Spatial Analysis of Perceived Health System Capability and Actual Health System Capacity for COVID-19 in South Africa


 
 People’s confidence in the health system's capability in managing the COVID-19 pandemic can determine public support, risk perceptions, and compliance to the required behaviors during the pandemic. Therefore, this paper investigated people’s perception of health system capability to manage the COVID-19 pandemic in different spatial areas across the country using data from an online survey.
 
 
 
 Multivariate logistic regression models were used to examine factors associated with people’s perception of the health system capability to manage the COVID-19 pandemic at the national and provincial levels. Spatial comparative analysis was conducted to contrast spatial density indicators of the number of hospitals, hospital beds, and ICU beds per given population across various provinces.
 
 
 
 Findings showed that South Africans had low confidence in the health system capability, with only two in five (40.7%) reporting that they thought that the country’s health system was able to manage the COVID-19 pandemic. Sex and knowledge on COVID-19 were significantly associated with the people’s perception of the health system capability to manage the pandemic at the national level and in four of the nine provinces.
 
 
 
 Overall, the findings of this study clearly highlight challenges facing the country’s health system, both perceived or real, that needed to be addressed as part of the preparation for the COVID-19 pandemic. Timeous implementation of a countrywide National Health Insurance (NHI) system is now more critical than ever in improving healthcare outcomes of the South African population beyond the existence of the COVID-19 epidemic.



Introduction
Coronavirus Disease 2019 (COVID-19) is a novel coronavirus caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV2). This novel coronavirus rst emerged as a causative agent that led to a respiratory disease outbreak in December 2019 in Wuhan, China (Jernberg 2020). South Africa recorded its rst con rmed case of COVID-19 on March 5 2020. Since then the number of con rmed cases have grown exponentially as there are currently (as of rst week of July 2020) more than 200 000 con rmed cases across the country. Health system capability and capacity to deal with this COVID-19 pandemic has been under pressure and scrutiny globally. Clinically, the COVID-19 disease is characterised by a dry cough, fever, fatigue and shortness of breath. In severe cases, patients present with severe respiratory illness such as a pneumonia or acute respiratory distress syndrome (ARDS) which could both lead to death (Zhang et al. 2020). According to recent presentation made to South African Members of Parliament by the Director General of the National Department Health, there is a total of 119 416 hospital beds available from both private and public hospitals across South Africa (du Toit and Cowan 2020; van der Heveer 2020). There are also 3 318 Intensive Care Unit (ICU) beds available, but it is projected that ICU beds could exceed 14 700 at the highest peak of COVID-19 infection. With regard to ventilators, the public healthcare system and private healthcare system currently has a total of 3 216 ventilators. It is estimated that at highest peak of COVID-19 outbreak, 7 000 ventilators will be required to deal with the virus, a shortage of 3 784 (du Toit and Cowan 2020).
In South Africa, inequalities exist in access to health services between public and private sectors, as well as within the public sector itself, especially between urban and rural areas  Robin et al. 2018). The public healthcare system supports around 84% of uninsured population while the private sector only deals with 16% of the people with medical aids across the country, even though its annual per capita healthcare expenditure is almost 10 times more than the public sector's (Benatar 2013). Those who can afford medical insurance are predominantly white and those who cannot are black African. These inequalities seen in health care access in South Africa are indicative of the racial segregation policies carried over into the present from the apartheid government (Gradin 2019). The COVID-19 outbreak has dramatically highlighted the necessity for a more integrated healthcare system (Nkonki and Fonn 2020). The COVID-19 pandemic presents signi cant opportunity for the National Health Insurance (NHI) to be piloted in real-time. In March 2020, the Competition Commission has published a COVID-19 block exemption for the health sector for promoting better coordination, sharing of information and standardisation of practice across the entire healthcare sector as well as promoting agreements between the National Department of Health and the private healthcare sector (DTI 2020). Spatial accessibility of healthcare facilities is also important factor in the ght against COVID-19 outbreak. Spatial accessibility is often concerned with the complex relationship between the spatial separation of the population and the supply of healthcare facilities and thus has a strong underlying geographic component (Black et al. 2004;Mokhele et al. 2012;Munoz and Källestål 2012;Khashoggi and Murad, 2020). Spatial accessibility is regarded as a primary deciding factor of access to healthcare and healthcare utilization (Siegel et al. 2016;Ma et al. 2018).
In addition, people's con dence on the healthcare system in managing COVID-19 outbreak is also crucial in managing the outbreak. For instance, in India, it was reported that some people escaped hospitals or quarantine and this had been alluded mainly to the con dence de cit in the public health system in many parts of the country (Chetterje 2020). Therefore, this paper explores South Africans' perceptions of the country's health system capability to manage COVID-19. It further assesses the country's current health system capacity in the various geographic areas and draws comparisons with perceptions of the health system capability to manage COVID-19 pandemic.

