Nudging COVID-19 Vaccine Uptake by Changing the Default: A Randomized Controlled Trial

Introduction

Although vaccination is acknowledged as the most promising way to reduce the risk of severe COVID-19 infection,^1,2 areas of higher vaccine hesitancy threaten global efforts to bring the pandemic under control.^3,4 Among the tools used to promote public health, nudging interventions have gained popularity in various domains, including dietary change, ⁵ exercise, ⁶ organ donation, ⁷ and vaccine uptake (for specific references in this area, see the “Discussion” section). These interventions function by structuring the environmental context in which a decision is made (i.e., the “choice architecture”) such that it “alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives.” ⁷

In this study, we investigated whether COVID-19 vaccine uptake could be increased by exploiting the status quo bias, ⁸ that is, the tendency to maintain the current state of affairs, even when it has not been actively chosen and change requires only minimal action. To this end, we compared the COVID-19 vaccination rate following two default options: one in which a vaccination appointment has been scheduled (with opting out possible) and another in which vaccination appointments are available (with opting in required to vaccinate). The latter is the standard in many, if not all, countries, whereas the former has not been reported as used in COVID-19 vaccinations.

Method

Our study involved 2000 adults, ages 50 to 59 years (see Figure 1), residing in the Autonomous Province of Trento (Italy), where vaccinations are managed by the Provincial Authority for Health Services (APSS). We focused on this population because its constituents were found to be particularly reluctant to vaccinate in the preceding months (see Figure 2), even though they were at relatively high risk of severe disease (0.6% death rate for COVID-19 ⁹ ). The sample size was determined by the number of appointments that could be prearranged by the APSS without attenuating regular vaccination activity. Exclusion criteria comprised professionals subject to mandatory COVID-19 vaccination requirements (e.g., physicians or nurses), residents of extended-care facilities (where vaccination is arranged internally), individuals who had already received or scheduled an appointment for the COVID-19 vaccination, and those for whom vaccination was not recommended due to prior COVID-19 infection.

OPEN IN VIEWER

Figure 1

Flow chart of study participants (and their characteristics).

OPEN IN VIEWER

Figure 2

COVID-19 vaccination coverage for individuals ages 50 to 59 years in the province of Trento (the target population from which the study sample was drawn) alongside coverage rates for older adults in the same geographical area, as well as the rates for the peer cohort across the entirety of Italy, at the beginning of data collection (July 26, 2021).

To maintain traceability and confidentiality, eligible participants were allocated a unique ID code. They were then assigned to one of two groups by gender- and age-stratified random sampling from the APSS database by an independent third party, who had no further involvement in the study, using the R package randomizeR. ¹⁰ Participants were unaware of the intervention (i.e., they were not informed of the existence of the two group assignments). The study team did not have access to the allocation list, and analyses were conducted without unblinding the assignments.

On July 23, all participants were sent, by post, a letter from the prevention department of the APSS reminding them of the importance and availability of COVID-19 vaccination. The letter was identical for the two groups, apart from appointment-scheduling information. For the opt-in group, this section constituted instructions on how to schedule an appointment for the vaccination (via web or by contacting their family doctor). For the opt-out group, it comprised date, time, and location information for a scheduled vaccination appointment, together with instructions on how to modify the arrangement, if desired. All participants were offered the BNT162b2 Pfizer/BioNTech vaccine.

Coders who were blind to group assignment abstracted vaccination records out of the APSS database for a 19-day interval between July 26 (i.e., 3 days after the letters were sent, which is a conservative estimation for local letter delivery and subsequent appointment scheduling) and August 13 (i.e., the last day on which appointments for the opt-out group were scheduled, a Friday before a public holiday weekend).

Results

Forty-three participants were excluded because it turned out they had received the COVID-19 vaccination outside the province shortly before the beginning of the study. The final sample, therefore, comprised 1957 individuals (47% female; M_age = 54 years, SD = 2.9 years): 978 in the opt-in group and 979 in the opt-out group (see Figure 1).

Results showed that 128 of 979 participants in the opt-out group (13.1%; 95% confidence interval [CI], 11%–15.4%) received the vaccine during the considered interval, compared with 97 of 978 participants in the opt-in group (9.9%; 95% CI, 8.1%–12%); this difference of 3.2 points represents a 32% relative increase in vaccination in the opt-out group compared with the opt-in group. A chi-square test showed that the proportions of participants in the two groups who accepted the vaccine were significantly different, χ²(1, 1957) = 4.79, P = 0.029. The results of a logistic regression analysis on vaccine uptake, which included group and gender as categorical predictors, showed that participants in the opt-out group were more likely to get vaccinated than those in the opt-in group (odds ratio [OR] = 1.367, 95% CI, 1.032–1.809; P = 0.029) and that males were more likely to get vaccinated than were females (OR = 1.403, 95% CI, 1.057–1.863; P = 0.019).

Discussion

Our results show that the simple switching of the default option from opt in to opt out generated significantly greater COVID-19 vaccine uptake in individuals who had not accepted any previous opportunities to vaccinate. The obtained results indicate that the status quo can exert a powerful influence on the decision to vaccinate, even in reluctant populations and even when opting out comes at no cost and involves no extra effort (since participants could ignore the vaccination appointment with no consequences).

Prior nudging interventions aimed at increasing vaccination coverage by acting on the default provided encouraging, though variable, results. For example, they have been shown to lead to greater influenza immunization among university workers (12 percentage points, corresponding to a 36% relative increase) ¹¹ and patients (2.8 percentage points, corresponding to a 6.7% relative increase) ¹² but not among health care employees. ¹³ To our knowledge, this study documents the first application of this specific nudging technique to COVID-19 vaccination efforts, following a number of suggestions for its implementation.^14,15 Other behavioral nudges—such as small incentives, advising of personal or social benefits, and reminders designed to make vaccination more salient and easier to book—have been reported to modulate COVID-19 vaccination either not at all or with increases up to 7.4 percentage points (corresponding to a 45% relative increase).^16–18

Our results are therefore in line with the most successful improvements in acceptance of various vaccines reported in the literature. Notably, we observed an effect of the default, even despite various factors that could have interfered with such an effect: reluctance of the target population, the distinct possibility that both versions of the letter could have served as reminders, the measurements being made during the peak summer holiday season (in Italy, July 15–August 15), and the intervention occurring in the midst of a then-new green pass policy in Italy (announced at the end of July 2021 and implemented on August 6).

Overall, these results confirm that vaccine hesitancy does not necessarily arise from strong concerns about vaccination but may well be rooted in procrastination or in difficulties in following through on the vaccination acquisition procedure. ¹⁹ Of course, the acceptability and effectiveness of analogous initiatives at other times and in other settings cannot be taken for granted since, as with other strategies, it would require an understanding of all complexities and characteristics of the target populations. ²⁰ Moreover, it is not obvious how changing the default might interact with different nudging interventions or financial incentives to encourage vaccination. Future research might explore these issues as well as how to encourage long-term trust and how to foster more deliberative adherence to vaccination, by complementing nudging with strategies such as education ²¹ and communication campaigns ²² aimed at clarifying the individual and collective benefits of vaccination.