Study design. We conducted a randomized controlled cross-over trial consisting of two 8-week interventions separated by one-month washout period on 43 patients with type 2 diabetes. The trial including recruitment and interventions was conducted in spring and summer 2015. The sample size was determined according to a previous publication  for detection of a mean between-group difference of 18 mg/dl for change of total cholesterol, considering a cross-over design, a statistical power of 90%, and error of 5%.
Subjects. Participants were 43 patients with type 2 diabetes diagnosed by an endocrinologist according to diagnostic criteria of having fasting blood glucose ≥126 mg/dl . The participants were collected from Nader Kazemi Diabetes Clinic in Shiraz, Iran. Patients were excluded in case of malignancy, organ failure, pregnancy or lactation, smoking, fasting blood glucose >200 mg/dl, use of chemotherapy, immunosuppressive, or anti-inflammatory medications, use of hypertriglyceridemia medications, nutritional supplements, insulin therapy, change of oral hypo-glycemic and cholesterol-lowering medications, hospitalization, and affection by infectious diseases or acute medical conditions during the intervention. All participants gave written informed consent. The study was performed according to guidelines of 1964 Declaration of Helsinki and its later amendments. The trial was approved by the Ethics Committee of Shiraz University of Medical Sciences (approval number. CT-P-9386-8456) and registered in the Iranian Registry of Clinical Trials (IRCT 2015062122364N2).
Intervention. Participants were allocated to the sequence of either honey-control or control-honey by block randomization using a block size of 4. Two investigators enrolled the participants and an independent investigator generated the random allocation sequence and assigned the participants to the arms. The first sequence was performed for 8 weeks. Then, a washout period was run for one month, after which the second sequence of the intervention was carried out. The ethics committee of our institution did not permit having a control treatment such as glucose and sucrose because of their adverse effects on type 2 diabetes patients. The cross-over design of the study may compensate for the lack of control product because in the cross-over design each subject acts as its own control.
In honey condition, participants were recommended to consume 50 g/day honey plus dietary recommendations and in control condition they received only dietary recommendations. Honey was from milk vetch (Astragalus bisculatus) flowers with carbohydrate composition of 39.6% fructose, 33.3% glucose, and 3.0% sucrose as assessed by our biochemical laboratory according to the standard method . The honey was natural; for which no heating process or sugar feeding for bees had been performed. Special measuring cups were provided for participants to ensure consumption of right amount of honey. The patients were asked to divide 50 g honey into 3 portions and consume the portions between meals with drinks like tea, coffee, or milk.
Weight maintenance diets were prescribed for all participants according to their age, sex, height, weight, body mass index, and physical activity. To ensure the suitability of weight maintenance diets a 3-week run-in period was used before the intervention. The subjects were asked to continue their diet, physical activity, and lifestyle during the study. In order to equalize energy intake between the two groups, the participants in the honey group were asked to reduce consumption of cereals by 2 servings per day. By this way, the energy deficit resulted from exclusion of 2 servings of cereals (bread or rice) was substituted with the energy obtained from consumption of honey (150-160 kcal/day). It is worth noting that the amount of the prescribed honey was within the range of recommendations for free sugar in healthy diet [15, 16] and according to the guidelines of the American Dietetic Association for adults with diabetes . Patients were recommended to abstain from all kinds of sugar-sweetened foods and beverages. Subjects were contacted biweekly to be reminded for recommendations regarding diet and honey consumption and emphasized to keep physical activity unchanged. Also, diary sheets were provided for the subjects and they were asked to mark the sheets after each time of honey consumption.
Assessments. Dietary intakes, physical activity, anthropometric, and biochemical parameters were assessed at the beginning and the end of each sequence of the intervention. Blood was collected at baseline and the end of each sequence of the study in the morning after a 12-h overnight fast. Serum was separated by centrifugation at 3000 rpm for 15 min and serum samples were frozen and kept in -70 until the end of the study. The primary outcome measure was malondialdehyde and the secondary endpoints were total antioxidant capacity and hs-CRP.
Glucose and lipid fractions were measured by an auto-analyzer (BT 1500, Biotecnica Instruments, Italy) using specific kits (Pars Azmoon, Tehran, Iran). Concentration of malondialdehyde, a byproduct of unsaturated fatty acid peroxidation was measured, as a marker of oxidative stress, through thiobarbituric acid reactive substance (TBARS) method as described by Satoh . In TBARS method, acidic condition and heat (90-100 °C) stimulates malondialdehyde to react with thiobarbituric acid, yielding the pink colored malondialdehyde-thiobarbituric acid adducts which can be quantified colorimetrically at 530 nm after extraction in organic solvents such as butanol.
Total antioxidant capacity of serum was determined using the ferric reducing capacity of plasma (FRAP) assay . Briefly, 50 µl of the serum was added to 1.5 ml of freshly prepared and pre-warmed (37 °C) FRAP reagent (300 mM acetate buffer, pH = 3.6, 10 mM tripyridyl-s-triazine in 40 mM HCl, and 20 mM FeCl3.6H2O in the ratio of 10:1:1) and incubated at 37 °C for 10 min. The absorbance of the sample was read against reagent blank (1.5 ml FRAP reagent and 50 µl distilled water) at 593 nm. Standard solutions of Fe2+ in the range of 100 to 1000 mM were prepared from ferrous sulfate (FeSO4.7H2O) in distilled water. The data was expressed as mM ferric ions reduced to ferrous form per ml (FRAP value). Levels of serum adiponectin and hs-CRP were measured by commercially available kits (IBL International, Germany) using immunoturbidimetry technique according to the manufacturer's instructions.
Trained dietitians carried out dietary assessments. Diet including all foods and drinks consumed during 3 days (2 random weekdays and one weekend day) was evaluated by 24-h recall. Physical activity was assessed by self-reported validated international physical activity questionnaire and expressed as metabolic equivalent task (MET) in min/day . Dietary intakes were analyzed using Nutritionist IV, version 3.5.2 (Hearst Corp., San Bruno, CA). Because Iranian food composition tables contain limited number of foods, nutrient compositions were determined using the US Department of Agriculture food composition database. For traditional Iranian foods, Iranian food composition tables were used.
Statistical analysis. Statistical analyses were performed with SPSS version 19 (SPSS Inc., Chicago, IL, USA). Data are presented as means ± standard deviation (SD). Baseline and final values in each group were compared with paired t-test. Between-group alterations in the study outcomes were evaluated by analysis of covariance (ANCOVA) with adjustments for the corresponding basal values. Between-group comparisons of dietary intakes were tested with independent t-test. The normality of data was assessed by Kolmogorov-Smirnov test and abnormal data were log transformed where needed. P < 0.05 was considered as the significant level.