Drug shortages are an “urgent public health crisis”, according to the American Medical Association (2020). The Food and Drug Administration (FDA) reported 41 new shortages of drugs and vaccines in 2021 and 83 shortages that continued from previous years (FDA, 2022). Addressing drug shortages is a priority among policymakers in the United States.3 While a common policy response in the face of shortages is recommending the rationing of supply (Hantel et al., 2019), there is limited empirical evidence on the effects of such policies. Understanding these effects is important, because the welfare effects of a shortage depend not only on the effect on quantities, but also on the allocation of quantities and on patient and provider behavior.

In this article we examine the effects of a vaccine shortage and subsequent rationing on the welfare and behavior of patients. We study this question in the context of an 18-month shortage of the pediatric Haemophilus Influenzae Type B (Hib) vaccine. The Hib vaccine shortage began in December 2007 when manufacturer Merck announced it could not guarantee the sterility of its Hib vaccine and stopped production. At the time, there was only one other approved Hib vaccine maker in the U.S., Sanofi Pasteur, which supplied about half the market. Because of the shortage, the Centers for Disease Control and Prevention (CDC) recommended delaying Hib booster doses in favor of primary series doses (CDC, 2007).

Measuring the welfare effects of a shortage presents empirical challenges, but several features of the Hib vaccine market allow us to overcome these obstacles. First, in other markets it is difficult to estimate counterfactual product demand in the absence of a shortage, due to demand fluctuations, substitution across products, and, in the case of drug markets, changes in market size resulting from variation in disease incidence. The Hib vaccine, however, is recommended for all children. Thus, the market size is straightforward to calculate. Moreover, pediatric immunization rates are remarkably stable over time and other childhood vaccines are recommended to be received on a similar schedule to the Hib vaccine, providing additional information about what uptake of the Hib vaccine might have been in the absence of a shortage. Second, in other markets it is challenging to determine if rationing during a shortage is efficient, because the value of the product to each consumer is unknown. For the Hib vaccine series, however, primary doses have higher marginal benefit than booster doses (Griffiths et al., 2012),4 allowing us to see whether rationing encourages higher-value uses.

A shortage of the Hib vaccine is also important to study in its own right. The Hib vaccine is highly effective and provides protection against Hib bacterial infections, which can cause severe brain damage, nerve damage, and death. Before the Hib vaccine was available, Hib was a leading cause of childhood meningitis and pneumonia in the United States, and approximately 20,000 children had serious Hib disease and about 1000 died annually (CDC, 2022). After the Hib vaccine became widely used, incidence of Hib disease rapidly declined by more than 99 percent and has remained low. Notably, from 2009 to 2018 there were only 36 Hib cases in patients younger than age 5 recorded by CDC surveillance sites (Oliver et al., 2020).

We first estimate the effects of the Hib shortage and rationing on vaccination uptake in the short- and long-run. This allows us to examine how well providers adhered to the CDC recommendation to delay the booster dose and whether children caught up after delayed vaccines. The CDC recommended delaying care to prioritize the higher-value primary series vaccine, assuming that delayed children would catch up. While these types of recommendations are common, there is little observational evidence on whether providers adhere to the recommendations or on the long-term consequences of delayed vaccination.

We next consider how a shortage alters care decisions by patients.5 There are many decisions a patient could make in response to a shortage which have different costs and spillovers. A patient could wait to see a doctor until the shortage has resolved, which might delay other care that would have otherwise been given at the same visit. Alternatively, a patient could receive other recommended preventive care on schedule and arrange a later visit for the Hib vaccine, adding to crowding in the health care system and inconvenience for patients. Finally, a patient could search for a different provider who has Hib vaccine available and is willing to administer it. Seeing an additional provider—who is less familiar with the patient’s history—fragments care. Fragmented care delivery has been shown in other contexts to increase health care costs and reduce quality of care (Agha et al., 2019, Agha et al., 2023).6

Finally, we conduct provider-level analyses to explore supply-side factors that impacted the shortage. For these analyses we explore whether providers who used the Merck Hib vaccine prior to the shortage differently reduced their primary and booster doses relative to those using the Sanofi vaccine prior to the shortage. This helps us to understand supply frictions and whether compliance with the CDC recommendation was uniform. Likewise, we explore whether there were differential impacts among providers operating in counties where mostly Merck doses were used prior to the shortage. This would suggest localized supply issues or differential information dissemination regarding the CDC recommendation.

We conduct these analyses using commercial insurance claims data from the Merative™ MarketScan® Research Databases, 2004–2017. We compare children who were of age to receive Hib vaccine doses during the shortage period to children from earlier or later birth cohorts. The sharp timing of the shortage combined with the recommended vaccination schedule generates clear predictions about which cohorts the Hib vaccine shortage affected. Our identification strategy assumes that in the absence of the shortage, outcomes for the shortage-exposed cohorts would have been similar to outcomes for non-exposed children. We support this assumption by showing that pre- and post-shortage outcomes are stable across cohorts and by showing that other childhood vaccines recommended to be received on a similar schedule as Hib, but which were not in shortage, did not experience the same changes during the shortage period.

We also show that our results are robust to difference-in-differences specifications in which the one vaccine recommended on the identical schedule (pneumococcal vaccine) is used as a control for the Hib vaccine. To the extent that the shortage had negative spillover effects on the uptake of the pneumococcal vaccine, this will bias our difference-in-differences estimates towards zero.

