The Vaccine Trust Gap — How Measles Came Back (2026 Data)

The core finding of this report can be stated in one sentence. In middle- and high-income countries, the return of vaccine-preventable disease tracks the erosion of trust in health institutions more closely than it tracks the spread of any particular falsehood. The United States illustrates the speed. In 2025 the country confirmed 2,289 measles cases and three measles deaths, the first since 2015 [5]. By 24 September 2026 the count for the new year had already reached 3,659, and 95% of patients were unvaccinated or of unknown vaccination status [6]. Two weeks later it passed 4,080 [7]. A country that declared measles eliminated in 2000 now faces a formal review of that status by the Pan American Health Organization, postponed to November 2026, with the clock running from an outbreak that began on 20 January 2025 [8]. Elimination is lost when one chain of the same viral lineage circulates without interruption for 12 months or more [40].

The global aggregate looks, at first glance, like slow recovery. According to the WHO and UNICEF estimates of national immunisation coverage published in July 2026, 90% of infants received a first dose of diphtheria, tetanus and pertussis vaccine in 2025, and 85% completed the three-dose series ✓ Established [1]. Measles coverage was weaker: 84% for the first dose and 77% for the second, against the 95% threshold needed to prevent outbreaks [1]. Some 7.3 million infants started the schedule but dropped out before their first measles dose, and 57 countries reported large or disruptive measles outbreaks during the year [1]. The WHO’s own measles report, released in November 2025, estimated 11 million infections and 95,000 deaths in 2024, most of them children under five, with 59 countries experiencing large outbreaks — nearly triple the number in 2021 [23].

Those aggregates conceal two different problems that require two different responses. The first is access. UNICEF’s executive director, Catherine Russell, attributes the plight of millions of unprotected children to conflict, displacement and poverty [1], and UNICEF’s flagship 2023 report found that in the poorest households one child in five is zero-dose, against one in twenty among the wealthiest [3]. The second problem is different in kind. In middle- and high-income countries where vaccines are free and on the shelf, coverage is slipping anyway. The 2026 coverage release singled out South Africa, where first-dose coverage has fallen 20 percentage points since 2019, and Bosnia and Herzegovina, where measles coverage dropped 23 points in a single year ✓ Established [1]. That second problem — refusal and delay in places that can deliver — is the subject of this report.

The American school-entry data show what the second problem looks like up close. During the 2024–25 school year, coverage among kindergartners fell for every reported vaccine, to 92.5% for measles, mumps and rubella and 92.1% for diphtheria, tetanus and acellular pertussis ✓ Established [4]. Exemptions from one or more vaccines rose to 3.6% from 3.3% the year before, the highest level ever recorded, and 17 states reported exemption rates above 5% [4]. Roughly 286,000 kindergartners entered school without documentation of a completed measles series [4]. MMR coverage has now sat below the 95% federal target for five consecutive years. None of these children lacked access in the ordinary sense: the vaccines were available, funded and recommended. What changed was the willingness of parents to accept them.

Survey data from the United States trace the attitudinal side of the same curve. In 2024 Gallup found that 40% of Americans considered it extremely important for parents to vaccinate their children, down from 58% in 2019 and 64% in 2001 ✓ Established [10]. The share calling it extremely or very important fell from 94% in 2001 to 69% [10]. The decline was steepest on one side of the political spectrum: only 26% of Republicans and Republican-leaning adults now rate childhood vaccination as extremely important, half the 52% recorded in 2019, against 63% of Democrats [10]. A health behaviour that was once a near-universal norm has become, within five years, a marker of political identity. That shift matters because the institutions recommending vaccines are themselves now politically contested, which means confidence in the product rises and falls with confidence in whoever is in charge of it.

