key: cord-0725125-bn3iwkgy
authors: Nshakira-Rukundo, Emmanuel; Whitehead, Anne
title: Changing perceptions about COVID-19 risk and adherence to preventive strategies in Uganda: Evidence from an online mixed-methods survey.
date: 2021-11-16
journal: Sci Afr
DOI: 10.1016/j.sciaf.2021.e01049
sha: 0aee5b0fa8b2e689e12f8b16d097848b38d4dd2f
doc_id: 725125
cord_uid: bn3iwkgy

Since the COVID-19 pandemic started, countries have enacted a series of non-clinical preventive mechanisms aimed at slowing the rate of spread. However, these mechanisms can be effective only when they are correctly followed and only when individuals still believe the risk of COVID-19 is high enough to warrant following them. As risk perceptions decline, individuals are more likely to relax in following preventive measures and the rate of spread might increase. This study assesses the determinants of changes in perceptions of COVID-19 risk and the determinants of adherence to preventive measures in Uganda. Logistic regression results show that age, access to information and being supportive of preventive measures strongly predicts keeping higher risk perceptions and adhering to preventive actions. Qualitative results show that risk perceptions are also influenced by economic stress, citizens’ level of confidence in the government, local political climate and the extent of proliferation of misinformation about COVID-19.

In responding to the fast-spreading COVID-19 virus, governments imposed non-clinical pandemic control policies such as social distancing, lockdowns and curfews. Evidence suggests that these measures curtailed the fast spread of the virus 1,2 . However, these measures also were associated with worsening socioeconomic conditions 3 and associated mental health challenges and fatigue 4, 5 . Studies around the world have shown that where higher risk perceptions existed, knowledge and attitudes towards prevention and containment were high 6 . However, some section of the population still espouses low-risk perceptions and prevention mechanisms 7 .

As the pandemic drags on and prevention policies double down in the more deadly second and third waves, change in perceptions will be expected as economic 3 and social anxiety 8 increases. Moreover, the increase in misinformation 9,10 is likely to further negatively affect adherence to prevention strategies. This has been the experience in previous pandemics 11 . While many studies have investigated knowledge, actions and practices (KAP) during the coronavirus pandemic 12 , fewer studies have investigated whether perceptions about the pandemic negatively change given the increase in pandemic fatigue 13 . Moreover, changing perceptions might derail the progress made on prevention and curtailing the spread. Developing countriesespecially those in which previous pandemics exposed health systems weaknesseswould be at higher risk 14 further underscoring the necessity of positive risk perceptions and prevention.

In this study, we focus on Uganda. We are interested in two main questions. First, we study the correlates of adherence to preventive strategies set by the government. One study in this regard is Amodan et al 15 who assessed the determinants of adherence to preventive measures in Uganda. Secondly, we assess the correlates of change in risk perceptions of COVID-19. Using an internet survey, we qualitatively and quantitatively highlight the extent of mostly negative perception changes and adherence to preventive strategies. We show access to information, local politics and economic situations influence how people perceive the risk of COVID-19 and their patterns of adhering to preventive actions. We find some indication of misinformation proliferation, but also notice that experience of previous epidemics provides individuals with reference points to build trust in the health system. Section 2 provides a brief background of COVID-19 in Uganda and methods and data analysis strategy are in Section 3. We provide results in Section 4, elaborate on them in discussion Section 5 and conclude with Section 6.

Uganda registered her first COVID-19 case on 21 March 2020 and by January 2021 had registered close to 40,000 cases and over 320 deaths. To curtail the spread, the country implemented lockdown and social distancing strategies in March 2020 16 . Though the country's case burden remains and is predicted to remain low 17 , the economic effects have been deeply regressive. In rural Uganda, non-farm income declined by 60 percent 18 and increasing food insecurity was recorded among a greater share of the population 19 . While knowledge, attitudes and practices about COVID-19 were higher among Ugandans 12 , some cohorts of the population registered various harmful misconceptions 20 that might affect adherence to prevention strategies. In 2020, Uganda also had election campaigns for the January 2021 general elections and individuals were more likely to renege on standard prevention mechanisms during campaign events. The country, therefore, presented an appropriate opportunity to study adherence and perceptions changes that would be very useful for other countries. Moreover, Uganda has had several outbreaks of haemorrhagic fevers in the last two decades 21 , which were successfully handled before they became full-fledged epidemics 22 . It is argued that Uganda's model of pandemic response can be helpful to other countries of low resource settings 23 .

