Misinformation about COVID-19 vaccines cost the United States $2 billion from hospitalizations in 2021

Abstract

Introduction

Misinformation is a major public health threat as it leads to unnecessary illnesses, deaths, and costs to society. In 2021, misinformation about COVID-19 vaccines was rampant, where a large portion of the US population believed in vaccine misinformation and refused vaccination.

Methods

This study utilized a microsimulation model of COVID-19 vaccination, cases, hospitalizations, and deaths to estimate the cost of vaccine hesitancy that was fueled by misinformation in the United States. The analysis compares a baseline model with scenarios where misinformation was removed.

Results

Misinformation was estimated to contribute to nearly 2.3 (2.04–2.5) million cases and 66 000 (59 300–72 500) hospitalizations in 2021. Misinformation resulted in $2 billion ($1.79–$2.33 billion) in extra costs of hospitalization and 45 000 (40 800–50 000) avertable deaths. Misinformation on vaccine hesitancy cost up to $229 million ($226–$231 million) in California, $173 million ($171–$175 million) in Texas, $171 million ($169–$173 million) in Florida, and $144 million ($143–$146 million) in New York State in 2021. Montana, Nevada, Colorado, Arizona, Wyoming, New Mexico, and Alaska faced the highest ($10–$14) per person costs of misinformation.

Conclusion

While the COVID-19 pandemic has subsided, misinformation remains. Combating misinformation on a large scale and building trust in health institutions and science is essential to prevent unnecessary costs and improve population health.

Introduction

The World Health Organization (WHO) called the current health information landscape an “infodemic,” which they define as an overabundance of information during a health event such as the COVID-19 pandemic. It is also characterized by intentional attempts to spread misinformation to hinder public health measures.¹ One of the priorities of the US Department of Health and Human Services, Office of the US Surgeon General, is to combat health misinformation, which is considered a serious threat to public health. Misinformation is defined as “information that is false, inaccurate, or misleading according to the best available evidence at the time.”² Misinformation is a challenge as it results in misallocation of resources, rejection of public health measures, increased vaccine hesitancy, and a negative impact on both physical and mental health.^2,3 Misinformation can lead people to behaviors that endanger their health. During the COVID-19 pandemic, misinformation slowed the impact of interventions and decreased the willingness to vaccinate.^3,4

The United States was considerably impacted by COVID-19, where over 1 million people died from the virus by mid-2022, incurring trillions of dollars in costs to the health care system and individuals.^5,6 The COVID-19 vaccine was initially introduced in the United States in December 2020. At the time it was predicted to avert over half of the COVID-19 cases, hospitalizations, and deaths in the first 9 months of introduction.⁷ However, misinformation about COVID-19 vaccines was widespread and hesitancy to vaccinate was high.⁸ When vaccination became widely available in the United States in 2021, vaccine hesitancy was decreasing globally but increasing in the United States.⁹ COVID-19 vaccine hesitancy was found to cluster in certain groups, especially by age, region, sex, religion, and race.^8,10 Relying on social media as a news source, where misinformation was pervasive, was associated with lower intention to vaccinate with COVID-19 vaccines in 2021.^11-13

Vaccines are cost-effective, and the development of COVID-19 vaccines was marked as an important breakthrough in reducing deaths and halting the pandemic.^14,15 Misinformation that contributed to COVID-19 vaccine hesitancy had an impact on health outcomes and costs for treatment. To our knowledge, the impact of COVID-19 vaccine misinformation has not yet been studied in the United States. This study utilizes observed real-world data on vaccinations, cases, deaths, misinformation beliefs, and costs to examine the health and economic impact of misinformation on COVID-19 vaccine hesitancy in the United States in 2021.

Data and methods

We created a microsimulation model to simulate weekly COVID-19 vaccinations, cases, hospitalizations, and deaths in the United States in 2021. The model simulates people living in 50 US states, plus Washington, DC, and Puerto Rico, in 6 age categories. The model begins on January 1, 2021, with everyone starting as unvaccinated. Based on age-specific data on belief in misinformation, the population is broken down into vaccine accepting or vaccine hesitant due to misinformation. Those who are vaccine accepting could transition to vaccinated based on weekly vaccination rates specific to each age group and state. Probabilities for COVID-19 disease, hospitalization, and death differed for vaccinated and unvaccinated persons. Rather than simulate transmission, the model used observed data for each health state specific to US state and age groups, which were applied weekly for 52 weeks. The model tracked the number of vaccinations, cases, hospitalizations, hospital costs, and deaths from COVID-19.

