Uptake of Recommended Vaccines During Pregnancy Among Publicly and Privately Insured People in the United States, December 2020–September 2022

Abstract

Objectives. To estimate maternal COVID-19, influenza, and pertussis vaccine uptake during pregnancy by insurance type and identify factors characterizing those vaccinated and unvaccinated.

Methods. We conducted a US cohort study of pregnant individuals (for pregnancies ending December 11, 2020–September 30, 2022) using insurance claims data. We calculated vaccination probability using Kaplan-Meier methods and identified factors associated with vaccination through binomial regression with inverse probability weights.

Results. Among 695 887 pregnant individuals (median age = 32 years for privately and 27 years for publicly insured), the cumulative probability of COVID-19 vaccination was 43.0% (privately insured) and 11.8% (publicly insured). We observed marked disparities between influenza (33.2% vs 14.2%) and pertussis (70.3% vs 42.8%) vaccination. Only 6.8% (privately insured) and 1.1% (publicly insured) received all 3 vaccines. COVID-19 and influenza vaccination odds were lower among drug and tobacco users. People with high-risk medical conditions, particularly the publicly insured, commonly were vaccinated.

Conclusions. Marked vaccine uptake disparities exist between privately and publicly insured pregnant people. Understanding structural barriers, particularly for Medicaid enrollees, is critical to improving maternal vaccine access. (Am J Public Health. 2025;115(3):354–363. https://doi.org/10.2105/AJPH.2024.307899)

Vaccine-preventable diseases, including COVID-19,^1,2 influenza,^3,4 and pertussis,⁵ can cause severe illness during pregnancy and infancy. Vaccination against these diseases during pregnancy directly protects the pregnant person and indirectly protects infants via transplacental passage of antibodies. This has led maternal vaccines to be characterized as “two-for-one” strategies.⁶

In the United States, recommendations for influenza vaccines during pregnancy were made in 1960⁷ and updated in 2004.⁸ Pertussis vaccines were recommended during pregnancy—optimally between 27 and 36 weeks of gestation—in 2012.⁹ COVID-19 vaccines were authorized for use in the United States in December 2020 and recommended for those most at risk for severe disease, including health care workers, older people, and those with immune-compromising conditions. At the time, the Advisory Committee on Immunization Practices (ACIP) recommended not withholding COVID-19 vaccines from at-risk pregnant people; however, explicit recommendations for all pregnant people to receive the COVID-19 vaccine were not made until August 2021.¹⁰

Although the safety and effectiveness of maternal vaccination are well established,^11–14 coverage is chronically suboptimal.^15–20 A survey of pregnant adults reported that 65% received at least 1 dose of COVID-19 vaccine before or during pregnancy.¹⁵ Earlier studies estimating COVID-19 vaccine uptake during pregnancy ranged from 16% in the United States¹⁷ to 37% in Canada.¹⁸ For maternal influenza and pertussis vaccination during pregnancy, coverage has remained relatively stable, with one third of pregnant people receiving an influenza vaccine and just over half receiving a pertussis vaccine.¹⁹ Low uptake of recommended vaccines has been observed among younger people, non-Hispanic Black and Hispanic people, and those experiencing higher socioeconomic disadvantage.^15,17 Few studies have explored differences by insurance type, with some reporting reduced uptake of influenza^15,21 and pertussis^15,22 vaccines among publicly insured individuals.

Monitoring maternal vaccine uptake during pregnancy remains essential to identify gaps in coverage by disease, time, and population. This is particularly important as additional vaccines and therapeutics for pregnant people and infants are introduced.²³ We estimated maternal COVID-19, influenza, and pertussis vaccine uptake during pregnancy by insurance type and identified factors that characterized those vaccinated versus unvaccinated.

METHODS

We followed the REporting of studies Conducted using Observational Routinely collected health Data statement.²⁴

Study Design, Setting, and Participants

We conducted a national cohort study of pregnant individuals in the United States using routinely collected administrative health data from insurance claims. Study participants were pregnant people—including women and transgendered people with the ability to become pregnant—aged 18 to 49 years with evidence of at least 1 pregnancy ending between December 11, 2020 (i.e., the COVID-19 vaccine authorization date)²⁵ and September 30, 2022. We use the term “maternal” to indicate exposures during pregnancy or attributes of a pregnant person and respectfully acknowledge that this may encompass pregnant noncisgendered individuals.

