COVID-19 Vaccine Hesitancy in Caregivers of Hospitalized Children From 2020 Through 2023

We performed a repeated cross-sectional, survey-based study at a large, urban, freestanding quaternary care children’s hospital in Houston, Texas, as well as 2 satellite pediatric community hospitals. We recruited a convenience sample of caregivers of children ages 6 months to 18 years admitted to general inpatient pediatric services from December 8, 2020, to April 5, 2021 (season 1 [S1]), and from November 30, 2021, to March 15, 2022 (season 2 [S2]). We targeted caregivers of hospitalized children ages 6 months to 11 years during the enrollment period spanning October 26, 2022, to March 15, 2023 (season 3 [S3]); after noting lowest vaccine uptake in younger age groups, we adjusted our sampling to reframe our focus on caregivers of children aged 6 months to 11 years. Supplemental Figure 3 shows a COVID-19 timeline by study season. Caregivers were excluded if they did not speak English or Spanish, if they had already enrolled in the study, or if the child was in Child Protective Services custody. Subjects were also excluded if they were SARS-CoV-2–positive or had a SARS-CoV-2 polymerase chain reaction (PCR) pending during S1 and S2 for study personnel protection. The university’s institutional review board approved this study.

We identified participants using daily census reports. Surveys were administered from 12 through 8 Pm on days when study investigators were available. Caregivers had the option of completing the survey in English or Spanish on an electronic tablet (using Research Electronic Data Capture^9,10 ) or verbally with assistance from a study investigator. Surveys took ∼10 to 15 minutes to complete.

We designed a 32-item survey (Supplemental Fig 4) to assess demographic information and caregiver attitudes, behaviors, and beliefs about COVID-19 and the COVID-19 vaccine. Survey questions included demographics such as caregiver gender, age, marital status, educational level, relationship to child, race/ethnicity, child age, child’s health insurance status, and household income, as well as caregiver attitudes, behaviors, and beliefs about COVID-19 and the COVID-19 vaccine. We also asked questions about personal contact to someone infected with COVID-19, how the pandemic has affected parental decisions regarding routine childhood vaccines, and caregiver intent to vaccinate themselves and their children. The survey was developed in English on the basis of content expertise and adaptations from existing literature.^11,12  We also included questions about contact with someone infected with COVID-19 and how the pandemic affected decisions regarding routine childhood immunization. Some items regarding the COVID-19 pandemic and COVID-19 vaccine were added to the survey on the basis of their relevance during S2 (2021–2022) and S3 (2022–2023).

The survey was translated into Spanish by qualified study members. We performed semistructured interviews with 46 English and 13 Spanish-speaking caregivers of hospitalized children to assess the face validity of the survey as previously described.¹³  Interviews assessed if the survey was easy to complete, understandable, culturally appropriate, and reasonable in length.^14–18  A 7- to 10-day test–retest reliability was conducted with 21 English-speaking caregivers demonstrating good reliability. We assessed general VH using the validated Parent Attitudes about Childhood Vaccines (PACV) survey.¹⁹ 

The PACV was scored using the method established by Opel et al.¹⁹  An unweighted score was generated to obtain a raw total PACV score, which was then converted to a 0 to 100-point scale using simple linear transformation.²⁰  General VH was defined as a final score of ≥50.

Given the enrollment of caregiver children aged 6 months to 11 years during S3, caregivers of children aged ≥12 who participated in S1 and S2 were removed from the analysis. All analyses were performed using R (RStudio Team [2020]. RStudio: Integrated Development for R. RStudio, PBC, Boston, Massachusetts). Our primary outcome was to identify trends in COVID-19 disease and vaccine attitudes, behaviors, and beliefs. Our secondary outcome was to identify associations with caregiver decision or intent to give their child COVID-19 vaccine. Responses were summarized by frequencies with proportions and compared by study season and vaccine status. We used χ² test for categorical variables. Multivariable logistic regression was used to identify variables associated with caregiver decision or intent to give their child SARS-CoV-2 vaccination. Controlling for collinearity, survey variables with P values of <.2 from univariable analyses were considered for our multivariable regression models. Backward selection by P value was used to select the reduced multivariable logistic regression analysis with calculated odds ratios and corresponding 95% confidence intervals. Variables with P values of <.05 were considered significant.

Across study seasons, 4072 caregivers were screened, 2428 met inclusion criteria, 1632 were approached, and 1507 were enrolled (Fig 1). Data from 107 (36%) caregivers of children aged ≥12 years in S1 and 109 (28%) from S2 were excluded from the analysis, given the focus on caregivers with children ages 6 months to 11 years during S3. Demographics by enrollment season are found in Table 1. Caregivers identified as Hispanic/Latino (49%), non-Hispanic white (25%), Black or African American (17%), Asian American (7%), and 2% other. Nineteen percent of caregivers screened positive for general VH across seasons.