Data

Table 1 lists the data parameters and sources used in the model. Table S1 provides further details on the data points extracted and data transformations that were made. All data utilized are publicly available. The population for each state and age group was taken from the 2020 US Census.¹⁶ The 6 age groups used in the model are as follows: age group 1, 12–17 years; age group 2, 18–24 years; age group 3, 25–39 years; age group 4, 40–49 years; age group 5, 50–64 years; and age group 6, 65 years and over. The proportion of the population who believed in misinformation and were vaccine hesitant, broken down by age group, was taken from a nationally representative sample from the United States.⁸ The Centers for Disease Control and Prevention (CDC) data for daily new fully vaccinated individuals per state and age group were summed per week and weekly probabilities of becoming fully vaccinated for each state-age combination were calculated.¹⁷ Cumulative COVID-19 cases and deaths were converted to new cases and deaths per week for 2021.^18,22

Table 1.

Data used to model COVID-19 vaccinations and outcomes.

Data	Breakdown	Source
Population	State, age	US Census Bureau, 202016
Vaccination coverage	State, age, week	CDC, 202117
Belief in COVID-19 misinformation	Age	Ognyanova et al8
	Age (model age group):	One misinformed belief:	Two or more misinformed beliefs:
	18-24 y (1-2)	12%	7%
	25-44 y (3)	11%	12%
	45-64 y (4-5)	7%	7%
	65+ y (6)	5%	3%
% Unvaccinated by belief in COVID-19 vaccine misinformation		Ognyanova et al8
	No. of misinformed beliefs:	% Not vaccinated:
	1	51%
	2 or more	69%
COVID-19 incidence	State, week	CSSE at Johns Hopkins University18, Dong et al19
COVID-19 hospitalizations	State, age (adult/children), week	US Department of Health and Human Services20 (new confirmed COVID-19 hospitalizations)
% of COVID-19 hospitalizations that were at the ICU	Age	Calculated from Stokes et al21
	Model age group:	Estimate:
	12-17 y (1)	17.5%
	18-24 y (2)	13.5%
	25-39 y (3)	14.5%
	40-49 y (4)	16.2%
	50-64 y (5)	18.9%
	65+ y (6)	14.8%
COVID-19 deaths	State, week	CSSE at Johns Hopkins University,22 Dong et al19
Breakdown of cases by COVID-19 vaccinated status	January, April, July, November	Johnson et al23
Breakdown of deaths by COVID-19 vaccinated status	January, April, July, November	Johnson et al23
Vaccine effectiveness at preventing hospitalizations	January-June, July-December	Lauring et al24
Cost of hospitalizations	State	Fair Health, 2024 (noncomplex)25
Cost of ICU hospitalizations	State	Fair Health, 2024 (complex)25

Sources of each data item are noted in the table.

Abbreviations: CDC, Centers for Disease Control and Prevention; CSSE, Center for Systems Science and Engineering; ICU, intensive care unit.

The risk of COVID-19 infection and death among vaccinated vs. unvaccinated populations was taken from CDC data from 2021 and varied at 4 time points over the year, corresponding to periods of dominance by different COVID-19 strains, which impacted the effectiveness of the vaccine.²³ The periods were pre-Delta strain, which was applied from January–May, Delta emergence from June–July, Delta predominance from July–November, and Omicron strain emergence for December.²³ The risk ratios were used to break down the observed cases per age-state combinations into vaccinated and unvaccinated weekly cases, then weekly probabilities of COVID-19 illness for vaccinated and unvaccinated groups were calculated.^18,19,23

