, water supply, water treatment and safe storage, household disinfection and hygiene promotion for cholera outbreaks. The longitudinal study revealed a decrease in trends of diarrheal diseases with non-bloody and bloody diarrhea being the main cause of morbidity and mortality, respectively. The highest number of cases were recorded in 2016 and lowest 2019 with more cases in children under five years. Notably, most cases were recorded during the rainy season. First level hospitals recorded the highest number of cases and deaths compared to other health facilities. The scoping review found five frameworks for mitigating risk of diarrheal diseases including hygiene improvement framework, community led total sanitation, global action plan for pneumonia and diarrhea, participatory hygiene and sanitation transformation, and sanitation and family education. None of these frameworks was specific for waterborne diarrheal diseases. These frameworks were used to propose a draft framework. Validation of the draft framework helped to improve the tool as the health workers and experts suggested several issues included in the final framework. The final framework consisted of the following elements: problem identification; identification and quantifying of risks; identification of evidence-based intervention(s); assessment of intervention(s) in target community; selection and adoption of intervention(s); implementing selected intervention(s); monitoring and evaluation; sustainability and system support factors. The developed framework is envisaged to help mitigate risk of waterborne diarrheal diseases in peri-urban areas of Lusaka Zambia if implemented and ultimately improving public health in Zambia and related settings.
Characteristics of adult patients who are lost to follow-up in antiretroviral roll out clinics – Gauteng, South Africa
(University of Pretoria, 2014-11) Kinkel, F.; Wolvaardt, Jacqueline Elizabeth (Liz); hfkinkel@foundation.co.za; Molefe, Thuthukile
The global commitment by governments throughout the world to scaling up access to Antiretroviral Therapy (ART) in response to the crisis imposed by the HIV epidemic has resulted in a large number of people living with Human Immune-deficiency Virus (HIV) worldwide. According to statistics provided by the World Health Organization (WHO), there were approximately 35 million people living with HIV (PLWHIV) in 2012.1This large number of PLWHIV observed in recent years reflects the life-prolonging benefit effects of ART.
The potential risks of long term exposure to low concentrations of antiretrovirals in treated and untreated water sources in Gauteng, South Africa
(University of Pretoria, 2020) Taylor, Maureen B.; De Jager, Christiaan; Ramalwa, Ntsieni Rahab
Access to safe and affordable drinking water and sanitation is highlighted in the Sustainable Development Goal Target 6.1 while Target 6.3 addresses the release of hazardous chemicals into water sources. Pharmaceuticals and personal care products in treated drinking water have been receiving growing attention from environmental and health organizations worldwide because they are more frequently being detected in water sources. The fact that pharmaceuticals are manufactured with the intention to cause biological effects continue raising concerns about the impact of unintentional exposure to pharmaceuticals on human health. Despite the relatively fast growing numbers of studies on the prevalence and potential risk associated with pharmaceuticals in potable water, few studies that have addressed the potential human health risks associated with ingestion of low doses antiretrovirals (ARVs) through drinking water. The aim of the study was to assess the potential risks posed by long-term exposure to trace levels of ARVs in treated and untreated water sources in South Africa (SA), more specifically the Gauteng Province. A review of national and international literature was conducted to determine the extent and risks posed by ARV contamination in water sources globally. From the review it was evident that there is paucity of data on pharmaceuticals in water sources worldwide, including Africa. Where such data was available, pharmaceuticals targeted and detected in each investigation were country-dependent and linked to the most commonly used drugs or antivirals in the region, e.g. oseltamivir in Japan, with only a few reviews reporting on the presence and fate of ARVs in environmental samples. From a review of global human immunodeficiency virus (HIV) epidemic it was evident that SA uses more ARVs per capita compared to any other country fighting the HIV/acquired immunodeficiency syndrome (AIDS) epidemic with 71% (5 million) of adults living with HIV on combination antiretroviral therapy (cART). From 2003 to 2019 the drugs used in the first-line regimen for adults were the most used