Title: Risk Perception related to COVID-19 among Sub-Sahara Africans: A Web-based Comparative Survey of local and diaspora residents

Perceived risk towards the coronavirus pandemic is a key to improved compliance with public health measures to reduce the infection. This study investigated how Sub-Saharan Africans (SSA) living in their respective countries and those in the diaspora perceive their risk of COVID-19 outbreak and the factors associated. people’s knowledge about COVID-19 and encouraging positive attitude towards the mitigation measures. Such interventions should target older participants and non-healthcare workers.


Introduction
Risk perception refers to people's subjective assessments of the possibility of outcomes that may follow undesirable events such as disasters and pandemics [1]. Assessment of risk perception is important in health and risk communication as it relates to individuals understanding of risks, their feelings and how they deal with those risks. The ongoing novel coronavirus SARS-CoV2 (COVID-19) pandemic has caused enormous global mortality and public health devastation [2],while renewing our collective awareness of global pandemics [3]. The 2014 West African Ebola Virus Disease (EVD) pandemic was limited to African countries, and severe acute respiratory syndrome (SARS) of 2002-03 limited to Asian countries, COVID-19 has been a global and unprecedented 'black swan' event [4,5]. COVID-19 infection is highly contagious and mortality caused by the virus has exceeded 2.4 million deaths as at 15th of February 2021 ─ more than any of its predecessors [6]. It is therefore no surprise that, countries are in a race towards developing an effective vaccine [7,8].
The COVID-19 virus poses a considerable global threat [9], hence, the World Health Organization (WHO) immediately raised awareness of healthcare workers around the world [10]. The WHO has also raised funds globally and developed Strategic Preparedness and Response Plans (SPRP) to support and protect poorer countries with weak healthcare systems [11]. The goal of the SPRP was to control infection, limit transmission, communicate key information, provide early acute care, and minimize disastrous economic and social effects. National governments locked-down their populations, stopped the mobility of goods and services and closed all schools and universities as well as state and international borders with many employees working from home [12][13][14][15]. Nonetheless, the success of these mitigating measures is dependent upon the public's readiness to comply, which in turn is inspired by their risk perceptions about the pandemic [16]. According to the extended parallel process model (EPPM) of health communication, the two cognitive assessments that may ensue after a person's exposure to a health risk are related to the perceived threat it poses and the perceived e cacy of the recommended advice [17]. This is critical to behavioral changes.
Globally, devastating pandemics such as COVID-19 can provide valuable opportunities to learn about human risk perception and attendant behavior [18,19] and how ndings from such studies can be used to inform allocation of resources within such countries, and within international multilateral organizations and agencies such as the WHO [20,21]. Such studies can also provide an evidence base for formulation of (public) health and risk policy. Severe outcomes from natural disasters are often in uenced by the level and distribution of economic resources and income within the population of a country (or region) [22,23] and well-resourced quality institutions [24]. Several seminal bodies of literature highlight the role of resources or the lack of it in societal responses to disasters [25] and show how positive psychology can contribute to community development during disasters [26].
Risk perception lead to decision-making and adjustment to the risks in a number of ways: they can move out of harm by self-isolation; they can self-protect; or they can prepare in various ways for a disaster, or ignore public warnings [27]. Quinn et al showed that people's attachment to their place of residence has an effect on how people perceive disaster related risk [28]. Hence, risk perception in uences how individuals behave and respond to recommended health behaviours [29]. There are limitations to surveys such as this to assess risk perceptions and compare the perceptions from sub-Sahara African (SSA) residents in and outside Africa. Ideally, one would like to be able to hold political institutions and the cultural context constant to be able to make comparisons. The ndings of this study will highlight the implications of the analysis for what we might expect for Africans living in Africa and in the diaspora (living outside Africa) and policy implications in disaster risk management in general. For policy makers tasked with communicating risk, this research can provide a particularly fruitful lens in addressing the emotional underpinnings of adaptation behaviour.

Methods
Design and Setting of the study This was an online survey created in Survey monkey to assess risk perceptions of Africans. The study was conducted between April 27th and May 17th 2020 corresponding to the mandatory lockdown period in most SSA countries. The survey instrument was adapted and developed from the WHO recommended questions [30] and have been used in previous studies [30]. At this special period of lockdown and restricted mobility, it was not feasible to undertake a conventional Africa-wide community-based sampling survey. A one-page project information statement, which doubled as a recruitment poster, was posted/reposted to WhatsApp and Facebook chat groups and individual accounts together with an e-Link to the online survey. The information sheet and poster contained a brief introduction on the background of the study, its objectives, procedures, the voluntary nature of participation, the declaration of anonymity, privacy and con dentiality as well as instructions for completing the questionnaire.
We also posted the poster and questionnaire on various websites and o cial accounts of several local organisations and individuals.
Survey questionnaires were also sent out by email to selected groups and individuals in all the target countries relying on the authors' networks with collaborating academics and local people.

