Effects of a smartphone-based chatbot intervention on influenza and COVID-19 vaccine uptake among South Asians

Cho Lee Wong, Dorothy Ngo Sheung Chan, Winnie Kwok Wei So, Carmen Wing Han Chan, Kai Chow Choi, Huiyuan Li, Shelby L. GarnerNpj Digit. Medicine 2025 · DOI

TL;DR

This study demonstrates that a smartphone-based chatbot intervention significantly increased influenza vaccine uptake and intention to vaccinate among South Asian ethnic minorities in Hong Kong. While the intervention effectively reduced vaccine hesitancy, it did not show a similar impact on COVID-19 vaccine uptake, likely due to changing pandemic conditions and guidelines.

What Problem Does This Paper Address?

Vaccine-preventable diseases like influenza and COVID-19 pose a significant global health burden. Despite the proven efficacy of vaccines, vaccination rates, particularly among ethnic minority groups, can be stagnant or low due to various barriers including inadequate communication, lack of trust, low literacy, and language differences. South Asians in Hong Kong represent a significant and growing ethnic minority group, and there's a lack of data on their vaccination rates. This paper addresses the need for effective, accessible interventions to improve vaccine uptake and reduce hesitancy in such underserved populations.

Methods

The study employed a two-arm, cluster-randomised, wait-list controlled trial.

  • Intervention Development: A smartphone-based chatbot was developed, informed by the 5 C model of vaccine hesitancy (confidence, complacency, constraints, calculation, and collective responsibility). The chatbot delivered educational messages and reminders twice weekly for four weeks, covering topics like disease epidemiology, vaccine options, addressing misconceptions, and practical information on booking appointments. It was available in English, Urdu, and Nepali, with an optional voice narration feature. An on-demand feature allowed users to connect with trained research assistants.
  • Trial Design: Six community centres and associations were cluster-randomised into either an intervention group or a wait-list control group. Participants were South Asians aged 18 or older who had not received influenza or COVID-19 vaccines in the past year.
  • Data Collection: Assessments were conducted at baseline (T0), immediately post-intervention (T1), and three months post-intervention (T2). Vaccination uptake was objectively verified through vaccination records. Intention to vaccinate and vaccine hesitancy (using the 5 C Scale) were measured via questionnaires.
  • Statistical Analysis: Generalised estimating equations (GEE) were used to compare outcomes between groups, accounting for clustering and repeated measures. Holm's multiplicity adjustment was applied for multiple comparisons.

Datasets

The study utilized data collected from 754 South Asian individuals approached at six community centres and associations in Hong Kong. Of these, 610 participants were included in the final analysis (306 in the intervention group and 304 in the control group). No new benchmarks were introduced; existing validated scales for vaccine hesitancy were used.

Key Insights and Ideas

  • Chatbots as a Scalable Solution: The study highlights the potential of smartphone-based chatbots as a scalable and accessible tool to deliver health information and promote vaccination, especially for hard-to-reach ethnic minority groups.
  • Targeted Messaging is Key: The intervention's success in increasing influenza vaccine uptake suggests that tailored, culturally relevant messaging can effectively address specific barriers faced by South Asian communities.
  • Influenza vs. COVID-19 Uptake Discrepancy: The significant impact on influenza vaccine uptake contrasted with a lack of impact on COVID-19 vaccine uptake is a crucial finding. This suggests that external factors, such as evolving pandemic conditions and public health guidelines, can override the effectiveness of even well-designed digital interventions for certain vaccines.
  • Addressing Vaccine Hesitancy Components: The chatbot was particularly effective in improving confidence and calculation related to vaccination, indicating that providing credible information and helping individuals weigh risks and benefits are critical for reducing hesitancy.

Key Results

  • Influenza Vaccination Uptake: Significantly higher influenza vaccination rates were observed in the intervention group compared to the control group immediately after the intervention (57.8% vs. 1.3%, p < 0.001) and at three months post-intervention (68.0% vs. 2.0%, p < 0.001).
  • Intention to Vaccinate: The intervention group showed a significantly greater increase in the intention to receive influenza and COVID-19 vaccines at both T1 and T2 compared to the control group.
  • Vaccine Hesitancy: All measures of vaccine hesitancy showed greater improvements in the intervention group. Notably, participants in the intervention group reported a significantly greater increase in confidence and calculation at T2.
  • COVID-19 Vaccination Uptake: No statistically significant differences were found between groups regarding COVID-19 vaccine uptake. This was attributed to low overall uptake in both groups during the study period, likely influenced by the easing of pandemic restrictions and revised vaccination guidelines.

Limitations and Future Work

  • Generalizability: The study was conducted in Hong Kong and recruited only South Asians from India, Nepal, and Pakistan, limiting the generalizability of findings to other ethnic groups or geographical locations.
  • Selection Bias: Cluster randomisation might introduce selection bias due to the limited number of clusters and potential heterogeneity within them.
  • Subgroup Analyses: The relatively small sample size within each cluster precluded subgroup analyses.
  • Control Group: The control group received passive usual care, and future trials could benefit from an active control group (e.g., standardised text-based information) to better isolate the chatbot's effects.
  • Follow-up Period: The short follow-up of three months may not capture long-term effects.
  • Chatbot Sophistication: The current chatbot is rule-based; future research could explore the impact of more advanced AI-powered chatbots.
  • Cost-Effectiveness: Incorporating a cost-effectiveness analysis would provide a more comprehensive understanding of the intervention's sustainability.
  • COVID-19 Uptake Factors: Further research is needed to explore the specific factors influencing COVID-19 vaccine uptake, especially during periods of low infection and evolving public health recommendations.

Summary updated

BibTeX
@article{wong2025effects,
  author = {Cho Lee Wong and Dorothy Ngo Sheung Chan and Winnie Kwok Wei So and Carmen Wing Han Chan and Kai Chow Choi and Huiyuan Li and Shelby L. Garner},
  title = {{Effects of a smartphone-based chatbot intervention on influenza and COVID-19 vaccine uptake among South Asians}},
  journal = {Npj Digit. Medicine},
  year = {2025},
  doi = {10.1038/s41746-025-02200-1},
}
Figure