and obese participants without T2DM (weight-reducing phase only)
Summary of findings 1 presents the effects of low-carbohydrate weight-reducing diets compared to balanced-carbohydrate weight- reducing diets (weight-reducing phase only) in overweight and obese participants without T2DM on short-term (three to <12 months) and long-term (≥ 12 months) changes in body weight, as well as on the number of participants per group with weight loss of at least 5% over the long term, long-term change in DBP and LDL cholesterol, and short-term constipation.
Key details about studies in this comparison, including characteristics of the intervention, population, outcomes and method of synthesis, are included in the Overview of Synthesis and Included Studies (OSIS) table (Table 6).
Primary outcomes
Change in body weight (kg) at three to < 12 months
GRADE assessment suggests that low-carbohydrate weight- reducing diets probably result in little to no difference in change in body weight over three to 8.5 months compared to balanced-carbohydrate weight-reducing diets in overweight and obese participants without T2DM (moderate-certainty evidence, downgraded once for risk of bias).
Average weight reductions from baseline over the short term ranged from 12.18 kg to 4.45 kg with low-carbohydrate weight- reducing diets and from 11.34 kg to 2.3 kg with balanced- carbohydrate weight-reducing diets in the 37 trials that reported this outcome. The meta-analysis showed little to no difference in weight reduction on average between low-carbohydrate and
balanced-carbohydrate diet groups (MD -1.07 kg, (95% confidence interval (CI) -1.55 to -0.59, I2 = 51%, 3286 participants, 37 RCTs, moderate-certainty evidence, Analysis 1.1) over three to 8.5 months. This lack of a difference in effect was confirmed by sensitivity analyses, including only trials with 'some concerns' of bias overall (Analysis 1.6; no trials at overall low risk of bias), those at low risk of attrition bias (Analysis 1.7) and those without reported diet/food industry funding (Analysis 1.8).
Subgrouping by similarity of energy prescription (Analysis 1.2), extent of carbohydrate restriction (Analysis 1.3), diagnosed cardiovascular event or disease at baseline (Analysis 1.4) or gender (Analysis 1.5) did not suggest important clinical differences in average effects between subgroups, with mean differences in weight reduction in overweight and obese participants at three to <
12 months ranging between 0.25 kg and 2.71 kg across the various subgroups. Subgrouping by similarity of energy prescription and extent of carbohydrate restriction explained some heterogeneity.
Heterogeneity was reduced (I2 = 0%) in the subgroup including only trials with similar energy prescriptions/approaches to energy restriction in both diets, and in the subgroup including only trials investigating low-carbohydrate diets (> 50 g to 150 g per day or < 45% of total energy intake). We did not downgrade for inconsistency. The subgroup including only trials with no energy prescriptions or unrestricted energy prescriptions in both diets had small total numbers of participants (n < 400). This was also the case for the subgroup of trials only including men.
Nine trials reported this outcome in an unusable format (e.g. in figures or as percentage change in weight from baseline) and usable data were not provided when authors were contacted (Table 4).
The funnel plot (Figure 3) shows most trials had similar standard errors of intervention effect estimates limiting the ability to assess asymmetry. The mean difference in fixed-effect analysis, which gives less weight to small studies (-0.91 kg, 95% CI -1.21 to -0.62), is similar to the mean difference in random-effects meta-analysis (-1.07 kg, 95% CI -1.55 to -0.59, Analysis 1.1), which gives more weight to smaller studies, suggesting that any small-study effects have little effect on the intervention effect estimate.
Low-carbohydrate versus balanced-carbohydrate diets for reducing weight and cardiovascular risk (Review) 26
Figure 3. Funnel plot for change in weight at 3 to < 12 months (Analysis 1.1) in comparison 1
-10 -5 0 5 10
MD 0
2
4
6
8
10
SE(MD)
Thirteen of the 37 trials had follow-up at three months from baseline, nine at four months, 13 at six months, one at 26 weeks (about 6.5 months), and one trial at 34 weeks (about 8.5 months) Change in body weight (kg) at ≥ 12 months
GRADE assessment suggests that low-carbohydrate weight- reducing diets probably result in little to no difference to change in body weight over one to two years compared to balanced-carbohydrate weight-reducing diets in overweight and obese participants without T2DM (moderate-certainty evidence, downgraded once for risk of bias).
Long-term average weight reductions in the 14 trials that reported this outcome ranged from 13.1 kg to 2.46 kg in low-carbohydrate diet groups and from 11.6 kg to 1.7 kg in balanced-carbohydrate diet groups. The meta-analysis of the mean difference in weight reduction between the low-carbohydrate and balanced-carbohydrate diets demonstrated little to no difference
at one to two years (MD -0.93 kg, 95% CI -1.81 to -0.04, I2 = 40%, 1805 participants, 14 RCTs, moderate-certainty evidence, Analysis 1.9). Sensitivity analyses including only trials at low risk of attrition bias (Analysis 1.10) and trials without diet/food industry funding confirmed this lack of a difference in effect (Analysis 1.11). Other planned sensitivity analyses included fewer than two studies and thus were not performed.
