High and timely coverage rates in publicly-provided and funded vaccination programs are central policy goals. Given coverage rates below the recommended targets for central childhood vaccinations and recurrent outbreaks of preventable diseases in high-income countries with large-scale public vaccination programs (Shetty, 2010, Galles et al., 2021, Plans-Rubió, 2021), an active debate centers around the design of polices to increase (timely) adherence to childhood vaccinations in these settings.

In this paper, we zoom in on one widely used policy, namely reminder systems. Reminders raise the attention of individuals towards a specific action, such as getting a vaccination (Gravert, 2021). Simple reminders only shift focus towards the action without changing the receivers’ attitudes, while more sophisticated reminders can include messages that potentially alter the receivers’ attitudes and beliefs. Both dimensions may be important in the case of childhood vaccination programs, where parents may be inattentive to the recommended vaccination schedule, reluctant to comply with specific vaccinations, or both.1

Existing research from behavioral economics has documented that reminder systems are effective tools for raising vaccination coverage (Szilagyi et al., 2000, Bronchetti et al., 2015, Busso et al., 2015, Hirani, 2021, Milkman et al., 2021).2 The majority of this evidence on the effectiveness of vaccination reminders comes from field experiments that study the extensive margin (receiving a reminder or not) and factors in questions of specific reminder timing and content (e.g., distributing pure reminders or reminders that include different types of messages) (Szilagyi et al., 2000). Given its experimental nature, this research typically focuses on specific samples of patients (often drawn from individual health care providers) and specific vaccination episodes.

This paper complements the empirical literature on reminders based on field experiments by studying a series of consecutive large-scale reminder policy changes in the national Danish childhood vaccination program. This public program covers all resident children between the ages of three months and 12 years, and it offers an encompassing series of vaccinations across eight vaccination rounds. To identify the causal impact of different aspects of reminder design across children of different ages and across vaccination episodes, we exploit variation in reminder design caused by three different reminder policy reforms in regression discontinuity designs (RDD). We use administrative data on the full population of families in Denmark, which in combination with the population-wide reforms allow us to zoom in on three empirically relevant dimensions of vaccination reminder design: the timing and format of reminders, differences in reminder effectiveness across different vaccines (e.g., boosters vs. non-boosters and vaccinations in earlier vs. later childhood), and spillovers to other health behaviors.

Specifically, we exploit changes in the Danish vaccination reminder system in 2014, 2017 and 2019, which differentially impacted families around cut-off dates exclusively defined by children’s dates of birth. The 2014 introduction of the national reminder system established a retrospective postal reminder only for non-compliant families, i.e., families with children lacking at least one of the vaccinations offered in the public program at three specific follow-up ages (between nine and 24 months after the recommended vaccination round). In 2017, digital retrospective reminders replaced the postal reminders. In 2019, reminders changed to their current form of being prospective, i.e., reminding all parents about each upcoming vaccination round. Additionally, non-vaccinated children in the 2019 system receive a follow-up reminder four weeks after the recommended vaccination age and the new prospective reminder includes more explicit messages on the positive externalities and social desirability of vaccinations.3

Exploiting the variation in reminder design locally in a set of RDD analyses, we generate five main findings on the importance of timing and format of reminders, their impact across well-established and new vaccinations, and finally spillover effects to other health behaviors: First, the 2014 introduction of retrospective vaccination reminder letters primarily impacted the coverage for the infant and toddler vaccinations recommended in the first two years of life. We show that the reminder was most effective for the closest-spaced vaccine (recommended at age 15 months, i.e., nine months prior to reminder receipt), which we interpret as a proximity effect rather than vaccine-specific impact of the 2-year reminder letter. For the vaccination coverage among older children (age 6.5 and 14 at retrospective reminder receipt), we document that the 2014 reminder was only moderately effective or ineffective. Second, switching from a retrospective postal reminder to a retrospective digital reminder letter did not impact the coverage rate of childhood vaccinations in Denmark. This change, however, dramatically decreased the yearly costs of the system.

Third, the 2019 reform illustrates that prospective reminders have the potential to boost timely vaccinations (with consequences also in the longer run): We find short-term impacts of changing the retrospective reminder to a prospective reminder on the timing of adherence for the early vaccinations (and catchup of the control group for these vaccines with high baseline coverage and with high salience due to frequent interactions with the health care system). However, we find that the introduction of prospective reminders increased not only the timely but also longer-run uptake of later childhood vaccinations. Specifically, prospective reminders increased the uptake of the age four measles, mumps, rubella (MMR) booster by 4.4 percentage points and the age 12 human papilloma virus (HPV) vaccination coverage rate for girls by 4.1 percentage points two years and one month after the recommended vaccination age (relative to the control group, who around that time have received their retrospective reminder). Focusing on timing, we find that one year after the recommended vaccination age (i.e., at a time without any reminder for the control children) the MMR booster coverage is 5.2 percentage points higher and the HPV coverage rate is 18.8 percentage points higher under the 2019 policy.4 Thus, reminding all parents in advance of childhood vaccinations increases timely coverage with lasting consequences (even in the presence of later, retrospective reminders). This finding may indicate that later childhood vaccinations are less salient than the early infant and toddler vaccinations, potentially due to less frequent interactions with the primary health care sector for children in these age groups. They also indicate that having parents react timely is key for achieving high coverage in the long run.5