The study, therefore, sought to establish four key issues; namely (1) the extent of adherence to preventive strategies, (2) determinants of adherence to prevention strategies, (3) extent of negative changes in risk perceptions between March when the first case was recorded and July 2020, three months into the pandemic in the country and (4) determinants of perception changes. In one dimension, we contribute to research elaborating on adherence to prevention mechanisms in Uganda 12 . However, this goes further to assess the extent to which individuals change perceptions from higher risk to lower risk perceptions.

To study these questions, we used a self-administered public opinion survey conducted in Uganda. The survey was run online on the Survey Monkey platform and circulated on Twitter, Facebook and WhatsApp between 6 th and 15 th July 2020. As an incentive for participation, respondents were entered into a raffle draw to win internet bundles. Altogether, 1353 individuals responded. In this analysis, we use 1195 observations for which all the relevant information was present.

Our main analysis deduces from two main questions. First, to establish the level of adherence to prevention strategies imposed by the governments, we asked respondents the question "How serious are you about following COVID -19 rules like physical distancing and wearing a mask?" The question had four responses categorised as 1= very relaxed, 2 = somewhat relaxed, 3= somewhat serious and 4 = very serious. At the analysis level, we recorded the responses to this question into a dummy variable corresponding to 1= seriously following COVID-19 rules and 0= otherwise (relaxed about following the rules). To assess the extent of change in perceptions, respondents answered the following question. "How much do you fear the COVID -19 disease now (July) compared to how you felt in March" The responses were 1 = "I fear COVID -19 less now (July) than I did in March" and 0 = "I fear COVID -19 more now (July) than I did in March". Individuals responding that they feared the diseases less in July than they did in March 2020 would then provide a proxy reduced risk aversion and hence a negative change in perceptions regarding the risk posed by COVID-19. As both questions were then analysed as binary outcomes, we implemented two separate logistic regressions of the following form.

Where was a dummy outcome corresponding with 1 if individual was following COVID-19 rules and 0 otherwise or 1 if individual feared COVID-19 less in July 2020 than they did in March 2020. is a vector of covariates and is the error term. In the two logistic regressions, we control for age, gender, education level, household expenditure, and information access, employment status represented by two dummy variables for formal sector and self-employment and a dummy of working from home. We also control for the perception of how helpful presidential COVID-19 addresses on television were, relief support from the government, level of approval of lockdown and curfew prevention mechanisms. We also include regional dummies to account for geographical variation. We report 95% confidence intervals and robust standard errors. Summary results are in Supplementary tables 1 and 2 for the two outcomes respectively.

The questionnaire also provided space for individuals to give reasons why they responded as they did concerning their perceptions of COVID-19 fear/ risk. We extracted this qualitative data and using simple thematic analysis, we analyse it to accompany quantitative analysis.

First, we show the results of adherence to COVID-19 prevention strategies. The strategies assessed in this study were social distancing and mask-wearing. Figure 1 below shows the descriptive results of how seriously people considered preventive measures. We found that just over 63% of the respondents self-reported that they were seriously adhering to COVID-19 regulations. A large proportion of respondents (36.6%) reported that they were not adhering to these regulations, indicating they were either very relaxed or somewhat relaxed. Secondly, we assessed the prevalence of perception changes. Out of a total of 1323 individuals who responded to this question, 85% said they feared COVID-19 less in July than they did in March 2020. While not all the perception change is a true reflection of an increase in risk loving and less prevention, it is a very close proxy. The question provided an option for respondents to give a reason why they felt the way they did in regards to fear of COVID-19. About 8.5% (102 respondents) provided this additional qualitative data. Incorporating this qualitative data, we confirm that a change in perceptions is most likely negative and makes individuals feel less COVID-19 risk. We coded each of the 102 qualitative responses on how much they were negative (less preventive behaviour) neutral or positive (more preventive behaviour) perception changes. Sixty-one percent had negative perception changes compared to only 10 percent whose perceptions were more positive, more risk-averse and hence predictive of higher prevention actions.

