Cost-Effectiveness of Maternal Vaccination to Prevent Respiratory Syncytial Virus Illness

To evaluate the projected impact of year-round and seasonal maternal vaccination with RSVpreF, we developed a decision analytical model of RSV disease simulating the short- and long-term impacts of RSV infection on infants in the following 2 scenarios: no vaccination and seasonal maternal vaccination over a 1-year time frame. For the base case scenario, we assumed a 50% uptake of RSVpreF in the US birth cohort administered between the beginning of week 32 and the end of week 36 of pregnancy (ie, 32 weeks and 0 days gestation to 36 weeks and 6 days gestation). We did not include the prevention of transmission, so the incremental cost-effectiveness ratios (ICERs) were invariant to uptake. RSV disease and economic outcomes were estimated by using the societal perspective, but we also explored a health system perspective.

With a uniform distribution of US annual births, the model simulates the number of infants undergoing their first RSV season using monthly time steps. Infants face a seasonal risk of RSV infection, and the risk of medically attended RSV LRTI and RSV hospitalization differs on the basis of the specific month of birth and age of the child (Supplemental Fig 4 and Table 1). We included RSV-associated disease outcomes, along with resource utilization, such as outpatient visits, emergency department (ED) visits, hospitalizations, and premature deaths among infants (Fig 1). Likewise, we included potential vaccine-associated adverse events, such as injection site reactions, systemic reactions, and serious adverse events to the vaccinated pregnant individual along with potentially increased risk rates of premature delivery for the infant.¹¹  Each of the vaccine-associated adverse events, as well as the RSV disease outcomes, had associated costs and health-related quality of life losses.

Model inputs for the annual incidence of RSV were based on several recent epidemiological studies of inpatient, ED, and outpatient disease burden associated with RSV (Table 1).^2–4,12  Because these sources did not distinguish between upper and LRTIs, we relied on previous estimates that 100% of RSV-associated hospitalizations are due to LRTI, and 50% to 65% of ED and 30% to 65% of outpatient visits associated with RSV are due to LRTI (Table 1).¹³  For our model, we assumed that RSV-associated deaths happen only among those infants hospitalized, and estimates are based on RSV mortality rates for children aged 0 to 23 months hospitalized with RSV.^14–16  We did not include the long-term sequelae associated with RSV infection (eg, asthma). Seasonality by month came from prepandemic data (2015–2019) on the proportion of RSV cases per month across the whole United States from the National Respiratory and Enteric Virus Surveillance System (Supplemental Fig 4).¹⁷  Although the RSV season can differ by geographic region, national averages were used in this analysis. We did not distinguish between RSV subtypes.

RSVpreF efficacy data were from the reported phase 3 clinical trial (Table 1).¹¹  We assumed that efficacy follows a sigmoid decay from months 0 to 5.9 after birth and is equal to 0% at month 6 (Supplemental Fig 5A). The assumptions of reduced maternal vaccine efficacy over time and no efficacy after 6 months were based on the trial showing no significant efficacy after 6 months and efficacy for infants aged 0 to 180 days compared with infants aged 0 to 90 days; additionally, waning effectiveness data of other vaccines administered during pregnancy has been shown after 3 months of age.^11,18,19  For the base case scenario, the average efficacy in the first 180 days after birth matched the reported efficacy in the trial. We also explored other functional forms for waning efficacy for alternative scenarios. One was a constant rate for efficacy during the first 180 days (Supplemental Fig 5B). The authors of another analysis used more optimistic assumptions that the efficacy against RSV-associated hospitalization was equivalent to the reported efficacy against severe medically attended RSV LRTI for full-term infants, and efficacy declined moderately during the first 6 months and then declined to zero efficacy at 10 months (Supplemental Fig 5C).²⁰  The Supplemental Information includes more detail on the precise definitions of efficacy scenarios and how they comport with trial data.

Maternal vaccine adverse event rates for injection site and systemic reactions came from the phase 3 clinical trial reports¹¹  (Table 1). The trial did not observe differential rates of systemic reactions in the vaccine and placebo groups, but we included a rate of 1 in 1 million of a hypothetical serious adverse event. We also included a 41% chance of injection site reaction. The Supplemental Information includes the logic behind the serious adverse event and preterm birth assumptions.