Data Sources

We constructed the study cohort using 2 data sources. First, we assembled a claims-based cohort of commercially insured (herein referred to as “privately insured”) pregnant people from the Merative Marketscan Commercial Database (the “Commercial data set”). This database captures nationwide de-identified patient-level data from privately insured employees and their dependents. Second, we assembled a cohort of publicly insured pregnant people from the Merative Marketscan Multistate Medicaid claims database, capturing de-identified patient-level data for more than 47 million Medicaid enrollees. Collectively, these databases capture inpatient and outpatient medical encounters—including pregnancy-related procedures and tests, laboratory work and drug prescriptions, and vaccinations—recorded by physicians, employers, insurance companies, mail order prescriptions, and specialty pharmacies (supplementary material, available as a supplement to the online version of this article at http://www.ajph.org). Data on geographical area of residence were available in the Commercial data set, and data on race/ethnicity (i.e., White, Black, Hispanic, other, missing) were available in the Multistate Medicaid data set.

Data Preparation

We used a previously validated algorithm to identify pregnancy outcomes, pregnancy start date, and gestational age at pregnancy end.²⁶ We excluded pregnant people without prescription benefits or continuous enrollment in their insurance plan (supplementary material). We additionally restricted pertussis vaccine analyses to pregnancies ending at 27 weeks or later of gestation to align with recommendations.⁹ The assembled cohort represented pregnancies, not unique people. We included all pregnancy outcomes, including spontaneous or medical abortions, trophoblastic pregnancies, whether live or stillbirth, and other unknown outcomes of pregnancy.

Outcomes

We assessed uptake of all vaccines routinely recommended to pregnant people. For COVID-19, a person was considered vaccinated if they (1) received at least 1 COVID-19 vaccine during pregnancy, or (2) completed their primary COVID-19 vaccine course before pregnancy. We did not consider individuals vaccinated if only 1 dose of a 2-dose primary vaccine series was received before pregnancy.^15,20 In sensitivity analyses, we explored more sensitive and specific case definitions (supplementary material). For influenza, we considered a person vaccinated if at least 1 influenza vaccine was administered during pregnancy, irrespective of influenza season timing. For pertussis, we considered a person vaccinated if at least 1 pertussis vaccine was administered during pregnancy. We considered a vaccine administered during pregnancy if it was given between the estimated date of the last menstrual period and 1 day before the infant’s date of birth, inclusive.

Descriptive Factors

We explored several sociodemographic and maternal factors that characterized vaccinated versus unvaccinated people. These were insurance type; maternal and gestational age at pregnancy end; US region of residence (privately insured people only); race (publicly insured people only); tobacco, drug, or alcohol use complicating pregnancy, childbirth, or the puerperium; singleton or multiple pregnancy; whether the pregnancy was the outcome of assisted reproductive technology; history of preterm birth; and administration of other recommended maternal vaccines. Other factors included the presence of a high-risk medical condition in the 12 months before pregnancy (supplementary material).

Statistical Analyses

We used statistics (counts; medians; and minimum, maximum, and interquartile ranges) to describe study participants and their characteristics. For each vaccine, we calculated crude vaccine uptake as the number of pregnant people vaccinated as a proportion of all pregnancies ending during the period of interest. We used the Kaplan-Meier survival analysis to calculate the cumulative probability of vaccination for each vaccine over time and by insurance type. For COVID-19 vaccines, we estimated cumulative coverage in weeks since COVID-19 vaccine availability in the United States (December 11, 2020). Additionally, for all vaccines, we estimated the cumulative probability of vaccination by gestational week of vaccination. We censored unvaccinated persons at the week of pregnancy end. We estimated the cumulative probability of vaccination at time (t) as 1 minus the survival probability (S): $1-S\left(t\right)$.

