Introduction

During public health crises, information can shape individuals’ knowledge of and/or attitudes about health interventions, and ultimately inform their behaviors1,2,3,4. The COVID-19 pandemic, which occurred amidst widespread technology use and social media connectedness, was deemed an “infodemic” by the World Health Organization – i.e., it was characterized by an overabundance of information, both accurate and inaccurate, about the virus, its origins, and control measures, including vaccines5. Misinformation can contribute to vaccine hesitancy6,7,8, and the spread of vaccination misinformation accelerated during the COVID-19 pandemic9 which has been linked to decreased COVID-19 vaccination intent and behavior10,11.

The spread of misinformation on social media platforms, in particular, has been identified as a contributor to vaccine hesitancy. Social media can rapidly spread false information about vaccines, eroding trust in critical health communication institutions12. Vaccine hesitancy has been linked to following, sharing, and interacting with untrustworthy and low-quality online sources13. Vaccine uptake has also been found to be influenced by the type of information encountered on social media; those who received the vaccine were exposed to more pro-vaccination messages, while those who did not get vaccinated encountered significantly more anti-vaccination content14. While social media use has been linked to vaccine hesitancy, the extent to which social media contributes to vaccine hesitancy varies across platform and population15 so understanding this dynamic in diverse contexts is essential.

The COVID-19 information and misinformation landscape in low- and middle-income countries – and how it is associated with low COVID-19 vaccine uptake in these settings16 – remain poorly understood17. In Malawi, only approximately 1 in 4 people have received a COVID-19 vaccination18 and although misconceptions and conspiracy theories about the COVID-19 vaccine have been documented in Malawi and elsewhere in the region19,20,21,22,23, the broader information environment has not been well-characterized, nor has its association with vaccination behavior. Globally, much of the literature on the COVID-19 “infodemic” has focused on the dissemination of (mis)information online and via social media24,25, and this is also relevant in low-income countries where access to technology is rapidly changing.

Understanding what information people are exposed to, the sources and tone of this information, and how this information influences behaviors may help to inform vaccination promotion strategies in settings with low vaccine uptake. To this end, we conducted a survey to describe the information environment around COVID-19 and the COVID-19 vaccine in Malawi, and assessed how individuals’ exposure to information (from different sources, of differing tone) was associated with vaccine uptake. We find that people – especially men, people in cities, and people with greater education attainment—report exposure to numerous and varied sources of information about the COVID-19 vaccine; and people who hear more information (and most positively-framed information) have higher odds of having received the COVID-19 vaccine than their peers. Information heard on social media and via messaging apps is least commonly viewed as positive in tone. It is also very common to have heard COVID-19 vaccine misinformation, and people who agree with these conspiracy theories – although the minority of respondents – have much lower odds of having received the COVID-19 vaccine.

Methods

Study context

At the time of data collection (June 2022), Malawi had reported a total of approximately 86,000 confirmed cases of COVID-19 and over 2600 deaths attributed to virus. The COVID-19 vaccine was introduced in Malawi in March 2021 but uptake was low; as of December 2023, only 22% of the Malawian population had completed the primary vaccine series26. Vaccine hesitancy in Malawi, which contributed to low uptake, was exacerbated by the spread of misinformation and conspiracy theories on social media, as well as a general mistrust of the healthcare system27. Similar patterns of vaccine hesitancy driven by misinformation have been reported in other African and low-income countries20,22,28.

Study design, site and participant selection

We conducted a cross-sectional survey of adults presenting at 32 health facilities supported by Partners in Hope, a Malawian non-governmental organization that supports implementation of the national HIV program. Sites were selected to represent both public and faith-based health facilities in urban, peri-urban, and rural areas across all three of Malawi’s regions (Northern, Central, and Southern). Eligible respondents were at least 18 years old; visiting a selected facility for care in the outpatient department, the non-communicable disease clinic, or the HIV care and treatment clinic; and not acutely ill on the day of recruitment.

We used systematic random sampling (every second individual in the queue for a provider was approached) to invite individuals to participate in the survey. This analysis uses data from a study by our team that was focused on quantifying uptake of the COVID-19 vaccine, and it was powered to detect a 5% difference in vaccination coverage (chosen to represent a programmatically meaningful difference) between respondents with and without HIV with 95% confidence and 80% power29.