Data
This paper utilised data from Human Sciences Research Council (HSRC) COVID-19 online survey conducted from 27 March 2020 to 2 April 2020 in South Africa. The survey was open to all adult people (18 years and older) residing in South Africa, regardless of race, sex or nationality. The self-administered structured questionnaire (48 items) included socio-demographic, knowledge and infection control measures regarding COVID-19, risk of contracting COVID-19, perception of health system capability, hygiene practices, preparedness for self-isolation/quarantine, preparedness for lockdown and the use of social media in accessing information on COVID-19. The survey alerts were distributed on a data-free messaging platform, were distributed via various social media platforms, were posted on the HSRC organisational website and broadcasted on radio and TV stations. Ethics approval was received from the HSRC Research Ethics Committee (REC); (REC 5/03/20).

Measures
Socio-demographic variables used were sex, age, employment, knowledge score (continuous variable), risk perception and self-isolation or quarantine possibility. These were age group in years (18-29, 30-39, 40-49, 50-59, 60+), sex (male or female), race (Black African, Coloured, White, Indian/Asian), employment status (employed full time, employed informal/part time, student, unemployed, self-employed), and dwelling type (formal dwelling or informal dwelling). Knowledge score was developed from six knowledge variables about COVID-19, which were four variables on knowledge about transmission (COVID-19 is spread by direct contact with the virus from: (Select all that apply)) with options being (infected people coughing/sneezing, pets, touching face after being in contact with infected person, contact in public gathering with infected person, contaminated surfaces and I don't know); one variable on knowledge about incubation (After how long will an infected person show signs of being sick?) with options being (immediately, after 1-2 days, after 2-14 days, after 15-20 days, and I don't know); and one variable on knowledge about symptoms (Which of the following best describes the symptoms of COVID-19: (Select all that apply)) with options being (shortness of breath, body pain, sweating, headaches, coughs, sneezing, red-itchy eyes, runny nose, fever, and I don't know). Respondents were assigned a value of 1 for each of the four transmission modes identi ed, for correct identi cation of the incubation period of 2-14 days and for correct identi cation of fever, cough and shortness of breath as the main symptoms. The nal knowledge sum score ranged from 0 to 6. For risk perception, self-perceived risk of contracting coronavirus was assessed using the question "How do you rate your personal risk of contracting COVID-19?" with response options being 1=very high risk, 2=high risk, 3=moderate risk, 4=low risk and 5= very low risk. These responses were recoded into 1=low risk (very low risk, and low risk), 2=moderate risk and 3=high risk (very high risk and high risk). For perception of self-isolation or quarantine possibility, this was assessed using the question "Do you think you may end up in a situation of self-isolation or quarantine?" with response options being 1=yes, 2=no and 3=don't know. These responses were recoded into 0=no (no and don't know) and 1=yes (yes). The primary outcome variable, perception on health system capability in dealing with COVID-19, was assessed using the following question "Do you feel each of the following are able to manage the South African COVID-19 outbreak? SA health system" with options being 1=yes, 2=no and 3=don't know.
These were dichotomised into 0=no (no and don't know) and 1=yes (yes) for logistic regression models.

Statistical analysis
Data was analysed using Stata version 15.0 (Stata Corp 2017). The data were benchmarked using Statistics South Africa 2019's mid-year adult population estimates for generalisability of the ndings. Differences in estimates between the socio-demographic variables were compared using Chi-squared tests with a statistical signi cance level of p < 0.05. Logistic regression was performed to determine factors associated with the perception that the country's health system was capable to manage the COVID-19 outbreak. This was done at both national level and at provincial level to determine differences across provinces. Race group and dwelling type were excluded from the logistic regression analyses due to small cell sizes for Indian/Asian and Coloured communities in some provinces, namely the Northern Cape, Limpopo and Mpumalanga. A signi cance level of p < 0.05 was used.