We used binomial regression models with robust (sandwich) variance estimation to quantify variations in vaccine uptake across each descriptive factor of interest to characterize vaccinated versus unvaccinated participants. We used inverse probability weights (IPW) to standardize each estimate to the distribution of all other covariates when quantifying the marginal vaccine uptake per factor. We used logistic regression models to generate the probability (propensity) of each factor of interest. We calculated unstabilized IPWs as the inverse of the probability of having a factor of interest, given all other factors. In total, we considered 22 covariates (described in Descriptive Factors) when building these models and in the calculation of the IPWs.

We assessed covariate balance after weighting using standardized mean differences (SMD), with an SMD less than 0.10 considered adequate. We calculated crude and adjusted odds ratios (AORs) and 95% confidence intervals (CIs) to quantify the multivariable standardized vaccine uptake by factor, accounting for all other factors. We incorporated insurance type into the model as a product term to derive separate estimates for each insurance type and to evaluate heterogeneity between the groups. We performed data preparation and analyses in Stata version 14.0 (StataCorp, College Station, TX).

RESULTS

We identified 695 887 pregnancies ending between December 11, 2020, and September 30, 2022. After applying exclusion criteria, we included 568 071 and 555 656 pregnancies in the analyses of COVID-19 and influenza vaccines, respectively; and we included 477 734 pregnancies in the pertussis vaccine analyses (Figure A, available as a supplement to the online version of this article at http://www.ajph.org). Characteristics of the cohort are shown in Table A (available as a supplement to the online version of this article at http://www.ajph.org). The median maternal age was 32 years (range = 18–49 years) among privately insured people and 27 years (range =  18–49 years) among publicly insured people.

Among pregnant individuals who were publicly insured, just under half were White, one third were Black, and 8.4% were Hispanic. Among privately insured people, the geographical distribution roughly followed the US population, with approximately half of pregnancies identified in the South (45.3%) and one fifth in the West (56 493; 18.1%).²⁷ The median gestational age at pregnancy end was 38 weeks for both insurance types. Proportionally, more publicly insured people had medical records indicating tobacco or drug use that complicated pregnancy, obesity, or a history of preterm birth. Proportionally more privately insured pregnancies were a result of assisted reproductive technology. The presence of specific high-risk medical conditions varied by insurance type (Table A, available as a supplement to the online version of this article at http://www.ajph.org). After we applied the IPWs for our adjusted models (data not shown), the SMDs indicated good balance across covariates.

The uptake of maternal vaccines during pregnancy varied considerably by insurance type. Overall, 21.4% (95% CI = 21.3, 21.5) of privately insured people were vaccinated against COVID-19 during pregnancy compared with 4.9% (95% CI = 4.8, 5.0) of publicly insured people. We observed a similar pattern of disparity for influenza (24.8%; 95% CI = 24.6, 25.0 vs 12.4%; 95% CI = 12.2, 12.5) and pertussis (68.1%; 95% CI = 67.9, 68.3 vs 40.9%; 95% CI = 40.6, 41.1) vaccines. On average, uptake of these vaccines was nearly 56% lower (P < .001) for publicly compared with privately insured pregnant people. Very few people were vaccinated with all 3 recommended maternal vaccines (6.8%; 95% CI = 6.7, 6.9 for privately insured and 1.1%; 95% CI = 1.0, 1.1 for publicly insured; Table B, available as a supplement to the online version of this article at http://www.ajph.org).

Vaccine uptake also varied over time. Maternal COVID-19 vaccine uptake peaked at 36.6% in February 2022 for privately insured people and 10.6% in January 2022 for publicly insured people (data not shown). Using sensitivity analyses assessing COVID-19 vaccine uptake and a more sensitive case definition, we identified a 52.5% vaccination rate among privately insured people by September 2022 compared with only 17.8% among publicly insured people (supplementary material). We observed notable increases in the cumulative probability of COVID-19 vaccination near key pandemic milestones, including when the Centers for Disease Control and Prevention and the American College of Obstetricians and Gynecologists recommended that all pregnant people be vaccinated (August 2021) and following their recommendations for booster doses to be administered to the general population (November 2021).²⁵

These increases were more pronounced among privately insured people. The cumulative probability of COVID-19 vaccination reached 43.0% among privately insured and 11.8% among publicly insured people by September 30, 2022 (Figure 1). We also observed lags in the time of vaccination relative to gestational age by insurance type. Compared with the publicly insured, privately insured people had a higher probability of COVID-19 vaccination at each week of pregnancy (Figure 1).