for the management of HIV with tenofovir disoproxil fumarate, lamivudine (3TC), emtricitabine and efavirenz (EFV) used more widely from 2010-2019. A newly approved ARV, dolutegravir, was included in the first-line regimen from 2020. A systematic review, conducted to establish which ARVs have been detected in water sources in SA, revealed that all ARVs that have been used historically in the first-line (stavudine, 3TC, EFV, nevirapine) and in second-line (didanosine, ritonavir boosted lopinavir, zidovudine [AZT]) regimens have been detected in one or more water sources, including treated drinking water, surface water and wastewater influent and effluent. To establish whether the low concentrations of ARVs in drinking water posed a possible health risk to individual ingesting polluted drinking water, a risk assessment was conducted. The method comprised of five general steps: a) selection of ARVs to be assessed; b) derivation of acceptable daily intake; c) derivation of predicted no effect concentrations; d) Exposure assessment - determination of environmental concentrations; and e) risk calculation. The risk quotient values needed for the risk assessment were sourced from studies that utilised acceptable daily intake values derived from dose-response model studies. The present study showed that from the current levels of AZT, 3TC and abacavir (ABC) detected in drinking water sources in SA, the possible human health risk was insignificant, although harmful to aquatic species. The predicted no effect concentrations were not available for the other ARVs present in the water sources in SA. Overall, this study showed that selected ARVs, namely EFV, in water were harmful to aquatic species, while the current levels of AZT, 3TC and ABC detected in drinking water sources in SA posed an insignificant human health risk. The study has therefore provided new data on the potential human health risk posed by exposure to low levels of ARVs in treated water sources in SA.
Risk factors associated with treatment default in pulmonary tuberculosis patients in Tshwane, Gauteng : case control study
(University of Pretoria, 2013) Moodley, Saiendhra Vasudevan; Letebele-Hartell, K.E.; nmmuvhango@webmail.co.za; Muvhango, Ntshengedzeni Michael
Background: Tuberculosis is a curable disease. The challenge faced by many TB control programmes around the world is treatment non-compliance. Patients who default their treatment are at risk of clinical deterioration and development of multi-drug resistant tuberculosis. This study therefore aimed at determining the factors associated with tuberculosis treatment default in Tshwane district, Gauteng Province. Methods: The study was conducted on patients who were diagnosed with TB and registered for treatment in Tshwane health facilities in Gauteng Province. This was a case-control study, carried out in two phases. During phase 1 of the study, TB registers in the health facilities were reviewed retrospectively. All the defaulters/cases and randomly selected non-defaulters/controls were identified from the TB registers and reviewed. During the review, the following data was extracted from TB registers including: demographic information, patient’s address, treatment information including dates of TB registration, treatment initiation and completion and treatment outcome. During phase 2 of the study, patients were traced and after giving consent were interviewed using a questionnaire. Data was captured using Microsoft Excell and Epi Info and analyzed using Statistical software (STATA 9.0; StataCorp; College Station, TX). Univariate and multivariate unconditional logistic regression analysis to determine association and Kaplan-Meier method to determine probability of staying in treatment over time were applied. Results: Of the 1509 cases in phase 1 of the study, 50.8% (767) and 27.6% (417) defaulted TB treatment within the first and second months of treatment respectively. On multivariate analysis, factors found to be significantly associated with treatment default in phase 1 of the study were age (OR 1.46, CI: 1.23-1.73), male gender (OR 1.56, CI: 1.32-1.85) and co-infection with HIV (OR 1.38, CI: 1.12-1.70). In phase 2 of the study, factors associated with treatment default (bivariate analysis) were inadequate TB knowledge (OR 4.08, CI: 1.68-9.60), changing of residence (OR 5.83, CI: 298-11.5), poor attitudes of health care workers (OR 4.18, CI: 1.75-9.97) and taking treatment without supervision (OR 2.72, CI: 1.42-5.22). None of the risk factors in phase 2 of the study were significant during multivariate analysis. Conclusions: Many patients defaulted their treatment during the first two months of treatment (intensive phase). Health care workers will need therefore to educate their patients and emphasize the importance of staying on treatment during counselling.