In our model of the RSVpreF policy, we assumed that 50% of pregnant persons intend to vaccinate between the beginning of week 32 and the end of week 36 (the weeks recommended for vaccination).²¹  Although the fraction of pregnant persons vaccinated affects the overall population-wide outcomes, it does not affect the ICERs. We assumed births occurring within 2 weeks of vaccination received no protection against RSV illness (Supplemental Fig 6).²²  For those born premature, if they were born >2 weeks after vaccination, we assumed full protection.

Because the trial results did not reveal efficacy against upper respiratory tract infection (URTI), we assumed that RSVpreF vaccination had zero efficacy against RSV URTI.

The cost of RSVpreF was set to $295 on the basis of its private sector cost per dose.⁸  We assumed it would be provided at a routine obstetrics visit, so we did not include direct and indirect costs associated with an extra health care visit, but we did include the cost to administer the vaccine (Table 1). We included costs of palivizumab for high-risk infants (Table 1). We assumed RSVpreF immunization did not affect palivizumab utilization. Medical costs of adverse events were derived from a MarketScan analysis on adverse events associated with influenza vaccination.²³  We also included productivity losses from RSV disease and adverse events (Table 1).²⁴  RSV health care utilization costs were from a systematic literature review of RSV disease costs in infants (Table 1).¹  We also included caregiver productivity losses from infant infections, and infant productivity losses from RSV-related deaths (Table 1).^25–27  Because we take a societal perspective, we also included overall general health care costs including those not related to RSV (Supplemental Table 5).

We included both quality-adjusted life-years (QALYs) lost by the child and QALYs lost by caregivers during the child’s illness. A systematic review of the literature on the impact of RSV on child and caregiver quality of life²⁸  was used for our base case QALYs lost from RSV-related hospitalizations, ED visits and outpatient visits. We based the lower bound on a study by Regnier²⁹  and the upper bound was based on an unpublished preference survey of parents of children who had experienced RSV illness (Table 1). We also included QALYs lost from deaths on the basis of lost quality-adjusted life expectancy (Table 1).

The model simulated RSV LRTI-associated outpatient visits, ED visits, hospitalizations, and deaths to calculate total medical and indirect costs, QALYs lost and an ICER in terms of dollars per QALYs gained when comparing RSVpreF vaccination with no vaccination. We also reported number needed to vaccinate to prevent an undesirable RSV disease outcome (eg, number of pregnant persons needed to vaccinate to prevent an infant RSV-associated hospitalization or death) as well as the cost per RSV disease outcome averted (eg, cost per hospitalization or death averted). Costs are reported in 2022 US dollars, for a 1-year policy time frame, and costs and outcomes are discounted at 3%. We used Microsoft Excel 365 to calculate the results.

We conducted a variety of 1-way, two-way, and probabilistic sensitivity analyses (Supplemental Table 6). We conducted several scenario analyses with varying efficacy assumptions. We also evaluated a combination of assumptions on efficacy, hospitalization costs, and mortality risks. Because RSV is seasonal, we also examined several scenarios assuming RSVpreF would be administered at different times of the year. Probabilistic sensitivity analysis assigned distributions to all input parameters (see Table 1) and conducted 1000 Monte Carlo simulation iterations to calculate 95% credible intervals and to calculate uncertainty in overall results using cost-effectiveness acceptability curves.

Finally, we explored an additional scenario in which we assume the infant will also be receiving nirsevimab (regardless of infant risk; Supplemental Table 7).¹⁰  This is comparing a combined strategy of RSVpreF administration and nirsevimab with a strategy of nirsevimab administration alone (the base case analysis involves the assumption that nirsevimab is not administered). No studies are available that have directly compared antibody levels or effectiveness among those infants who received nirsevimab with infants born to people who received an RSV vaccine during pregnancy. Because we lack clinical evidence of combined RSVpreF and nirsevimab efficacy, we assumed the efficacy of combined RSVpreF and nirsevimab protection to be the higher efficacy of either RSVpreF or nirsevimab at a particular age of the infant at a particular time (Table 1 and Supplemental Fig 7).

Our model projects that vaccinating approximately half of pregnant persons with the RSVpreF vaccine during weeks 32 through 36 of gestation (Supplemental Fig 6) would avert 7571 (16%) hospitalizations each year (Table 2) while causing 725 501 pregnant persons to experience injection site reactions and 1.8 serious adverse events. To prevent 1 RSV LRTI hospitalization among infants would require 234 pregnant persons to be vaccinated during pregnancy with RSVpreF.

Although an additional $666 million would be spent on RSVpreF and its administration, $100 million in medical costs would be saved and $53 million in productivity costs would be saved, leading to an annual societal net investment of $513 million (Table 3).

