# The AuTOMATIC trial: a multicentre digitally-automated, Bayesian, adaptive, parallel, factorial randomised controlled trial of SMS reminders for childhood vaccination

## TL;DR

This study tested different SMS reminder strategies to see if they help parents vaccinate their children on time, finding only a modest improvement overall.

## Contribution

A novel Bayesian adaptive trial design evaluated multiple combinations of SMS reminder content and timing for childhood vaccination.

## Key findings

- SMS reminders were associated with a modest 6% increase in on-time vaccination compared to no reminders.
- Persuasive message framings were more effective than neutral ones, but no optimal combination was identified.
- The trial used digital automation and Bayesian methods to adaptively evaluate multiple intervention arms.

## Abstract

The estimated effectiveness of SMS (short message service) reminders for improving childhood vaccine coverage and timeliness has varied in previous studies. The observed heterogeneity in effectiveness may be explained in part by variation in reminder content or timing of the reminder relative to the vaccine schedule date. We sought to evaluate the effectiveness of a range of SMS reminders of varied content and timing for improving on-time childhood vaccination.

AuTOMATIC was a multi-centre Bayesian adaptive factorial randomised trial comparing four alternative SMS message framings at three alternative message timings versus a no reminder control strategy. Participants were parents of children registered with one of 20 primary care clinics Australia-wide and randomly assigned to one of 12 SMS reminder arms or to control. Reminders varied by framing of content (neutral, positive, risk-based or social benefit) and timing (14 days prior to the due date, on the due date, or 7 days afterwards). The primary endpoint was on-time vaccination, i.e. within 28 days of its scheduled date. Allocation probabilities were updated and stopping rules implemented over the trial according to pre-specified rules based on the posterior probability of effectiveness of each arm evaluated at interim analyses. Trial procedures were largely digitally automated. This trial was registered on Australian New Zealand Clinical Trials Registry (ACTRN12618000789268).

Between January 14, 2021 and February 26, 2024, 9993 parents were randomised and all were included in the primary analysis; between 380 and 1110 were assigned to each of the 12 SMS reminder arms and 637 to control. The adjusted odds ratio (aOR) of on-time vaccination for each of the 12 SMS arms compared to control ranged from 1.02 [95% CrI 0.76–1.34] to 1.53 [1.22, 1.92] with a pooled effect aOR of 1.29 [1.06, 1.55]. This pooled effect corresponded to a standardised difference of 6% [2%, 11%] in the proportion of on-time vaccinations.

On average, SMS reminders were associated with a modest increase in on-time vaccination compared to no reminder. There was evidence that neutral SMS reminders were less effective than persuasive reminders, but we were unable to identify a single best combination of reminder content framing and timing.

10.13039/100009860Ramaciotti Foundations, 10.13039/501100001232Royal Australasian College of Physicians, and the Western Australia Department of Health.