This report deliberately sets aside the mechanics of online misinformation, which OsakaWire examined in a separate investigation. The argument here is narrower and, the evidence suggests, more consequential. Falsehoods about vaccines have circulated since the first smallpox inoculations; what varies across countries and decades is how many people are prepared to believe them over their own doctors and health agencies. The sections that follow trace how that trust has been measured, how it has historically been broken — almost always by institutional conduct rather than by rumour — what its loss is now costing in measurable illness and money, which countries have lost ground fastest, and what the evidence says about rebuilding it. The record shows that trust can be lost in months and takes decades to restore, while the diseases it once held back return on a far shorter clock.

Heidi Larson, who founded the Vaccine Confidence Project at the London School of Hygiene and Tropical Medicine and has measured attitudes to vaccines across more than a decade, puts the distinction bluntly. In her account the world does not primarily have a misinformation problem; it has a relationship problem, and the misinformation is a symptom of not trusting [35]. The distinction is not semantic. If the problem is informational, the remedy is correction: fact-checks, platform moderation, better public messaging. If the problem is relational, the remedy lies in the conduct of the institutions themselves — whether they are competent, honest about uncertainty, independent of commercial and political pressure, and present in people’s lives through a trusted clinician. Correction aimed at a relationship problem tends to fail, because the correction arrives from the very source the audience has stopped believing.

The most comprehensive American measurement comes from a 50-state survey led by Roy Perlis and published in JAMA Network Open in 2024. Across 24 waves of internet surveys between April 2020 and January 2024, comprising more than 582,000 responses from over 443,000 adults, the proportion reporting a lot of trust in physicians and hospitals fell from 71.5% to 40.1% ✓ Established [11]. The decline was not evenly distributed. Lower trust was associated with being aged 25 to 64, female, less educated, lower-income, Black, or living in a rural area [11]. Most importantly for this report, greater trust was associated with a greater likelihood of being vaccinated against SARS-CoV-2 and influenza and of receiving a booster ◈ Strong Evidence [11]. Trust was not a background attitude. It predicted behaviour.

International data point the same way. A study of 177 countries published in The Lancet in February 2022 by researchers at the Institute for Health Metrics and Evaluation and the Council on Foreign Relations found that trust in government, trust in other people and lower corruption were associated with lower COVID-19 infection rates ◈ Strong Evidence [24]. The authors estimated that if every society had trust in government at least as high as Denmark’s, at the 75th percentile, the world would have recorded 13% fewer infections; matching that level of interpersonal trust would have meant 40% fewer [24]. In middle- and high-income countries, where supply was not the constraint, less corrupt governments achieved broader and faster vaccine coverage [24]. Trust, on this evidence, behaves less like a sentiment and more like a piece of health infrastructure.

Trust has also become partisan in a way that makes it unstable. Polling by KFF shows the pattern reversing after the change of administration in January 2025. Between 2023 and 2025 the share of Democrats expressing a great deal or fair amount of trust in the CDC to provide reliable vaccine information fell from 88% to 70%, and trust in the FDA from 86% to 67% ✓ Established [9]. Among Republicans it rose, from 40% to 51% for the CDC and from 42% to 52% for the FDA [9]. The agencies had not changed their scientific staff overnight. What changed was who led them. When confidence in a vaccine recommendation depends on the party affiliation of the official issuing it, the recommendation has lost its claim to neutral expertise, and every future change of government becomes a potential shock to coverage.

None of this began with the pandemic. The Wellcome Global Monitor, a 2018 Gallup survey of about 140,000 people worldwide, found that one in three people in France disagreed that vaccines are safe — the highest share of any country surveyed ✓ Established [25]. French scepticism varied little by age, gender or education, and the researchers connected it to a wider loss of confidence in government, in science and in the pharmaceutical industry [25]. France in 2018 had high internet use but no unusually large anti-vaccine media ecosystem compared with its neighbours. What it had was a long record of public-health controversies that had weakened faith in the institutions themselves. The French case is evidence that the trust deficit can be deep and stable long before social platforms amplify it.