Questionnaire
The questionnaire was divided into three sections, including demographics, knowledge, risk perception, feeling about self-isolation, attitude towards public health practices to mitigate the spread of COVID-19 (compliance) as presented in Table 1. Most of the items on the questionnaire that assessed the respondent's knowledge of COVID-19, required mostly a 'true' or 'false' or a 'yes' or 'no' response with an additional "I don't know" option. Each question used a binary scale and a correct answer was assigned 1 point whereas an incorrect/unknown answer was assigned 0 point. The knowledge score ranged from 0-19 points. These items have been validated elsewhere to have an acceptable internal consistency [31].
For the risk perception items shown in P1 − P6 of Table 1, each question used a Likert scale with ve levels and the scores ranged from 1 for 'lowest' and 5 for 'highest' with a maximum score range of 5 to 30 points. We determined the Cronbach's alpha coe cients of the perception items to be 0.84, which indicated a satisfactory internal consistency of perception items. Questions were asked on "How the respondents felt about self-isolation" (P7 − P12) with responses utilizing a Likert scale with ve levels and the scores also ranged from 1 (lowest) to 5 (highest). The Cronbach's alpha coe cient of the "How the respondents felt about the quarantine items" was 0.74, which indicated an acceptable internal consistency. Respondents were also asked about their attitude towards the public health measures put in place by the respective governments to reduce the spread of the virus in items A1-A8.

Characteristics of the participants
Participants were those living in South Africa, Nigeria, Ghana, Kenya, Tanzania and Malawi. Respondents in the diaspora including those living in the UK, USA, Australia, Canada, New Zealand and Germany were also included. Recipients were further encouraged to send on or 'snowball' the survey questionnaire to other WhatsApp groups that they know as well as to friends. Eligibility criteria included that respondents had to be of African nationality, aged 18 years or older and able to understand the contents of the poster/questionnaire, and agreed to participate in the study. How do you feel about the Self-isolation?

P7
I am worried about self-isolation.

P8
I am bored by self-isolation.

P9
I am frustrated by self-isolation P10 I am angry because of self-isolation.

P11
I am anxious about self-isolation.

P12
I am angry because of quarantine.
Attitude towards public health practices to mitigate the spread of COVID-19 (Compliance)

A1
Are you currently or have you been in (domestic/home) quarantine because of COVID-19?

A2
Are you currently or have you been in self-isolation because of COVID-19?

A3
In recent days, have you gone to any crowded place including religious events?

A4
In recent days, have you worn a mask when leaving home?

A5
In recent days, have you been washing your hands with soap and running water for at least 20 seconds each time?

A6
Since the government gave the directives on preventing getting infected, have you procured your mask and possibly sanitizer?

A7
Have you travelled outside your home in recent days using the public transport A8 Are you encouraging others that you meet to observe the basic prevention strategies suggested by the authorities?

Dependent variable
The dependent variable for this study was perception of risk for contracting COVID-19, which was categorized as continuous. The items utilized to measure risk perception of COVID-19 are shown in Table 1 (P1-P6). The response included very high, high, low, very low, and unlikely.

Independent variables
These included demographic A) characteristics of the participants which consists of age, gender, marital status, education, employment status, occupation (if employed), religion, if they lived alone, number of people living together in the household and place of current residence.

Sample size determination
The survey assumed a proportion of 50% with 95% con dence and 2.5% margin of error. This is because the main objective of this research was on COVID-19, and there are no previous studies from SSA that examined factors associated with risk perception of 2019-nCoV. An online sample size calculator was used, and we assumed a sample size of approximately 1921, including 20% non-response rate.

Statistical analysis
Scores for risk perception were calculated for each of the independent variables and treated as continuous variable with mean (± standard deviation) risk scores. The risk scores ranged from 1 to 30. Risk scores by independent variables were summarized using t-test for two categorical groups and oneway analysis of variance (ANOVA) for more than two categorical groups. Univariate linear regression analyses were conducted to assess the unadjusted coe cients (B) with 95% con dence intervals among SSA residents and residents in the diaspora. The adjusted coe cients (β) with 95% con dence intervals obtained from the multiple linear regression model were used to measure the factors associated with risk perception of COVID-19 among SSA residents and those in the diaspora. Only signi cant variables in the univariate analysis were used to build the regression model. Knowledge was included in the model because it is strongly related to attitude and practice while knowledge and attitude has been reported to be associated with practice ([32]) . Feeling about the practice of self-isolation during COVID-19 lockdown would help in identifying individuals who could develop mental health issue during the lockdown. Including attitude towards the mitigation practices in the model would in uence action to reduce the spread of the infection. All analysis were performed using Stata version 14.1 (Stata Corp. College Station United States of America) and a two-tailed p-value < 0.05 was considered statistically signi cant.

