CHICAGO – Despite being a cornerstone of contemporary diabetes management, carbohydrate counting had no significant effect on glycemic control in type 1 diabetes in a meta-analysis and systematic review involving 667 adults and children.

Pooled results from six randomized controlled trials showed an overall change in hemoglobin A_1c levels of just –0.3% points with a variety of carbohydrate counting methods, compared with other dietary interventions (P = .185), Ms. Kirstine Bell reported at the annual scientific sessions of the American Diabetes Association.

Overall, four of the six studies favored carbohydrate counting and all six showed a trend toward decreased risk of hypoglycemia, suggesting a stabilization of blood glucose levels.

Four of the six studies that measured quality of life showed an improvement, but only one reached statistical significance. In addition, there were no changes in insulin dose, weight, or fasting plasma glucose level with the strategy.

"We need additional evidence to support our clinical practice, particularly in children and adolescents" where carbohydrate counting is widely used, said Ms. Bell, a dietician, diabetes educator, and Ph.D. candidate at the University of Sydney, Australia.

She suggested that clinicians and patients need realistic expectations of the improvements in glycemic control achievable with carbohydrate counting. Evidence continues to grow – including a study presented during the same session – on the effect fat and protein can have on prandial insulin requirements. A prescribed meal plan also has been shown to lower HbA_1c irrespective of whether it includes carbohydrate counting.

The increase in popularity of flexible insulin therapy and more flexible eating choices means there’s a risk that carbohydrate counting may lead to unhealthy eating and food beliefs, Ms. Bell said. Patients may exceed nutritional recommendations for fats and proteins in an effort to avoid carbohydrates or rely too heavily on packaged foods because the labels make carbohydrate counting easier.

Ultimately, the efficacy of carbohydrate counting is limited by the skills of the patient, she observed. Greater accuracy and precision in carbohydrate counting is associated with lower HbA_1c, however, the literature shows a wide variation in counting skills. Skills and compliance were not measured in the studies, and subgroup analyses of different methods of carbohydrate counting were not possible due to the lack of studies, she said.

Session moderator Dr. Anastassios Pittas, codirector of the Diabetes Center at Tufts Medical Center in Boston, applauded the investigators for looking at the evidence behind something clinicians routinely use and take for granted, but joined an audience member in pointing out that there was a lot of heterogeneity in the results, which can influence the strength of the conclusion.

"We have to somehow change the amount of insulin we deliver each meal every day, depending on the circumstances, and I think carb counting is a method that makes the most sense," he said in an interview. "To change my practice I will need to see evidence against the practice, not evidence in favor of what I’m already doing, and the evidence has to be pretty clear."

Data were presented during the same session on one such potential new tool, called the food insulin index. Ironically, it was this research that Ms. Bell also authored that prompted her to conduct the meta-analysis in the first place, she said in an interview. If carb counting was to serve as the control, it was necessary to know how effective it truly was.

What she found after screening 294 potentially relevant studies, was only six quality randomized controlled trials of at least 3 months’ duration. Study quality averaged 7.7 on a 13-point scale, with 13 indicating the least risk of bias.

Five trials were in adults (n = 563) and one in children, aged 8-13 years (n = 104), and all were in the outpatient clinical setting. Controls received usual care, general dietary advice or low glycemic index dietary advice, while a variety of carbohydrate counting methods was used including 10- and 15-gram carbohydrate exchanges.

The results of the meta-analysis chip away at a cornerstone of current practice and are likely to spark debate, particularly when presented by a lowly Ph.D. candidate, albeit in two sessions at the meeting including the presidential oral session.

"I expected people to be quite defensive of their current practice, but at the same time, it highlights the need for more evidence" and "the need to be more aware of the limitation in practice," she told this news agency. "It’s been common to say patients aren’t compliant, or aren’t doing it accurately, or they need to work on their skills, but not necessarily to acknowledge that there may be other factors in play here that aren’t related to their actual ability to count carbohydrates, but to the method itself."

Ms. Bell reported no relevant disclosures.

pwendling@frontlinemedcom.com

Counting carbs comes up short in type 1 diabetes