Background/aims:
Nudges are non-judgemental prompts that aim to influence behaviour. This study aimed to determine whether a nudge delivered via mobile phone, soon after dinner reduces after-dinner snacking events as measured by continuous glucose monitors (CGMs) in people with type 2 diabetes (T2D).
Methods
Adults with T2D (N=67, 48% females, age 61.6±10.3 years, BMI 30.8±7.6 kg/m2) managed with diet or oral antidiabetic medications ≥3 months, and who habitually snack after dinner ≥3 nights/week were recruited, and seven picto-graphic nudges were co-developed with people with T2D. Following a 2-week lead-in phase to determine eligibility, 59 participants were enrolled and a nudge or no nudge was delivered after dinner over two weeks using a micro-randomised design (1:1 ratio) via a mobile phone APP (Intui Research, goAct, Australia). Dinner timing was determined by photographing the evening meal. Complete data blocks were defined as paired available CGM data on successful nudge delivery days vs non-nudging days. The primary outcome was incremental area under the CGM curve (iAUC) from 90-minute post-dinner until 4am. Mixed effects models were used to determine intervention efficacy. The post-hoc analysis was performed on iAUC and glucose peak frequency after excluding blocks without a stable CGM baseline, where participants failed to access nudges on time or patient protocol violation.
Results
CGM data were collected on 780 nights, 410 of which followed digital nudges. There were 386 complete blocks. Nudges were without effect on the primary outcome (β=0.07, 95%CI:[-0.46, 0.60], p=0.78). Post-hoc analysis of 157 blocks from 50 participants with per protocol data showed that glucose iAUC was lower (β=-0.69, 95%CI: [-1.33, -0.05], p=0.03) and glucose peak frequency, indicating decreased evening snacking events (β=-0.46, 95%CI: [-0.84, -0.08], p=0.02), on nudging vs non-nudging days.
Conclusion
This study provides promising evidence for potential benefits of just-in-time digital nudges to altered after-dinner snacking in T2D.