key: cord-0902614-k2ngw3i7 authors: Solis Arce, J. S.; Warren, S. S.; Meriggi, N. F.; Scacco, A.; McMurry, N.; Voors, M.; Syunyaev, G.; Malik, A. A.; Aboutajdine, S.; Armand, A.; Asad, S.; Augsburg, B.; Bancalari, A.; Nyqvist, M. B.; Borisova, E.; Bosancianu, C. M.; Cheema, A.; Collins, E.; Farooqui, A.; Fracchia, M.; Guariso, A.; Hasanain, A.; Kamwesigye, A.; Kreps, S.; Levine, M. E.; Littman, R.; Platas, M.; Ramakrishna, V.; Shapiro, J. N.; Svensson, J.; Vernot, C.; Vicente, P.; Weissinger, L.; Zhang, B.; Karlan, D.; Callen, M.; Teachout, M.; Humphreys, M.; Omer, S. B.; Mobarak, A. M. title: COVID-19 Vaccine Acceptance and Hesitancy in Low and Middle Income Countries, and Implications for Messaging date: 2021-03-13 journal: nan DOI: 10.1101/2021.03.11.21253419 sha: c983c104e92848e0a45d47c70e2fa8fb241fcb80 doc_id: 902614 cord_uid: k2ngw3i7 Background As vaccination campaigns are deployed worldwide, addressing vaccine hesitancy is of critical importance to ensure sufficient immunization coverage. We analyzed COVID-19 vaccine acceptance across 15 samples covering ten low- and middle- income countries (LMICs) in Asia, Africa, and South America, and two higher income countries (Russia and the United States). Methods Standardized survey responses were collected from 45,928 individuals between June 2020 and January 2021. We estimate vaccine acceptance with robust standard errors clustered at the study level. We analyze stated reasons for vaccine acceptance and hesitancy, and the most trusted sources for advice on vaccination, and we disaggregate acceptance rates by gender, age, and education level. Findings We document willingness to take a COVID-19 vaccine across LMIC samples, ranging from 67% (Burkina Faso) to 97% (Nepal). Willingness was considerably higher in LMICs (80%) than in the United States (65%) and Russia (30%). Vaccine acceptance was primarily explained by an interest in personal protection against the disease (91%). Concern about side effects (40%) was the most common reason for reluctance. Health workers were considered the most trusted sources of information about COVID-19 vaccines. Interpretation Given high levels of stated willingness to accept a COVID-19 vaccine across LMIC samples, our study suggests that prioritizing efficient and equitable vaccine distribution to LMICs will yield high returns in promoting immunization on a global scale. Messaging and other community-level interventions in these contexts should be designed to help translate intentions into uptake, and emphasize safety and efficacy. Trusted health workers are ideally positioned to deliver these messages. Introduction calculated combining all other answer options ("No", "Don't Know" and "Refuse") into a single reference category. We estimated this outcome for each study with robust standard errors and employed sampling weights where available. In addition to study-level estimates, we combined data from all studies other than the USA and Russia to calculate an aggregate estimate for all LMIC studies. For these analyses, each included study received equal weight and standard errors were clustered at the study level. Averages in the "All LMICs" group then reflect the expected share across studies. In this combined analysis we also estimated the underlying heterogeneity of vaccine acceptance across studies using the between studies variance estimator τ 2 from a random effects model. We also conducted subgroup analyses by gender, age and education level and reported differences between groups. For the "All LMICs" analyses we calculated the average of differences between subgroups within studies with standard errors clustered at the study level. We additionally examined reported reasons for COVID-19 vaccine acceptance and hesitancy, as well as the types of actors respondents would trust when making the decision about whether to take a COVID-19 vaccine. Among respondents who expressed willingness to take the vaccine, we asked about several possible reasons why they would take it. For respondents unwilling to take the vaccine, we asked about several possible reasons why they would not take it. Finally, we asked all respondents, regardless of their answers to other questions, whom they would trust most to help them decide whether to get a COVID-19 vaccine. We report estimates of agreement with reasons for vaccine acceptance/hesitance and trust in actors for individual studies and for the "All LMICs" group. CI 74.9-85.6), with a median of 78, a range of 30.1 percent points and an interquartile range of 9.7. Our estimate of τ 2 is 0.007 which implies a standard deviation over country averages of 0.084. The acceptance rate in every LMIC sample is higher than in USA (64. 6%, 61.8-67 .3) and Russia (30.4%, 29.1-31.7). We find limited evidence of variation across demographic subgroups in LMICs, as shown in Table 8 . Women are generally less willing to accept the vaccine (average difference about 4.3 points, significant at p < .01). Younger respondents (defined as aged <55 given younger-skewing populations in LMICs) are marginally more willing to take the vaccine, but this difference is not statistically significant. Less educated people were on average more willing to take the vaccine in LMICs, but this difference is not significant. To better understand the reasoning behind vaccine acceptance, we asked those who were willing to take the vaccine why they would take it. We summarize in Table 3 , with more details in Table 5 . The most common reason given for vaccine acceptance was personal protection against COVID-19 infection. The average across LMICs is 91% (86) (87) (88) (89) (90) (91) (92) (93) (94) (95) (96) . In every individual study, it ranks as the first reason. In distant second place, LMIC respondents reported willingness to take the COVID-19 vaccine in order to protect their families. The average across LMICs is 36% (28) (29) (30) (31) (32) (33) (34) (35) (36) (37) (38) (39) (40) (41) (42) (43) . In comparison to self-protection, protecting the community did not feature prominently in the stated reasons at all. Self-protection also ranked as the most commonly expressed reason for taking the vaccine in Russia (76%, [74] [75] [76] [77] [78] and the USA (94%, 92-95). This evidence contrasts with appeals to altruistic behavior and prosocial motivations in order to promote vaccine acceptance. 