Level 1: Likely reliable evidence
In a large trial several years ago, lifestyle changes were found to prevent type 2 diabetes. Now, a follow-up study shows that the beneficial effects are sustained after seven years.
The original study was a randomized trial of 523 patients aged 40-65 years who were overweight (BMI >25, average 31) and had impaired glucose tolerance (fasting glucose <140 mg/dL, glucose 140-200 mg/dL two hours after oral glucose 75 g) (N Engl J Med. 2001;344:1343-1350). Participants were randomized to lifestyle intervention or control. One patient with diabetes at baseline was subsequently excluded; 40 patients (8%) dropped out but were analyzed using intention-to-treat analysis with data censored at the last oral glucose tolerance test (OGTT).
The intervention consisted of individual one-hour sessions with a nutritionist seven times during the first year, then every three months; an exercise assessment and program plan; and a one-year membership at a local health club or exercise opportunity at an appropriate facility. Participants in the control group had yearly meetings with a nutritionist and a physician and received standard advice for the same type of dietary improvements and increased exercise as those in the intervention group. An OGTT was performed annually.
Diabetes was defined as fasting glucose >140 mg/dL or two-hour glucose >200 mg/dL confirmed with a second OGTT. The mean duration of follow-up was 3.2 years.
Comparing intervention vs. control, the average weight loss at one year was 4.2 vs. 0.8 kg (9.24 vs. 1.76 lb) and at two years was 3.5 vs. 0.8 kg (7.7 vs. 1.76 lb). The cumulative incidence of diabetes after four years was 11% vs. 23% (NNT 9).
The new study (Lancet. 2006;368:1673-1679) returned to these patients seven years later. After a median four years of active intervention, patients were followed for an additional median three years; 47 patients (9%) were lost to follow-up. The incidence of type 2 diabetes over the entire follow-up period was 4.3 per 100 person-years in the intervention group vs 7.4 per 100 person-years in the control group (P =.0001). Fewer participants in the intervention group developed diabetes (28.3% vs. 42.8%, NNT 7).