Survey domains and variable definitions

Survey questions reflecting the data used in this analysis are shown in Supplementary Note 1. Respondents were asked about their exposure to COVID-19 vaccine information: had they heard any information about the COVID-19 vaccine from thirteen different potential sources of information, spanning traditional media (newspaper, radio, television), health system and the government (health care provider, government/Ministry of Health), community figures (faith leaders, traditional medicine practitioners), personal relations (family members, friends, neighbors), and social media (posts on Facebook, Instagram or Twitter from a known person, or from an organization, company, or someone not known personally, and conversations/groups in a messaging app). For analysis, we counted the number of different information sources a person said they had been exposed to; and, for each information source, we estimated what proportion of respondents had heard information from that source. Respondents were also asked who they trusted the most to give them advice regarding the COVID-19 vaccine.

For each source that a respondent had heard COVID-19 vaccine information from, they were asked about the tone of that information: did they feel the information was overall positive, neutral/factual, negative, or a mix of positive and negative. Respondents could also answer “I don’t know.” We counted how many positive and negative information sources each person said they were exposed to and what proportion were reportedly positive or neutral/factual (i.e., number of positive or neutral/factual information sources divided by the total number of information sources exposed to).

We asked whether the respondents had heard each of ten common statements (circulating conspiracy theories in Malawi) about COVID-19 and the COVID-19 vaccine30,31. If a respondent said they had heard statement, we asked if they believed it to be true, false, or did not know whether it was true or false. Exposure to COVID-19 misinformation was characterized by counting how many statements (here classified as conspiracy theories) each respondent had heard; and endorsement of COVID-19 misinformation was quantified as the count of conspiracy theories believed to be true, and the proportion of heard conspiracies believed to be true (number of conspiracies believed to be true divided by the number of conspiracy theories heard).

Vaccine uptake was defined as having received any (1 or more) doses of any manufacturer’s COVID-19 vaccine available in Malawi before or at the time of the survey (Johnson & Johnson, AstraZeneca, or Pfizer).

Respondents were also asked about their sociodemographic characteristics: age, gender, marital status, place of residence, religion, educational attainment, employment status, and household income sufficiency over the prior year.

The survey instrument was developed in English and translated to Chichewa, the local language. It was then reviewed by bilingual (English/Chichewa) study team members to ensure clarity and meaning.

Data collection

The survey was administered face-to-face and was interviewer-administered in a private area within the health facility by a trained research assistant. All respondents provided oral informed consent for anonymous data collection prior to commencing the survey, and all responses were recorded using the SurveyCTO mobile data collection platform on Android tablets. All respondents received 4000 Malawi kwacha (approximately US$5) upon completing the survey.

Data were collected during May and June 2022. At the time of data collection, all adults 18 years of age or older were eligible to receive the COVID-19 vaccine in Malawi. Ethical approval for the study was obtained from the National Health Sciences Research Committee in Malawi (#2883) and the University of California Los Angeles Institutional Review Board (#22-000380).

Statistics and reproducibility

We described sample characteristics, the information environment (exposure to and tone of information sources), and exposure to and endorsement of COVID-19 misinformation. Chi-square tests, t-tests, and multivariable logistic regression models (all with two-sided tests of statistical significance) were used to assess whether these variables were associated with respondents’ sociodemographic characteristics and with their COVID-19 vaccination status; these were also converted to estimated marginal effects for ease of interpretation. All analyses were conducted in Stata v17 using the svyset command to account for clustering at the sampling (health facility) level.

Results

Per AAPOR standard definition for Response Rate 332, the survey had a response rate of 95.1%. A total of 944 individuals were approached for participation in this survey; 3 were found to be ineligible (<18 years of age) during screening (0.3% of the 898 screened), and 46 declined to be screened or to consent to participation (4.9% unknown eligibility) but we estimate that 45.86 of these individuals would have been eligible based on the eligibility rate among those screened. The remaining 895 individuals (99.7% of those known to be eligible, or 95.1% of those known or estimated to be eligible) completed the survey and are included in this analysis. Each completed survey lasted, on average, 29 min (median 28 min).

Nearly half (43%) of respondents had received at least one dose of a COVID-19 vaccine. Over half of the respondents were female (57%) and three-quarters (75%) were married. The majority of the respondents were residing in rural areas (82%), were employed (68%) and identified as Christian (91%) (Table 1).

Table 1 Sample characteristics (n = 895)
Full size table

Information environment

Exposure to information about COVID-19 vaccine

All but one survey respondent had heard information about COVID-19 vaccines from at least one source. Respondents had been exposed to information from a median of 7 sources (IQR 6–9). The most common sources were friends and neighbors, health care workers, and radio – each reported by >90% of respondents (Fig. 1, Supplementary Data 1). Receiving vaccine information from a traditional medicine practitioner was least common (3% of respondents). Approximately one-quarter to one-third of respondents said they had heard information on social media, messaging apps, television or newspapers.