11 The risks and benefits to personal well-being feature much more prominently in people's stated reasons for vaccine acceptance. Figure 2 summarizes the reasons given among respondents who said they were not willing to take a Covid vaccine. The most common reason expressed for reluctance to take the vaccine in LMIC studies was concern about side effects. For studies Uganda 1 (85.1% 80.7-89.6), Sierra Leone 2 (57.9%, 50.1-65.7), Sierra Leone 1 (53.5% 47.1-59.9) and Uganda 2 (47.3% 42.2-52.5), more than half of those respondents unwilling to take the vaccine mentioned this reason. Respondents in Russia (36.8%, ) and the USA (79.3%, 74.6-84), reported high levels of this same concern. While serious adverse events that are life-threatening or require hospitalization are very rare, with only .6% of respondents reporting at least one side effect in the Pfizer vaccine trial, 12 one potential explanation for the outsized concern about side effects could be the lack of widespread information about features of the vaccine at the time of data collection. Media coverage of the few cases of serious adverse events and spread of fake news may contribute as well. 13 Concerns about side effects could also be due to a concern about mild side effects from experiences with other vaccines. In the case of available COVID-19 vaccines, we now know that mild side effects are common but transient. These include fatigue, muscle pain, joint pain and headache, which were severe in fewer than 10% of people in the clinical trials of tens of thousands. Severe fever occurred in fewer than 2% of them. 14 Allergic reactions from the COVID-19 vaccine seem to be extremely rare. Data from trials of the Pfizer-BioNTech vaccines shows that anaphylaxis after reported administration occurs at a rate of 11.1 cases per million vaccine doses administered. 15 Our data reflects this -no more than 6% of respondents expressed concern about allergies in any of our LMIC studies. Other concerns that make many respondents unwilling to take the vaccine could be countered by accurately presenting the scientific data to the public. Studies Uganda 2 (31%, 25.9-36.2), Mozambique (29.7%, and Pakistan 1 (26%, 18-34) showed relatively high levels of skepticism about vaccine effectiveness. This is also true for respondents in Russia (29.6%, 28.1-31.1) and the USA (46.8%, ). Recent clinical trials reveal very high rates of vaccine efficacy, 16,17 so clearly communicating these results to the public is a high priority, given the skepticism we observe in our data. In contrast, conspiracy theories were rarely mentioned by respondents in any of our study samples, in spite of widespread popular discourse about antivaxxer movements and theories in higher-income countries. 18 Finally, respondents in some studies downplayed the seriousness of this disease, and listed this as a reason not to be vaccinated. Studies USA (39.3% 33.5-45), Pakistan 1 (29.4%, 20.9-37.9) and Nepal (20.4% 6.7-34.1) report high rates of lack of concern about getting seriously ill from the disease. The analysis above identifies the nature of the information gaps that any vaccine messaging should focus on, while in Figure 3 we try to identify the actors who are best placed to deliver those messages. We asked respondents about their most trusted source of information during the process of deciding whether to take the vaccine, because these sources are vital to disease control strategies during public health emergencies. 19 Results from Figure 3 are reproduced as Table 7 in Appendix A. We find striking consistency across countries. In all but one study, respondents identified the health system as the most trustworthy source to help them decide whether or not to take the COVID-19 vaccine (with the exception of Rwanda, where the government in general was identified as the most trusted source, with the health system a close second). Family and friends were the next most important reference points in most samples. Across samples, women were 3 percentage points more likely to rely on family and friends than male respondents though this difference is not significant at conventional levels ( Figure 4 in Appendix D). By contrast, endorsements by religious leaders or celebrity figures were not seen as important sources of influence in any sample other than Nepal. To our knowledge, this is the first study documenting rates of expressed COVID-19 vaccine acceptance and hesitancy in a large set of LMICs. Our findings show variable but broadly high levels of prospective COVID-19 vaccine acceptance across LMICs using data from 45,928 respondents in 13 original household surveys from Africa (Burkina Faso, Mozambique, Nigeria, Rwanda, Sierra Leone, Uganda), Asia (Bangladesh, India, Nepal, Pakistan), and Latin America (Colombia). Acceptance across LMIC averages 80.3, ranging between 66.5 and 96.6. We document considerably lower levels of acceptance in Russia and the United States. 