Looking into the qualitative reasons why individuals changed perceptions, with a few selected and informative quotations from the respondents, we highlight some of them. First, COVID-19 in Uganda was associated with sharp declines in living standards, income losses and increasing food insecurity 18, 19 . It therefore seems the dire economic situation was more prominent in individuals' risk assessment than the risk of the disease. One respondent put it:

"I don't fear it anymore. The current situation has proven that we need to coexist with it because people are dying from mental illnesses, police brutality, lack of food and access to health care services instead." Female respondent -18-25 years

The statement by this respondent also pointed to an important theme of the politics of pandemic response and how that affected how individuals think about the pandemic. In some countries, governments have used the pandemic as a toll for increased marginalisation of some sub-groups 24 . In Uganda, the pandemic response would not be separated from the politics of a very heated election season. Military and police brutality were observed with military and police enforcing regulations 25, 26 including the November 2020 events in which at least 50 individuals lost their lives 27 . Several respondents, therefore, felt that the risk associated with militarised pandemic response was possibly higher than of the disease itself. One male, aged between 26 and 34 years inferred that:

"I fear the local defence unit (LDU) more than COVID-19."

Agreeing with another male in the same age group who also commented that A female respondent between 35 and 44 years also intimated that:

"I know COVID -19 exists but the Ugandan government is taking undue advantage of it."

Corruption in Uganda's health sector 28 and in general is well documented 29 . The low expectation of pandemic management from a financial governance perspective, therefore, gives individuals less belief that the pandemic itself is dangerous if it can be used for economic gain.

The other reason for the negative perception changes is the relatively low case numbers in Uganda and Africa at large. Low COVID-19 mortality and morbidity have increasingly become a puzzle 30, 31 . The outcome of this in individual prevention behaviour and practices is that individuals and policymakers might develop complacency in pandemic management strategies. Because many individuals would not have encountered COVID-19 casualties in their social networks, their behaviour towards prevention would have regressed. We find this to be the case among some people in our sample. One young female respondent said: However, not all perception change is negative. We observed positive perception change in about 10% of those who provided additional qualitative statements. This was close to the general question on perception change in which 14 percent indicated that they feared COVID-19 more in July than they did in March. Two key themes seem to influence positive perception changes that induce more behaviour that is protective. The first is confidence in the health system, especially regarding its experience with handling previous viral diseases such as Ebola and Marburg virus outbreaks 22 . The success in handling previous Ebola and Marburg viruses not only provided best practices for other low resource countries 32 but also should potentially lay the ground for management of current and future pandemic outbreaks 33 . Some individuals felt confident that COVID-19 was and would continue to be handled successfully. One respondent commented: The second theme is continued vigilance with self-protection. Individuals who did not negatively change their perception also showed more adherence to self-protective preventive behaviour. One respondent wrote: These qualitative results reveal a couple of things. First, there is some substantial misinformation in developing countries about COVID-19. While this study does not seek to establish the level of misinformation, other studies in countries like Uganda find high levels of misinformation 34, 35 .

There is a high likelihood that such misinformation is increasing risky behaviour and nonadherence to preventive strategies. Secondly, we learn that political and socio-economic reasons are likely to increase pandemic apathy. This is more so in countries like Uganda that are characterised by reduced civic space, electoral seasons and pre-existing high levels of corruption and less government trust. However, we also note that there are still some islands of hope for individuals' faith in the health system, especially building on experiences from the previous management of viral haemorrhagic fevers.

We report results from two logistic regression models in Table 1 . Odds ratios are presented with 95% confidence intervals. One key factor that increased the odds of changing perceptionemployment in the formal sector. Employment in the formal sector, compared with not having a job, was associated with more than 50% higher odds of negatively changing perceptions (OR 1.505; 95% CI 0.937 -2.416). Across the regions, higher odds of perception changes were observed in Eastern and Northern regions and in areas in and around the capital, Kampala. Compared to individuals who listened to news only about once a week, those who listened to or watched the news at least daily or several times a day were twice (OR 2.066; 95% CI 1.211 -3.524) and 78 percent (OR 1.783; 95% CI 1.039 -3.060) more likely to adhere to preventive measures. Individuals who approved and those who strongly approved of lockdown measures were more than 2.1 times (OR 2.143; 95% CI 1.167 -3.936) and more than 2.2 times (OR 2.242; 95% CI 1.226 -4.099) more likely to adhere to preventive behaviour. Approving of curfew measures (which followed lockdown) was associated with higher odds on adherence to preventive behaviour. Odds of adhering to preventive actions increased from 60% (OR 1.687; 95% CI 1.174 -2.424) for individuals who disapproved of curfew measures to more than six times higher (OR 6.109; 95% CI 3.745 -9.967) for individuals who strongly approved of curfew. 