Spatial analysis
For spatial density indicators, secondary data for the numbers of hospitals, hospital beds and ICU beds was Spatial density indicators were developed to determine number of hospitals, hospital beds and ICU beds per given population across various provinces using ArcGIS 10. For spatial comparative analysis across perception of health system capability maps and density indicators of health system capacity maps, the graduated colour symbology and equal interval classi cation method was done using ve classes for all maps. The COVID-19 hospitals earmarked by the national government were also geocoded and mapped for spatial analysis of their distribution and geographic accessibility.

Demographics of respondents
The study sample from the online survey comprised 53 488 respondents. After benchmarking, females constituted 53.9% and Black Africans accounted for 76.4% followed by Whites at 10.8%. In terms of age composition, 29.9% were 18-29 years old followed by those aged between 30 and 39 years old (25.7%). Perception of health system capability in managing COVID-19 Table 1 highlights people's perceptions towards health system capability to manage COVID-19 outbreak across different socio-demographic variables and provinces. Overall, two in ve South Africans (40.7%) reported that they thought that the country's health system was able to manage the COVID-19 outbreak. The perception of health system capability to manage COVID-19 varied signi cantly across all socio-demographic variables (p < 0.01 for age and p< 0.001 for all other variables). Males had a signi cantly higher prevalence of perceiving the health system as capable than females. The elderly (70 years and older) had the lowest con dence on the health system's capability (29.0%). Black Africans had the highest con dence on the health system's capability (46.9%) while White participants had lowest (13.4%).
Those employed part-time or informally had the highest con dence (44.8%) on health system capability and selfemployed adults had the lowest (34.4%). The perception of health system capability was less prevalent among those who thought they were at moderate and high risk of contracting COVID-19 (around 38% each) and among those who thought they might end up in self-isolation or quarantine (34.1%). People residing in informal dwellings had a higher con dence on the health system's capability (55.7%) than those who lived in formal dwelling (40.0%).  Health system capacity Table 3 shows secondary data on total population, hospitals, hospital beds, and ICU beds by province. Gauteng, the most populated (smallest by geographic area) province had the highest number of hospitals, however these were mainly private hospitals (83) compared to 39 public hospitals. There were more Hospital beds in public hospitals than in private hospitals whereas the opposite was the case with regard to ICU beds across the country.  With regard to the number of ICU beds per 10 000 people, Limpopo and Mpumalanga fell under the lowest category of 0.10 to 0.28 ICU beds per 10 000 people (Fig. 3a). Only Gauteng, followed by Free State and Western Cape, had the highest ICU beds per 10 000 people. Fig. 3a further depicts the spatial distribution of hospitals earmarked to attend to COVID-19 patients. Almost all provinces, except Gauteng, seems to have unbalanced location of the COVID-19 hospitals. The authors are not aware of the factors that were considered when selecting the current COVID-19 hospitals, From a geographic point of view, it appears that the hospitals in each province are often located in some major cities which are not necessarily centrally located within their province. Therefore, these hospitals are not easily accessible to a large proportion of a province's population, especially if the patients will be transported by road. Hence, at least two hospitals per province could have resolved the skewness of this spatial distribution. KwaZulu-Natal in particular has all three COVID-19 hospitals around Pietermaritzburg, which raises some spatial concern in terms of physical accessibility of these COVID-19 hospitals. For the vulnerable population, the elderly, a similar pattern was noticed, with the exception of the Eastern Cape, which ranked in the second lowest category of numbers of ICU beds per 10 000 people but had one of the lowest numbers of ICU beds per 10 000 elderly people. Limpopo, Mpumalanga and the Eastern Cape had between 1.17 and 3.24 ICU beds per 10 000 elderly people (Fig.   3b).