In S1, 41% of caregivers intended to vaccinate their child against COVID-19. In S2, 69% of caregivers reported receiving COVID-19 vaccine versus 56% in S3 (P < .001). In S2 and S3, 10% and 19% of children received COVID-19 vaccine; 37% and 10% of children had not received the vaccine but the caregiver intended to vaccinate the child, and 53% and 71% of caregivers did not intend to vaccinate their child, respectively (P < .001). Figure 2 shows COVID-19 vaccine status by PACV score.

Top sources of COVID-19 vaccine information included doctors (57%, P = .24) and online/Internet (56%, P = .13) in S2 to S3. More caregivers were satisfied with information received from health care providers on COVID-19 vaccine in S3 (54% in S2, 62% in S3; P = .02). Across S2 and S3, caregivers preferred to receive information via conversations with their doctors (60%, P = .35) versus educational materials such as handouts (32%, P = .49). Fewer caregivers, however, reported being able to openly discuss their concerns about COVID-19 vaccine with their doctors in S3 (71% in S2, 64% in S3; P = .04). Before the hospitalization, 44% and 67% (P < .001) of caregivers in S2 and S3 talked to a health care provider about COVID-19 vaccine for their child (Table 2).

In S1, S2, and S3, 63%, 66%, and 51% of caregivers, respectively, knew someone who was hospitalized because of COVID-19 (P < .001). Moreover, 44%, 57%, and 49%, respectively, knew someone who died of COVID-19 (P = .03). In S1, S2, and S3, 68%, 74%, and 61% of caregivers reported being scared of their child getting COVID-19 (P < .001). Approximately one-third of caregivers in S2 and S3 reported believing that the COVID-19 vaccine was safe (P = .44). Almost half (47% in S2, 46% in S3; P = .76) expressed concerns that the COVID-19 vaccine is new, and that it was developed quickly (46% in S2, 45% in S3; P = .78). However, the majority agreed that the COVID-19 vaccine would play an important role in bringing the pandemic under control (56% in S1, 55% in S2, and 53% in S3; P = .63).

Belief that COVID-19 can be dangerous for children remained stable between S2 and S3 (76% vs 72%; P = .44), but fewer caregivers believed children who are otherwise healthy can die of COVID-19 (68% in S2, 52% in S3; P < .001). More caregivers in S3 believed the COVID-19 vaccine does not work (P = .02). In S3, compared with previous seasons, fewer thought vaccination prevents complications from COVID-19 (51% vs 41%; P = .005), and that vaccination prevents children from dying from COVID-19 (36% vs 29%; P = .03) (Table 2, Supplemental Table 4).

Supplemental Table 5 displays demographics by child COVID-19 vaccination status. On univariable analysis, child receipt of COVID-19 vaccine was significantly more likely in S3 (versus S1–S2), children 5 to 11 years old, privately insured children, caregivers aged >30 years, caregivers who were married/living with a partner, caregivers with some college or higher education, higher household income, and Asian American race. Child COVID-19 vaccination was significantly less likely with PACV score >50. The child being up to date with routine vaccines by parental report, the child receiving influenza vaccine, and caregiver receipt of COVID-19 vaccine were significantly more likely with caregiver decision to vaccinate and caregiver intent to vaccinate their child. There was no association between child vaccination status and caregiver religious or political beliefs.

Caregivers who reported concern that the COVID-19 vaccine was new or who were not scared of their child getting COVID-19 were significantly less likely to immunize or intend to immunize their child. Caregivers who were not going to vaccinate their child were significantly more likely to get information about COVID-19 from online sources, less likely to go to their doctors for information, and less likely to trust the information they receive from doctors on COVID-19 vaccine. Supplemental Tables 6 and 7 display additional caregiver experiences, beliefs, attitudes, and behaviors by child vaccination status.

In multivariable logistic regression, study season (2022–2023), older child age, higher household income, child receipt of influenza vaccine, caregiver receipt of COVID-19 vaccine, and caregiver not being “worried that the COVID-19 vaccine is new” were positively associated with child receipt of COVID-19 vaccine. Intent to vaccinate was negatively associated with study season (2022–2023), general VH (PACV ≥50), lack of child influenza and caregiver COVID-19 immunizations, caregiver not being “scared of child getting COVID-19,” and caregiver being worried that the COVID-19 vaccine is new (Table 3).