Fig. 1: Exposure to information about COVID-19 vaccine, by source and tone (n = 895).
figure 1

Bars represent the percentage of respondents who reported having heard information about the COVID-19 vaccine from each source; and among respondents who had heard information from that source, the percentage who characterized that information as positive or neutral, mixed or unknown, or negative in tone. Includes full sample of 895 respondents; 2 people did not answer the question about having heard information from family members, and 1 did not answer about having heard information from religious leaders.

Full size image

Men, urban residents, and respondents with a higher education level were exposed to a higher mean number of COVID-19 vaccine information sources (men 8.2 [SE 0.18] versus women 6.9 [SE 0.17]; t-test p < 0.001; urban 8.6 [SE 0.27] versus rural 7.2 [0.15]; t-test p < 0.001; secondary or higher education 9.3 [SE 0.17] versus primary school or less 6.8 [0.14]; t-test p < 0.001). No differences were seen in the mean number of information sources by age or marital status.

Among specific sources, information exposure gaps in traditional and social media were particularly wide for women versus men, and people with greater versus lesser educational attainment (Fig. 2). People in urban areas also much more commonly were exposed to vaccine information from social media compared to their rural counterparts (Fig. 2). Higher shares of older people, married people, men, people in rural areas, and people with greater educational attainment had heard information from health workers or the Ministry of Health, and from personal relations, than their younger, unmarried, female, urban, and less-educated peers – but in general these gaps were smaller than for other information sources (Fig. 2).

Fig. 2: Estimated percentage of each respondent group who received COVID-19 vaccine information from various sources (n = 895).
figure 2

Circles represent estimated margins based on odds ratio point estimates. Green circles represent respondents by educational attainment (some, or all, primary-level education; or some, all, or beyond secondary-level education). Orange circles represent respondents by gender (male; or female). Purple circles represent respondents by location of residence (rural; or urban). Blue circles represent respondents by current marital status (married; or unmarried). Yellow circles represent respondents by reported HIV status (living with HIV; or not living with HIV or status unknown). Gray circles represent respondents by age group (18–39 years; or over 40 years of age). Includes full sample of 895 respondents.

Full size image

Tone of information

Most people who had heard information about COVID-19 from health care workers and the government/Ministry of Health, and from traditional media (newspapers, television, radio), perceived this information as positive in tone (Fig. 1, Supplementary Data 1). Approximately one-fifth of people exposed to COVID-19 vaccine information on social media and messaging apps, 41% of people exposed to COVID-19 vaccine information from traditional medicine practitioners, and just over 10% of those exposed to COVID-19 vaccine information from peers (friends, neighbors, and community members) and family members, felt it was negative in tone.

Women reported that a greater share of their information sources about COVID-19 vaccine was negative in tone: on average, women considered 10.2% of their information sources about COVID-19 vaccines negative in tone (SE 1.2), versus 7.6% for men (SE 6.2) (t-test p = 0.053). No differences were seen by respondent educational attainment, area of residence, or marital status.

Most trusted information sources

Respondents were asked who they trust most for advice regarding the COVID-19 vaccine (Supplementary Table 1). A health care provider was selected most often (69.4%), followed by a community health worker (22.1%); family, friends, religious leaders, and celebrities were selected infrequently (0.2–5% of respondents each). No significant differences in trusted information were seen by gender, age, or education level. Unvaccinated individuals twice as commonly said they trusted family, friends, or a religious leader for COVID-19 vaccine information compared to individuals who had been vaccinated (10.6% versus 5.2%).

Association of COVID-19 vaccine information with vaccine uptake

Having received information about COVID-19 vaccines from more sources was positively associated with COVID-19 vaccine uptake, with each additional source increasing vaccination probability by 9% (OR 1.09, 95% CI [1.03–1.16]). Adjusting for gender, residence (urban/rural), age, HIV status, and education, this association persisted (aOR 1.08, 95% CI [1.01–1.16]). The adjusted odds of vaccination was much higher when respondents reported a greater proportion of COVID-19 information being positive in tone (aOR 3.53, 95% CI [1.42, 8.76]).