The European data also show that trust can recover, which is why the relational framing is not a counsel of despair. The 2022 State of Vaccine Confidence in the EU report, based on 25,143 members of the public and 3,012 health professionals across all 27 member states, found that 81.5% agreed vaccines are important, 85.6% that they are effective and 82.3% that they are safe ✓ Established [34]. Those levels had broadly returned to 2018 values after pandemic fluctuation [34]. But the same report identified a widening confidence gap between age groups: people aged 18 to 34 became less confident between 2018 and 2022 while those over 65 did not [34]. The generation now becoming parents is the one whose trust has weakened most — which means the coverage consequences are only beginning to arrive.

The longest-running natural experiment in medical mistrust began in 1932. For 40 years the US Public Health Service observed hundreds of Black men with syphilis in Macon County, Alabama, without treating them, even after penicillin became the standard cure. The study’s public disclosure in 1972 provides a sharp before-and-after line. Marcella Alsan and Marianne Wanamaker, in research published in the Quarterly Journal of Economics in 2018, found that the disclosure was followed by increased medical mistrust and reduced outpatient and inpatient contact with physicians among older Black men ✓ Established [33]. Life expectancy at 45 for Black men fell by up to 1.5 years in response, accounting for roughly 35% of the 1980 life-expectancy gap between Black and white men [33]. The vaccine-relevant lesson is that a betrayal by a public-health agency in one domain withdrew an entire population from preventive care across all domains.

The second case is a failure of scientific gatekeeping. In 1998 The Lancet published a paper by Andrew Wakefield and colleagues suggesting a link between the measles, mumps and rubella vaccine and autism. In England, MMR uptake slumped to around 80% by 2003–04 [36]. The journal formally retracted the paper only on 2 February 2010, after the UK General Medical Council found that the children in the study had been selectively recruited rather than consecutively referred, as the paper had claimed, and that Wakefield had committed multiple ethical breaches ✓ Established [37]. Twelve years elapsed between publication and retraction. The paper was dishonest, but its reach depended on the authority of the institution that published it — and the slowness of that institution’s correction allowed the doubt to settle into a generation of parents.

The Pakistan case matters because it isolates the mechanism with unusual clarity. Nothing about the vaccines changed; no new scientific claim emerged. An institution with no role in public health borrowed the form of a vaccination campaign for an unrelated purpose, and in doing so handed the most effective possible argument to those who already claimed that vaccinators were agents of foreign powers. Martinez-Bravo and Stegmann document that the decline followed the revelation in time and was concentrated where the political messaging landed [19]. In 2024 Pakistan reported 74 wild poliovirus cases and Afghanistan 25, while environmental surveillance detected the virus in 628 samples in Pakistan alone [32]. A global eradication programme that has run since 1988 remains stalled in the two countries where trust in vaccinators was most deliberately broken.

Japan shows how a cautious government decision can do lasting harm. The human papillomavirus vaccine entered Japan’s national immunisation programme in April 2013. Two months later, after sensational media reports of alleged adverse events, the Ministry of Health, Labour and Welfare suspended its proactive recommendation — without withdrawing the vaccine or finding evidence of harm [20]. Coverage among eligible girls fell from more than 70% to below 1% ✓ Established [20]. The recommendation was not restored until April 2022, and coverage has recovered only minimally since [20]. Modelling cited in subsequent coverage estimated that the gap could mean around 25,000 additional cervical cancer cases and more than 5,000 additional deaths among women born between 1994 and 2007 ◈ Strong Evidence [21]. The ministry’s nine-year hesitation was read by the public as confirmation that something was wrong.