Factors Associated with ART Initiation among Eligible HIV Positive Pregnant Women in Swaziland
(University of Pretoria, 2014) Louwagie, Goedele M.C.; Girdler-Brown, Brendan; cchouraya@pedaids.org; Chouraya, Caspian
Conclusions: ART initiation is very important for HIV positive women both for their own health and for prevention HIV transmission to their children. This study found that ART initiation among eligible pregnant women in Swaziland was statistically associated with the presence of partner support and favourable perceptions of benefits of ART after multiple logistic regression analysis. Stronger counselling and education for pregnant women and male involvement strategies need to be implemented as momentum gathers towards elimination of paediatric HIV by 2015.
Smoking-related health risk knowledge, and reactions to cigarette warning labels amongst South African adults
(University of Pretoria, 2020) Ayo-Yusuf, Olalekan A.; flavia.senkubuge@up.ac.za; Senkubuge, Flavia
Background Tobacco control is a public health concern. By 2020, it is estimated that seven out of every 10 people killed by smoking will be from developing countries. Smoking ranked third after unsafe sex/sexually transmitted disease and high blood pressure in a South African study. There are numerous smoking-related conditions. While it is known that knowledge and understanding of health risks by smokers may influence their smoking behaviour, few studies have been conducted in this regard in South Africa. Most countries communicate health risks to smokers through cigarette warning labels or media campaigns. Limited information is available on the effects of health warnings on smoking behaviour, particularly in South Africa, and on whether pictorial warnings will influence South Africans. Additionally, to date, no current nationally representative study has been conducted among South Africans to gather information on the acceptability of pictorial warnings, even though current tobacco legislation recommends that such warnings be implemented on tobacco packs. Aims and Objectives Aim: This thesis aimed to assess the knowledge of tobacco health risks among a population of South Africans and to determine the effectiveness of text-based health warning messages and pictorial warnings with brand design elements (branded) and pictorials without brand design elements (plain) on smoking behaviour. Objectives: The objectives of this study were: 1. To select pictorial health warning labels with brand design elements (branded) and without brand design elements (plain) on cigarette packs to prioritise for testing among South Africans. 2. To determine the reactions of a select sample of South Africans (non-smokers and smokers) towards text-only and pictorial (on branded and plain packs) cigarette health warning labels under a quasi-experimental condition. 3. To assess the factors associated with change in motivation and plan to quit smoking following experimental exposure to test cigarette packages with text-only and pictorial (branded and plain) warning labels among smokers. 4. To determine the knowledge of smoking-related health risks among a nationally representative sample of South Africans and potential reactions to the selected pictorial warnings on branded pictorial warning packs at population level. Methods The thesis consists of three interrelated parts as discussed below:  Part One: In Part One, secondary data analysis using data from the 2010 South African Social Attitudes Survey (SASAS) (n ~3000) was used, together with other published sources to select the prioritised health warning themes and messages for use in Part Two of the thesis.  Part Two: Part Two of the thesis was conducted in South Africa, in the Gauteng and Western Cape provinces respectively, so as to have greater representation of the South African population. Part Two used a quasi-experimental crossover design and mixed methods, including focus group discussions, to establish the effectiveness of text and pictorial health warnings with (branded) and without (plain) brand design elements on the change in motivation and plans of a smoker to quit, which are proximal determinants of actual change in behaviour i.e. quitting. Using a crossover design, participants were requested individually to assess each of the health warnings – text and pictorial (branded or plain package). After that, they completed a post-evaluation questionnaire after being exposed to all the health warnings. The post-evaluation questionnaire concluded the individual assessment of the health warnings. The same participants then proceeded to