Participants were recruited from the community by newspaper advertisements. After telephone screening, potential participants were scheduled for a "screening visit" which included informed consent approved by the local institutional review board, a medical history, physical examination and laboratory tests. The inclusion criteria were: diagnosis of type 2 diabetes mellitus > 1 year (confirmed by hemoglobin A1c > 6.0%), onset of diabetes after age 15 years, no history of diabetic ketoacidosis, age 18–65 years old, body mass index (BMI) from 27–50 kg/m2, and desire to lose weight. Exclusion criteria were: unstable or serious medical condition; significant co-morbid illnesses such as liver disease (AST or ALT > 100 IU/L), kidney disease (serum creatinine > 1.5 mg/dL), cancer; pregnancy; or nursing mothers. No monetary incentives were given.
If study criteria were met, participants were randomized to one of two treatment groups stratified upon BMI greater or less than 32 kg/m2 using a computer-generated list, and invited to attend the "baseline visit." (Measurements taken at the "screening visit" were used as the initial value in comparison testing for laboratory tests; measurements from the "baseline visit" were used as the initial value for other outcomes.) The intervention for both groups included group sessions, diet instruction, nutritional supplements, and an exercise recommendation. Group meetings took place at an outpatient research clinic every week for 3 months, then every other week for 3 months. If a participant was taking medication for diabetes or hypertension, a physician reviewed the blood glucose and blood pressure readings and made medication changes according to a pre-specified algorithm. Participants were encouraged to exercise for 30 minutes at least 3 times per week, but no formal exercise program was provided. Both groups received the same nutritional supplements known to have a mild lowering effect on blood glucose levels (vanadyl sulfate 200 mcg/day, chromium dicotinate glycinate 600 mcg/day, alpha-lipoic acid 200 mg/day) [17, 18].
Low-carbohydrate, Ketogenic Diet Group Intervention (LCKD)
Using a lay-press diet book and additional handouts, a registered dietitian instructed participants to restrict intake of dietary carbohydrate to fewer than 20 grams per day, without explicitly restricting caloric intake . Allowed foods were unlimited amounts of animal foods (i.e., meat, chicken, turkey, other fowl, fish, shellfish) and eggs; limited amounts of hard cheese (e.g., cheddar or swiss, 4 ounces per day), fresh cheese (e.g., cottage or ricotta, 2 ounces per day), salad vegetables (2 cupfuls per day), and non-starchy vegetables (1 cupful per day). Participants were encouraged to drink at least 6 glasses of permitted fluids daily. Drinking bouillon dissolved in water was recommended 2–3 times a day during the first two weeks to reduce possible side effects.
Low-glycemic index diet group intervention (LGID)
Using a lay-press diet book and additional handouts, a registered dietitian instructed participants to follow a low-glycemic index, reduced-calorie diet with approximately 55% of daily caloric intake from carbohydrate . The energy intake was individualized to be 2.1 MJ (500 kcal) less than the participant's calculated energy intake for weight maintenance (21.6*lean body mass + 370 kcal + activity factor) .
Primary outcome measure
Hemoglobin A1c was measured at baseline, week 12, and week 24. The primary outcome was change in hemoglobin A1c from baseline to week 24, using an immunoassay technique. The hemoglobin A1c provides an estimate of glycemic control for the previous 3-month period and is predictive of clinical outcomes .
Other outcome measures
All participants completed food records (5 consecutive days, including a weekend) at baseline, and during the intervention (weeks 4, 12, and 24). Participants were instructed how to document food record information and given handouts with examples of how to complete the records. A sample of completers (n = 8 for low-carbohydrate diet group; n = 7 for low-glycemic diet group) was selected for food record analysis based upon record detail. A registered dietitian analyzed the food records using a nutrition software program (Nutritionist Five, Version 1.6, First DataBank Inc., San Bruno, CA). Food record results were averaged over weeks 4, 12, and 24.
Wearing light clothing and no shoes, participants were weighed at each visit on the same calibrated scale. Body mass index was calculated as: (body weight in kilograms)/(height in meters)2. Systolic and diastolic blood pressures were measured in the non-dominant arm using an automated digital cuff (model HEM-725C, Omron Corp., Vernon Hills, IL) after sitting for 3 minutes. Two measurements were taken per visit and averaged for the analysis.
Other metabolic effects
Blood tests were obtained in the morning after at least 8 hours of fasting and processed by a commercial laboratory (Labcorp, Burlington NC). Glomerular filtration rate was estimated by using an equation containing the variables age, gender, race, and serum albumin, creatinine, and blood urea nitrogen (Modification of Diet in Renal Disease (MDRD) Study equation) . Twenty-four hour urine collections for protein were collected at baseline and at 24 weeks.
At all return visits, participants completed an open-ended side effects questionnaire. To enhance the description of side effects, participants completed a checklist of side effects commonly mentioned during weight loss studies at both the 20 and 24-week visit. These two measures were combined to report the proportion in each group who experienced an adverse effect at any time during the study.
At baseline and at all return visits, participants recorded all of their current medications with dosages and schedules.
Adherence with the diet and exercise recommendations was measured by self-report, food records, and urinary ketones [24, 25]. The delivery of the intervention and the assessment of outcomes were not blinded to the treatment assignment.
For categorical outcomes, comparisons between groups were performed using the chi square test or Fisher's exact test, as appropriate. For all continuous outcomes, comparisons were made using a t-test or Wilcoxon rank-sum test as appropriate, testing the difference between groups for the change from baseline to week 24. For the primary outcome variable, a completer's analysis and last observation carried forward (LOCF) were performed, and a multiple linear regression analysis adjusting for weight change was performed to determine if the change in hemoglobin A1c was independent of weight loss. A p value of ≤ 0.05 was considered statistically significant. Analyses were performed using SAS Statistical Software, Version 8.02 (SAS Institute Inc., Cary, NC). In order to detect a clinically meaningful change in hemoglobin A1c (absolute change of 1%, SD = 1.5) with 80% power (two-sided alpha of .05) in a completers analysis, a total of 60 participants was required. To protect for dropouts, 97 participants were recruited.
Role of the funding source
The investigators conducted the study independently of the funding source. The funding source had no involvement in conduct of the study.