The Philippines provides the most dramatic measured collapse. Following the 2017 controversy over Dengvaxia, a dengue vaccine whose manufacturer disclosed an elevated risk of severe disease for people with no prior dengue infection, the Vaccine Confidence Project surveyed 1,500 Filipinos. The share strongly agreeing that vaccines are important fell from 93% in 2015 to 32% in 2018 ✓ Established [18]. Agreement that vaccines are safe fell from 82% to 21%, and that they are effective from 82% to 22% [18]. Larson said the team had anticipated some decline but not one as dramatic as the results showed [18]. The controversy concerned one new vaccine; the collapse in confidence spread to the entire childhood schedule, including vaccines with half a century of safety data.

Read together, the five cases share a structure. In each, an institution with authority over health — or borrowing that authority — behaved in a way the public later judged to be dishonest, careless or self-interested. In each, the damage was not confined to the product or programme at issue but spread to unrelated vaccines and to health services generally. And in each, the recovery took far longer than the breach: decades for Tuskegee, a generation for MMR in England, nine years and counting for HPV in Japan, more than a decade for polio in Pakistan. This is the pattern against which the institutional upheaval in American vaccine policy since 2025 should be assessed. It is, in effect, a sixth case study whose outcome is not yet known.

The scale of what is at stake is set by what vaccines have already achieved. A WHO-led study published in The Lancet in 2024, marking 50 years of the Expanded Programme on Immunization, estimated that immunisation has saved at least 154 million lives since 1974 — six lives every minute for five decades — of whom 101 million were infants ✓ Established [22]. Each life saved represented an average of 66 years of full health, or 10.2 billion healthy years in total [22]. When the programme began, fewer than 5% of infants worldwide had access to routine immunisation [22]. Measles vaccination alone is credited with saving nearly 59 million lives since 2000, and measles deaths fell by 88% between 2000 and 2024 [23]. These are the gains now being eroded at the margin.

Europe has already paid a visible instalment. The WHO European Region, covering 53 countries in Europe and Central Asia, recorded 127,350 measles cases in 2024, double the 2023 total and the highest since 1997 ✓ Established [15]. Children under five accounted for 43% of cases, more than half of all patients required hospitalisation, and 38 deaths had been reported by 6 March 2025 [15]. Romania recorded the most infections, 30,692, followed by Kazakhstan with 28,147 [15]. These are not low-income settings with collapsing supply chains. They are countries with functioning health systems where the vaccine is available and free, and where coverage nonetheless drifted below the level at which a highly contagious virus can be contained.

In the United States the financial cost has been measured with some precision. A review of 18 studies reported in October 2025 put the mean cost of a measles case at $43,204, rising to $58,592 once public-health response costs are included; simply initiating a response to an outbreak costs around $297,747 ✓ Established [29]. Specific outbreaks give the range. Clark County, Washington, spent just over $3 million on a 2019 outbreak of 71 cases — about $47,479 per case, with no deaths and two hospitalisations [30]. The 2025 West Texas outbreak cost an estimated $10 million, mainly because it produced around 50 hospitalisations [30]. New Mexico’s smaller outbreak that year cost $5.4 million [31]. Every one of these sums was spent by local health departments whose budgets were not designed for it, and every one diverted staff from other work.

The deaths are fewer but they are the measure that matters. The three American measles deaths of 2025 included two unvaccinated school-aged children in Texas with no known underlying conditions [5]. The CDC listed at least one further measles death in 2026 by late September [6]. In England, a child died at Liverpool’s Alder Hey Children’s Hospital in July 2025 after contracting measles [17]. Romania declared a national measles epidemic on 5 December 2023 [38] and went on to record the largest national outbreak in Europe the following year [15]. Behind each death sit many more hospitalisations — more than half of all European patients in 2024 [15] — whose long-term consequences no case-cost estimate fully captures.