Demographics of respondents in Africa and in the diaspora
Of 1,969 respondents (55.1% male and 44.9% female) that completed the survey, majority were living in SSA (n = 1855, 92.8%) and 143 (7.2%) in the diaspora at the time of data collection. Figure 1 presents the mean and the 95% CI of risk perception scores towards COVID-19 based on respondents region of residence. There was no signi cant difference in the mean risk perception scores between the two groups (p = 0.117). Table 2 shows the demographics of SSA in Africa and in the diaspora with their mean (standard deviation) scores for perceived risk towards COVID-19. Compared to SSA residents, those living in the diaspora were younger, more often female, and less often married.
Perception of overall COVID-19-associated risk For those in SSA, the risk perception score was signi cantly lower in the 18-28 years age group (p = 0.003, Table 2) than in older age groups. Again, employment (p = 0.040) and higher level of education (p < 0.001, Table 2) had signi cantly higher risk perception scores than being unemployed with lower education, respectively. There was no signi cant difference in the risk perception scores based on gender, marital status, religion, occupation, and the number of people living together, among SSA residents. The risk perception score did not yield any signi cant difference on sociodemographic characteristics among participants living in the diaspora.

Discussion
This study found a comparable high risk perception score among residents living in SSA and those in the diaspora which were associated with increase in knowledge of COVID-19 and attitude towards the mitigation measures.
Older health care workers had higher risk perception scores compared to younger non health care workers.
Although having a lower education and not working during the pandemic was associated with a signi cantly lower risk perception of COVID-19 among local residents, this was nulli ed after adjusting for other demographic variables.
The nding that older individuals felt at greater risk of COVID-19 was in line with past studies showing that older individuals have signi cantly higher COVID-19 related severe complications and deaths than young individuals [33]. Public awareness of this information may explain the nding of lower risk perception for contracting the infection among younger respondents in SSA. As highlighted by Dillard et al [34], having a perceived low risk of infection can make young people become less compliant to the public health measures. This can in turn lead to higher COVID-19 infection [34], and ultimately passing the infection to the population more susceptible to COVID-19 related complications, since young people were shown to be more likely to transmit the virus than others [35]. In line with these ndings, some countries took stringent steps to limit young population from transmitting COVID-19 infection to older population [36][37][38][39] but recorded mixed success [39][40][41]. Rapid and proactive outreach programs targeted at young people in Australia and Canada might explain why the risk perception was similar between younger and older participants living in the diaspora in this study [42]. Such directed programs and policies should be implemented within the vulnerable groups in our local populations.
Studies have reported a high perceived risk of COVID-19 among African health workers [43][44][45] but did not compare between health and non-health workers. In a cross-sectional study conducted on 350 Ghanaians during the early stage of the outbreak, there was no difference in risk perception scores between health and non-health care workers [46]. In this study, high risk perception for contracting COVID-19 was associated with working in the health sector but this was only signi cant among those who were living in SSA. Firsthand experience with the virus is often linked to high-risk perception [47], higher knowledge of the disease among health care workers compared to the non-health workers might explain their higher perception of risk for contracting the infection. The lack of proper training on protective measures reported in previous studies by health workers in SSA countries [45] may explain the signi cant association found among local health care workers but not among those living in the diaspora. Again, the implementation of targeted policies may as well account for the lack of association among respondents living abroad.
In this study, knowledge about COVID-19 and positive attitude towards the mitigation measures were associated with high-risk perception of contracting the disease, both in SSA and the diaspora. Similar ndings have been reported in Ethiopia [48] showing that individuals who perceive higher risk are more likely to adopt protective measures, which in turn in uences the probability of infection [48,49]. However, the prevalence of misinformation about COVID-19 among SSA respondents [50] together with the immoderate psychological stress caused by these misinformation about COVID-19 due to the poor knowledge about the disease [31] can also lead to overestimation of negative risks or inaccurate risk judgement [29,51]. Hence, accurate information about the pandemic using the trusted media platforms, can help in accurate risk judgement and proper adoption of public health measures to control the spread of infection [31,52]. COVID-19 related morbidity and mortality vary disproportionately based on socio-demographic characteristics, for instance, males and older people have high mortality due to COVID-19 compared to females and young population [53]. Individual's behaviours towards safety measures have been linked to their level of perceived risk of a disease [34]. Adopting public health measures such as the use of nose mask in public areas and frequent hand sanitization, can lead to successful control of air-borne infectious diseases like COVID-19 [52]. Therefore, public health strategies for successful control of COVID-19 among SSAs may bene t from targeting the sub-population identi ed in this study.
That is, the unemployed, non-health care workers, the younger population and those with lower levels of education.
This study was limited by several factors, which restricts the generalizability of the present ndings. It was an online survey made available only in English language thus restricting respondents without access to the internet where internet penetration remains relatively low and some from French-speaking SSA nations [54]. However, the use of an internet-based methodology was the only reliable means to disseminate information at the time of this study. Notwithstanding these limitations, this study from the SSA region provided insight into the role of place of residence in mitigating the factors that in uence risk perception of COVID-19 among SSAs during the pandemic. The study used a robust analysis to control for potential confounders during the analysis in order to reduce the issue of bias.

Conclusions
In summary, this study explored the factors associated with the risk perception of contracting COVID-19 among SSAs particularly looking at the role of place of residence in peoples' level of risk perception. The ndings indicate that health communication and education strategies, which are designed to promote the adoption of preventive behaviours among SSAs, should focus on increasing knowledge about the disease and encouraging positive attitude towards the mitigation measures.
In addition, such programmes will bene t from targeting the unemployed, less educated and non-health care workers for optimum outcome. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests