Regional heatmap of the 3-item VCP confidence battery (importance · safety · effectiveness) — shock propagation and recovery dynamics after safety scares
Founded in 2010 by Heidi Larson at the London School of Hygiene & Tropical Medicine, the Vaccine Confidence Project (VCP) built the first systematic global instrument for tracking public trust in vaccines. Rather than asking a single vague question ("do you trust vaccines?"), the VCP battery decomposes confidence into three distinct, empirically separable beliefs — importance, safety, and effectiveness — which are then aggregated into a composite index that can be tracked, compared across regions, and used as an early-warning signal for immunization programs.
The core VCP instrument asks respondents to agree or disagree (strongly agree, agree, disagree, strongly disagree) with three statements:
1. Importance: "Vaccines are important for children to have." 2. Safety: "Overall I think vaccines are safe." 3. Effectiveness: "Overall I think vaccines are effective."
Each item is scored 0–3 (strongly disagree=0 to strongly agree=3), and the composite confidence index is computed as the proportion of respondents answering "strongly agree" or "agree" on each dimension, then averaged and rescaled to 0–100. A region scoring 100 would have near-universal agreement across all three dimensions; a region scoring near 0 would reflect widespread rejection of vaccine importance, safety, and effectiveness alike.
The three-item structure matters because confidence is not monolithic: a population can believe vaccines are effective but still doubt their safety (common after a localized adverse-event scare), or believe vaccines are safe but doubt their importance (common in low-disease-burden settings where the threat feels abstract). Tracking the sub-components separately lets health authorities diagnose which type of distrust they are facing and design a matched intervention rather than a generic "vaccines are good" campaign.
The VCP battery was scaled globally through partnership with the Gallup World Poll and, from 2018 onward, the Wellcome Global Monitor — the largest ever study of public attitudes toward science and health, surveying more than 140,000 people across 144–149 countries using nationally representative samples and professionally translated, cognitively tested questionnaires.
Methodological safeguards that make the index comparable across such a linguistically and culturally diverse sample set: • Back-translation: every language version is translated then independently translated back to English to check for semantic drift • Cognitive pretesting: pilot interviews verify that respondents in each country interpret "safe" and "effective" consistently • Probability sampling: household or phone-based random sampling ensures representativeness, not convenience sampling • Sampling weights: post-stratification weights correct for age, gender, urban/rural, and education skew
Larson et al. published the first large-scale application in "The State of Vaccine Confidence 2016: Global Insights Through a 67-Country Survey" (EBioMedicine, 2016), establishing the index as a standing global surveillance tool rather than a one-off academic exercise — subsequent "State of Vaccine Confidence" reports (EU 2018, 2020, 2022; US pre/post-COVID) use the identical battery, enabling true longitudinal comparison.
When the composite index is computed for every surveyed region and plotted as a heatmap, a striking pattern emerges: confidence is not a smooth global gradient but a patchwork shaped by history, media systems, and institutional trust. Some regions cluster persistently high; others sit persistently low for reasons that predate any single scandal.
The 2018 Wellcome Global Monitor found striking cross-national variation. Western Europe recorded the lowest safety confidence of any world region — France was the single lowest-scoring country surveyed, with only about one in three respondents agreeing that vaccines are safe, a legacy attributed to a cluster of controversies (the 2009 H1N1 vaccination campaign, a 1990s hepatitis B–multiple sclerosis scare, and broader distrust of pharmaceutical companies and government health messaging).
By contrast, South Asia and East Africa recorded among the highest confidence scores globally, often exceeding 95% agreement on importance and effectiveness — regions where the visible burden of vaccine-preventable disease remains high and vaccination programs are strongly integrated with trusted primary care and community health worker networks.
This heterogeneity is the baseline the heatmap visualization is built to show: a grid of regions colored red (low, <40) through amber (mid, 40–65) to teal (high, >65), reflecting real dispersion rather than uniform global trust.
Cross-national regression analyses (Larson et al., de Figueiredo et al., Lancet 2020 "Mapping global trends in vaccine confidence") identify recurring predictors of low baseline confidence:
• Prior scandal exposure: countries with a well-publicized vaccine safety controversy in the preceding two decades show persistently depressed safety scores, even years after the underlying claim was scientifically discredited • Political trust: confidence in vaccines correlates strongly with generalized trust in government and public institutions — where one is low, the other tends to be low • Religious and philosophical exemption culture: regions with organized anti-vaccination or natural-health movements (parts of Western Europe, pockets of North America) show lower baseline scores independent of disease burden • Health system contact: populations with frequent, positive contact with primary care and community health workers report higher confidence — the relationship with a trusted local provider is consistently the single strongest individual-level predictor of vaccine acceptance • Social media penetration: higher social media usage correlates with greater exposure to anti-vaccine content and, in several studies, modestly lower confidence, though causal direction is debated
The clearest real-world illustration of a confidence "shock" is the Philippines' Dengvaxia crisis. In December 2017, vaccine maker Sanofi Pasteur disclosed that its dengue vaccine could increase the risk of severe disease in individuals who had never previously been infected with dengue. The disclosure triggered a media firestorm, a suspended national program, congressional hearings, and a collapse in vaccine confidence that spread far beyond the dengue vaccine itself.
Dengvaxia had been rolled out in the Philippines in 2016 to roughly 830,000 schoolchildren as the first licensed dengue vaccine anywhere in the world. In November 2017, Sanofi announced new clinical trial follow-up data showing that in people with no prior dengue infection, vaccination could prime the immune system in a way that made a subsequent natural infection more severe (antibody-dependent enhancement).
The Philippine Department of Health suspended the program within days. What followed was less a scientific debate than a media and political spectacle: sensationalized coverage linked dozens of unrelated child deaths to the vaccine before autopsy results were available, congressional hearings amplified worst-case framing, and public health officials struggled to communicate the nuanced actual risk (small, and specific to dengue-naive individuals) against a backdrop of viral outrage.
The VCP index captured the result with unusual clarity: agreement that vaccines are important collapsed from roughly 93% in 2015 to roughly 32% in 2018 — one of the largest and fastest confidence declines ever recorded in the survey's history.
The most consequential feature of the Dengvaxia shock was that it did not stay contained to the dengue vaccine. Confidence collapsed across the entire immunization program — including MMR, which has no scientific relationship whatsoever to Dengvaxia. This is the "spillover distrust" or contagion mechanism the heatmap ripple animation is designed to visualize: a shock at one epicenter propagates outward through shared media channels, social networks, and a general erosion of trust in the health authority that endorsed the original product.
Mechanistically, the ripple can be modeled the way epidemiologists model contagion: an initial shock amplitude at the epicenter, a propagation rate governed by media reach and network density, and a decay with distance (geographic, social, or topical) from the source. Regions with high media/social connectivity to the epicenter see a delayed but still substantial confidence drop; more isolated regions are partially insulated.
The downstream public-health cost was severe and measurable: national MMR coverage fell from roughly 80% to roughly 60% between 2017 and 2019, and the Philippines recorded over 47,000 measles cases with more than 600 deaths in 2019 — a resurgence directly linked by WHO and Philippine DOH investigators to the Dengvaxia-driven collapse in general vaccine confidence, not to any defect in the measles vaccine itself.
Once a shock has propagated through the population, confidence indices typically follow a predictable recovery pattern: a fast initial partial rebound as acute media coverage fades, followed by a much slower, long-tailed climb back toward (but often not fully reaching) the pre-shock baseline.
Repeated survey waves following documented confidence shocks (Dengvaxia, the 1998 Wakefield MMR-autism fraud in the UK, Japan's 2013 HPV vaccine scare) fit reasonably well to a simple exponential recovery form:
C(t) = C_base − Δ·e^(−t/τ)
Where C_base is the pre-shock baseline, Δ is the magnitude of the initial confidence drop, t is time since the shock, and τ is a region- and event-specific recovery time constant. Fitted values of τ across documented cases range widely — from roughly one year for smaller, quickly-resolved local scares to four or more years for scandals involving perceived institutional cover-up or ongoing litigation.
Importantly, the fitted asymptote is frequently below C_base: many populations do not fully return to their pre-shock confidence level even after a decade, leaving a persistent "confidence scar" that raises the population's vulnerability to the next shock.
This asymmetry — fast collapse, slow rebuild — mirrors a well-established pattern in the trust and risk-perception literature. Paul Slovic's foundational work on the "asymmetry principle" of trust (Slovic, 1993, "Perceived Risk, Trust, and Democracy") showed that negative, trust-destroying events are more visible, more cognitively available, and more heavily weighted by the public than positive, trust-building events of equal informational value. A single dramatic adverse-event report can undo years of routine, unremarkable safe vaccination.