the next part of the evaluation, which was the focus groups. Focus group discussions were conducted – among the same participants who had previously completed the individual assessment of the health warnings – to establish and examine the prevailing attitudes, perceptions, understanding and behaviours among the target groups of non-smokers and smokers towards the tobacco health warning labels. Participants were allocated to one of 12 categories of focus groups of 10 each (n=960) according to their race, gender, age, and smoking status. Focus groups were conducted until saturation of the focus groups, where no more new information was obtained. Participants rated their responses (on a scale of 1 to 5) using previously validated measures of effectiveness, grounded in the constructs of persuasive communication theory, namely “attention,” “communication,” “identification” and “effect.” After the participants’ responses to the health warning labels had been analysed, the labels were revised. Focus groups were then held among a smaller select sample of participants only in Gauteng who were requested to assess the revised pictorial health warning messages: with brand design elements (branded) and those without brand design elements (plain). A structural equation model (SEM) was also constructed to understand the pathway from reaction to health warnings to changes in planning to quit.  Part Three: Part Three included secondary data analysis, using data from the 2016/17 South African Social Attitudes Survey (SASAS), which is a nationally representative household survey to assess the state of knowledge of tobacco health risks among South Africans in 2016/17 (n ~3000). Using a self-administered questionnaire, the data obtained included socio-demographics; tobacco use; and participants’ reaction about whether “plain” packs, as shown on a “show card” to each participant, could make smokers think about quitting. Another structural equation model (SEM) was constructed this time to understand the pathway from response to exposure to health warnings and quit attempt at population level. Results This thesis showed that although South Africans have some knowledge of smoking- related health risks, this knowledge differs by the type of smoking-related health risk. South Africans have particularly limited knowledge of the vascular (hypertension, impotence, and stroke) smoking-related health consequences, compared to their knowledge about the risks related to cancer and/or respiratory diseases. Furthermore, although the participants were not the same individuals, there was no increase in the overall knowledge of smoking-related health risks among South Africans who participated in the 2010 SASAS and those who participated in the 2016/17. These results on knowledge indicate an urgent need to implement interventions that will increase South Africans’ knowledge of tobacco-related health risks, such as pictorial warnings. The quasi-experimental study using a crossover design. There were 767 participants, with a response rate of 79.9%. There were about equal numbers of smokers and non- smokers. Before exposure to test health warnings, the majority who smoked indicated they were not planning to quit (64.6%). Overall, out of all the 20 health warnings that were evaluated before the revision of the health warnings, text-only health warnings were ranked lowest. Pictorial warnings, regardless of package design, ranked higher than text-only health warnings. However, participants most often indicated that the pictorial warnings on packages without the brand design elements (plain packages) were more effective than the pictorial warnings on packages with the brand design elements (branded packs). The pictorial warning that was ranked as the most effective before revision was the abortion picture on plain packaging, with a mean rating score of 3.92 (SD=0.40). After revision of the pictorial warnings, which now also included lung cancer warnings, as suggested by participants, the most effective warning was lung cancer on plain packaging with a mean score of 3.77 (SD=0.68). Smokers felt that the pictorial warning on abortion, regardless of the pack (plain pack mean=3.88, SD=0,49; branded pack mean 3.88, SD=0,45) was most effective in motivating smokers to quit smoking or think about quitting. The top five pictures selected as the most effective among the pictorial warnings after revision were those related to lung cancer (62.9%), gangrene (45.2%), impotence (44.4%), abortion (34.7%), and oral disease (21.8%). After controlling for potential confounders, some of the factors that