Measles is the leading indicator, not the whole bill. Because it needs roughly 95% coverage to be contained [1], it surfaces within a year or two of a fall in confidence, while other consequences take longer to appear. Japan’s HPV gap will show up as cervical cancers diagnosed in the 2030s and 2040s among women who were schoolgirls in 2013 ◈ Strong Evidence [21]. Polio persists in Pakistan and Afghanistan, with the virus detected in hundreds of environmental samples each year [32]. Pertussis, rubella and invasive bacterial diseases sit behind measles in the queue. The dropout between first-dose diphtheria and first-dose measles — 7.3 million infants globally in 2025 [1] — is a measure of how many children enter the system and are then lost to it.

There is an asymmetry in the timing that makes the trust deficit unusually dangerous as a policy problem. Confidence can fall within months, as the Philippine data show [18]. Coverage follows within one or two birth cohorts, and the most contagious diseases return within a few years. Recovery runs on a different clock: Japan restored its HPV recommendation in 2022 and coverage has barely moved [20], while England, where MMR uptake slumped to around 80% after the Wakefield paper [36], has persistently failed to reach the 95% threshold since, with second-dose coverage at 84.4% in 2024 [17]. Political cycles, meanwhile, run at four or five years. Any decision that spends public trust for short-term advantage therefore creates costs that will be paid by successor governments, and by children not yet born when the decision was made.

Canada was the first country in the Americas to fall. An outbreak that began in New Brunswick in October 2024 spread across several provinces and infected more than 5,000 people, concentrated in under-vaccinated communities ✓ Established [16]. In November 2025 the Pan American Health Organization confirmed that the same strain had circulated for more than 12 months and revoked the elimination status Canada had held since 1998 [16]. Canada can regain it only after transmission of the outbreak strain is interrupted for at least a year [16]. The Canadian case is instructive because national coverage figures had not signalled an emergency. The outbreak exploited clusters of low coverage within a well-vaccinated country — exactly the pattern that institutional distrust produces, since it concentrates in communities rather than spreading evenly.

The United Kingdom lost its status on 26 January 2026 ✓ Established [17]. The WHO’s decision rested on 2,911 laboratory-confirmed measles cases in England in 2024, the most in any year since 2012, and on sustained transmission that researchers at the London School of Hygiene and Tropical Medicine attributed to a decade-long decline in routine coverage [17]. Coverage with the second MMR dose stood at 84.4% in 2024, down from 89.3% in 2014 and well below the 95% required [17]. The UK has now lost elimination status twice, and the government has responded partly with a schedule change: from January 2026 young children in England are offered a combined vaccine that adds chickenpox protection to measles, mumps and rubella [17]. Whether a more attractive product can offset a decline rooted in confidence remains untested.

Romania shows how far coverage can fall inside the European Union. National measles vaccination coverage above 97% declined steadily from 2010, and in 2022 reached its lowest level of the decade: 83.4% for the first dose and 71.4% for the second ✓ Established [38]. The Ministry of Health declared a nationwide epidemic on 5 December 2023 [38], and in 2024 Romania reported 30,692 cases, the largest national total in the WHO European Region [15]. The Romanian decline began long before the pandemic and long before the current generation of social platforms, and researchers have linked it to a combination of access gaps in poorer communities and eroding confidence in the health system — the two problems this report separates, here compounding each other.

The United States is the largest country now under review. PAHO announced in January 2026 that it would assess the measles elimination status of both the United States and Mexico, later postponing the review to its regular commission meeting in November 2026 while US authorities completed whole-genome sequencing of outbreak strains [8]. The analysis window for the United States runs from 20 January 2025, the onset of the West Texas outbreak, which infected more than 760 people before the state declared it over in August 2025 [5] [8]. The question for the commission is technical: whether the outbreaks of 2025 and 2026 represent one continuous lineage or several separate introductions. The fact that the question has to be asked at all, 26 years after elimination was declared, is the political finding.