Fourth, by studying the impact of the prospective reminder policy for a new vaccination, the human papilloma virus (HPV) vaccine for boys, we provide an important case study for the scope of reminder effects for new, less salient vaccines in increasingly complex vaccination systems. In Denmark, the HPV vaccine was well established for girls by 2019 (though at times subject to public controversies) but only introduced in the public program for boys in that year. We find strong gender heterogeneity in the impact of the 2019 reminder reform on HPV vaccination coverage. With the prospective reminder letter, the large gender gap (close to 20 percentage points) in HPV vaccination uptake among all eligible girls and all eligible boys was virtually closed. This finding suggests that prospective reminders (providing timely information to families) are especially effective for vaccines that are not yet well-established in public vaccination programs.6

Fifth, we ask whether vaccination reminders promote other family health behaviors. Focusing on the introduction of prospective reminders in 2019, we find an eight percentage point increase in the share of children receiving the recommended routine preventive health check at the general practitioner (GP) at age four (scheduled at the same age as the MMR booster vaccine). Our findings suggest that vaccination reminders make this type of GP care more salient to parents (most likely through interactions with the GP office when scheduling vaccinations). In terms of other spillovers, we study older siblings’ HPV vaccination uptake and mothers’ uptake of cervical cancer screening (a behavior that may be directly impacted by reminders for children’s HPV vaccination, as HPV infections are directly linked to this type of cancer). We do not find any spillovers of prospective vaccination reminders.

Our paper contributes to three streams of the literature: First, a large literature studies reminders systems in a range of settings, including health care, charity donations or financial decision making. Recent examples within the area of health behaviors include studies on the impact of reminders related to COVID-19 to foster vaccination uptake and social distancing (Dai et al., 2021, Li et al., 2021, Falco and Zaccagni, 2021, Cappelen et al., 2021, Chang et al., 2021). In line with the findings in this existing literature, we find that reminders can stimulate desired health behaviors at modest costs. Even at high baseline coverage rates for childhood vaccinations in the publicly funded Danish program, our study indicates that there is scope for relatively light interventions to foster positive health investments.

Second, we contribute to a literature that focuses on the determinants of vaccine uptake. Several important channels have been identified in this research, among them infection risks (Philipson, 1996, Bauch and Earn, 2004, Quadri-Sheriff et al., 2012, Böhm et al., 2016, Oster, 2018, Schaller et al., 2019), peer effects (Karing, 2018, Sato and Takasaki, 2019), religious beliefs (Lahav et al., 2021), societal recommendations and sanctions (i.e. mandatory vaccinations for daycare entry) (Carpenter and Lawler, 2019, Lawler, 2020), the perceived risk and severity of side-effects (Hansen and Schmidtblaicher, 2019, Carrieri et al., 2019, Gørtz et al., 2020), timely interaction and advice from health professionals (Hirani and Wüst, 2022), financial incentives (Banerjee et al., 2010, Campos-Mercade et al., 2021) and, finally, reminder systems (Szilagyi et al., 2000).

Studying national reminder reforms in population data from Denmark, we complement earlier work on vaccine reminders that has predominantly focused on samples drawn from specific providers. Work on childhood vaccination reminders has primarily been based on RCT studies comparing prospective reminders to no reminders as control condition (Dini et al., 2000, Ahlers-Schmidt et al., 2012, Bangure et al., 2015, Busso et al., 2015, Hofstetter et al., 2015, Niederhauser et al., 2015, Domek et al., 2016, Haji et al., 2016, Kazi et al., 2018).7 We complement these small-scale RCT studies from primarily US and low-income countries by studying a national-scale implementation of reminder reforms, by exploiting population data, and by considering consecutive changes to reminder design. This last feature allows us to directly zoom in on the importance of reminder timing and frequency. Finally, while studies usually fix the mode of reminder delivery (with a focus on text message reminders), we also investigate the importance of the mode of delivery, comparing postal and digital reminders.

While our data and setting allow for a study of reminder impacts in a general and non-experimental setting, our estimates based on local RD designs are arguably specific to the respective Danish reform setting. As detailed below, Denmark (similar to many other European welfare states) provides an encompassing system of childhood preventive care and vaccinations for free. Interestingly, our main point estimates are well in line with findings from experimental work on childhood vaccination reminders from very different settings: Szilagyi et al. (2000) review reminder policies in the context of childhood vaccinations across five randomized controlled trials. They report a median effect of 12.3 percent on vaccination coverage rates. This estimate is comparable to our effects of introducing prospective reminder letters on MMR coverage (7.2 percent) and female HPV coverage (12.9 percent) one year after the recommended vaccination age.

A third and final contribution of our work is an examination of potential spillover effects of vaccination reminders. Reminders raise attention towards vaccinations and may also impact other health behaviors or family members. Evidence about reminder letter spillovers is instrumental for policy design in the light of a recent literature on the importance of spillovers in health behaviors within families exposed to other interventions or shocks (Al-Janabi et al., 2016, Daysal et al., 2019, Fadlon and Nielsen, 2019, Einav et al., 2020, Hodor, 2021). We find limited positive spillovers of reminders, and these are concentrated on preventive care at the GP for focal children that can typically be scheduled together with vaccines.