FIGURE 1—

Cumulative Probability of COVID-19 Vaccination During Pregnancy by Insurance Type and (a) Week Since Vaccine Availability (December 11, 2010) and (b) Gestational Age in Weeks at Vaccine Administration: United States, December 2020–September 2022

Influenza vaccine uptake varied by seasonal vaccine availability, peaking at 39.5% in December 2020 for the privately insured and 22.5% for the publicly insured (Figure 2). Privately insured people also had a higher probability of influenza vaccination throughout each week of pregnancy, with the gap between privately and publicly insured people widening over time relative to gestational age. After 40 weeks gestation, 33.2% of privately insured people were vaccinated against influenza during their pregnancy compared with 14.2% of publicly insured people (Figure 3).

FIGURE 2—

Percentage of People Vaccinated During Pregnancy, by Insurance Type and Month and Year of Pregnancy End, Against (a) Influenza or (b) Pertussis: United States, December 2020–September 2022

FIGURE 3—

Cumulative Probability of Vaccination During Pregnancy, by Gestational Age in Weeks at Administration, of (a) Influenza and (b) Pertussis: United States, December 2020–September 2022

Pertussis vaccine uptake remained relatively stable over the study period, peaking at 70.4% in October 2021 for privately insured people and 45.3% in December 2021 for publicly insured people (Figure 2). Consistent with COVID-19 and influenza vaccination rates, we observed disparities between privately and publicly insured people throughout pregnancy in pertussis vaccination rates: the cumulative probability of vaccination after 40 weeks of gestation was 70.3% among privately insured compared with 42.8% among publicly insured people (Figure 3).

In our adjusted models, we identified several factors more likely to describe vaccinated versus unvaccinated people, with some notable differences between insurance types. For both insurance types, people with a record of tobacco or drug use complicating pregnancy had a reduced odds of COVID-19 and influenza vaccination. This finding was also reflected among privately insured people regarding pertussis vaccination. Similarly, for both insurance types, COVID-19 vaccination was higher among older people. This was also observed among privately insured people with regard to influenza and pertussis vaccination. Among publicly insured people, Hispanic people had higher odds of COVID-19 (AOR = 1.50; 95% CI = 1.38, 1.62) and influenza (AOR = 1.95; 95% CI = 1.85, 2.05) vaccination compared with the referent group (White). Among privately insured pregnant people, compared with those living in the Northeastern United States, people in all other US regions had lower odds of COVID-19 vaccination, most notably in the South (AOR = 0.74; 95% CI = 0.70, 0.78; Figure 4; Figure B, Figure C, available as a supplement to the online version of this article at http://www.ajph.org).

FIGURE 4—

Characteristics of People Vaccinated During Pregnancy Against COVID-19 by Insurance Type: United States, December 2020–September 2022

Note. AOR = adjusted odds ratio; ART = assisted reproductive technology; CI = confidence interval; PTB = preterm birth. All factors listed without a reference group are binary 1 = yes; 0 = no outcomes, and the reference group is 0 = no. For race, the category listed as “other” is commonly used in Medicaid claims data to represent people of Asian descent. Missing/NA represents data from the Commercial data set for which data on race was not available. For region, missing/NA represents data from the Medicaid data set for which region was not available. Unknown represents data from the Commercial data set for which the region was recorded as “unknown.”

We observed variations in the odds of vaccination among people with high-risk medical conditions. Publicly insured people with immunocompromising conditions had higher odds of influenza vaccination than did people without these conditions (AOR = 1.41; 95% CI = 1.22, 1.60); by contrast, privately insured people with these conditions had reduced odds of influenza vaccination (AOR = 0.85; 95% CI = 0.76, 0.95). Differences in the estimates between these 2 groups were statistically significant (P < .001). We observed similar results for other high-risk medical conditions, such as liver disease and obesity (Figure 4; Figures B and C, available as a supplement to the online version of this article at http://www.ajph.org).

