This systematic review and meta-analysis explored type 2 diabetes remission, weight loss and HbA1c with low-carbohydrate diets (LCDs) or very-low-carbohydrate diets (VLCDs) compared with (mainly low-fat) control diets. Remission was defined as HbA1c <48 mmol/mol (6.5%) or fasting glucose <7 mmol/L, with or without glucose-lowering medication. LCDs were defined as <130 g of carbohydrates per day (<26% of calories as carbohydrates), while VLCDs were defined as <50 g of carbohydrates per day (<10% of calories). Previous type 2 diabetes remission studies have used either a very-low-calorie liquid diet (Lean et al, 2018) or a low carbohydrate diet (Hallberg et al, 2018). Previous meta-analyses have included a mixture of low- and moderate-carbohydrate diets containing carbohydrate intakes ranging from 20 g/day up to 45% of calories as carbohydrates (Snorgaard et al, 2017), and have demonstrated HbA1c reduction in those with type 2 diabetes but did not explore remission.
In total, 23 randomised controlled trials and 1357 people with type 2 diabetes were included in the present analysis. The authors included both published and unpublished studies, using LCDs or VLCDs of at least 12 weeks’ duration. Twelve of the studies explored the impact of VLCDs and 11 the effects of LCDs, and the two groups were combined in the pooled analyses. Eighteen of the 23 studies used low-fat diets as the control, meaning that these results cannot be extrapolated to situations where other diets, such as the Mediterranean diet, are being used. LCDs may reduce hunger and result in lower caloric intake, which may confound results.
Remission rates were reported in eight studies. When remission was defined as HbA1c <48 mmol/mol irrespective of diabetes medication, the pooled diets (LCDs and VLCDs) were 32% more likely to achieve remission than the control diets (remission rates 57% vs 31%). However, when remission was defined as HbA1c <48 mmol/mol without medication, the remission rates were much lower and not significantly greater in the LCDs than in controls. In the 14 studies that included people using insulin, remission rates with the LCDs were lower than amongst those that did not include people using insulin. In the 18 studies that reported on weight loss, the combined LCDs achieved 7.41 kg greater weight loss compared with control diets. In the 17 studies assessing HbA1c, the LCDs achieved greater reductions (5 mmol/mol [0.47%]) than the control diets at 6 months, but there was no difference at 12 months. Overall, weight loss at 6 months was better in people on LCDs compared with VLCDs, and this was attributed to poorer adherence with the more challenging VLCD eating pattern. Those who were adherent to VCLDs had greater weight loss than those who were not, and they lost more weight than those on LCDs.
Reductions in glucose-lowering medications, triglycerides and insulin resistance were secondary outcomes, and these were also significantly reduced with the LCDs and VLCDs. There was some evidence of worsening of quality of life and LDL-cholesterol levels in two studies, although these effects were not significant.
Any reduction in weight, HbA1c or medication requirements in people with type 2 diabetes, even if remission is not fully achieved, is likely to be beneficial. This meta-analysis suggests it may be better to recommend more sustainable LCDs rather than VLCDs if it seems likely that adherence will not be achievable with the stricter carbohydrate levels.
Click here to read the study in full.