Clin Invest Med. 2000 Oct; 23(5): 311-7.

Herbal tea in the treatment of diabetes mellitus.

Ryan EA, Imes S, Wallace C, Jones S.

Department of Medicine, University of Alberta, Edmonton.

OBJECTIVE: To evaluate the effects of a native herbal tea in patients with type 2 diabetes. DESIGN: Randomized, placebo-controlled, single-blind study. SETTING: The Metabolic Centre at the University of Alberta Hospitals. SUBJECTS: Forty volunteers with type 2 diabetes. INTERVENTIONS: After a 1 month "run-in" period, subjects drank 250 mL/d of either the herbal tea or a placebo tea for 10 days, and were followed up for a further 4 weeks. OUTCOME MEASURES: A responder analysis defined as a 10% change in mean blood glucose levels based on 4 capillary glucose readings daily. Secondary end points included changes in HbA1c, fructosamine and response to a meal challenge using Ensure. RESULTS: The responder analysis showed no benefit from the herbal tea. Fructosamine levels before and after tea therapy decreased significantly in both study groups. Mean HbA1c levels and incremental areas under the glucose curve (AUC) in the meal challenge did not change in either study group. These data were reanalysed in hyperglycemic subjects with HbA1c levels greater than 120% of normal. The responder analysis and HbA1c levels did not change in either group. Mean (and standard deviation) fructosamine levels, before and after tea therapy, were significantly lower in the herbal tea group than in the placebo tea group (361 [98] versus 338 [100] micromol/L, p < 0.01 compared with 338 [60] versus 323 [49] micromol/L, p = 0.08). In the hyperglycemic subgroup the mean AUC during the meal challenge, before versus after tea therapy, was 776 (369) versus 639 (331) mmol/L (p = 0.22) in the herbal tea group and 433 (125) versus 420 (173) mmol/L (p = 0.90) in the placebo group. CONCLUSIONS: Although the responder analysis failed to show an effect of the herbal tea, the data suggest there may be a short-term benefit from the tea in subjects with poor glycemic control.