were independently associated with higher odds of having a positive change in planning to quit smoking after exposure to health warnings were self-identifying as Indian/Asian (OR=2.70; 95% CI=1.11-6.58) compared to black African; being employed (OR=3.94; 95% CI=1.98- 7.83) as compared to being unemployed; and indicating spending money on cigarettes rather than food (2.62; 1.41-4.88). The SEM depicting pathways to changes in planning to quit after exposure to cigarette health warnings fit the data well (comparative fit index=0.997; normed fit index=0.975; root mean square error of approximation=0.026). SEM confirmed that current text-only warnings were less likely directly to influence changes in planning to quit (β= -0.29). Greater changes in planning to quit were directly influenced by a higher rating of the branded packaging (β=0.25). Current smoking in 2017 was 19.3% (n=607), with only 49.6% (n=292) planning to quit and 59.9% (n=345) having attempted to quit in the past 12 months. Of the respondents, 70.8% (n=2071) have never or rarely read the current text-only health warnings, but 85.7%(n=2495) agreed that text-only warnings (as shown) were easy to understand. Of those who agreed that packs with pictures would make smokers think of quitting, 54.4% (n=1030) thought the current displays of cigarette packs inside stores and shops could encourage young people to take up smoking. Only 42.0% (n=273) of the current smokers indicated that adding pictures to cigarette packs as shown would make them think about quitting, but 61.2% (n=385) agreed that displaying the pictorial warnings would encourage the youth not to smoke. Those with an educational status lower than Grade 12 (44%; n=747) and those with an educational status higher than Grade 12 (46.4%; n=196) agreed that adding pictures would make smokers think about quitting. Although the majority of respondents agreed that the text warnings shown to them were easy to understand, only 15.1% (n=509) felt that these warnings would make a smoker think of quitting, whereas 41.9% (n=1301) felt that the pictorial warnings would make smokers think of quitting. After controlling for potential confounders, the factors that were independently associated with higher odds of believing that pictorial warnings on a “plain” pack would stop a smoker who wants a cigarette were an educational status of more than 12 years of schooling (OR=1.71; 95% CI=0.74-3.93) as compared to 12 years or less of education, believing that displaying cigarette packs in shops was a form of advertisement (3.27; 1.91-5.60), exposure to smoking at work (2.29; 1.29-4.07) and having attempted to quit smoking within the last 12 months (1.95; 1.11-3.41). The SEM on path to attempt to quit fitted the data well (comparative fit index=0.986; normed fit index=0.956; root mean square error of approximation=0.028). Smokers’ perceived health risk directly influenced their quit intention (β= 0.21), which in turn was positively directly associated with having actually attempted to quit (β= 0.43). Conclusion Assessing the effectiveness of cigarette pack health warning labels is a matter of public health importance, given the significant burden of disease associated with tobacco use. This thesis is the first of its kind in South Africa and comes at a time when legislation is being amended to include pictorial warnings and plain packs. The thesis provides evidence that pictorial warnings, particularly on plain packaging, would be effective in South Africa. The findings reported in this thesis were used to assist in providing information on the implementation of pictorial warnings in South Africa that are evidence-based and tailored to the South African market. These pictorial health warnings could therefore potentially save lives by increasing cigarette smokers’ motivation to quit and eventually quitting tobacco use. These findings suggest that adding pictorial warnings to the current cigarette packs in South Africa is more likely to prompt quitting and deter the youth from taking up smoking than text-only warnings. Most countries communicate health risks to smokers through cigarette warning labels or media campaigns. Limited information is available on the effects of health warnings on smoking behaviour, particularly in South Africa and, whether pictorial warnings will influence South Africans. Additionally, there is also no current nationally representative study conducted among South Africans to inform on the acceptability of pictorial warnings although, current tobacco legislation recommends that such warnings be implemented on tobacco packs.
