Attitudes Toward COVID-19 Vaccines Among Pregnant and Recently Pregnant Individuals

This survey study evaluates changing attitudes regarding COVID-19 vaccines by vaccination status and race, ethnicity, and language.


Introduction
][9][10][11][12][13] Young children are also at risk of severe disease and death from COVID-19, with infants under 6 months old with the highest rates of severe COVID-19 outcomes. 14,15[18][19][20] As the COVID-19 public health emergency ended, these trends continued.As of July 29, 2023, Vaccine Safety Datalink (VSD) surveillance found just 16.2% of pregnant people aged 18 to 49 years had received a COVID-19 booster vaccine, with only 8.3% of Black pregnant people and 9.6% of Latino pregnant people vaccinated during pregnancy. 21Assessing attitudes toward COVID-19 vaccines among pregnant and recently pregnant people is critical to public health messaging and clinician counseling. 22We conducted 2 cross-sectional surveys of distinct samples of pregnant and recently pregnant individuals during the latter portion of the COVID-19 pandemic.Our primary aims were to assess trends in attitudes regarding COVID-19 vaccines by (1) self-reported vaccination status and (2) race, ethnicity, and preferred language.

Overview
The Colorado Multiple Institutional Review Board (COMIRB) approved this survey study; participating sites' institutional review boards ceded oversight to COMIRB.COMIRB granted a waiver of written consent for study participation because the survey's introduction language made it clear a respondent's choice to complete the survey indicated consent to participate.This work was part of a larger project assessing vaccination attitudes and status among pregnant and nonpregnant VSD members over time. 21Distinct cohorts of pregnant and recently pregnant members were sampled over time, with the first survey administered from November 1, 2021, to February 1, 2022, and the second survey administered from October 1, 2022, to February 1, 2023.Respondents were asked to report their COVID-19 vaccination status and share their attitudes toward COVID-19 infection and monovalent COVID-19 vaccines (wave 1) or bivalent Omicron booster vaccines (wave 2).Selfreported vaccination status was considered the criterion standard, as with prior attitudinal surveys in the VSD. 23We followed the American Association for Public Opinion Research (AAPOR) reporting guideline for survey studies, applying Response Rate Definition 6. 24

Setting
The VSD is a collaboration between the US Centers for Disease Control and Prevention (CDC) and 13 integrated health care systems (called sites). 25Vaccination data were derived from the electronic health record (EHR) and reconciled routinely with state immunization information systems. 26VSD members represent over 3% of the total US population. 27Eight VSD sites contributed data: Denver Health (Colorado), Kaiser Permanente Colorado, Marshfield Clinic (Wisconsin), HealthPartners (Minnesota), Kaiser Permanente Washington, Kaiser Permanente Northwest (Oregon), Kaiser Permanente Northern California, and Kaiser Permanente Southern California.

Study Participants: Identification of Pregnant Persons
9][30] We used this algorithm to identify adults aged 18 to 49 years who were pregnant any time between December 11, 2020, and

Study Participants: Sampling Procedures
Within eligible cohorts, we conducted stratified sampling at each VSD site using 6 mutually exclusive strata defined by the following EHR factors: COVID-19 vaccination (unvaccinated or Ն1 vaccines), race (Black or other race and ethnicity, which included American Indian or Alaskan Native, Asian, Native Hawaiian or Pacific Islander, and White), and preferred language (English or Spanish).We oversampled individuals roughly 2:1 whose EHR indicated they were unvaccinated or self-identified as Black race or preferred Spanish language, as prior surveys of pregnant VSD members have shown lower response rates in these groups. 23Spanish-speaking participants were sampled from Denver Health and Kaiser Permanente Southern California. 32Separate samples were created for each wave.

Sample Size
The target sample size for each survey wave was determined by a related project outcome: the accuracy of EHR vaccination data vs self-reported vaccination.Assuming variable response rates by vaccination status and race, 23 a 60% negative predictive value (EHR-unvaccinated people reporting they were unvaccinated), and a sample of 907 unvaccinated individuals and 593 vaccinated individuals for each survey wave, the study was powered to achieve a 2.0% CI around the vaccination confirmation rate and an 8.0% CI around the nonvaccination confirmation rate in each wave.
Estimated CIs were corrected for the anticipated response rate.

Survey Design and Cognitive Interviews
[42][43] One question probed perceptions of trusted information sources, and as response categories changed between waves, we have presented data for this question from wave 2 only.We included questions about race, ethnicity, household income, household size, and highest educational attainment.For respondents, self-reported race and ethnicity were the reference standards 44 ; EHR data were used when survey data were missing.[18][19][20] Draft surveys underwent a first round of revisions from VSD site content experts.Afterwards, the survey was translated into Spanish by a certified bilingual translator.

Survey Administration
Prospective participants received up to 10 survey invitations via postal mail, e-mail, and telephone calls, per tailored survey design best practices. 47Survey administration was consistent across VSD sites, except 1 site required participants to receive a presurvey letter with an opportunity to opt out and prohibited contact by email or phone.Outreach stopped after survey completion or if a person opted out.Surveys were hosted online via Research Electronic Data Capture Software. 48spondents received a $25 gift card.We obtained a waiver of written consent, but participants could opt out by email, in writing, or by phone.