Hospitalization data were available per state and in 2 age groups: 18 years and older and 17 years and younger.²⁰ Hospitalizations were broken down into complex (intensive care unit [ICU]) and noncomplex cases using data on the proportion of hospitalizations that required ICU treatments and applied by age groups in the model.²¹ The effectiveness of the vaccine at protecting against COVID-19 hospitalizations and deaths was entered in the model as a proportional reduction in the probability of those outcomes in the vaccinated group. We utilized the vaccine effectiveness at preventing hospitalization from a prospective observational study at 2 time points in 2021.²⁴ Requiring ICU hospitalization was not directly impacted by vaccine status in the simulations. However, as a proportion of noncomplex hospitalizations, ICU hospitalizations were indirectly affected by the vaccine effectiveness at preventing hospitalizations. To capture the uncertainty in the data inputs, a range of 10% was incorporated around data on numbers of vaccinations, cases, deaths, hospitalizations, and ICU hospitalizations.

COVID-19 vaccine misinformation counterfactuals

To estimate the impact of COVID-19 misinformation, 2 data points on belief in misinformation and the percentage of misinformed people who did not get vaccinated were utilized. The estimates of belief in misinformation by age group were taken from a Covid States Project report, where a nationally representative sample of US residents were asked about their belief in 4 false statements about COVID-19 vaccines. Specifically, individuals were asked if they believed that COVID-19 vaccines will alter people's DNA, contain microchips that could track people, contain lung tissue of aborted fetuses, and/or can cause infertility. The vaccination status of respondents was also ascertained.⁸ In the first counterfactual, we focused on people who believed 1 false COVID-19 statement. This was 12% for model age groups 1 and 2, 11% for age group 3, 7% for age groups 4 and 5, and 5% for age group 6. The second counterfactual focused on people who believed 2 or more false statements: 7% for model age groups 1 and 2; 12% for age group 3; 7% for age groups 4 and 5; and 3% for age group 6. We broke down the vaccination status by individuals’ belief in misinformation using the same report.⁸ Of those who held only 1 misinformed belief, 51% were not vaccinated by the end of 2021–2022. Of those who held multiple misinformed beliefs, 69% were not vaccinated.⁸ As these are distinct populations, we estimated a third scenario of the impact of removing all misinformation. These three scenarios provide a range around the possible impacts of misinformation.

In the counterfactual scenarios, the effect of misinformation was removed, and those who were not vaccinated at baseline because of misinformation were set to be vaccine-accepting and could become vaccinated. Increasing the acceptance is the only change made in the counterfactual scenarios. The rates of vaccination, cases, hospitalizations, and death stayed constant. Changes in infection because of an increase in vaccinated people were not modeled. Vaccinations per week were capped in the counterfactual scenario at the highest number of vaccinations in a week in each state in the observed real-world data. This ensured that the model stayed in line with the health care system's capacity to vaccinate.

Model outcomes

Results were tracked throughout the model for the number of vaccinations, cases, hospitalizations, and deaths per state for each week. Baseline model outcomes were compared with observed data for number of vaccinations, cases, hospitalizations, and deaths per state for validation. Results for the counterfactual scenario of no misinformation were subtracted from the baseline to demonstrate the impact of reducing misinformation about COVID-19 vaccines. The differences in hospitalization costs between the baseline and no-misinformation scenario per 100 000 population per state were mapped.

Hospitalization costs

Costs for complex and noncomplex inpatient hospitalizations for the baseline and counterfactual scenarios were calculated. Median and average costs were retrieved by state from a Fair Health analysis.²⁵ Allowed values for in-network charges were extracted as a conservative estimate for hospitalization costs for COVID-19. The number of non-ICU hospitalizations per state was multiplied by the state-specific cost for noncomplex COVID-19 hospitalizations and the number of ICU hospitalizations by the complex hospitalization cost. Costs for Puerto Rico were estimated as 31.1% lower than the average US state costs for hospitalization.²⁵ Costs were varied probabilistically using a gamma distribution varying cost inputs within 20% of the point estimates where we present the range around the outcomes.