Japan and the Republic of Korea complete the picture from the other side of the income distribution. Both were among the five countries where UNICEF found the perceived importance of childhood vaccines fell by more than 33% after the pandemic began ✓ Established [3]. Japan’s HPV experience demonstrates the cost of a government that hesitates in the face of public alarm: a recommendation suspended in 2013 and restored only in 2022, with coverage that has barely recovered since [20]. Korea and Japan have high-quality health systems and among the world’s highest life expectancies. Their inclusion on UNICEF’s list is a reminder that the trust deficit is not confined to polarised Western democracies, and that an advanced health system offers no automatic immunity from it.

Across these five countries the pattern is consistent. Coverage rarely collapses suddenly; it erodes by a few percentage points over a decade, concentrated in particular communities, until it crosses the threshold below which measles can sustain itself. In the UK the slide was 89.3% to 84.4% over ten years [17]; in American kindergartens it has been five years below 95% [4]; in Romania it was 97% to 83% [38]. In each country the slide was visible in published data long before the outbreak that made it news. What was missing was not information about the risk but the institutional standing to persuade hesitant parents to act on it — and, in several cases, the political will to treat a slow decline as an emergency before it became one.

France offers the clearest test of compulsion. Faced with the lowest vaccine confidence in the world, the government extended mandatory infant vaccination from 3 to 11 vaccines from 1 January 2018 [26]. The effect on coverage was immediate. First-dose meningococcal C coverage rose from 39% in 2017 to 75% in 2018, and first-dose coverage with the hexavalent vaccine, which includes hepatitis B, rose from 92.3% to 97.8% ✓ Established [26]. Coverage of vaccines not covered by the law also increased, which the evaluators read as evidence that the accompanying communication campaign had effects beyond the mandate itself [26]. The French case suggests that a clear, unambiguous institutional signal — these vaccines are required because they matter — can itself be a form of communication.

Australia took a financial route. Its No Jab, No Pay policy, introduced in 2016, removed non-medical objection as a basis for receiving family welfare payments and childcare subsidies [27]. In one regional evaluation, the annual catch-up vaccination rate rose from an average of 6.2% before the policy to 9.2% in 2016 and 7.8% in 2017 ✓ Established [27]. A national analysis published in The Lancet Regional Health in 2025 found that removing conscientious objection produced a significant increase in childhood vaccination above the pre-intervention trend, but that the effect of monetary sanctions varied with socio-economic disadvantage [28]. Sanctions moved families who were late or disorganised more than families who were opposed. That distinction is the crux of the policy debate.

The United States moved in the opposite direction. In June 2025 the Secretary of Health and Human Services, Robert F. Kennedy Jr., dismissed all 17 members of the Advisory Committee on Immunization Practices and appointed replacements, many of them vaccine sceptics, despite having told Congress he would leave the committee alone ✓ Established [12]. On 5 December 2025 the new panel voted 8 to 3 to end the universal recommendation of a hepatitis B dose at birth, in place since 1991, for infants of mothers who test negative, even as members acknowledged there was no evidence the birth dose causes harm [12]. The American Academy of Pediatrics warned of heartbreaking consequences [13]. On 5 January 2026 the CDC cut the universally recommended childhood schedule from 18 diseases to 11, citing Denmark as its model, while stating that insurers would continue to cover all the vaccines without cost-sharing [14].

The trust implications of these changes run in two directions at once, which is what makes them dangerous. To parents already sceptical, a federal decision to downgrade vaccines reads as an admission that earlier recommendations were wrong, deepening the suspicion that justified their hesitation. To parents who trusted the old system, the replacement of expert advisers by appointees reads as politicisation, and the KFF data show that this group’s trust in the CDC fell from 88% to 70% in two years ✓ Established [9]. Pediatric bodies warned that the January schedule change would increase confusion and lower vaccination rates even with insurance coverage intact [14]. A policy designed, on its own account, to align American practice with peer countries has made the source of vaccine advice itself a subject of partisan dispute.