Statistical Analysis
We defined respondents as people who completed the first survey question: "Have you received a COVID-19 vaccine?"We used the Pearson χ 2 test to compare survey respondents to nonrespondents on sociodemographic variables available via EHR using 2-sided tests, considering a P value less than .05significant.Among respondents, we calculated weighted descriptive statistics for any selfreported COVID-19 vaccination (ie, Ն1 dose ever), self-reported monovalent booster vaccination among those who had received at least 1 prior COVID-19 vaccine (wave 1), and self-reported Omicron booster vaccination among those who had received at least 1 prior COVID-19 vaccine (wave 2).For attitudinal and sociodemographic measures, we compared respondents by self-reported vaccination status (unvaccinated vs receipt of Ն1 dose) and by 3 mutually exclusive race, ethnicity, and language groups (non-Hispanic White, non-Hispanic Black, and Spanish-speaking Hispanic of any race).
Missingness in the dataset was low (<10%), and unknown or missing categories were included in analyses.We used the Rao-Scott χ 2 test for weighted tables using 2-sided hypothesis testing, considering a P value less than .05significant.We included a finite population correction and incorporated inverse probability weighting to account for sampling and response probability by VSD site, vaccination status, and oversampling of non-Hispanic Black and Spanish-speaking Hispanic people.When considering both waves, individual survey weights were adjusted to reflect an average annual population. 49Analyses were conducted using SAS version 9.4 (SAS Institute).Data were analyzed from May 2022 to September 2023.
By EHR, 877 of these were unvaccinated, 551 identified as non-Hispanic Black, and 510 preferred Spanish language.In wave Table 2 compares the weighted estimates for self-reported sociodemographic characteristics in ever-vaccinated respondents and unvaccinated respondents across waves.Generally, sociodemographic characteristics did not change among vaccinees from wave 1 to wave 2, but unvaccinated individuals in wave 2 were more likely to be older, identify as White or other races, have an associate degree (or higher), and have household income above 200% of the federal poverty level (Table 2).Ever-vaccinated respondents differed from unvaccinated respondents in many respects in wave 1 (eTable 1 in Supplement 1) but only by educational attainment by wave 2 (eTable 2 in Supplement 1).

Perceptions of COVID-19 Vaccines and Vaccine Safety Across Waves
Table 3 provides weighted estimates of COVID-19 vaccine safety perceptions across waves, stratified by vaccination status.Among those reporting 1 or more COVID-19 vaccination, we observed a 34% relative decrease in the proportion in agreement that COVID-19 vaccines are safe for pregnant people and a 32% relative decrease in the proportion agreeing that COVID-19 vaccines are safe for a pregnant person's baby (Table 3).There was a 41% relative decrease (χ 2 1 = 10.3;P < .01) in the proportion of those ever-vaccinated agreeing that most pregnant people should get a COVID-19 vaccine.Among unvaccinated respondents, attitudes toward COVID-19 vaccines were largely unfavorable and did not change across waves (Table 3).Attitudes toward COVID-19 vaccines differed by vaccination status in both waves (eTable 3 and eTable 4 in Supplement 1).In wave 1, desiring to wait until after pregnancy and perceived vaccine adverse effects were the top reported reasons for being unvaccinated (eTable 3 in Supplement 1).

JAMA Network Open | Infectious Diseases
b Wave 1 of the survey was conducted from November 2021 to February 2022.
c Wave 2 of the survey was conducted from October 2022 to February 2023.
d Self-reported race and ethnicity was used as the criterion standard; if unavailable, electronic health record data for race and/or ethnicity were used.
e Other race included individuals whose race was reported as "other" by survey self-report or by electronic health record (EHR).
f Estimated per 2021 Federal Poverty Level Standards, based on family size and yearly income.

Trust in the CDC and Health Care Practitioners
In the second survey wave, respondents' most trusted sources for information about COVID-19 and COVID-19 vaccines varied by vaccination status (eTable 4 in Supplement 1).Among ever-vaccinated participants, 34.4% (95% CI, 18.5%-50.3%)trusted the CDC the most for this information, and a d Response categories were "not at all/not very safe" (negative), "not sure/prefer not to answer" (neutral), and "very/somewhat safe" (positive).
e Response categories were "very/somewhat hesitant" (negative), "not sure/prefer not to answer" (neutral), and "not too/not at all hesitant" (positive).
f Response categories were "will probably/definitely NOT get" (negative), "not sure/ prefer not to answer" (neutral), and "will probably/definitely get" (positive).

