Text messages improved four kinds of diabetes self-care, but many trials had weak designs.
HelpedThe short version
Text-message programmes helped people exercise more than the other groups did.

- 17
- studies pooled
- 8,971
- adults with type 2 diabetes in 37 trials
The review
Who took part
8,971 adults with type 2 diabetes in 37 trials from 20 countries. Average ages in the trials ranged from about 43 to 67, and men made up from 25% to 86% of each trial.
What they did
The text-message groups got health messages by text, either alone or as one part of a wider programme. Most other groups got usual care, and the rest got a different active programme.
What happened
Exercise, healthy eating, medicine use and blood sugar checks each improved more than in the other groups, in 5 to 17 trials. Overall, the blood sugar test (HbA1c) fell, and the authors rated that evidence as moderate. A check for missing studies found signs of publication bias, and allowing for it made the fall smaller.
What the review found
- Helped
Text-message programmes helped people exercise more than the other groups did.
Pooled analysis of 17 studies · vs usual care, or a different programme · ⚠ most studies had serious flaws; whole programmes were tested, so no single part can be credited
- Helped
Text-message programmes helped people eat more healthily than the other groups did.
Pooled analysis of 17 studies · vs usual care, or a different programme · ⚠ most studies had serious flaws; whole programmes were tested, so no single part can be credited
- Helped
Text-message programmes helped people take their medicine as planned more than the other groups did.
Pooled analysis of 9 studies · vs usual care, or a different programme · ⚠ most studies had serious flaws; whole programmes were tested, so no single part can be credited
- Helped
Text-message programmes helped people check their blood sugar more often than the other groups did.
Pooled analysis of 5 studies · vs usual care, or a different programme · ⚠ most studies had serious flaws; results look too good to be true; whole programmes were tested, so no single part can be credited
How solid is this?
- Large group
- Whole programmes were tested, so no single part can be credited
Source
Effectiveness of text message-delivered health behaviour intervention on HbA1c change in adults with type 2 diabetes mellitus: a systematic review and meta-analysis of randomised controlled trials (2026)
DOI: 10.1080/17437199.2025.2595126 (opens in a new tab)

