Fourteen RCTs were included in the review (1,268 participants, range nine to 322). Study duration ranged from three to six months. Seven RCTs scored one or two out of five on the Jadad scale, seven scored three out of five.
Patients using continuous glucose monitoring systems had a significantly greater decrease in HbA1c from baseline compared with self monitoring of blood glucose (WMD -0.26, 95% CI -0.34 to -0.19; 14 RCTs). There was no evidence of statistical heterogeneity (Ι²=0%).
Devices that offered real-time glucose readings for continuous glucose monitoring had a significantly greater decrease in HbA1c from baseline compared with self monitoring of blood glucose (WMD -0.27, 95% CI -0.34 to -0.19; eight RCTs), but the difference was not statistically significant for devices that provided retrospective readings. Results of other subgroup analyses were similar to those of the main analysis. Results were similar for studies of children and adolescents and studies of adults.
Similar results were reported for end of trial HbA1c levels, but there was significant heterogeneity associated with these results. A significantly higher proportion of patients using continuous glucose monitoring systems achieved target HbA1c levels compared with patients using self monitoring of blood glucose (OR 2.14, 95% CI 1.41 to 3.26; numbers-needed-to-treat 7.40, 95% CI 4.70 to 17.43; four RCTs).
There was a significant reduction in hypoglycaemic events in the continuous glucose monitoring group compared with the self monitoring of blood glucose group (SMD -0.32,95% CI -0.52 to -0.13; four RCTs), but there was no significant difference in the frequency of severe hypoglycaemic episodes.
Only a few trials reported adverse events; the most common was mild reactions at the sensor implantation site in the continuous glucose monitoring group. Only a small number of patients reported experiencing more severe reactions (painful itching, severe pain during sensor implantation, skin abscess formation or cellulitis). The risk of ketoacidosis episodes was rare and not significantly different between groups. Three studies reported technical problems related to the use of continuous glucose monitoring systems.
There was no evidence of significant publication bias.