8 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 13, 2021. ; Patterns of COVID-19 vaccine hesitancy are not well predicted by existing measures of concerns about the safety of other vaccines (e.g., the Wellcome Global Monitor shown in Table 1 ). Compared with other vaccines, COVID-19 vaccine acceptance is lower and more variable across LMIC samples. This suggests that concerns may apply specifically to COVID-19 vaccines rather than to vaccination more broadly. Our study also documents reasons why respondents express intentions to take (or not take) a COVID-19 vaccine. The main reason expressed for willingness to take such a vaccine was to protect oneself. The most common reasons offered by those unwilling to take the vaccine were concerns about safety (side effects) and efficacy. Across all contexts, health care workers were the most trusted source of information about vaccines. Our study samples offer an important window into the motivations underlying COVID-19 vaccine acceptance in LMICs, but our data are not fully nationally representative. Random digit dial samples and follow-up phone surveys, while necessary during a global pandemic, do not include individuals who reside outside coverage areas, who do not own or cannot operate cell phones, or who choose not to respond to telephone surveys. Care should also be taken in any attempt to extrapolate to the population level from the samples representative of narrow subpopulations. If intentions reported in our LMIC samples translated into actual vaccination uptake, the rates would far exceed the range of what would be required for COVID-19 herd immunity (40-67% in recent estimates). 20,21 However, reported intent may not materialize into actual vaccine adoption. 22 The high salience of COVID-19 due to extensive media coverage and government mitigation efforts and excitement around vaccine release may have increased reported intention. 23 Results from the first COVID-19 vaccine Phase 3 clinical trial were announced before some surveys were fielded; during others, subsequent approvals were granted. The fast-moving information environment may change people's perceptions about vaccines by the time they are widely available in LMICs. Nonetheless, our findings provide some specific guidance on how to design messaging to boost COVID-19 vaccine acceptance and uptake in LMICs. Our data have implications for both what the content of the message should be, and who should deliver the message. First, high levels of trust in the advice of health workers and governments on COVID-19 vaccine decision-making suggest that social and behavioral change communication (SBCC) strategies that engage local health workers may be particularly effective tools to encourage timely and complete vaccine uptake, and to combat remaining vaccine hesitancy. The literature has explored messaging strategies to promote welfare-improving behaviors, with considerable attention paid to celebrity endorsements. 24 Our data strongly support the view 25 that those with the most relevant expertiseas opposed to celebrities or general opinion leaders -are most trusted on this specific topic and are therefore best positioned to deliver the message. Second, the average COVID-19 vaccine acceptance rate across our LMIC samples is high, approximately 80.3%. Given such positive intentions, there may be high returns to investing in straightforward "last-mile" nudges that help citizens convert intentions into actions. Reminder messages from healthcare providers and messages alerting patients that vaccines have been reserved for them at an upcoming appointment may provide a low-cost encouragement to initiate and complete two-dose COVID-19 vaccinations, as was found in two recent large-scale studies in 9 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 13, 2021. ; the United States. 26, 27 Similarly, childhood vaccination reminders plus cash incentives in Kenya substantially increased full immunization, 28 and cash and in-kind incentive programs in Nigeria and India have also proven effective. 29, 30 Third, high coverage rates of existing vaccines, coupled with respondents' reliance on friends and family as information sources, suggest that the general pro-vaccination stance of many citizens could be leveraged to convert intent to uptake. Social learning strategies and norm-setting are powerful drivers of information diffusion and behavior change in many related sectors. 31 Social signalling of positive attitudes toward COVID-19 vaccines may also help shift social norms toward even greater immunization acceptance and two-dose completion in the community at large. 32 Finally, our findings offer guidance on the specific content of vaccine messaging that is likely to be most persuasive. Messaging should highlight the high efficacy rates of the COVID-19 vaccines currently on the market in reducing or eliminating disease, hospitalizations and death. Alluding to clinical data that addresses people's concerns about potential side effects should be prioritized. Messaging should also emphasize the direct protective benefits of the vaccine to the adopter. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 13, 2021. ; JSo, SW, NMe, AS, NMc, GS, MV and AM are co-first authors. DK, MC, MT, MH, AMM and SO are co-last authors. AMM and SO are also the corresponding authors. DK, AMM, MT, NMe, MC, and MV conceived of the study and provided overall guidance. SAb and NMe led the literature search, with input from AS, NMc, SW, AMM, AM and JSo. SW, NMe, AS, NMc, MV, GS, AA, SA, BA, AB,EB, CMB, AC, EC, MF, AG, AK, SK, RL, MBN, MP, JSh, JSv, PV, LB, BZ, MC, SAs, AC, AF, AH, MC, MT, and MH oversaw data collection as part of other research efforts. SW, NMe, and MT coordinated the project across study samples. The following verified the underlying data for individual study samples: EC (Burkina Faso, Colombia, Rwanda, Sierra Leone 1), BA (India), AS and RL (Nigeria), AG and MBN (Uganda 1), CMB and MH (Uganda 2), NMe and MV (Sierra Leone 2), GS (Russia), MF (Mozambique), AF and JSh (Pakistan 1), SAs (Pakistan 2), CV (Nepal), and NMc (USA). JSo, GS, MH and SA collated and processed all datasets used for the analysis. NMe, MH, AMM, JSo, GS, SW, AS, EC, EB, MT, MV and NMc did the data interpretation with guidance from SO and AM. JSo, GS, EC and MH verified final