We report a robust, uniquely designed, repeated cross-sectional study eliciting COVID-19 and COVID-19 vaccine beliefs, attitudes, and behaviors of caregivers among hospitalized children in the Southern United States. Although 41% of caregivers reported intent to vaccinate their child in S1 (before vaccine availability for children), only 10% and 19% of children had received vaccine in S2 and S3 when vaccine was available. Our vaccination rates are consistent with national data; 9% of US children <2 years (10% in our study), 11% of 2- to 4-year-olds (15% in our study), and 40% of 5- to 11-year-olds (33% in our study) received at least 1 dose of COVID-19 vaccine through March 15, 2023.²  Our study provides insight into why caregivers are declining COVID-19 vaccine for their children, as well as how caregivers prefer to receive information about COVID-19 and COVID-19 vaccine.

Given the staggered EUA for pediatric COVID-19 vaccines, it was expected that more children would have received vaccine in S3 compared with S2. Unfortunately, caregivers reported significantly less intent to vaccinate their children in S3 (10%) versus S2 (37%). This general trend has been observed in the existing literature. A nationally representative sample in March 2020 showed that ∼80% of Americans would accept COVID-19 vaccine for themselves and their children.²¹  Surveys of US parents from February to March 2021 found that only 46% to 48% of caregivers intended to vaccinate their children.^7,22  A 2021 study found that parents cited full Food and Drug Administration (FDA) approval as the major factor that would increase their likelihood of vaccinating their children in the future.⁸  Unfortunately, a surge in vaccine uptake after full FDA approval has not materialized in adults nor in older children.

Previous studies suggest that reasons for low uptake and increasing COVID-19 VH include concerns regarding vaccine safety and novelty.^7,8,23–27  We found that these concerns remain prevalent; 50% of caregivers across S1 to S3 reported fear of their child “getting COVID-19 vaccine” and concern that the vaccine is new. In S2 to S3, only a third believed that the vaccine was safe. Of note, S1 enrollment ended before media coverage of rare COVID-19 vaccine adverse events, including blood clots, myocarditis/pericarditis, and Guillain-Barre syndrome.²⁸  Furthermore, rare vaccine adverse events associated specifically with messenger RNA vaccines (myocarditis/pericarditis) have not been reported in children ages 6 months to 11 years. This suggests that both initial and continued fear surrounding COVID-19 vaccine is not grounded on facts or safety monitoring data. It is important that health care providers acknowledge and address caregiver fear directly, explain that COVID-19 vaccines are the most studied vaccines in history, and emphasize the safety profile in children aged 6 months to 11 years.

Caregiver COVID-19 VH has also been attributed to concerns regarding vaccine effectiveness and a lack of perceived threat of COVID-19 infection in children.^23,24,26,27  We found these beliefs were more prevalent over time. In S3, more caregivers believed the COVID-19 vaccine was ineffective and fewer thought children were at risk for death from COVID-19. Yet, COVID-19 can cause respiratory failure, myocarditis, and multisystem inflammatory syndrome in children.⁶  Moreover, in 2021 to 2022, COVID-19 was the leading infectious cause of death among children and young people aged 0 to 19 years.⁴  COVID-19 vaccine protects against infection, medically attended disease, hospitalization, and multisystem inflammatory syndrome in children.^6,29  Caregivers should receive consistent messaging from health care providers highlighting the real threat COVID-19 poses to children, as well as the data demonstrating protection against symptomatic infection and serious illness after vaccination.⁶ 

Similar to previous studies, we identified additional associations with child receipt of COVID-19 vaccine and caregiver intent to vaccinate their child.^7,8,25,30,31  Caregivers with a PACV score ≥50 were less likely to report intention to vaccinate their child against COVID-19, as expected. Notably, similar to what has been described with influenza VH, caregivers with COVID-19 VH had PACV scores that spanned the spectrum of general VH such that COVID-19 VH is not unique or exclusive to caregivers who are hesitant about vaccines in general.³²  Having said that, caregiver COVID-19 vaccination status and child influenza vaccination status were strongly associated with child receipt of vaccine and caregiver intent to vaccinate. All caregivers should receive a presumptive and clear recommendation to vaccinate regardless of likelihood to immunize based on these associations. Additional effective communication strategies will need to be developed and deployed when counseling a caregiver who is hesitant despite a strong recommendation to immunize.