This paper contributes to the broader literature of COVID-19 prevention and in particular contributes to studies in Uganda documenting how the country has dealt with the pandemic, especially regarding risk perceptions and adherence to prevention mechanisms. We find that age, being a woman, listening to news and approving of pandemic response mechanisms were correlated with reducing the odds of changing risk perceptions and adhering to preventive practices.

The findings on the age and behaviour of older people are not at all surprising. While developing countries like Uganda generally have younger populations, older people are more at risk of COVID-19. Older people are therefore more likely to consider the disease deadly than younger people and therefore maintain risk-averse behaviour rather than relax their perception of risk. Older people are also more likely to continue adhering to prevention behaviour than younger people.

The results regarding employment corroborate the evidence of job and income losses across the country 18 . Moreover, our qualitative evidence also shows that economic pressures associated with the side effects of lockdown policies were imposed detrimental effects on households than the immediate risk of COVID-19. Therefore income and employment security were key in people's response and risk perception 36 .

In principle, one would expect no differences between men and women. However, we observe that women were more likely to adhere to prevention behaviour compared to men, similar to another study in Uganda 15 despite men believing they were more at risk than women 20 . We advance two hypotheses to explain this gender-based difference. First, possibly due to the nature of men's occupations and level of involvement, it is logical for them to believe they are more at risk than women 20 . Men are the main breadwinners for their households and often take on more risky occupations and activities than women. These activities are likely to affect their ability to adhere to social distancing and other preventive measures. Secondly, men often have poor healthseeking behaviour 37 which might also influence their risk perceptions and COVID-19 prevention behaviour. Our findings, in this case, resonate with those in Pakistan, where it was also found that men generally had lower COVID-19 risk perception than women 38 .

We further observe that individuals who consume more news, especially traditional news media like radio, television or newspapers are more likely to take COVID-19 more seriously. Access to news remains a challenge in developing countries. While radio stations have increased in countries like Uganda, radio ownership remains at less than 60 percent of the population and less than 20 percent own a television 39 . Moreover, equity in access to news remains a challenge with more skewness to urban residents and men. However, a larger proportion of the population lives in rural areas, which might also be underserved with health facilities. This might imply that a prolonged pandemic might unevenly affect rural, underserved and unconnected communities.

Suffice to mention that we do not find a significant correlation between receiving news from social media and perception of COVID-19 seriousness. In our sample, 84% received news from a social media platform, so we could rule out the hypothesis that social media penetration was higher than average among survey respondents, likely because the survey was primarily distributed through Twitter, WhatsApp and Facebook. However, we find a negative correlation between social media and changing perspectives about COVID seriousness. Though the correlation is not statistically significant, it suggests that receiving news from social media might discourage people from taking COVID-19 seriously. This is in line with the recent literature on misinformation in pandemics, especially through social media 40 . Moreover, government public health organisations are more likely to transmit COVID-19 related information on traditional media platforms 41 which could explain the strong positive association.

Approving of lockdown restrictions is likely associated with the socioeconomic class as poor people are more likely to be affected than richer people 42 . We, therefore, think whereas poor people might also consider the disease serious, their precarious socioeconomic conditions cannot allow them to stay without work for extended periods to strictly observe lockdown and social distancing measures. Moreover, in developing countries like Uganda, this vulnerability is worsened by a large informal sector and the absence of unemployment insurance.

One key limitation of our work is our internet sampling strategy that might limit the generalisability of the results. Generally, internet-based surveys have become increasingly important during the COVID-19 pandemic 43 . However, they might not be precisely representative due to low internet penetration rates in developing countries because of low internet penetration and subscription. Respondents are therefore more likely to be male, urban and more educated than the average population 44 . We, therefore, limit our interpretation to an online population rather than the general population.