Misinformation about COVID-19 vaccines

Respondents reported having heard a median of 5 of the 10 surveyed conspiracy theories. The most commonly heard were “The spread of COVID-19 is a deliberate attempt to reduce the size of the global population” (80.7% of respondents [SE 16.2]) and “The COVID-19 virus is a hoax” (74.1% of respondents) [SE 15.7] (Fig. 3, Supplementary Data 2).

Fig. 3: Prevalence of COVID-19 conspiracy theory beliefs across demographics.
figure 3

Heat map displaying the percentage of survey respondents (overall, by gender, and by place of residence) who reported hearing of 10 common COVID-19 conspiracies (n = 895) Color scale spans from white (smallest proportion) to dark red (highest proportion).

Full size image

Men, respondents with higher educational attainment, and urban respondents were exposed to more conspiracy theories: men had heard 5.51 theories [SE 0.21] while women had heard 4.76 [SE 0.16] (t-test p = 0.0001); urban residents had heard 5.70 [SE 0.31] and rural residents had heard 4.95 [SE 0.17] (t-test p = 0.002), and individuals with a secondary school education or higher had heard 5.85 [SE 0.16] while those with primary school or less had heard 4.79 [SE 0.18] (t-test p < 0.001).

Most respondents (72%) did not believe any of the 10 conspiracy theories to be true. Among those who did believe any to be true (n = 254), they endorsed (believed to be true) two conspiracy theories on average. Younger age groups, urban respondents, and respondents with higher education attainment endorsed significantly more conspiracy theories than their older, rural, and less-highly educated counterparts (Supplementary Table 2).

The most commonly endorsed conspiracy theories were “The government is exaggerating the number of COVID-19 deaths” (22.7%), “Big pharmaceutical companies created COVID-19 to profit from vaccines” (17.3%), “The spread of COVID-19 is a deliberate attempt to reduce the size of the global population” (16.9%) and “Bill Gates has created COVID-19 in order to reduce the world population” (15.2%) (Table 2).

Table 2 Endorsement of conspiracy theories and association with COVID-19 vaccine uptake
Full size table

Association with COVID-19 vaccine uptake

Endorsing more misinformation was associated with lower odds of COVID-19 vaccine uptake (aOR 0.78, 95% CI [0.68–0.89]) (Table 2). Among specific conspiracy theories, respondents who believed that COVID-19 vaccines can change your DNA had the lowest odds of vaccination (aOR 0.16, 95% CI [0.05, 0.58]), followed by those who believed that you can get COVID-19 from the vaccine (aOR 0.22, 95% CI [0.06–0.83]). Believing that the government is exaggerating the number of COVID-19 deaths and that deaths due to COVID-19 are being intentionally hidden by government were not associated with vaccine uptake; and believing that the COVID-19 vaccine would cause infertility or is being used for mass sterilization were borderline significant.

Discussion

These findings indicate that there is a wide, heterogeneous, nuanced, and influential information environment about vaccines in Malawi. Malawians are exposed to many different information sources – although this varies across subgroups – that are both pro- and anti-vaccination, and this information environment was associated with COVID-19 vaccination status in this sample. We also found that exposure to misinformation, measured here by familiarity with conspiracy theories, was very common. Although few respondents believed these theories to be true, those who did were less likely to be vaccinated.

Respondents had heard COVID-19 vaccine information from an average of seven different sources – most commonly peers, health care providers, and the radio. As most respondents felt that most sources were positive or factual in tone, this suggests that hearing more information generally means hearing more encouragement for vaccination behavior. We are unable to assess if this relationship is driven by active information-seeking versus passive information-receipt – but many of the most-common sources (health care providers, radio, religious leaders, government/Ministry of Health) have platforms for widespread information dissemination, so respondents may have been exposed to this information without their active pursuit of it. The presence of such a diverse information environment suggests that a multitude of communication strategies may be needed to reach large audiences, including those who actively seek information and those who passively hear messages.

Our results show that not all information sources are the same. Information from health care providers, the Ministry of Health, traditional media, and church or religious leaders was mostly perceived as positive; while information from peers (friends, neighbors, and community members) was largely mixed in tone; and social media, messaging apps, and traditional medicine practitioners were mostly perceived to be negative in tone. Other studies have similarly identified social media as a source of medical misinformation in LMICs33 and as a potential correlate of COVID-19 vaccination behavior in Malawi34, to which these results add support.

Most respondents perceived traditional media as sharing COVID-19 vaccination information that was positive in tone; this may reflect the use of these platforms for disseminating official Ministry of Health or World Health Organization messages. This finding underscores the importance of distinguishing messenger from channel and medium when reporting about health information environments35.