The model projects a net gain of 1294 QALYs, 943 from the children and 351 from their caregivers because of lower RSV burden (Table 4). The societal cost of RSVpreF vaccination for pregnant persons is $396 280 per QALY gained and $67 735 per RSV LRTI-associated hospitalization averted (Table 2). From a health system perspective, the cost is $437 607 per QALY gained (Supplemental Table 8).

We conducted many sensitivity and scenario analyses (Supplemental Table 6). The results are sensitive to a wide number of inputs and assumptions (Fig 2). Specifically, for the first 1 percentage point increase in the risk of prematurity from the vaccine (ie, from 0% to 1%), the societal ICER increased almost 4 times (up to >$1.4 million per QALY saved) mostly because of the dramatic losses in QALYs (Supplemental Fig 8). If quality of life losses from RSV were measured by using the highest QALY loss values for children and their caregivers for all types of RSV LRTI events, then the ICER dropped to $95 313 per QALY gained. However, when RSV QALY losses were measured by using their lowest values for children and when caregivers did not experience any QALY losses, the societal cost increased up to $794 009 per QALY gained. Other influential factors in order of the magnitude of their effect on cost-effectiveness (Fig 2) were the RSVpreF vaccine cost per dose (Fig 3), inpatient medical costs, the potential effect of RSV on URTI, and RSV mortality.

The alternative assumption of constant 6-month vaccine efficacy (Supplemental Fig 5B) did not have a considerable impact on the ICER, as it reduced the societal cost to $361 679 per QALY gained, ∼8.8% reduction (Supplemental Table 9). However, the alternative assumption of more optimistic efficacy (Supplemental Fig 5C) did have a bigger impact on the ICER, dropping the cost to $282 287 per QALY gained, ∼28% reduction (Supplemental Table 9).

Supplemental Figure 9 shows a scenario analysis varying vaccine efficacy, hospitalization cost, and RSV mortality. If RSV hospitalizations are more expensive or associated with higher mortality, maternal RSVpreF vaccination becomes more cost-effective. In special circumstances, combining high vaccine efficacy with high hospitalized mortality (1%), and high cost per hospitalization ($45 000), RSVpreF vaccination could potentially be cost-saving (ie, cost per QALY saved <0).

In the probabilistic sensitivity analysis, we found 3% of simulations had a societal cost <$100 000 per QALY saved, 8% of simulations had a societal cost <$200 000 per QALY saved, and 53% had a cost <$500 000 per QALY saved (Supplemental Fig 10).

We also evaluated scenarios related to the timing of RSVpreF administration (Supplemental Fig 11). The ICER of RSVpreF varied widely on the basis of the month of administration (Supplemental Fig 11A). Specifically, the ratios of cost per QALY saved when RSVpreF is administered to pregnant persons right before or at the start of the RSV season were much lower. With the complex interaction of seasonality, decreasing hospitalization risk with age, and decreasing vaccine efficacy after birth, the most cost-effective month to administer RSVpreF is anticipated to be in November, right at the start of the expected RSV season, with a societal cost of $107 544 per QALY gained (Supplemental Fig 11A). Administration of RSVpreF during the range of months from September through January would cost $163 513 per QALY gained (Supplemental Fig 11B).

Lastly, we simulated a scenario with RSVpreF vaccination added in an environment in which nirsevimab was expected to be administered (ie, comparing RSVpreF vaccination and nirsevimab administration to the infant with only nirsevimab administration). In this scenario, the ICER increases dramatically, because RSVpreF offers marginal additional protection beyond what might be expected from nirsevimab alone. In that case, the best month to vaccinate pregnant persons would be in April (because this model involves the assumption that nirsevimab will be administered in October for children born in April, and RSVpreF would result in protection during the tail end of the RSV season during April); however, the ICER was still high at $2.4 million per QALY saved (Supplemental Fig 12). The ICERs of RSVpreF administration during other months are even higher in this scenario.