Several mechanisms compound the asymmetry specifically for vaccine confidence: • Media half-life mismatch: a scandal generates an acute, high-volume media spike; reassuring follow-up data (e.g., "no causal link found") receives a fraction of that coverage • Generational imprinting: cohorts who came of age during a scandal (e.g., UK parents during the Wakefield era) often retain depressed confidence for decades, requiring generational turnover rather than pure elapsed time to fully normalize • Institutional trust spillover: once the endorsing health authority itself is doubted, subsequent reassurances from that same authority carry less persuasive weight, slowing the recovery further
Because the index is granular and regionally resolved, health authorities can direct scarce risk-communication resources precisely at the bottom decile of regions rather than running a diffuse national campaign — the same principle underlying WHO's Risk Communication and Community Engagement (RCCE) pillar of outbreak response.
Rather than treating vaccine confidence as a uniform national problem, the granular VCP index lets program managers identify exactly which regions (and, within the three-item battery, which specific dimension — importance, safety, or effectiveness) are lagging, then match the intervention to the diagnosis:
• Low safety score, high importance/effectiveness → targeted safety-communication and adverse-event transparency campaigns • Low importance score in a low-disease-burden setting → campaigns emphasizing continued disease risk and outbreak case studies • Broad collapse across all three items following a scandal → rebuild via trusted local messengers rather than top-down messaging from the same authority whose credibility triggered the collapse
This is the operational logic of WHO's RCCE pillar, formalized during Ebola and COVID-19 responses: community engagement, two-way listening, and locally-led messaging outperform generic top-down public information campaigns, particularly in already-distrustful populations.
A consistent finding across VCP and related implementation research: a direct, personal recommendation from a trusted healthcare provider is the single strongest predictor of vaccine uptake in virtually every population studied — stronger than any mass-media campaign.
Evidence-backed intervention types, ranked by typical local effect:
1. Community health worker / trusted-messenger outreach: door-to-door or clinic-based engagement by messengers embedded in the community (not external experts) — often the largest and most durable local confidence gains 2. "Prebunking" (inoculation theory): exposing people to a weakened form of a misleading argument before they encounter it in the wild, so they can recognize and resist it — shown in multiple RCTs (Roozenbeek & van der Linden, 2019 onward) to outperform reactive fact-checking, which can backfire via the "familiarity backfire effect" (repeating a myth to debunk it makes the myth more memorable) • Motivational interviewing by providers: a structured, non-confrontational conversational technique shown in clinical trials to increase parental vaccine acceptance more than didactic fact-sheets • Avoiding myth-repetition: messaging research consistently finds that leading with the false claim, even to refute it, is less effective than leading with the correct fact and only briefly noting the myth exists
Repeated survey waves — the annual and biennial "State of Vaccine Confidence" reports for the EU and US, and successive Wellcome Global Monitor rounds — turn the single-snapshot heatmap into a genuine early-warning surveillance system, revealing which regions recover, which stabilize at a permanently lower equilibrium, and how the COVID-19 pandemic reshaped the entire global landscape.
A single heatmap snapshot cannot distinguish a region that is stably confident from one that just happens to be measured between shocks. Longitudinal tracking through repeated "State of Vaccine Confidence" waves resolves this: the EU reports (2018, 2020, 2022) and equivalent US tracking show that most regions oscillate within a relatively narrow band absent a shock, punctuated by sharp, event-driven drops with the slow asymmetric recovery described in Stage 4.
The COVID-19 pandemic produced the largest global natural experiment in vaccine confidence to date, with genuinely divergent regional trajectories: several countries saw confidence in vaccines generally rise (driven by pandemic salience and successful mRNA vaccine rollout experience), while others — particularly where COVID-19 vaccination itself became politically polarized — saw measurable declines that also spilled over into routine childhood immunization confidence, echoing the Dengvaxia spillover pattern at a global scale.
The core policy argument for maintaining a standing global confidence index, made explicitly in Larson's original 2016 EBioMedicine paper, is that confidence erosion is detectable well before it shows up as a coverage decline or an outbreak. Building the three-item battery into routine, low-cost, high-frequency polling (rather than one-off academic surveys) allows national immunization programs to:
• Detect early warning signs of an emerging shock in a specific region or demographic before it becomes a full-blown crisis • Pre-position trusted-messenger and RCCE resources in regions showing early softening, rather than reacting after a coverage drop is already visible in administrative data • Evaluate intervention effectiveness directly, by comparing index trajectories in treated vs. untreated regions • Benchmark national performance against comparable countries using a common, validated instrument rather than incompatible ad hoc surveys
The index has since been adopted, in whole or adapted form, by WHO regional offices, the European Centre for Disease Prevention and Control, and numerous national immunization technical advisory groups (NITAGs) as a standing part of outbreak-preparedness surveillance.