What does rebuild confidence is slower and less visible. The Perlis survey found that trust in physicians and hospitals — not in distant agencies — was the variable most clearly associated with whether people were vaccinated [11]. Larson’s work points in the same direction: the remedy for a relationship problem is a relationship, typically with a clinician, nurse or community health worker who is known and present [35]. The European recovery between 2020 and 2022 occurred without any single dramatic intervention [34]. The common thread in the evidence is that people extend trust to institutions they experience as competent, consistent and honest about uncertainty, and withdraw it from institutions that appear to change their position for reasons other than evidence.

The historical cases also show that the speed of institutional correction matters. The Lancet took 12 years to retract the Wakefield paper [37]. Japan’s health ministry took nine years to restore its HPV recommendation [20]. In both cases the delay was itself interpreted as information — evidence that the institution was uncertain, or had something to hide. Rapid, transparent correction of genuine errors, and equally rapid defence of established findings against political pressure, are the institutional behaviours most likely to preserve trust. Neither is cheap. Both require agencies that are insulated enough from short-term politics to say unwelcome things, and accountable enough that the public believes them when they do.

The strongest objection to the trust framing comes from low-income settings. A study of access to COVID-19 tools across 14 low-income countries found that low vaccination rates there had little to do with hesitancy and much to do with supply and delivery ◈ Strong Evidence [39]. UNICEF’s own data show that zero-dose children are concentrated among the poorest households [3], and UNICEF’s executive director identifies conflict, displacement and poverty as the main reasons millions remain unprotected [1]. On this view, a policy debate centred on trust risks diverting attention and money from the larger global problem, which is getting vaccines to children who would accept them if they could reach a clinic.

A second objection holds that the post-pandemic dip is already reversing. The EU’s 2022 confidence report found that public agreement on the importance, effectiveness and safety of vaccines had returned to 2018 levels ✓ Established [34]. Globally, the number of zero-dose children fell by around 750,000 in 2025, and first-dose coverage reached 90% [1]. In the United States, 92.5% of kindergartners still had a completed MMR series in 2024–25 [4], meaning the overwhelming majority of parents continue to vaccinate despite the polling. Advocates of this reading argue that attitudinal surveys exaggerate behavioural change, because answers to pollsters increasingly express political identity rather than intended action.

The two readings are less opposed than they first appear, because they describe different populations. Globally, by headcount, the larger problem is access: millions of children in conflict zones and poor rural districts never reach a vaccinator [1] [3]. In middle- and high-income countries the larger problem is confidence, and it is visible precisely where supply is not the constraint, as South Africa, Bosnia and Herzegovina, Romania, the UK and the United States all show [1] [15] [17]. The error is to apply one diagnosis to both. Funding delivery in Canada would not have prevented the New Brunswick outbreak, and a trust campaign in a conflict zone would not reach a child whose clinic has been destroyed.

The second disagreement — whether recovery is under way — turns on timing. The EU recovery data run to 2022 [34]. The US coverage data run to the 2024–25 school year [4], before the federal advisory overhaul of June 2025 and the schedule change of January 2026 [12] [14]. Every measured behavioural effect of the 2025–26 American policy shift therefore lies in the future. Recovery advocates are right that most parents still vaccinate; trust-crisis advocates are right that the share has fallen for five consecutive years and that the institutions most able to reverse the trend are themselves now in dispute. Both statements are consistent with the data. Neither settles what happens next.

A third contested question concerns compulsion ⚖ Contested. France and Australia demonstrate that mandates and sanctions raise coverage [26] [28]. Critics argue that compulsion converts hesitant parents into hostile ones and hands opponents a liberty argument, and the Australian evidence that sanctions work unevenly across income groups gives that concern some empirical footing [28]. Supporters reply that France’s coverage rose even for vaccines the law did not cover [26], suggesting the mandate strengthened rather than weakened the social norm. The evidence available supports mandates as an effective coverage tool in countries with functioning delivery systems; it does not settle their long-run effect on trust, which no study has yet tracked over a full generation.