Our model reveals that year-round maternal vaccination of half of the pregnant persons with RSVpreF during 32 through 36 weeks’ gestation will decrease RSV LRTI medical events (ie, outpatient and ED patient visits, hospitalizations, and deaths) but will also increase societal costs, resulting in $396 280 per QALY saved. RSVpreF has the potential to be cost-effective under certain conditions and in specific situations. The magnitude of QALYs lost because of RSV illness can raise or reduce the ICER. The vaccine cost per dose and mortality associated with RSV illness also can impact the ICER. In addition, the month in which RSVpreF is administered has a significant impact on the ICER. The ACIP recommendation was limited to seasonal administration (ie, September to January for most of the United States), in part to maximize cost-effectiveness and benefits.⁹  Administration in September through January had an ICER of $163 513 per QALY saved. Finally, if vaccination increases the risk of prematurity, the societal costs per QALY saved with RSVpreF vaccination will substantially increase.

To our knowledge, there are no published cost-effectiveness studies on maternal vaccination to prevent RSV disease in US infants. There has been, however, another unpublished manufacture-sponsored cost-effectiveness analysis that was also discussed and summarized for ACIP consideration and decision-making.²⁰  Although there were considerable similarities in the modeling approach and sources of inputs, the authors of this other analysis reported substantially lower ICERs (ie, ∼$85 000 per QALY saved). The main reason for the difference in ICER with our model is the selection of inputs and the adoption of critical assumptions. Among the most influential assumptions, the manufacturer’s analysis adopts (1) a relatively higher initial efficacy, (2) a longer assumption for the duration of protection, with a linear decay in efficacy finally reaching 0% at 9 months, and (3) higher medical costs for hospitalization and ED and outpatient visits.

In our base case analysis, we focused on comparing RSVpreF with no vaccination. This baseline scenario does not consider the use of nirsevimab. In our analysis of nirsevimab compared with no immunization, we found a lower ICER of $153 517 per QALY saved, suggesting that it would be more cost-effective compared with RSVpreF alone.¹⁰  Caution is urged in directly comparing the results of these models because the model inputs are based on efficacy trials with different definitions of outcomes, and the duration of the protection of both products remains unknown. In an environment in which nirsevimab is expected to be used, adding RSVpreF vaccination in addition to nirsevimab administration dramatically increases costs, with only marginal improvements in health. However, this analysis relies on speculative assumptions because there are no trials that have measured the combined efficacy of administering nirsevimab to an infant born to a person who had been vaccinated with RSVpreF. The Centers for Disease Control and Prevention recommends the use of either RSVpreF in pregnant people or nirsevimab in infants, but administration of both products is not needed for most infants.⁹ 

Our analysis is subject to several limitations. As a model-based analysis, the results are subject to model structure and input choices. No data on RSV outcomes in pregnant persons from the clinical trials data are available, and the model involves the assumption that the RSV vaccine does not prevent the transmission of RSV. Because the RSVpreF vaccine reduces RSV disease in individuals other than infants, the cost-effectiveness of RSVpreF vaccination would improve. However, those of childbearing age are generally at low risk of RSV, and pregnancy is not known to be a risk factor for severe outcomes with RSV.^30,31  The model also does not include specific explicit clinical risk groups that stratify infants at increased risk for severe RSV illness because the vaccine was designed to be administered to a general population. With that said, our model inputs are designed to represent the general population of infants. Our analysis also does not include any impact of RSVpreF vaccination on disease transmission dynamics. Although a 2016 model-based analysis of a theoretical RSV vaccine reveals that the protection of children may reduce RSV infections in other unvaccinated populations,³²  there is no evidence the RSVpreF vaccine reduces transmission. If maternal RSVpreF vaccination were to reduce population RSV transmission, then our ICER would be expected to decrease. Additionally, infant RSV antibody levels could be reduced if the mother had a poor immune response to the vaccine (eg, immunocompromising conditions) or had a condition associated with reduced transplacental antibody transfer (eg, HIV).³³  We were also uncertain of a variety of inputs, particularly related to vaccine effectiveness, QALYs lost from RSV, and prematurity. These and other assumptions can dramatically change the cost-effectiveness of RSVpreF. Additional research on these assumptions may be valuable.

A seasonal maternal vaccination program with RSVpreF aimed at preventing RSV LRTI in infants is likely to decrease the disease burden of RSV LRTI but at a sizable societal cost. RSVpreF has the potential to be cost-effective in specific circumstances, particularly when administered at the ideal gestational and seasonal time.

ACIP

Advisory Committee on Immunization Practices

ED

emergency department

ICER

incremental cost-effectiveness ratio

LRTI

lower respiratory tract infection

QALY

quality-adjusted life-year

RSV

respiratory syncytial virus

RSVpreF

respiratory syncytial virus bivalent prefusion F maternal vaccine

URTI

upper respiratory tract infections