Declining trends in vaccine confidence across sub-Saharan Africa: A large-scale cross-sectional modeling study

Abstract

Current WHO/UNICEF estimates of routine childhood immunization coverage reveal the largest sustained decline in uptake in three decades with pronounced setbacks across Africa. Although the COVID-19 pandemic has induced significant supply and delivery disruptions, the impact of the pandemic on vaccine confidence is less understood. We here examine trends in vaccine confidence across eight sub-Saharan countries between 2020 and 2022 via a total of 17,187 individual interviews, conducted via a multi-stage probability sampling approach and cross-sectional design and evaluated using Bayesian methods. Multilevel regression combined with poststratification weighting using local demographic information yields national and sub-national estimates of vaccine confidence in 2020 and 2022 as well as its socio-demographic associations. We identify declines in perceptions toward the importance of vaccines for children across all eight countries, with mixed trends in perceptions toward vaccine safety and effectiveness. We find that COVID-19 vaccines are perceived to be less important and safe in 2022 than in 2020 in six of the eight countries, with the only increases in COVID-19 vaccine confidence detected in Ivory Coast. There are substantial declines in vaccine confidence in the Democratic Republic of Congo and South Africa, notably in Eastern Cape, KwaZulu-Natal, Limpopo, and Northern Cape (South Africa) and Bandundu, Maniema, Kasaï-Oriental, Kongo-Central, and Sud-Kivu (DRC). While over 60-year-olds in 2022 have higher vaccine confidence in vaccines generally than younger age groups, we do not detect other individual-level socio-demographic associations with vaccine confidence at the sample sizes studied, including sex, age, education, employment status, and religious affiliation. Understanding the role of the COVID-19 pandemic and associated policies on wider vaccine confidence can inform post-COVID vaccination strategies and help rebuild immunization system resilience.

Keywords: Africa; Bayesian statistics; COVID-19; vaccine confidence; vaccine hesitancy.

Figure 1.

Declining trends in vaccine confidence across the sub-Saharan countries DRC, Ivory Coast, Kenya, Niger, Nigeria, Senegal, South Africa, and Uganda. The estimated percentage of respondents for each country agreeing that vaccines in general (a) and the COVID-19 vaccine specifically (b) are important, safe, and effective in 2020 and 2022. Percentages are only shown if the 95% highest posterior density interval for the change in agreement over time excludes zero.

Figure 2.

Subnational estimates of temporal changes in vaccine confidence from 2020 to 2022. Model-based estimates of the sub-national changes in vaccine confidence from 2020 to 2022 given as a difference in percentage points. Estimates are provided for these differences in perceptions toward vaccine importance (a,b), safety (c), and effectiveness (d). Blue colors denote a decrease in confidence since 2020, while pink colors denote an increase. Grey colors signify that no samples were collected in that region in either 2020 or 2022 and so a difference cannot be computed.

Figure 3.

Subnational estimates of temporal changes in COVID-19 vaccine confidence from 2020 to 2022. Model-based estimates of the sub-national changes in COVID-19 vaccine confidence from 2020 to 2022 given as a difference in percentage points. Estimates are provided for these differences in perceptions toward vaccine importance (a), safety (b), and effectiveness (c). Blue colors denote a decrease in confidence since 2020, while pink colors denote an increase. Grey colors signify that no samples were collected in that region in either 2020 or 2022 and so a difference cannot be computed.

Figure 4.

Socio-demographic determinants of vaccine confidence in sub-Saharan Africa. Socio-demographic associations across all countries are shown in (a) and (b) for general and COVID-19 vaccine confidence (respectively) for 2022 data only. (All log-odds ratios in 2020 had 95% highest posterior density intervals – HPDI – that included zero, and these results are not displayed. A filled dot denotes that the 95% HPDI excludes zero.) Changes in socio-demographic associations with vaccine confidence are shown in (c) and (d) where log-odds ratios less (greater) than one signify that the socio-demographic group has become less (more) confident over time with respect to the baseline category: only log-odds ratios whose 95% highest posterior density interval excludes zero are shown. Colours represent country-question combinations and are indicated in the legend below the four plots.