Discussion
Our study ndings shows that, overall, people reported having low con dence in the South African health system, with only two out of ve people indicating they thought the health system would manage to deal with COVID-19 pandemic. In similar studies conducted in Denmark and Finland, respondents reported high rates of con dence in their health systems to deal with COVID-19 with 86% and 78% respectively (Norrestad 2020a; 2020b).. Black Africans, those employed part-time or informally and informal dwellers had higher con dence in health system capability while Whites, self-employed and those residing in formal dwellings had lower con dence in health system capability to deal with COVID-19. This demonstrates clear divisions based on socio-economic status, geolocation as well as some racial differences in peoples' perceptions. The majority of the Black Africans rely on the public health system almost exclusively therefore this increased con dence could well be due to the fact that they would not have alternative options in terms of access to health care, traditional forms of treatments aside. The lower con dence shown here by White respondents could be directed at the public health system and not the private which is usually very well resourced. The majority of Whites have access to medical insurance and therefore private medical care.
Also concerning is that, the elderly (70 years and above), those who perceived themselves at high risk of contracting COVID-19 and those who thought they might end up in self-isolation or quarantine had lowest con dence in health system capability to manage COVID-19 pandemic. People in older age groups are among those that are classi ed as highly vulnerable to COVID-19 pandemic around the world both in terms of infections and fatalities (WHO 2020).
Sex and knowledge on COVID-19 were signi cantly associated with the perception of the health system's capability to manage the pandemic at the national level and in four of the nine provinces. Females were less likely to have con dence on health system's capability to manage COVID-19 outbreak than their male counterparts. The nding that men have more con dence in the health care system is surprising since previous studies reported that men generally access health care services much less than women do (van Heerden et al. 2015). It is possible that women lacking con dence in the health system's capability could be associated with how they perceive the quality thereof. Future research would need to examine perceptions of health system's capability together with the quality. Age and self-isolation or quarantine possibility key at national level and in three provinces while employment and risk perception were signi cant predictors at national level and in two provinces. The vulnerable group, elderly (60 years and older), were key determinant of people's con dence on health system capability to deal with COVID-19 pandemic at national level and in Western Cape. People's risk perceptions (which are in uenced by age and degree of perceived vulnerability) and knowledge about a public health problem often result in lower con dence in managing the problem.
Spatial accessibility has been previously reported to represents an important barrier to accessing healthcare services (Munoz and Källestål 2012). Spatial accessibility of COVID-19 hospitals in our ndings raises some concerns regarding the readiness to deal with COVID-19 pandemic, especially considering that patients might have to be transported through roads. These could be addressed by at least adding extra temporary hospitals earmarked for COVID-19 in each province with one COVID-19 hospital. Provinces have already started setting up this additional health care facilities in preparation for the expected surge in the number of cases and the potential need for hospitalisation across the country. Mostly rural provinces such as the Eastern Cape and KwaZulu-Natal should also receive special context speci c planning for the provision which addresses challenges such as long distances between villages and available hospitals as well as poor road infrastructure. These extra hospitals should be allocated based on the geographic location and population distribution as well as the emerging trends of the pandemic whereby the number of cases and epicentres have been shifting since the outbreak of COVID-19. The Western Cape Province initially only had one COVID-19 earmarked hospital in the early stages of the epidemic when our study conducted. This later changed dramatically with now the province being regarded as the epicentre for the epidemic with leading number of con rmed cases and number of deaths, and this has had to be revisited and more hospitals being allocated as well as temporary hospitals being setup at the International Convention Centre for the ght against the COVID-19 pandemic. Other provinces with big urban metros such as the Eastern Cape, Gauteng and KwaZulu-Natal had to also readjust their plans as they all initially had three hospitals each that were earmarked for the response.
Our study nding shows that number of hospitals per population and number of hospital beds per population across the country are generally lower compared to some international statistics from both public and private hospitals such as in Canada, Italy and Japan (Wilson et al. 2018). The current national average of 5.6 ICU beds per 100 000 in South Africa is slightly lower compared to those of some European countries such as the United Kingdom (6.6), France (9.7) and Italy (12.5). Turkey and Germany had one of the highest reported numbers with 46 and 29.2 ICU beds per 100 000 population (WHO 2020). As indicated earlier, the country currently has lower ICU beds available compared to the number of ICU beds that may be required (3318 vs 14 700) at the highest peak of COVID-19 infection (du Toit and Cowan 2020). This means that there are currently on average 0.56 ICU beds per 10 000 people across the country. At the highest peak, worst case scenario, national average of 2.50 ICU beds per 10 000 people will be required. If we take this into account, this means that there is a lot to be done in all provinces to reach the required national average. In particular, urgent attention is needed to increase the ICU beds numbers in Eastern Cape, KwaZulu-Natal, Western Cape and Gauteng as they are currently at around 0.30, 0.51, 0.75 and 0.96 ICU beds per 10 000 people respectively. These four provinces require urgent attention as they are currently leading with con rmed cases of COVID-19 pandemic.
Findings from this paper shows that there were some similarities between people's perceptions on healthcare systems capability and the actual health systems capacity to deal with COVID-19 pandemic in some provinces. For instance, Free State was predominantly at upper band in terms of proportions of people who had con dence on the health systems capability and the actual health system capability to deal with the pandemic. Eastern Cape on the other hand was at the lower end with regard to people's con dence in health systems capability and health system capacity. Eastern Cape appeared at upper bound only on number of hospital per 100 000 and number of hospital beds per 10 000 population. Surprisingly, Western Cape had the lowest score on people's con dence on health system capability in managing COVID-19, yet it was among the top for almost all density indicators of health systems capacity except for the density on number of ICU beds per 10 000 elderly people where it was it the middle of the range. Western Cape also had the highest signi cant proportion of elderly people based on the online survey results. In addition, results showed that the elderly (60 years and above) was signi cant predictor of people's con dence in health system capability (OR = 0.49, p < 0.05). This might have attributed to the lowest con dence on health system capability in this province. Although North West and Limpopo had the highest proportions of people who felt that the health system was capable to deal with COVID-19 outbreak, they were among the lowest in all health system capacity indicators.
The outbreak of the COVID-19 epidemic showed and demonstrated the urgency and the importance of the NHI implementation, especially in addressing the inequity access to healthcare that exists between private and public health care. The COVID-19 pandemic has laid bare the health disparities that exists between different communities that exists in the country due to historical structural systems brought about by both colonialism and Apartheid race based special planning and service provision. The NHI is now more than ever our only hope in correcting the mistakes of the past. This outbreak provides the rst and best opportunity for South Africa to reengineer the country's health system to be united and integrated as much as possible between the public and the private health care systems. The COVID-19 block exemption for the health sector is an important step towards the testing the implementation of NHI. The government and all role players in the country should embrace this opportunity and make it real by ensuring that all systems that are necessary for the NHI implementation are put in place and expanded upon.
This study does have some limitations. One of the main limitations was that there was no clarity on the online questionnaire on whether South African health system included both private and public sector. So there might be underestimation or overestimation on the people's con dence on health system capability to deal with COVID-19 depending on how they understood and experience the health system. Respondents may also have been responding on their perceptions of the public health care system regardless of whether they utilise those services or not. Other than the above mentioned limitations, ndings from this paper gives a rough picture of what are people's perception on the South African health system capability and the current state of health system capacity to deal with COVID-19 outbreak at provincial and national levels across the country.