Unlike previous studies, our study did not identify associations between COVID-19 VH and political or religious affiliation or race and ethnicity.³³  A previous study with comparable non-Hispanic Black respondents found that caregivers who did not intend to vaccinate their child were more likely to be Black.²³  Although earlier studies also showed lower COVID-19 vaccine acceptance among people who identified as Hispanic,^34,35  later studies showed higher vaccine uptake among Hispanic populations.^7,8  However, given that Black or African American and Hispanic or Latino communities have been disproportionately affected by COVID-19, increased resources and support are still needed to combat COVID-19 and VH in minority patients and communities.^35,36 

Our study identified a troubling trend in caregiver trust in health care providers. National surveys of US parents from 2021 found that 72% to 89% of parents agreed their children’s health care provider is a trustworthy source of information on COVID-19 vaccine.^7,8  By comparison, 55% of caregivers in our study agreed they trust information on COVID-19 vaccine from doctors. Fewer caregivers in S3 felt they could openly discuss their concerns about COVID-19 vaccine with doctors. Notably, compared with caregivers who had or were going to vaccinate their children, fewer caregivers who did not intend to immunize their children reported trust in the information received from physicians and that they could discuss their concerns about COVID-19 vaccine with doctors. Given the nature of the survey question insofar as it alludes to physicians in general, it could be that our findings are at least in part because of the difference between trust in a doctor specifically as compared with trust in doctors generally. However, health care institutions must actively work to recover any trust lost during the pandemic to optimize child health outcomes overall and for COVID-19. Importantly, the majority of caregivers, even among those who did not intend to vaccinate their children, still wanted to receive information on COVID-19 vaccine by talking with their doctor. We should therefore take advantage of this preference, listen to caregiver concerns, and further research and implement interventions that optimize provider–caregiver vaccine-related communication.

Finally, our study supports exploring the inpatient setting for vaccination of children against COVID-19. We conducted this study in caregivers of hospitalized children, both “low-risk” children and children with underlying medical needs who are at higher risk of severe illness and complications from COVID-19.³⁷  We found that the majority who intend to vaccinate their children are willing to give the vaccine before discharge if recommended by their doctor, but that few were offered COVID-19 vaccine in the hospital. Hospitalizations thus provide an excellent opportunity to counsel families and vaccinate children. Hospitalists can play an active role in improving vaccine uptake. Best practice alerts and standing orders for COVID-19 vaccine should be studied in a hospital setting.

Our study has several limitations. First, this was a convenience sample, and by virtue of not being a random sample, our results might not be fully representative of all eligible caregivers of hospitalized children. Bias was minimized as much as possible by only enrolling during similar months and within specified time frames. Although our survey aimed to capture the evolution of attitudes, behaviors, and beliefs on COVID-19 and COVID-19 vaccine in caregivers of hospitalized children, because the same caregivers weren’t surveyed each season, it is possible that changes in responses may reflect differences in participants rather than being representative of trends over time. We also enrolled caregivers at 3 pediatric hospitals within a single health care organization in the Southern United States, so our results might not be generalizable to other settings or geographic areas. Additionally, survey development was conducted primarily, but not exclusively, in English. Our survey was limited to English- or Spanish-speaking caregivers, leading to the exclusion of 39 caregivers across study seasons. Still, our survey population was drawn from 1 of the most racially and ethnically diverse metropolitan areas in the nation, and we enrolled caregivers of hospitalized children at both our urban/quaternary-care pediatric hospital and at our 2 community pediatric hospitals to mitigate this limitation. Third, for infection control and preservation of personal protective equipment purposes, any caregiver of a child who was SARS-CoV-2–positive or who had a SARS-CoV-2 PCR pending was not eligible for enrollment during S1 and S2. However, only 14% of caregivers in S1 and 11% in S2 were excluded for these reasons; most PCR tests resulted within a 24-hour period, and caregivers became eligible once their child no longer required enhanced respiratory precautions. Finally, our study was limited by enrollment completion before full FDA approval of SARS-CoV-2 vaccine in children aged <12 years, so we were unable to determine how this affected beliefs and attitudes.

Only a third of caregivers have vaccinated or intend to vaccinate their children against COVID-19. Compared with earlier studies, intended vaccine uptake is lower. Vaccine novelty and perceived lack of need continue to be factors associated with vaccine refusal. Additionally, influenza vaccine acceptance and caregiver COVID-19 vaccine status remain positively associated with COVID-19 vaccine uptake in children. The inpatient setting is an important opportunity to ameliorate VH and improve vaccine uptake.

We thank Dr Huay-Ying Lo and Dr Kathryn Ban for their assistance coordinating access to our targeted patient population. Lastly, we thank Texas Children’s Hospital and the Pediatric Hospital Medicine faculty and staff for allowing us to complete our study across affiliated institutions.