Inattention or reluctance? Parental responses to vaccination reminder letters

Abstract

This paper studies non-adherence in the Danish Childhood Vaccination Program using a nationwide introduction of a vaccination reminder letter policy and administrative data from 2011–2017. First, I provide causal estimates of how the reminder letter policy affects vaccination adherence using a Regression Discontinuity Design (RDD). Second, I link parental responses to the reminder letter to parents’ causes for being non-adherent. I find that the reminder letter policy positively affects adherence. However, 72% of non-adherent parents are non-responsive to the reminder letter indicating that reluctance and not inattention is the leading cause for non-adherence. Thus, other policies beyond reminder letters – such as mandatory vaccination laws – are necessary to substantially increase vaccination coverage.

Introduction

Socio-economic differences in child and later-life health are well-documented (Case et al., 2002, Mackenbach et al., 2008, Chetty et al., 2016, Kreiner et al., 2018) but our understanding of their origins remains limited. One suggested explanation is differences in parental investments and health behavior. This paper addresses an important part of preventive health behavior by studying non-adherence in the free of charge and voluntary Danish Childhood Vaccination Program.

Non-adherence in vaccination programs in developed countries poses a major challenge for public health (Black and Rappuoli, 2010, Shetty, 2010) and scepticism concerning vaccinations is becoming more widespread although scientific evidence strongly documents the importance and safety of existing vaccinations (Centers for Disease Control and Prevention, 1999, Offit et al., 2002, Elliman and Bedford, 2003, Stern and Markel, 2005). Thus, knowledge about the differential causes for vaccination non-adherence is of crucial importance in order to increase coverage rates and understand and mitigate inequalities in health behaviors.

This paper studies the effects of a reminder letter policy on subsequent vaccination behavior, sibling spill-overs and participation in other parts of the preventive care system beyond the vaccination program. To assess the driving forces of vaccination non-adherence, I propose a decomposition into shares of responsive, non-responsive and delaying parents based on their reaction to the reminder letter and provide a framework that links parental responses to the reminder letter to causes for non-adherence.¹

I exploit a national reform, which introduced a vaccination reminder letter policy on May 15, 2014, in a sharp Regression Discontinuity Design (RDD) to estimate the causal effects of the policy. Parents with a child registered as lacking at least one scheduled vaccination at age 2 receive a reminder letter. The letter only informs parents that their child has not received all recommended vaccinations and highlights that vaccinations are voluntary. I provide evidence for the validity of the design by showing that a wide range of pre-determined covariates are well-balanced and that there is no bunching across the cut-off. Moreover, the results are robust to bandwidth selection, estimation method and pass various placebo tests.

I find that the reminder letter policy increases adherence by 50% for non-adherent children a year after receiving the reminder letter. The response is immediate as only vaccination behavior in the quarter following the intervention is affected. The timing of the response suggests that the reminder letter mainly affects inattentive parents. The resulting coverage rates show that the reminder letter policy pushes the rate above the minimum herd immunity threshold for measles but falls short for pertussis (whooping cough). Furthermore, I estimate positive effects of the reminder letter on participation in a GP preventive care program but no spill-overs on participation in later vaccination rounds. I detect negative sibling spill-overs suggesting a cost of information as parents postpone vaccinations knowing they will be reminded.

Using my estimates, I decompose the group of parents of non-adherent children at age 2. I find that 8.7% are responsive, while 19.2% and 72.1% are delayers and non-responsive, respectively. Moreover, by evaluating heterogeneity, I discover that responsiveness to the reminder letter is stronger among not first-time parents, parents with an education in a health- or childcare-related field, parents of children with unproblematic births and parents without a university degree. The finding, that parents with an education in a health- or childcare-related field are non-responsive to the reminder letter, indicates that responsive parents were non-adherent due to inattention. Furthermore, as responsive parents have lower education levels, the reminder letter policy reduces socio-economic differences in vaccination adherence. Finally, a simple comparison of the effect on adherence and the cost of the policy shows that the vaccination reminder letter policy costs an estimated 1246 DKK (166 EUR) per additional fully-vaccinated child.

The paper contributes twofold to the understanding of vaccination adherence. First, I extend the literature on the impact of recall systems and reminder letter policies. While research on pro-vaccination campaigns finds modest to no effect (Leader et al., 2009, Chanel et al., 2011, Sadaf et al., 2013, Nyhan et al., 2014, Dubé et al., 2015, Buttenheim et al., 2016, Baskin, 2018),² studies highlight that reminders and recall systems improve coverage in a wide variety of contexts (Vann and Szilagyi, 2005, Harvey et al., 2015). Closely related, Suppli et al., 2017, Suppli et al., 2018 evaluate the effects of the reminder letter policy in Denmark using the same natural experiment in a before-after strategy. They use a before-sample one year prior to the intervention and an after-sample one year post the intervention. They find that during a 6-month follow-up period 2264 more vaccinations are administered to children in the treatment group. I extend their analysis of the impact of the reminder letter policy in four distinct ways: (i) I address causality more rigorously by dealing with time trends and seasonality in a RDD, (ii) I relate the impact on adherence to herd immunity thresholds, (iii) I study other types of preventive care participation beyond the vaccination program and (iv) I test for persistent effects by studying younger siblings of the treated children.

Second, I provide evidence on the factors influencing vaccination non-adherence. Existing evidence has identified factors such as parents’ perception of infections and side-effects risks, religion, ideology and social pressure (Tickner et al., 2006, Grabenstein, 2013, Wombwell et al., 2015, Larson et al., 2016, Amin et al., 2017, Karing, 2018). Recently, a growing literature highlights misinformation as a major factor causing parents to opt out of vaccination programs (Anderberg et al., 2011, Chang, 2018, Hansen and Schmidtblaicher, 2019, Carrieri et al., 2019, Gørtz et al., 2020, Qian et al., 2020).³ The economic literature furthermore shows the difficulties in reaching full coverage due to free-riding (Philipson, 1996, Geoffard and Philipson, 1997, Oster, 2018): When a substantial share of the population is covered, the immunized individuals not only protect themselves but also lend protection to individuals without vaccinations. Consequently, the private benefit of vaccinations might be low in a society with relatively high coverage rates. My results indicate that reluctance is a leading cause for non-adherence in this setting and that other policies (beyond reminder letters) are necessary to substantially increase vaccination coverage (preferably after identifying the reasons for reluctance).

The paper proceed as follows: In Section 2, I provide background on The Danish Childhood Vaccination Program and describe the introduction and purpose of the policy. Section 3 covers my empirical strategy. Section 4 describes data and descriptive statistics. Section 5 presents results and section 6 performs a simple cost-effectiveness analysis. Finally, Section 7 concludes.