Results

Table 2 shows the results of the baseline model compared with the counterfactual scenarios where COVID-19 misinformation was removed. In the baseline model, it was estimated that there were 202 million (95% Confidence Interval (CI) 182–222 million) fully vaccinated individuals in the United States by the end of 2021. The number of cases of COVID-19 totaled 34.7 million (95% CI 31.2–38.1 million). These cases resulted in an estimated 2.5 million (95% CI 2.3–2.8 million) hospitalizations across the country and nearly 429 000 (95% CI 386 000–472 000) deaths. We estimated direct hospital costs at over $78 billion (95% CI $71.6–$86.2 billion) for the whole year, making the average cost per hospitalization across all states approximately $31 000.

Table 2.

Model outputs for the burden and cost of COVID-19 vaccine misinformation in 2021.

	Baseline	Removing 1 misinformed belief	Impact	%	Removing multiple misinformed beliefs	Impact	%	Impact of removing all misinformation- driven hesitancy	%
Total vaccinations	202 133 122	210 945 813	8 812 691	4	212 797 056	10 663 934	5	19 476 625	10
Cases	34 673 851	33 646 125	−1 027 726	−3	33 429 614	−1 244 237	−4	−2 271 963	−7
Hospitalizations	2 546 328	2 516 499	−29 829	−1	2 510 256	−36 072	−1	−65 901	−3
Hospitalizations with ICU	406 331	401 594	−4737	−1	400 557	−5774	−1	−10 511	−3
Deaths	428 884	405 350	−23 534	−5	407 095	−21 789	−5	−45 323	−11
Total costs	$77 865 009 366	$76 946 798 222	−$918 211 144	−1	$76 750 837 170	−$1 114 172 196	−1	−$2 032 383 340	−3

Source: Authors’ own analysis of data resulting from the modeling described in the article; sources for the data inputs are noted in Table S1. Vaccine misinformation beliefs include the following: (1) the COVID-19 vaccines will alter people's DNA; (2) the COVID-19 vaccines contain microchips that could track people; (3) the COVID-19 vaccines contain lung tissue of aborted fetuses; (4) the COVID-19 vaccines can cause infertility, making it more difficult to get pregnant. Total costs include direct costs of hospitalizations and ICU stays.

Abbreviation: ICU, intensive care unit.

With more people vaccinated in the counterfactual scenarios, the model estimated a total reduction of 2.3 million (95% CI 2.04–2.5 million) cases in 2021, 7% fewer. The burden of cases due to COVID-19 misinformation was split evenly, with 1 million (95% CI 925 000–1.13 million) among those who believed in 1 misinformed statement and 1.2 million (96% CI 1.1–1.4 million) among those with 2 or more misinformed beliefs. Most of the case reduction happened in the late summer to fall waves of COVID-19 in 2021 when there was a large difference in the number vaccinated in the scenarios compared with baseline, and the incidence of COVID-19 was high.

Removing the effect of misinformation for those with just 1 misinformed belief resulted in a decrease of 29 000 (95% CI 27 000–33 000) hospitalizations, $919 million (95% CI $895 million–$1.2 billion) in hospitalization costs, and 24 000 deaths (95% CI 21 000–26 000) (Table 2). The model estimated a reduction of 36 000 (95% CI 32 000–40 000) hospitalizations when removing misinformation for those with 2 or more misinformed beliefs. That resulted in nearly $1.1 billion (95% CI $890 million–$1.4 billion) less in costs of hospitalization, including costs of reducing nearly 5000 (95% CI 4200–5300) complex ICU hospitalizations. Increasing vaccination by removing this misinformation resulted in nearly 22 000 (95% CI 19 600–24 000) fewer deaths. If the effect of all misinformation was removed, the benefits would double, averting approximately $2 billion (95% CI $1.79–$2.33 billion) in hospitalization costs, 66 000 (95% CI 59 300–72 500) hospitalizations, and 45 000 (95% CI 40 800–50 000) COVID-19 deaths in 2021.

Figure 1 compares the proportion of the US population aged 12 years and older who were vaccinated in the baseline and counterfactual scenarios where misinformation was removed (Figure 1A) alongside the COVID-19 cases per week in 2021 (Figure 1B). Observed data on vaccination and cases from the CDC are depicted for comparison to the baseline. Changing those with 1 misinformed belief to vaccine-accepting resulted in 4% more fully vaccinated people—211 million (95% CI 190–232 million) overall. Changing those who believed in multiple COVID-19 falsehoods to vaccine-accepting resulted in another 5% increase in vaccination. The largest increase in vaccination in the scenarios was seen in the spring from May to June 2021.