datasets and analysis. JSo and GS did the data analysis and produced output figures with input from MH, AMM, DK, SW, EC, MV, NMe and MT. MH supervised the data analysis. JSo, SW, NMe, AMM, AS, NMc and MV wrote the first draft of the manuscript, with guidance from AM and SO. JSo, SW, NMe, AS, NMc, MV, SAb, EB, MP, JSh, PV, BZ, MC, MT, MH, AMM and SO revised the manuscript. All authors approved the final version of the manuscript. All authors had full access to all the data used in this study and had final responsibility for the decision to submit for publication. We declare no competing interests. Individual participant data (de-identified) that underlie the results reported in this article, analytic code and replication files will be available immediately following publication to no end date for anyone who wishes to access the data and use it for any purpose. A replication exercise is available here https://wzb-ipi.github.io/covid_vaccines/replication.html. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Evidence before this study COVID-19 vaccine acceptance has been widely studied in high-income countries. Much less evidence exists for LMICs, and data that are available 33,34 systematically exclude the estimated 66% of individuals in LMICs who do not use the internet. 35 Searches of PubMed and the Cochrane Database of Systematic Reviews using the terms "vaccine hesitancy", "low-and middle-income countries", "trust in vaccines", "immunization campaign", "vaccination incentives" and "vaccination policy" to select studies investigating the determinants of COVID-19 vaccine uptake and policy-led actions to increase it and restricted to studies published between Jan 1, 2020 and Jan 31, 2021 identified two studies addressing COVID-19 vaccine acceptance in LMICs, and no any studies comparing uptake for the COVID-19 vaccine with general attitudes toward vaccines. This study documents COVID-19 vaccine acceptance across ten LMICs and identifies key sociodemographic predictors, combining analyses of data from 15 distinct studies that cover a total of 45,928 individuals. To date, no comparable quantitative mapping of COVID-19 vaccine acceptance in LMICs has been released. By extending the analysis to attitudes toward vaccinations in general, and by asking respondents to specify reasons for their acceptance or refusal, our study offers novel insights that may help inform country-specific policies to smooth the path to vaccine acceptance in LMICs. As mass immunization campaigns are deployed across the world, our analysis offers cause for optimism regarding potential uptake of the COVID-19 vaccine in LMICs. Our findings suggest that policies to encourage widespread uptake should focus on converting intentions to take the vaccine into action. Our analysis of reasons for hesitancy and most trusted sources for advice about vaccination suggests that communication campaigns focusing on vaccine safety and efficacy delivered through trusted health workers may be particularly effective in persuading those who are still hesitant. While acceptance of COVID-19 vaccinations is high across our sample of LMICs, acceptance of vaccines in general is even higher, highlighting opportunities to leverage existing pro-vaccine attitudes and norms to encourage uptake and address remaining hesitancy. Social signaling of vaccination status may also be effective in demonstrating local acceptance of safety and efficacy claims. 14 Wadman M. Public needs to prep for vaccine side effects. 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) 14 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Insight I. Impact of conditional cash transfers on routine childhood immunizations in North West Nigeria. 2020. https://files.givewell.org/files/DWDA%202009/NewIncentives/ IDinsight_Impact_Evaluation_of_New_Incentives_Final_Report.pdf (accessed Feb 4, 2021) . Banerjee AV, Duflo E, Glennerster R, Kothari D. Improving immunisation coverage in rural india: Clustered randomised controlled evaluation of immunisation campaigns with and without incentives. BMJ 2010; 340. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 1 presents an overview of vaccination beliefs and incidence across countries in our sample. Columns 2-5 use data from the Wellcome Global Monitor 2018. Column 2 shows the percentage of respondents who are parents and report having had any of their children ever vaccinated. Columns 3-5 show the percentage of all respondents that either strongly agree or somewhat agree with the statement above each column. All percentages are obtained using national weights. Columns 6-8 use data from the World Health Organization on vaccine incidence. Columns 6-8 report the percentage of infants per country receiving the vaccine indicated in each column. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. Figure 1: Acceptance rates overall and broken down by respondent characteristics. If a COVID-19 vaccine becomes available in [country], would you take it? Figure 1 presents average acceptance of the COVID-19 vaccine across studies and subgroups within studies. For each study, we summarize sampling information in parentheses in the following way: First, we indicate whether the geographic coverage of the sample is national or subnational. If the coverage is subnational we provide further details. Second, we list the number of observations included in the study. In the plot, points represent the estimated percentage of individuals who would take the vaccine. "No", "Don't know" and "Refuse" are taken as a single reference category. Bars around each point indicate a 95% confidence interval for the estimate. An estimate of average acceptance for all studies in LMICs (excluding USA and Russia) is also shown. 