We find evidence of broad adherence to the CDC recommendation to delay Hib booster doses and prioritize primary doses. Among shortage-exposed cohorts, there was only a 4 percentage point reduction in children receiving their primary doses, while there was a 26 percentage point reduction in children receiving their booster dose. Our results also show that following the shortage there was significant catch-up vaccination, although it was imperfect. Years after the shortage ended shortage-exposed children were fully caught up on the primary series, but remained 4 percentage points less likely to have ever received a booster dose.7

Supplemental analyses using nationally representative National Immunization Survey-Child data, 2005–2015, suggest that during the shortage Hib doses were distributed fairly equitably across the population. We find no significant differences across race/ethnicity, household income, or maternal education. We also find that the shortage had similar effects on vaccination uptake for children regardless of whether they were privately insured, supporting the external validity of our MarketScan results.

We next examine how patients altered care decisions in the face of the shortage. We show that shortage-exposed children were about 3 percentage points less likely to be up-to-date at 18 months on the vaccine recommended to be received on the same schedule as Hib (pneumococcal vaccine) and they made 0.3 more provider visits for vaccinations by age 5 than children in surrounding cohorts. These results suggest some patients delayed their preventive care visits during the shortage, while others made additional visits to receive the missed Hib dose. We also find that children in shortage-exposed cohorts were 3 percentage points more likely to switch providers during the Hib vaccine series, consistent with patients searching for new providers in order to obtain the Hib vaccine. Extrapolating these coefficient estimates to the entire population suggests that patients were delayed receiving more than 160,000 pneumococcal vaccine doses, and there were more than 1.5 million extra provider visits and more than 140,000 provider switches.

Finally, our provider-level analyses show that the depth of the shortage varied significantly across providers. We find that providers who mostly used Merck Hib vaccines prior to the shortage reduced administration of primary series doses by about 25 percentage points (relative to the number of pneumococcal vaccines they gave) in the first six months of the shortage, and by 9 percentage points in the shortage’s last year. However, for providers who used mostly the Sanofi vaccine pre-shortage, we find no reduction in the number of primary series doses given during the shortage. This suggests provider-level supply constraints may have been an issue throughout the shortage. For the booster series, we find both types of providers had similar levels of compliance in terms of rationing the booster doses, but areas with more pre-shortage Merck providers reduced their booster doses more quickly at the onset of the shortage.

This article builds on several important literatures. First, we provide novel evidence of the short- and long-run immunization effects of a vaccine shortage. The existing literature on the impacts of drug and vaccine shortages have primarily been surveys of doctors and pharmacists and have focused on short-run effects (Tucker et al., 2020).8 Focusing solely on the short-run may over- (or under-) state the effects of the shortage on vaccine coverage in the population if catch-up vaccination occurs (or supply frictions persist) after the end of a shortage.

Most closely related to our work, Santibanez et al. (2012) and White et al. (2011) examine the short-term effects of the Hib vaccine shortage and find evidence that the primary and booster series quantities fell substantially at the onset of the shortage. While our findings that primary and booster doses fell during the shortage by 4 and 26 percentage points, respectively, are similar to this existing literature, we also show that substantial catch-up vaccination occurred. By the time shortage-exposed cohorts reached age 5, they were as likely to be up-to-date on the primary series, and were only 4 percentage points less likely to be up-to-date on the booster dose, relative to surrounding cohorts. Understanding these long-run impacts of a shortage on vaccination rates is important as they determine population level immunity.

We also contribute to the literature by providing the first evidence on the broader healthcare utilization effects of the Hib vaccine shortage. By showing that shortage-exposed children were 3 percentage points less likely to be up-to-date on the pneumococcal vaccine at 18 months, 3 percentage points more likely to switch providers during the Hib vaccine series, and that they made 0.3 additional vaccinations visits by age 5, we capture costs of the vaccine shortage that have previously been unexamined.9 We are aware of no previous studies of broader healthcare utilization effects of a pediatric vaccine shortage.10

This article additionally expands the literature examining the demand-side of the vaccine market. Our findings on the effects of the CDC rationing recommendations are consistent with evidence from Lawler, 2017, Lawler, 2020 showing that, in non-shortage contexts, CDC vaccination recommendations can be effective at impacting immunization uptake.11 Similarly, our findings complement existing evidence of important spillover effects of other vaccination shocks (Andersson et al., 2021, Carpenter and Lawler, 2019, Schaller et al., 2019).

Our study also contributes to the literature showing how government policies can exacerbate or mitigate drug and vaccine shortages. Although rationing recommendations are a common policy response, existing work has primarily focused on the impacts of reimbursement rates (Ridley et al., 2016, Yurukoglu et al., 2017), or interventions targeting manufacturers (Woodcock and Wosinska, 2013, Lee et al., 2021). Recent evidence from the COVID-19 pandemic shows that governments can be effective at rationing vaccines (Kim and Lee, 2022). More broadly, by demonstrating that providers follow CDC rationing recommendations during pediatric vaccine shortages our work also relates to the literature examining physician adherence to practice recommendations (Alalouf et al., 2019, Buchmueller and Carey, 2018).