Discussion
Among diverse pregnant and recently pregnant VSD members during the latter part of the COVID-19 public health emergency, we observed significant changes in perceptions of COVID-19 vaccine safety over time.First, we observed a significant change over time in respondent groups' perceptions of COVID-19 vaccine safety for pregnant persons and their infants.While we did not survey the same individuals at 2 time points, the general trends we observed among those who had received at least 1 COVID-19 vaccine and among racially, ethnically, and linguistically diverse groups are concerning.
Substantial evidence continues to accrue supporting COVID-19 vaccine safety for pregnant people, 10,50-52 but the concomitant spread of misinformation 53,54 may partially explain our declines in perceived safety.A Kaiser Family Foundation survey ending in June 2023 found that 27% of respondents believed COVID-19 vaccines have been shown to cause infertility; 68% of respondents were uncertain whether the claim was definitely true or definitely false. 55Interestingly, 94% of adults in the Kaiser survey reported "a great deal" or "fair amount" of trust in their physician to make the appropriate vaccination recommendations for them, with the next most highly trusted information source being the CDC. 55Our data suggest more modest levels of trust in the CDC and physicians with notable differences by race, ethnicity, and language preference.Future work could study perceived safety of COVID-19 vaccines as a function of message tailoring for diverse groups.d Response categories were "not very/not at all safe" (negative), "not sure/prefer not to answer" (neutral), and "very/somewhat safe" (positive).
e Response categories were "very/somewhat hesitant" (negative), "not sure/prefer not to answer" (neutral), and "not too/not at all hesitant" (positive).
g Response categories were "will probably/definitely NOT get" (negative), "not sure/ prefer not to answer" (neutral), and "will probably/definitely get" (positive).

JAMA Network Open | Infectious Diseases Attitudes
Toward COVID-19 Vaccines Among Pregnant and Recently Pregnant Individuals August 31, 2021(wave 1), or January 1, 2022, and August 1, 2022 (wave 2).As time elapsed between sampling and surveys going into the field, respondents included currently and recently pregnant people.Eligible individuals had continuous health insurance JAMA Network Open.2024;7(4):e245479.doi:10.1001/jamanetworkopen.2024.5479(Reprinted) April 8, 2024 2/15 Downloaded from jamanetwork.comby guest on 04/18/2024 enrollment during the same periods for each wave, except at Denver Health, which uses empanelment as a proxy for enrollment. 31We excluded people with an International Statistical Classification of Diseases, Tenth Revision, Clinical Modification code for adverse pregnancy outcomes (eg, spontaneous abortion, anencephaly), those with possible data errors (eg, simultaneous administration of multiple COVID-19 vaccines), and individuals who had opted out of research.
2, 123 690 people were eligible and 1456 (1.2%) were sampled.By EHR, Figure.Timeline of Survey Waves 1 and 2 in Relation to the Centers for Disease Control and Prevention (CDC) and American College of Obstetricians and Gynecologists (ACOG) COVID-19 Vaccine Emergency Use Authorizations (EUAs) and Subsequent Recommendations for Monovalent and Bivalent (Omicron) Booster COVID-19 Vaccines in Pregnant Individuals language.Abbreviation: NC, not calculated.aVaccineSafety Datalink sites contributing data were geographically located in California, Colorado, Minnesota, Oregon, Washington, and Wisconsin.bWave 1 of the survey was conducted from November 2021 to February 2022.cWave 2 of the survey was conducted from October 2022 to February 2023.dOtherrace included individuals whose race was reported as "other" by electronic health record.JAMA Network Open | Infectious DiseasesAttitudes Toward COVID-19 Vaccines Among Pregnant and Recently Pregnant Individuals JAMA Network Open.2024;7(4):e245479.doi:10.1001/jamanetworkopen.2024.5479(Reprinted) April 8, 2024 5/15 Downloaded from jamanetwork.comby guest on 04/18/2024 those with an

Table 3 .
Weighted Estimates of Attitudes About COVID-19 and COVID-19 Vaccines Among 652 and 575 Pregnant or Recently Pregnant Persons in Wave 1 and 2, Respectively, in the Vaccine Safety Datalink, Stratified by Vaccination Status a,b Missing values were included in the calculation of the weighted proportions.P values were calculated comparing the probability of responding in the group with the most vaccine favorable attitudinal group between wave 1 and 2. P values were adjusted with age, education level.The absolute and relative difference and P values are calculated based on the difference in weighted proportion of the most vaccine favorable attitudinal group.
b c

Table 4 .
Weighted Estimates for Attitudes Regarding COVID-19 and COVID-19 Vaccines Among 652 and 575 Pregnant or Recently Pregnant Persons From Wave 1 and 2, Respectively, in the Vaccine Safety Datalink, Stratified by 3 Mutually Exclusive Ethnic, Racial, and Linguistic Groups of Interest a,b (continued) Questions about COVID-19 vaccines differed from wave 1 (November 2021 to February 2022) to wave 2 (October 2022 to February 2023); in wave 1, questions were asked about "COVID-19 vaccines" (ie, original monovalent mRNA and viral vector vaccines).In wave 2, questions referred specifically to the "COVID-19 Omicron booster vaccines" (ie, bivalent booster vaccines).Missing values were included in the calculation of the weighted proportions.Missingness accounted for between 0% and a maximum of 4% of the weighted responses.P values were calculated comparing the probability of responding in the group with the most vaccine favorable attitudinal group between wave 1 and 2. P values were adjusted with age, education level.The absolute and relative difference and P values are calculated based on the difference in weighted proportion of the most vaccine favorable attitudinal group.
a b c