18 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 3 shows percentage of respondents mentioning reasons why they would take the Covid-19 vaccine. The number of observations and percentage correponds only to people who would take the vaccine. Respondents in all countries could give more than one reason. A 95% confidence interval is shown between parentheses. Studies India, Pakistan 1 and Pakistan 2 are not included because they either did not include the question or were not properly harmonized with the other studies. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 100 0 25 50 75 Why would you not take the COVID-19 vaccine? Figure 2 shows the percentage of respondents mentioning reasons why they would not take the COVID-19 vaccine. In the plot, points represent the estimated percentage of individuals that would not take the vaccine or do not know if they would take the vaccine for each possible response option. Bars around each point indicate a 95% confidence interval for the estimate. An estimated average for all studies in LMICs is also shown. Size of points illustrates the number of observations in each response option. Studies India and Pakistan 2 are not included because they either did not include the question or were not properly harmonized with the other studies. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Figure 3 : Trusted actors and institutions, broken down by expressed willingness to take a COVID-19 vaccine. Which of the following people would you trust MOST to help you decide whether you would get a COVID-19 vaccine? Figure 3 shows histograms of actors and institutions respondents say they would trust most to help them decide whether to take the COVID-19 vaccine. Respondents were only permitted to select one most trusted actor or institution. Studies India, Mozambique, Pakistan 1, Pakistan 2 and Uganda 1 are not included because they either did not include the question or were not properly harmonized with the other studies. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Appendix A: Tables from results Table 4 shows percentage of respondents willing to take the COVID-19 vaccine as plotted in Figure 1 . A 95% confidence interval is shown between parentheses 22 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 5 shows percentage of respondents mentioning reasons why they would take the Covid-19 vaccine. The number of observations and percentage correponds only to people who would take the vaccine. Respondents in all countries could give more than one reason. A 95% confidence interval is shown between parentheses 23 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. Table 7 shows percentage of respondents that mention actors who they would trust the most to help them decide whether to get a COVID-19 vaccine. For all countries the questions was asked regardless if respondent would take a vaccine, would not take it, does not know or does not respond. For India respondents were able to mention more than one actor, for the rest of countries only one actor was allowed. While rows should sum to 100%, rounding makes number slightly above or below. A 95% confidence interval is shown between parentheses. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. If a COVID-19 vaccine becomes available in [country], would you take it? Figure 4 shows histograms of actors and institutions that respondents say they would trust most to help them decide whether or not to take the COVID-19 vaccine. Respondents were only permitted to select one most trusted actor or institution. Responses are broken down by acceptance of the COVID-19 vaccine. 28 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 8 shows the results of subgroup mean differences. Subgroup differences were generated considering only LMICs. The differences in means for gender and age do not include the Uganda 1 study, which only included female respondents under the age of 55. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 9 show the percentage of observations that are not missing values for each variable included in Figure 1. 30 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Appendix B: Question wording and answer options per study Table 13 presents question wording and answer options from answers used in Figure 1 to get estimated vaccine acceptance. Answer options are separated by a semicolon. In India options 'Yes, only for free' and 'Yes, even if I have to pay' are both recoded as 'Yes'. In Pakistan 2, 'Absolutely yes' is recoded as 'Yes', 'Neutral' is recoded as 'Don't know' and 'Absolutely no' is recoded as 'No'. In Russia, 'Yes, if a Russian vaccine will be available' and 'Yes, if an imported vaccine will be available' are both recoded as 'Yes'. In USA 'Definitely yes' and 'Probably yes' are recoded as 'Yes', and 'Probably not' and 'Definitely not' are recoded as 'No' 31 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Table 10 presents question wording and answer options used in Table 2 to get an estimated percentage of reasons to take the COVID-19 vaccine. Columns 'Protection: self', 'Protection: family' and 'Protection: community' show the answer options that were recoded in each category. Answer options are separated by a semicolon. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; Why would you not take it? I will not take a vaccine because I am concerned about side effects I will not take a vaccine because I am not concerned about the risk associated with me/my relatives getting COVID-19ne is I will not take a vaccine because I am not concerned about the risk associated with me/my relatives getting COVID-19 I will not take a vaccine because they are not effective . I will not take a vaccine because I don't like needles . I will not take a vaccine because I don't have time I will not take a vaccine because I don't think COVID exists I will not take a vaccine because of other reasons; I will not take a vaccine because my community objects it; I will not take a vaccine because I don't have symptoms; I will not take a vaccine because I am immune; I will not take a vaccine because it is provided by foreign aid; I will not take a vaccine because I Cost or difficulty of getting the vaccine Table 11 presents question wording and answer options used in Figure 2 to get an estimated percentage of reasons not to take the COVID-19 vaccine. Columns 3-10 show the answer options that were recoded in each category. Answer options are separated by a semicolon. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. Table 12 presents question wording and answer options used in Figure 3 to get the percentage of respondents mentioning each actor or instituion that they would trust to decide whether to get the COVID-19 vaccine. Columns 3-8 show the answer options that were recoded in each category. Answer options are separated by a semicolon. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Sampling Weights: Post-stratification weights are computed to adjust for differential attrition between the first and second rounds of the RDD panel, weighting on gender, region, and educational attainment. Original Study Design: Randomized controlled trial, with complete census of households within 142 slums (September to December 2017), and a series of household and caretaker surveys, objective measurements, incentivized behavioural measurements, and a Structured Community Activity, collected for a sub-set of 100 slums between April 2018 and September 2019. Intervention: Catchment areas of CTs were randomly allocated to two interventions. The first intervention aimed at community toilet improvements by offering caretakers the choice of a grant to be spent for improvements in the facility. Following the grant, caretakers were offered a large 37 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; financial reward conditional on the cleanliness of the facility. The second intervention added to this CT improvement awareness creation among potential users through face-to-face information sessions, leaflets, monthly reminders using voice messages sent to mobile phones, and posters hung in the CTs. Sampling Frame: A two-step sampling was applied, first, study households from the main study sample were sampled, then households from the whole slum population were added. Intervention: The first study was dedicated to analyzing the impacts of interventions targeting microentrepreneurs in urban markets on financial inclusion and literacy. The second study focused on the role of information to counteract the political resource curse after a substantial natural gas discovery. Sampling Frame: The first initial sample was selected by in-field random sampling in 23 urban and periurban markets in Maputo and Matola. Stratification was based on the gender of the respondent 2 Original study: http://catiabatista.org/bsv_mm_urban.pdf 3 Original study: https://www.aeaweb.org/articles?id=10.1257/aer.20190842 38 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted March 13, 2021. ; and on the type of establishment (stall vs. store). The second initial sample was selected to be representative of 206 communities in the province of Cabo Delgado, randomly drawn from the list of all 421 polling locations in the sampling frame, stratified on urban, semiurban, and rural areas. This survey in this paper was done by phone. Original Study Design: Initial baseline data was collected in-person in July of 2019, and 5 rounds of phone survey data were collected between August 12, 2019 and January 4, 2020. Sampling Frame: The phone survey sample includes 2,636 rural households in the districts of Kailali and Kanchanpur, which represent the set of households that responded to phone surveys from an original sample of 2,935 households. This sample was constructed by randomly sampling 33 wards from 15 of the 20 sub-districts in Kailali and Kanchanpur and selecting a random 97 villages from within those wards. At the time of baseline data collection in July of 2019, 7 of these 97 villages were dropped from the sample due to flooding. Households belong to the bottom half of the wealth distribution in these villages, as estimated by a participatory wealth ranking exercise with members of the village. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Original Study Design: Initial data was collected from a subset of the sample in December 2019 (in person survey) and July -Aug 2020 (phone survey) as part of an experiment testing the effects of a brief radio program on inter-religious animus. A random walk procedure and random sampling were used within households to recruit a representative sample of adults in Kaduna town. The rest of the sample was recruited for the study in Aug 2020 by purchasing phone lists for residents of Kaduna State. Intervention: The study examines the effects of a radio program and a TV drama on inter-religious animus. The subset of the sample in the radio study was randomly assigned to listen to a brief radio program on one of the following topics: (1) an inter-religious storyline, (2) an intra-religious storyline, and (3) a message about maintaining safe health practices. All respondents in the sample participated in a study examining the effect of viewing an inter-religious storyline unfolding over a full season of a popular TV drama, Dadin Kowa. The season aired from Aug -Oct 2020. A third of the sample were encouraged to watch Dadin Kowa, a third were encouraged to watch the TV station Africa Magic Hausa at the same time Dadin Kowa aired, and a third were in the treatment-as-usual group. All participants received a weekly incentivized SMS quiz from Aug -Oct 2020. COVID-19 Survey Design: This survey is not primarily about COVID-19, but was designed as an endline survey to follow the TV drama intervention described above. The goal of this survey is to measure a range of attitudinal outcomes related to Christian-Muslim relations (including prejudice, intergroup threat perceptions, dehumanization, and support for the use of violence, among others). We included nine of the standardized COVID-19 vaccine-related questions collected specifically for this vaccine acceptance study in the final module of the endline survey. Sampling Frame: 950 respondents in the sample were recruited in person through a random sampling procedure in the Kaduna metropolitan area (pre-COVID). The remaining 1,700 respondents were recruited into the study over the phone from lists of phone numbers of Kaduna state residents that were purchased from a private vendor. Survey Dates: November 18 -December 18, 2020. Sample size, tracking and attrition: All 1,834 individuals who completed the endline survey are included. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Survey Dates: July 24 to September 9, 2020 Sample size, tracking and attrition: Sample includes 1,473 respondents. Sampling Weights: Post-stratification weights are computed to adjust for the sampling process, which involved stratifying first on 27 police stations, then within each police station on beats, then PPS sampling within beats using Asiapop population data. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Sampling Weights: Post-stratification weights are computed to adjust for differential attrition between the first and second rounds of the RDD panel, weighting on gender, region, and educational attainment. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Sampling Weights: Post-stratification weights are computed to adjust for differential attrition between the first and second rounds of the RDD panel, weighting on gender, region, and educational attainment. IRB Approval: This research was approved via IPA IRB Protocol 15592, and Sierra Leone Ethics and Scientific Review Committee approval (no approval number, letter available upon request). Sierra Leone, Towns that are Candidates for Rural Electrification Nation-wide sample, International Growth Centre (IGC), Wageningen University & Research, Yale Research Initiative on Innovation and Scale (Y-RISE), WZB Berlin Social Science Center and Columbia University Project Title: Sierra Leone Rural Electrification (SLRE) Target Population: Households in 195 rural towns across all 14 districts of Sierra Leone. Of these, 97 villages were selected to benefit from an electrification program. Original Study Design: Initial baseline data was collected during late 2019 and early 2020 as part of a study to assess the impact of Rural Electrification in rural towns in Sierra Leone. Intervention: The Government of Sierra Leone (GoSL) in collaboration with the United Nations Office for Project Services (UNOPS) and international donors is implementing the Rural Renewable Energy Project (RREP). In its first wave, during 2017, the project provided stand-alone solar photovoltaic powered mini-grids to 54 communities across the country. Construction of mini-grids in a further 43 towns is ongoing. In RREP communities, engineers construct 6kW-36kW power 44 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 13, 2021. ; Uganda Subnational sample, International Growth Center, Trinity College Dublin, Stockholm School of Economics and Misum, Institute for International Economic Studies, Stockholm University Target Population: Women from semi-rural and rural villages across 13 districts in Uganda (Iganga, Kayunga, Mbale, Mityana, Apac, Dokolo, Gulu, Adjumani, Koboko, Maracha, Nebbi, Soroti, Kumi) . Original Study Design: Initial baseline data was collected in 2016 as part of a large cluster randomized controlled trial, with the aim of selecting households likely to have children during the study period. Four criterias for selection were thus used, in descending order of importance: the household has a woman that is currently pregnant, or aged 16-30 years old, with a young child less than three years old, and/or married (formally or informally). In each household, the respondent was chosen as the female household head or the primary female health care giver of the household if the household head could not be found. COVID-19 Survey Design: The data was collected through multiple rounds of phone surveys. The variable measuring age was constructed by approximation, using the baseline data from 2016 and adding 4 years to the 2016 measure. When the baseline respondent was replaced, the initial age information was deleted. Sampling Frame: Households were selected within 500 clusters (the village of the household). Survey Dates: September 21 to December 06, 2020. Sample size, tracking and attrition: Out of 2,743 respondents, 1752 were included, provided that they answered the main question about vaccine uptake. Sampling Weights: None. Uganda Subnational sample, WZB Berlin Social Science Center and Columbia University, NYU Abu Dhabi, Innovations for Poverty Action (IPA) Target Population: All residents of Kampala who are Ugandan citizens, above the age of 18, and agree in principle to attend a short citizen consultative meeting. Original Study Design: Baseline data was collected between July and October 2019 for an intervention that randomized citizen attendance to a set of 188 consultative meetings organized across Kampala. The meetings were organized to collect citizen preferences for the design of a forthcoming municipal citizen charter. The study also aimed to assess patterns of political inequality in meeting participation, dynamics, and outcomes, as well as study the subsequent effects on prosociality of being incorporated in this participatory process. 