Digital behaviour change interventions to increase vegetable intake in adults: a systematic review

**Background:** Low vegetable consumption is a leading modifiable risk factor for non-communicable diseases globally, with 88% of populations across 162 countries having inadequate intake in 2020. Digital interventions offer scalable delivery models, but their effectiveness for increasing vegetable intake specifically (separate from fruit) is unclear. This systematic review aimed to describe effectiveness of digital interventions for increasing vegetable intake in adults and examine links with co-design, personalisation, behavioural theories, and policy frameworks.

**Methods:** A systematic search of MEDLINE, Embase, PsycINFO, Scopus, CINAHL, Cochrane Library, INFORMIT, IEEE Xplore, and clinical trial registries was conducted for studies published January 2000 to August 2022. Eligible studies included RCTs, pseudo-RCTs, and pre-post interventions targeting community-dwelling adults. Digital interventions were defined as those using apps, websites, computer programs, mobile games, SMS, social networking services, or wearable devices. The primary outcome was change in vegetable intake. Risk of bias was assessed using Cochrane RoB 2, RoB 2 CRCT, and ROBINS-I tools. The NOURISHING framework was used to map policy domains.

**Key Results:** From 1,347 records, 30 studies were included. Risk of bias was high or serious in 25 of 30 studies (83%). Only 8 interventions (27%) were effective at improving vegetable intake. Effective interventions showed increases of 0.29 to 1 serve/day. Embedding behavior change theories was more common in effective interventions (89%, 8/9) versus ineffective ones (61%, 12/21). Stakeholder involvement in design was also more frequent among effective interventions (44%, 4/9) compared with ineffective ones (38%, 8/21), though only one ineffective intervention used true co-design. Personalisation was present in 67% (6/9) of effective interventions versus 81% (17/21) of ineffective ones, but the degree and nature of personalisation varied considerably. All interventions mapped to the NOURISHING behavior change communication domain; only one ineffective intervention also mapped to the food environment domain (using economic incentives). No studies addressed the food system domain. Vegetable intake was a primary outcome in 63% of studies (19/30). Among effective studies reporting serves/day, improvements ranged from 0.29 serves/day to 1 serve/day. One study reported 87% of participants improved intake versus 29% in the control group (p<0.001). Another pre-post study reported a 3.75-point increase in vegetable score (Global Diet Quality Index, p=0.005).

**Clinical Implications:** The review found that most digital interventions to increase vegetable intake are ineffective, with only about one-quarter showing statistically significant improvements. Embedding behavior change theories and involving stakeholders in design may increase effectiveness, though the evidence is limited by high risk of bias across most studies. The under-utilisation of comprehensive co-design methods represents a missed opportunity to ensure personalisation approaches meet target population needs. The absence of interventions addressing food environment factors (e.g., affordability, access) suggests future digital interventions should integrate strategies at both individual and environmental levels. The heterogeneity in outcome reporting (serves/day, grams/day, diet quality scores) prevents quantitative synthesis and highlights the need for standardised reporting of vegetable intake separately from fruit. The predominance of female, mid-to-older age, and non-rural populations limits generalisability, and future research should prioritise diverse and underserved populations to avoid exacerbating health inequities through the digital divide.