We also observed differences in vaccine uptake between insurance types among people with a history of preterm birth. Publicly insured people with a history of preterm birth had slightly higher odds of COVID-19 (AOR = 1.15; 95% CI = 1.07, 1.24) and pertussis (AOR = 1.16; 95% CI = 1.12, 1.20) vaccination; by contrast, privately insured people with a history of preterm birth had reduced odds of being vaccinated against these diseases (COVID-19: AOR = 0.89; 95% CI = 0.84, 0.94; pertussis: AOR = 0.82; 95% CI = 0.78, 0.86). Among both insurance types and across all vaccines examined, individuals vaccinated with any 1 maternal vaccine had higher odds of receiving other maternal vaccines (Figure 4; Figures B and C).

DISCUSSION

In this large cohort study, we found considerable variations in the uptake of COVID-19, influenza, and pertussis vaccines among pregnant people. Very few people received all 3. There were marked disparities in uptake of these vaccines between privately and publicly insured people, with exceptionally low uptake among Medicaid enrollees. This was most apparent for COVID-19 vaccination: uptake reached only 11.8% among publicly insured people compared with 43.0% among privately insured people. We identified several factors that characterized the vaccinated versus unvaccinated people and noted the differences by insurance type.

In the United States—as in other parts of the world—the COVID-19 vaccine rollout was staged, initially focusing on those at highest risk for severe disease. Pregnancy was not a contraindication to vaccination for people who met eligibility criteria. In our study, most pregnant people vaccinated in the first 8 months of rollout were privately insured; this likely reflects the eligibility of pregnant health care workers for early vaccination because of their employment status. However, when COVID-19 vaccines were recommended for all pregnant people, observed coverage gaps between privately and publicly insured people persisted. This was despite national policy enabling the provision of COVID-19 vaccines for all, including Medicaid enrollees with limited benefit packages, who generally did not have access to the full range of ACIP-recommended vaccines.²⁸ Observed disparities in vaccine coverage between privately and publicly insured people were not unique to COVID-19; they were also evident with influenza and pertussis vaccines, both of which have long-established programs supported by robust evidence of safety and effectiveness^11,12 and—particularly for maternal pertussis vaccination—high levels of acceptability among patients and providers.²⁹

Our findings point to a systematic lack of access to vaccines among publicly insured people. Barriers to uptake of ACIP-recommended vaccines among adult Medicaid enrollees include variations or restrictions in vaccine coverage policies by state; limited provider support, access, and availability; and inadequate support and education for beneficiaries.²⁸ Although restrictions were lifted for Medicaid enrollees during the COVID-19 pandemic²⁸—theoretically removing barriers to vaccine availability—our results suggest that other barriers to accessing vaccines may have affected COVID-19 vaccine uptake in this population.

In pandemic and nonpandemic settings, governments and health care providers may need to tailor vaccine delivery approaches to meet the specific needs of Medicaid beneficiaries and other underserved populations. This could involve addressing logistical barriers to accessing care, implementing financial policies or incentives to remove cost barriers, and developing culturally and linguistically sensitive support and communication material.³⁰ The proposed Vaccines for Adults program currently under consideration as part of the federal budget would increase the availability of ACIP-recommended vaccines to all Medicaid beneficiaries and uninsured individuals. Nonetheless, broader access barriers need to be addressed as part of local implementation.

Consistent with other studies,^15,21 we found that maternal influenza vaccine coverage was low among both publicly and privately insured pregnant people. Coverage estimates for pertussis vaccination were moderately high among privately insured pregnant people (reaching 70.3% at > 40 weeks gestation). Although uptake of pertussis vaccine among publicly insured people was lower (42.8% at > 40 weeks gestation), it was high relative to the uptake of COVID-19 and influenza vaccines among this group. These results demonstrate that moderate to high levels of vaccine uptake during pregnancy can be achieved irrespective of insurance type. Variations in uptake by vaccine may reflect differences in attitudes toward the diseases against which they are designed to protect, rather than a hesitancy to be vaccinated during pregnancy.³¹ Further work is required to understand the factors driving the relative success of the maternal pertussis vaccination program in the United States and to characterize structural or attitudinal factors driving vaccine-specific differences. Emerging insights could strengthen existing maternal vaccination programs and support the delivery of new maternal and neonatal therapeutics against other vaccine-preventable diseases, such as respiratory syncytial virus.²³