Figure 1.

Monthly outcomes for vaccinations (A) and cases of COVID-19 (B) across model scenarios. Source: Authors’ own analysis of data resulting from the modeling described in the article; sources for the data inputs are noted in Table S1.

Figure 2 shows the costs of COVID-19 vaccine misinformation per 100 000 population (12 years and up) per state (Figure 2A) compared to the total costs per state (Figure 2B). While the states with the largest population held the most shares of the overall costs, some states with less population, where hospitalizations were high and vaccinations lagged, experienced more costs of misinformation per person in the state. Most states (81%) experienced misinformation costs above $500 000 per 100 000 population. States with the highest per-population costs of misinformation were Montana, Nevada, Colorado, Arizona, Wyoming, New Mexico, and Alaska, ranging from $1 million to $1.4 million per 100 000 people in the state. The median total cost per state was $24 million, with some high outliers. The states that had the highest costs due to misinformation were California with $229 million (95% CI $226–$231 million), Texas with $173 million (95% CI $171–$175 million), Florida with $171 million (95% CI $169–$173 million), New York with $144 million (95% CI $143$146 million), and Pennsylvania with $107 million (95% CI $105–$108 million).

Figure 2.

Cost of misinformation by state in 2021. Source: Authors’ own analysis of data resulting from the modeling described in the article; sources for the data inputs are noted in Table S1. (A) Costs of COVID-19 misinformation per 100 000 population. (B) Total costs of misinformation per state. Abbreviation: USD, US dollars.

Discussion

Misinformation about vaccinations predates COVID-19. However, the scale of the pandemic and rapid vaccination campaign made this an unprecedented situation for the spread of misinformation. The results of this study show that increasing vaccination among those who believed 2 misinformed statements could have reduced COVID-19 cases by 1.2 million in 2021. This would have resulted in fewer hospitalizations and deaths, yielding direct hospitalization cost savings of $1.1 billion. Removing misinformation among people who believed 2 or more misinformation statements would have resulted in nearly 1 million fewer cases and a $918 000 reduction in hospitalization costs. As the populations who believed 1 as opposed to 2 or more misinformed statements are distinct populations, increasing vaccination in both would nearly double the impact. This means that, when all misinformation is removed, the US would have benefitted from cost savings of $2 billion.

Costs of misinformation had a substantial impact ($3.2–$229 million) in each state. Our results showed that the cost of misinformation-driven vaccine hesitancy was born by all US states. While New York and California had high proportions of residents vaccinated in 2021, hospitalizations among people who believed misinformation still cost each of those states millions that year. Alaska’s and Nevada's misinformation costs per population were high primarily because they had the highest COVID-19 hospitalization costs in the country,²⁵ so every avertable hospitalization due to misinformation had a greater economic impact there. Avertable hospitalizations and deaths are devastating to patients and their families and can leave them with medical debt. In 2021, the waivers for patient out-of-pocket care for COVID-19 hospitalizations were dropped, meaning that the costs of hospitalizations resulting from misinformation about COVID-19 vaccines were born not only by public and private health insurance but also by the patients and their families.^26,27

Misinformation and its impacts go beyond that of COVID-19 vaccines. While the emergency status of the COVID-19 pandemic ended in early 2023, the erosion of trust and skepticism towards science that was borne from misinformation around the COVID-19 pandemic remains.^28,29 People who believe in misinformation trust science less and are more likely to use unproven medical measures.³⁰ Misinformation decreases the credibility of information circulating during times of unease, increases mental and physical fatigue, and can lead to misallocation of resources and spending on unapproved and illegal substances.³ This study provides an example of the health and economic costs of misinformation for a single year and disease, but misinformation about other health topics is also detrimental. For example, misinformation campaigns and advertisements with false information were shown to impact the uptake of pre-exposure prophylaxis for HIV treatment.^31,32