In this study, we utilise an internet survey to shed more light on the extent of changing perception regarding the risk of COVID-19 in Uganda. We found that about 85% of the sample were worried about COVID-19 more in March than in July 2020. We further found that just about 63% of the sample were either somewhat seriously or very seriously following prevention recommendations such as regularly using facemasks and social distancing. Epidemiologists have predicted that, due to some advantages such as younger populations and warmer weather, African countries like Uganda might be spared a full-blown pandemic 31 . However, complacency associated with relaxing risk perceptions and associated preventive measures could be very dangerous. We found that being an older individual, receiving COVID-19 related news especially from traditional media platforms, and approving of prevention policies such as lockdowns and curfews, were predictive of maintaining preventive behaviour and higher COVID-19 risk perception. However, these results should be read with the caveat that internet surveys are less representative of the average population.

A country level analysis measuring the impact of government actions, country preparedness and socioeconomic factors on COVID-19 mortality and related health outcomes

Effective containment explains subexponential growth in recent confirmed COVID-19 cases in China. Science (80-)

Falling Living Standards during the COVID-19 Crisis: Quantitative Evidence from Nine Developing Countries

COVID-19 pandemic and psychological fatigue in Turkey

Living with Chronic Fatigue Syndrome during lockdown and a global pandemic

Risk perceptions of COVID-19 around the world

Changes in risk perception and self-reported protective behaviour during the first week of the COVID-19 pandemic in the United States

COVID-19 in Africa: socio-economic impact, policy response and opportunities

Irrational beliefs differentially predict adherence to guidelines and pseudoscientific practices during the COVID-19 pandemic

Misinformation making a disease outbreak worse: outcomes compared for influenza, monkeypox, and norovirus

Institutional trust and misinformation in the response to the 2018-19 Ebola

Community Drivers Affecting Adherence to WHO Guidelines Against COVID-19 Amongst Rural Ugandan Market Vendors. Front Public Heal

Intervention fatigue is the primary cause of strong secondary waves in the COVID-19 pandemic

Health-system resilience: reflections on the Ebola crisis in western Africa

Level and determinants of adherence to COVID-19 preventive measures in the first stage of the outbreak in Uganda

Early cases of SARS-CoV-2 infection in Uganda: epidemiology and lessons learned from risk-based testing approaches

Predicting the impact of COVID-19 and the potential impact of the public health response on disease burden in Uganda

Household response to an extreme shock: Evidence on the immediate impact of the Covid-19 lockdown on economic outcomes and well-being in rural Uganda

COVID-19 implications on household income and food security in Kenya and Uganda: Findings from a rapid assessment

Misconceptions on COVID-19 Risk Among Ugandan Men: Results From a Rapid Exploratory Survey

Repeated outbreaks of viral hemorrhagic fevers in Uganda

Uganda ' s experience in Ebola virus disease outbreak preparedness

Uganda as a role model for pandemic containment in Africa

Antidemocratic and Exclusionary Practices: COVID-19 and the Continuum of Violence

COVID-19, Public Authority and Enforcement

Young People, Inequality and Violence during the COVID-19 Lockdown in Uganda. London

Death toll from riots rises to 50. The Daily Monitor

Challenges in Multi-Level Health Governance: Corruption in the Global Fund's Operations in Uganda and Zambia

Elite corruption and politics in Uganda

The relatively young and rural population may limit the spread and severity of COVID-19 in Africa: A modelling study

Why is there low morbidity and mortality of COVID-19 in Africa?

What can we learn from Uganda on fighting deadly disease outbreaks? World Bank Blogs

COVID-19 Response in Sub-Saharan Africa: Lessons from Uganda

COVID-19 Misinformation spread in eight countries: Exponential growth modeling study

Misinformation sharing and social media fatigue during COVID-19: An affordance and cognitive load perspective

Lockdown measures in response to COVID-19 in nine sub-Saharan African countries

Men and health help-seeking behaviour: Literature review

COVID-19 risk perception and coping mechanisms : Does gender make a difference ?

Uganda Demographic and Health Survey

Misinformation During the Coronavirus Disease 2019 Outbreak

Seeking Formula for Misinformation Treatment in Public Health Crises: The Effects of Corrective Information Type and Source

The economic impact of COVID-19 lockdowns in sub-Saharan Africa

Social media as a recruitment platform for a nationwide online survey of COVID-19 knowledge, beliefs, and practices in the United States: Methodology and feasibility analysis

Survey sampling in the Global South using Facebook advertisements

The authors have received no funding for this work.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.