We also described maternal and sociodemographic characteristics of vaccinated versus unvaccinated participants. Although some characteristics are shared between privately and publicly insured unvaccinated people (e.g., tobacco use, younger maternal age), several other characteristics were more dominant in the publicly insured. For example, vaccination was generally higher among those with high-risk medical conditions, likely reflecting increased contact with health care providers during pregnancy and in line with recommendations for people with preexisting medical conditions to be vaccinated. Reasons for the observed differences between privately and publicly insured individuals are unclear but may be attributable to attitudinal differences toward vaccination among providers and patients in the 2 insurance types. These coverage gaps warrant further investigation, as do our contrasting findings related to vaccine coverage among people with a history of preterm birth. Observed differences in vaccine uptake by insurance type may be explained by increases in vaccine hesitancy in populations of higher socioeconomic position.³²

Regardless of insurance type, provider hesitancy to recommend vaccines during pregnancy has been implicated in low vaccine coverage, particularly for COVID-19^10,32 and influenza.³³ Patient-level vaccine hesitancy also plays a role, with more hesitancy against influenza and pertussis vaccines during the COVID-19 pandemic period.¹⁵ Vaccine hesitancy can be overcome, but it requires a multimodal approach that addresses both provider- and patient-level barriers. Provider recommendations are the most important predictors of vaccination during pregnancy,²⁹ even when patients previously refused vaccine.³⁴ Other provider-level strategies include ensuring adequate vaccine stock in obstetric practices, implementing standing orders, and strengthening provider to patient discussions that focus on safety and benefits to the infant.³⁴

Limitations

Our use of national data that captured comprehensive medical information from a large cohort of publicly and privately insured pregnancies provided a robust platform to assess maternal vaccine uptake, contrasting with single-site studies or those drawing on cross-sectional or panel survey data. Nonetheless, several limitations exist.

First, misclassification of vaccination status is likely in claims data, with 1 study demonstrating that claims data may underestimate COVID-19 vaccination status by up to 15 percentage points.³⁵ Additionally, the timing of vaccine exposures relative to pregnancy may be misclassified because of the algorithm used to estimate gestational age at pregnancy end. Conversely, misclassification of pregnancy start and end dates is likely to be low, with 1 study demonstrating high agreement between claims-based algorithms and physician adjudication of electronic medical records.²⁶ Misclassification may have influenced the magnitude of our estimates (in either direction); however, it is unlikely to explain all disparities observed. Underascertainment of COVID-19 vaccination may have occurred because of the rapid rollout, but our coverage estimates (including results from our sensitivity analyses) are not unlike other studies.^15,17,20 Our influenza and pertussis vaccine coverage estimates also align with previously reported national estimates,^19,22 providing reassurance about external validity.

Second, our results may be affected by selection bias, as we did not exclude pregnancies ending early (e.g., owing to early termination). People whose pregnancies end early have reduced opportunity for vaccination and may systematically differ from people whose pregnancies end at later stages of gestation. Nonetheless, inclusion of early losses reflects the general pregnant population and offsets sampling bias introduced by including only healthy pregnancies ending in live birth or closer to term.

Third, in our analyses describing characteristics of those vaccinated, we were unable to include factors not captured in claims data, limiting our ability to draw inferences about the role of race and region. Nonetheless, Medicaid data are considered a surrogate for socioeconomic status, as they correlate with several socioeconomic measures. Further, we included several factors commonly unreported in other studies, such as maternal medical history and tobacco and drug use.

Fourth, our data sources precluded the assessment of vaccine uptake among uninsured pregnant people. It is likely that coverage in this group is lower than in the Medicaid sample.

Public Health Implications

We examined uptake of 3 vaccines recommended during pregnancy in a pandemic setting and identified considerable disparities in uptake by vaccine and insurance type. Very few people were vaccinated with all 3 recommended vaccines. Structural and attitudinal barriers to the accessibility, availability, and uptake of maternal vaccines—particularly among publicly insured people—should be explored. The complexity of structural barriers facing socioeconomically disadvantaged pregnant people likely requires multimodal solutions, including reform in the way public health interventions are delivered to this population. Enhanced vaccine promotion and delivery efforts will be needed to ensure maternal vaccination equity.