We need to address misinformation and build back trust in science. In some contexts, creating trust in institutions could be a way to overcome the effects of misinformation.³³ But the government itself might not be the best messenger for health information as low levels of trust in government persist.³⁴ Sometimes building trust might be in opposition to the desired public health outcome, such as when communicating transparently about vaccine side effects can decrease vaccination intention but improve trust in science.³⁵ On a patient level, doctors need to understand that the information their patients are getting from the internet affects their health decisions, and they need to be given time to guide patients where to find accurate health information online.³⁶

Interventions to address misinformation and vaccine hesitancy are being tested, but there is not a one-size-fits-all method for the issue, especially given that misinformation comes in many categories from different sources.³⁷ A systematic review that identified 119 different interventions to combat COVID-19 misinformation found that interventions that used debunking offer some optimism, but due to the difficulty in defining and measuring misinformation, it is not easy to compare these interventions.³⁸ Two studies in Germany that tested the effect of debunking and prebunking on COVID-19 vaccine misinformation found that debunking impacted people's beliefs in misinformation, but that impact did not extend to also change their intent to vaccinate or their perceptions about vaccines.^39,40 Another study in Germany similarly found that debunking did not increase one's willingness to vaccinate, but an intervention where benefits of vaccination were repeatedly highlighted over time did eventually increase willingness.⁴¹ Deeply held misinformed beliefs can be highly entrenched, where correcting people's false beliefs may not be enough to prompt them to vaccinate. Even after misinformation has been corrected, the perception or feeling of unease regarding vaccines often remains.

Other studies have estimated the impact of COVID-19 vaccination and vaccine hesitancy in the United States. An analysis of hospital claims data in the United States estimated that 1200 hospitalizations and nearly $1 billion per day could have been averted by completely removing vaccine hesitancy during months in 2021 when COVID-19 transmission was relatively low.⁴² Of that, $50–$300 million is attributable to mis- or dis-information.⁴² In Canada, vaccine hesitancy fueled by misinformation was estimated to cost over 300 million Canadian dollars in 2021 in direct hospitalization and ICU costs.⁴³ Our model, which was populated with observed data for vaccinations, cases, hospitalizations, and deaths, looked specifically at the portion of vaccine hesitancy that was driven by misinformation and did not look at removing all vaccine hesitancy in the United States. Removing misinformation-driven vaccine hesitancy showed a 10% increase in fully vaccinated individuals overall, but removing all vaccine hesitancy would result in a substantially larger impact.

Assumptions and structural decisions were made in this study that created some important limitations when discussing the model results. Namely, while the data on belief in vaccine misinformation are linked to vaccination status, we only have data on misinformation status by age group, not specific to each individual US state or territory. This means that the variations in difference between the baseline and the scenarios removing misinformation are based on the age structure of each state and the state's observed vaccination rates. The data on misinformation utilized are nationally representative, but this may mean that the results between states are less pronounced. Moreover, some characteristics make people more susceptible to believing misinformation, including belonging to some religious groups, political partisanship, race/ethnicity, and underlying trust in science and institutions.^44-46 These would be important for policymakers to consider when planning interventions but were not included in this analysis. This study does not model the transmission of COVID-19, which would be affected by increases in vaccination. Instead, we utilized the same incidence of COVID-19 for vaccinated populations in both the baseline and counterfactuals and changed the proportion of the population that is vaccinated. This could lead us to a more conservative estimate of the impacts of increased vaccination. Finally, we only included direct medical costs of hospitalization, and not indirect costs such as productivity losses due to premature deaths. Including those indirect costs would vastly increase the cost of misinformation in the results.

Conclusion

Widespread misinformation in the United States was linked to COVID-19 vaccine hesitancy and refusal to vaccinate. This devastatingly cost the country over 45 000 lives and nearly $2 billion from hospitalizations in 2021 alone. While the COVID-19 pandemic has subsided, the impacts of the infodemic remain and extend to other public health issues. Beyond addressing COVID-19 vaccine hesitancy, combating misinformation on a large scale and building trust in institutions and science would be impactful both to save lives and prevent avoidable costs to society.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This work was supported in part by a research grant from an Investigator-Initiated Studies Program of Merck, Sharp & Dohme Corporation. The opinions expressed in this article are those of the authors and do not necessarily represent those of Merck, Sharp & Dohme.