Conclusion
To the best of our knowledge this paper was the rst to investigated perception of the South African health system capability and its actual health system capacity to respond to the COVID-19 pandemic and its subsequent spread. Overall, most people reported having low con dence in the health system's capability in managing COVID-19 outbreak and that may be due to having the wrong perceptions or may be due to prior bad experiences in their interaction with health care facilities and services. Sex and knowledge on COVID-19 were signi cantly associated with the perception of the health system's capability to manage the pandemic at the national level and in four of the nine provinces. Females were less likely to have con dence on health system's capability to manage COVID-19 outbreak than their male counterparts. Overall, the ndings of this study clearly highlights the challenges on the country's health system both perceived or real that needed to be addressed as part of the preparation for the COVID-19 pandemic. Urgent policy interventions and implementations are recommended for continued increase in the number of ICU beds across the country with particular focus in Western Cape, Eastern Cape, KwaZulu-Natal and Gauteng. Timeous implementation of a countrywide NHI system is now more critical than ever in improving health care outcomes of the South African population even beyond the existence of the COVID-19 epidemic Declarations Consent Informed consents were obtained from participants before participants proceeded to the online survey questions. Participants were informed that their participation was voluntary and that they could withdraw from the online survey at any time should they wish so.
Author Contributions SPR conceived the study and TM conceptualised the paper.  Percentage of people who indicated they believe that the health system was able to manage the COVID-19 outbreak