PM2.5 chemical composition, source apportionment and geographical origin of air masses in Pretoria, South Africa
(University of Pretoria, 2020) Wichmann, Janine; Molnar, Peter; abifaz@yahoo.com; Adeyemi, Adewale Adekunle
Background: Air pollution is one of the major problems being faced by most of the big and industrial cities of the world and has become a major environmental threat over the last few years. This environmental threat has gained more attention because of its increased health effects on humans which includes morbidity and mortality. It has various adverse effects, such as increased pulmonary infections, respiratory diseases, acute illnesses and hospitalizations and can eventually lead to death. The need for source identification and abatement strategies for air pollution is crucial in order to meet the SDG goal to reduce pollution by 2030. Therefore, this study investigated the characterisation of PM2.5, source apportionment, origin of air masses into Pretoria alongside the association between air pollutant and hospital admissions due to respiratory diseases. Method: This study was divided into primary and secondary data collection phases. For the primary data collection, daily 24-hour PM2.5 samples were collected every third day between 18 April 2017 and 17 April 2018 at an urban background site in Pretoria. A total of 122 PM2.5 samples and 25 duplicate PM2.5 samples were collected and analysed for particulate mass, soot, black carbon (BC), organic carbon (OC) and 18 trace elements. Source apportionment analysis was conducted on this dataset using the positive matrix factorisation method. Air mass trajectories, as a surrogate for distant sources of PM2.5, were estimated using the HYSPLIT model (version 4.9). The daily average trajectories were calculated backwards for 72 h and used for cluster analysis. The clustering algorithm coupled in HYSPLIT was based on the distance between a trajectory endpoint and the corresponding cluster mean endpoint. The secondary data, daily hospital admissions, PM10, NO2 and SO2 data, used for this project were obtained from a private hospital group and the South African Air Quality Information System, managed by the South African Weather Services. The time-stratified case-crossover epidemiology study design and conditional logistic regression models were applied to investigate the association between PM10, NO2, SO2 and respiratory disease (RD) hospital admissions during the study period 1 January 2011 to 30 November 2014. Results: The annual mean for PM2.5 (n =122 days) was 21.1µg/m3 (range 0.7 - 66.8 µg/m3). The highest PM2.5 mean value was recorded during winter, which was significantly higher than autumn, spring, and summer (p<0.0001). No significant difference between weekdays and weekend (P>0.9567) was observed. Most exceedances of PM2.5, when compared with daily World Health Organization (WHO) guidelines and South African standards were observed in mid-autumn and winter. Soot, BC and OC followed the same trend as PM2.5 concentration. Average S (1480 ng/m3) concentration was the highest among elements detected, followed by Si, Fe, K and Ca, in that order. Seven sources and their contributions to the total PM2.5 were identified and quantified. These included vehicle exhaust – 8.6%, and base metal/ pyrometallurgical - 0%, soil dust -13.2%, secondary Sulphur – 31.4%, vehicle exhaust – 12.5%, road traffic – 7.3%, coal burning -27.2%, while the percentage of PM2.5 specie in the base metal/ pyrometallurgical factor was 0%. The identified source factors exhibited seasonal variations, coal burning and secondary Sulphur being the highest during winter while soil dust and road traffic were lowest during summer. Five transport clusters were identified during the 1-year study period: National Limpopo (Nat-LP), transboundary (TB), Easterly-Indian Ocean, South Easterly-Indian Ocean and South Westerly-Atlantic Ocean. In addition to this, 85% of the transport clusters were of local and transboundary origin, 15% were long-range transport, while cluster 1 had the highest PM2.5 concentration. Cluster 1 can be attributed to main source of pollution contributing to the PM level at the sampling site due to the activities going on in the region, such as biomass burning, coal mining. Of the 17,647 hospital admissions in Pretoria, 51.8% (n=9,147) were women and 61.6% (n=10,870) were 0-14-year old. In the unstratified analysis, a 10 g/m3 increase in PM10 was associated with statistically insignificant increase of 0.2% (-0.7%; 1.2%) in RD hospital admissions; no significant association was observed for NO2 during cold days. Significant association between SO2 and RD hospital admissions was observed for females and male patients during warm and cold days. Conclusion: This project contributes to the very few source apportionments studies of PM2.5 in Africa and specifically South Africa. Coal burning remains one of the main sources that should be addressed. Late autumn and winter season recorded the highest concentration. The risks of RD hospital admission due to PM10 exposure in Pretoria were higher on warm days than on cold days. The apportioned sources and the origin of air masses from this study align with the known existing sources in the country. Oceanic influences, local and transboundary sources (Southern African countries) contribute to the air masses passing over Pretoria, therefore, abatement strategies are paramount to reduce the level of pollution during this time. The findings of the study can be of help to the government in the formulation of air pollution guidelines as a measure to mitigate the effect of air pollution on the environment. In addition to this, if there is strict compliance to the already formulated regulation on the identified sources, this will significantly reduce the effect of air pollution in our cities. Lastly, the outcome of this project will help the South African government in their air quality management plan that are reviewed regularly.

Browse

Recent Submissions