1/3 of the sample was randomly allocated to control, while 2/3 of respondents were invited to attend a consultative meeting. The consultations took place between November 2019 and February 2020 across Kampala divisions. Intervention: The intervention consisted of attendance at the consultative meeting organized a few months after baseline data collection. A further randomization allocated ½ of the invited participants to a meeting moderated by a local bureaucrat, while the remaining ones attended a meeting moderated by a neutral discussion leader. The COVID-19 survey sample comprises the 2,189 respondents to the baseline who were selected on the basis of their residence in the city. Having received permission 46 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted March 13, 2021. ; • Number of confirmed cases: 1,720,063 as of November • Number of deaths: 29,654 as of November Adult internet users who reside in one of 61 federal subjects (federal cities, oblasts, republics, krais and autonomous okrug) of Russia. The regions included in the study are Republics Original Study Design: N/A It contains a number of questions on the personal experience, norms and values, trust in government institutions, provision of social services, and mass media use The sample was enrolled from the pool of Russian online survey company OMI (Online Market Inof 25,558 recruited respondents, 22,125 completed the survey. Among 22,125 respondents who completed the survey, 20,821 were enrolled from the general pull of the survey company respondents, while the remaining 1,304 respondents were enrolled among residents of cities with populations below 100,000 and rural areas Sample size, tracking and attrition: 22,125 respondents who completed the survey with the vaccine acceptance module included Sampling Weights: Post-stratification weights are computed to match marginal population distributions of age, gender and education with target proportions coming from the 2019 Yearbook and IRB Approval: This study was approved via Columbia University IRB Protocol IRB-AAAT4453 mini-grids that provide reliable power year-round. Electricity is free for schools and clinics. Residential and commercial users can acquire connections from commercial operators Village Sampling Frame: Household data was collected in 195 towns across all 12 districts of Sierra Leone. The GoSL selected 97 towns with (planned) mini-grids. We used Propensity Score Matching to select 98 control communities. Within communities, respondents were randomly selected from a census roster stratified by occupation status of farmers, business owners and other occupations [47 percent, 47 percent and 7 percent COVID-19 Survey Design: The goal was to assess households' degree of economic vulnerability in the face of the COVID-19 pandemic Phone surveys were attempted to all 195 rural communities from the baseline survey. The total baseline household sample comprised 7047 respondents. We recontacted all baseline respondents that listed a phone number (4,594 respondents) and obtained informed consent for the phone survey 110 respondents, in 186 towns for a tracking rate of 46 percent. Sampling Weights: None IRB Approval: Approval was secured from Uganda COVID-19 Experience • First confirmed case • Total cases: 741 as of • Total deaths: 0 as of June 18, 2020 and 63 as of pdf to re-contact these individuals, we coordinated a 3-wave panel throughout the summer and fall of 2020, with respondents contacted via phone. The goal was to assess households' degree of economic vulnerability in the face of the COVID-19 pandemic and respondents' evaluations of performance of Sampling Frame: The 2,189 respondents to the baseline were randomly selected from a sampling frame of all buildings in Kampala, for which information about their geographical coordinates was available. After randomly selecting a set of candidate structures Sample size, tracking and attrition: Of the 2,189 respondents which we aimed to contact Sampling Weights: None The study was approved by IPA Global IRB (protocol number 15018) on and by the Mildmay Uganda Research Ethics Committee Tufts University COVID-19 Experience • First confirmed case • Total cases: 14,499,637 as of • Total deaths: 281,678 as of December Target Population: Nation-wide sample of adult internet users recruited through the market research firm Lucid COVID-19 Survey Design:This survey was part of a panel study on attitudes toward COVID-19 technologies and public health surveillance Lucid partners with a network of companies that maintain relationships with research participants by engaging them with research opportunities. While Lucid does not provide probability samples of the U.S. adult population, its quota samples approximate the marginal distributions of key demographic characteristics. Recent validation exercises have found that Lucid samples approximate nationally representative samples in terms of demographic characteristics and survey experiment effects Survey Dates In the main question regarding intention to take the vaccine, approximately 10% of respondents (184) did not answer Sampling Weights: Post-stratification weights are computed to match marginal population distributions of income, age, education, gender, race and region among the US adult population IRB Approval: This study received approval from the Cornell University IRB under Protocol #2004009569 Sample size, tracking and attrition: Sample includes 977 respondents from the second round of a panel. In the first round conducted between June 6 to 15, 2020, 1,356 individual surveys were contacted through Random Digit Dialing (RDD) from the sampling frame of all mobile phone numbers in Burkina Faso. 2,313 working numbers yielded 1,383 eligible respondents for a completion rate of 98% of eligible respondents.Sampling Weights: Post-stratification weights are computed to adjust for differential attrition between the first and second rounds of the RDD panel, weighting on gender, region, and educational attainment.IRB Approval: This research was approved via IPA IRB Protocol 15608, and the Burkina Faso Institutional Ethics Committee for Health Sciences Research, approval A13-2020.