The final contested point is methodological. Polls such as Gallup’s and KFF’s measure stated attitudes and trust, and in a polarised country those answers carry partisan signal [9] [10]. Coverage data measure behaviour but arrive with a lag of a year or more and are now thinner: only 18 national immunisation surveys were submitted to the WHO and UNICEF in 2025, against 50 the year before [1]. The honest conclusion is that the direction of travel is well established in high-income countries — confidence down, exemptions up, coverage below threshold, measles returning — while the size of the next step, especially in the United States, is not. A policy debate that treated either reading as settled would be claiming more than the evidence supports.

First, the reversal is real and measurable in high-income countries. Measles cases in the United States reached their highest level in 35 years in 2026 [7]; Canada and the UK lost elimination status within three months of each other [16] [17]; Europe recorded its worst measles year since 1997 [15]. Second, the decline in these countries is driven primarily by confidence rather than access, as shown by falling coverage in places where vaccines are free and available [1] [4]. Third, the confidence that matters most is trust in institutions and clinicians, which predicts vaccination behaviour more reliably than knowledge of the facts ◈ Strong Evidence [11] [24]. These three findings rest on independent data sources across multiple countries and do not depend on any single study.

Fourth, the historical record shows that trust is most often broken by institutional conduct rather than by rumour: a research programme in Alabama [33], a journal in London [37], an intelligence operation in Pakistan [19], a ministry in Tokyo [20], a vaccine rollout in Manila [18]. Fifth, the costs of broken trust are large, delayed and borne by people who had no part in the decision — infants too young for vaccination, children whose parents were persuaded, women who will develop cancers decades from now [15] [21]. A single American measles case costs around $43,204 to treat [29]; an outbreak can cost a county millions [30]. The value of trust is visible only when it is gone.

The open question is institutional. The United States has, within 12 months, replaced its vaccine advisory committee, reversed a 34-year-old newborn recommendation and cut its universal childhood schedule from 18 diseases to 11 [12] [14]. Each step was presented as a response to lost public confidence. The historical evidence suggests that abrupt changes of this kind by the institutions responsible for vaccine advice tend to deepen distrust rather than repair it, because they teach the public that recommendations follow politics. That outcome is not yet proven for the American case, and this report treats it as contested [4]. But the burden of evidence from Japan, the Philippines and England points one way.

The wider lesson concerns how governments account for trust. It does not appear in health budgets, yet the 177-country evidence implies it moved pandemic infection rates by margins of 13% to 40% [24]. It cannot be bought quickly, yet it can be spent quickly. And the people who spend it — officials responding to a news cycle, agencies reluctant to admit error, governments using public-health institutions for other ends — rarely bear the cost. Those costs fall later, on children, and on successor governments forced to pay for outbreaks they did not cause. The 154 million lives that immunisation has saved since 1974 [22] were built on an assumption of institutional trust that a single generation has proved capable of eroding.

What would a serious response look like. The evidence suggests four elements. Keep delivery and access strong, because the access problem is real and global [1]. Use mandates where delivery systems are good, accepting that they lift coverage without curing distrust [26] [28]. Invest in the clinicians and community health workers who carry most of the trust that remains [11] [35]. And protect the independence of the bodies that make vaccine recommendations, so that their advice survives changes of government. The last element is the hardest, because it asks governments to constrain themselves. It is also the one the historical cases most clearly support.

The vaccine trust gap is not, at bottom, a story about vaccines. The products are among the most studied interventions in medicine, and the case for them is as strong as it has ever been [22] [23]. It is a story about whether people believe the institutions that recommend those products, and about what happens when they stop. The data from 2025 and 2026 show that the answer arrives quickly and in measurable form: in case counts, in hospital admissions, in the elimination certificates that countries spent decades earning. Rebuilding that belief will take longer than losing it did. The record suggests it can be done, but only by institutions that behave, consistently and visibly, in ways that deserve it.