These rates suggest that the majority of youths with type 2 diabetes will have to go on insulin or another form of combination therapy a few years after their diagnoses, the researchers wrote.
Why the failure rates in children are higher than those seen in adult trials should be the subject of further research, they added.
Hirst and colleagues did find that metformin plus Avandia was superior to metformin alone in terms of lowering HbA1c, and that the lifestyle-plus-metformin group provided an "intermediate" benefit but didn't differ significantly from either of the other two treatments.
The effects didn't appear to be because of differences in adherence and couldn't be explained by baseline characteristics, body mass index (BMI), or differences in insulin secretion, insulin sensitivity, or body composition, Hirst and colleagues wrote.
BMI did differ significantly according to treatment group over time, with metformin plus lifestyle offering the best weight loss and metformin plus Avandia prompting the greatest weight gain, but again, this did not affect treatment outcomes, they said.
In subgroup analyses, combination therapy appeared to be more effective in girls than boys, and metformin alone was less effective in blacks than it was in whites or Hispanics.
Overall, serious adverse events occurred in about 19 percent of patients, and were greatest in the metformin-plus-lifestyle group: 25 percent versus 18 percent with metformin alone and 15 percent for metformin plus Avandia.
The majority of adverse events (87 percent) were not considered to be related to the study treatment, but hospitalizations accounted for more than 90 percent of serious adverse events, including severe hypoglycemia, nonfatal transient lactic acidosis, and asthma exacerbation.
The authors noted that rosiglitazone had no effects on bone mineral content or fracture rates, but that this result should be interpreted with caution given the small sample size.
It's not clear if the benefits of combination therapy were due to the Avandia, more general effects of the thiazolidinedione class, or some other feature of combination therapy, the researchers said.
Although Allen had called the results discouraging, he warned that by no means do they "put the nail in the coffin" of lifestyle modification for young type 2 diabetes patients.
The differences between metformin plus Avandia and metformin plus the lifestyle intervention weren't significant, he noted, and changes in eating and activity didn't reduce patients' weight as much as they should have in the trial, "so the feasibility of lifestyle change was evaluated more than its effect," he wrote.
Instead, youths need to be taken out of a "sedentary, calorie-laden environment" to control their diabetes, Allen wrote.