In fact, peers (friends, neighbors, and community members) were the most prevalent source of COVID-19 vaccine information in this survey but were among the least-positive and least-trusted sources for this information. The importance of social networks, including in sharing negative information about vaccines, has been identified in other settings as well36,37,38.

We found that people with dominantly positive information environments were more likely to be vaccinated against COVID-19; this adds to an emerging literature showing the impact of positive information environments39,40. The results are also reflective of the broader literature on the harmful impacts of negative vaccine information9,24,25,27.

Although most people in our sample did not endorse the conspiracy theories we asked about (despite having heard them), those who did had decreased odds of vaccination. This highlights the need for greater monitoring and action against vaccine information that is intentionally misleading24. We also found greater endorsement of conspiracy theories among younger, more educated, and urban respondents. This offers a possible explanation of findings from elsewhere in Africa that younger41 and urban42 respondents are more likely to be vaccine hesitant or believe in certain COVID-19 conspiracy theories. These demographic groups may need targeted and culturally-appropriate efforts to counter misinformation. Effective approaches to address vaccine hesitancy should address and correct both negative and false information, particularly as spread through social networks both online (social media, messaging apps) and in real life. More research is needed to identify effective and locally-relevant strategies for this in Malawi and similar contexts43; one promising approach is a WhatsApp-based counseling intervention for countering COVID-19 fake news trialed in Nigeria44. An automated chatbot on Facebook Messenger that addressed individuals’ concerns about the COVID-19 vaccine in Kenya and Nigeria was found to increase respondents’ willingness and intention to get vaccinated45. Further, accuracy nudges (i.e., prompts on social media that ask users to consider to truthfulness or reliability of the post before interacting with it) have also been shown to decrease the frequency in which people share false information on social media in Kenya and Nigeria46.

Echoing previous research47,48, we found that men, urban respondents, and individuals with more education had more diverse information environments (i.e., exposure to more sources). These groups also heard more conspiracy theories, and urban respondents as well as those with greater educational attainment also said they believed more conspiracy theories were true (as did younger respondents). This may reflect greater internet/mobile penetration and media access among these groups. Given our findings that exposure to more information was associated with vaccine uptake, these results offer hypotheses to explain why acceptability of the COVID-19 vaccine has been found to be higher among men and people with more education49,50,51,52.

There is a need to close the health information gap that we found for women, rural residents, and those with less education. This may be best achieved by expanding and diversifying the information environment reaching these groups, with a particular emphasis on sources that disseminate positive vaccine information – e.g., radio, television, or health care workers. Previous research has emphasized the importance of trust in information sources for promoting vaccine uptake and suggests leveraging trusted figures to effectively combat vaccine misinformation53,54. A study in India found that SMS messages about COVID-19 prevention from credible sources increased health-preserving behaviors, like hand-washing and reporting health symptoms to community health workers55. Further, in this study, as in many others in Africa28,52,56, health care workers were the most trusted source of COVID-19 vaccine information. This suggests that health care workers should be positioned as spokespeople and vaccine champions, and trained to share accurate and persuasive vaccination information during clinical encounters.

We note several limitations of this study. The study respondents were recruited at health facilities hence our sample overrepresents people with stronger trust in health care services, and those who had received a COVID-19 vaccine. Likewise, the characteristics of this sample (demographics, socioeconomics) may not be representative of the Malawian population. Additionally, vaccination status was ascertained based on self-report, which may have overestimated coverage due to social desirability bias. Third, respondents may have interpreted some survey questions differently, and we did not collect qualitative or other data to disentangle this; for example, some people may have interpreted negative information as meaning vaccine hesitancy while others may have interpreted it as meaning misinformation or conspiracy theories. Future research should work to better-define these terms in different contexts, to develop a more nuanced understanding of how information tone is associated with people’s attitudes and behaviors. Lastly, there may have been reporting bias as the survey was administered orally by a research assistant at a health care facility, so respondents may have been reluctant to endorse conspiracy theories or may have over-emphasized the positive information heard from health workers or from the government/Ministry of Health.

Conclusion

Our findings provide important insights on how people’s vaccination decisions may be shaped by information sources, tone, and content. More research is needed on this important topic, particularly in low-income countries where the media environment is changing rapidly. These results suggest that targeted interventions that leverage prominent and trusted sources of COVID-19 vaccine information, are designed to reach all socio-demographic groups, and combat misinformation may be successful at increasing vaccine uptake in Malawi.