Two trials reported this outcome in an unusable format (e.g. in figures or as percentage change in weight from baseline) and usable data were not provided when authors were contacted (Table 4). The funnel plot (Figure 4) suggests that smaller studies may be missing.
The mean difference in fixed-effect analysis, which gave less weight to small studies (-0.81 kg, 95% CI -1.43 to -0.19), was similar to the mean difference in the random-effects meta-analysis (-0.93 kg, 95%
CI -1.81 to -0.04, Analysis 1.9), which gave more weight to smaller studies, suggesting that any small-study effects had little effect on the intervention effect estimate.
Figure 4. Funnel plot for change in weight at ≥ 12 months (Analysis 1.9) in comparison 1
-10 -5 0 5 10
MD 0
1
2
3
4
5
SE(MD)
Ten of the 14 trials had follow-up at one year, one at 15 months, one at 74 weeks (about 18.5 months), with the two longest trials having follow-up at two years.
Number of participants per group with weight loss of at least 5% at three to < 12 months
The number of participants achieving at least 5% weight loss over the short term in the three trials reporting this outcome ranged from 14 to 41 with low-carbohydrate weight-reducing diets and from eight to 35 with balanced-carbohydrate weight-reducing diets. The meta-analysis of the risk ratio (RR) of the number of participants per group with weight loss of at least 5% at three to six months was 1.30 (95% CI 1.00 to 1.68, I2 = 47%, 161 participants, 3 RCTs, Analysis 1.12) when comparing low-carbohydrate and balanced-carbohydrate diet groups. Two of the three trials that reported this outcome had follow-up at three months and the other at six months.
Number of participants per group with weight loss of at least 5% at ≥ 12 months
GRADE assessment suggests that the evidence is very uncertain about whether there is a difference in the number of participants per group with weight loss of at least 5% at one year, when comparing low-carbohydrate weight-reducing diets with balanced-carbohydrate weight-reducing diets in overweight and obese participants without T2DM (very low-certainty evidence;
downgraded once for risk of bias and twice for imprecision).
Two studies reported on this outcome at one year (RR 1.11, 95%
CI 0.94 to 1.31; I2 = 17%; 137 participants, 2 RCTs, very low- certainty evidence, Analysis 1.13), with the number of participants achieving at least 5% weight loss ranging from 28 to 30 in the low-carbohydrate weight-reducing group and from 27 to 29 in the balanced-carbohydrate weight-reducing group. Another included trial also measured this outcome at two years but it was not reported in a usable format (Table 4).
Secondary outcomes
For comparison 1, no trials reported on the number of participants per group with reduction in BMI (kg/m2) of at least 5% at ≥ 12 months.
Change in BMI (kg/m2) at three to < 12 months
Twenty-one trials (n = 1517 participants) reported on short-term BMI change for comparison 1, with five of the 20 trials having follow- up at three months, five at four months and 11 trials at six months.
Across the trials, average reduction in BMI from baseline ranged from 4.7 kg/m2 to 1.8 kg/m2 with low-carbohydrate diets and from 3.7 kg/m2 to 0.98 kg/m2 with balanced-carbohydrate diet groups.
We did not pool these results into an overall effect estimate due to substantial heterogeneity (I2 = 53%, Tau2 = 0.18; Chi2 = 42.52, df = 20 (P = 0.002)) but rather presented the individual effect sizes between the low-carbohydrate and balanced-carbohydrate diets per study (Analysis 1.14).
Low-carbohydrate versus balanced-carbohydrate diets for reducing weight and cardiovascular risk (Review) 28
Change in BMI (kg/m2) at ≥ 12 months
Average reduction in BMI from baseline over the long term in the five trials reporting this outcome ranged from 5.2 kg/m2 to 1.65 kg/
m2 in low-carbohydrate diet groups and from 4.1 kg/m2 to 0.92 kg/
m2 in balanced-carbohydrate diet groups. When low-carbohydrate weight-reducing diets were compared to balanced-carbohydrate weight-reducing diets, the difference in BMI change on average at one to two years was -0.61 kg/m2 (95% CI -0.99 to -0.23, I2 = 0%, 490 participants, 5 RCTs, Analysis 1.15). Four trials had follow-up at one year and the remaining trial at two years.
Number of participants per group with reduction in BMI (kg/m2) of at least 5% at three to < 12 months
A single small study reported on this outcome at three months and the risk ratio was 1.57 (95% CI 0.79 to 3.12, 55 participants, 1 RCT, Analysis 1.16) with 14 participants in the low-carbohydrate weight-reducing group and eight participants in the balanced- carbohydrate weight-reducing group achieving at least 5%
reduction in BMI.
Change in DBP (mmHg) at ≥ 12 months
GRADE assessment suggests that low-carbohydrate weight- reducing diets probably result in little to no difference to change in DBP compared to balanced-carbohydrate weight-reducing diets over one to two years in overweight and obese participants without T2DM (moderate-certainty evidence, downgraded once for risk of bias).
Over one to two years, average change in DBP from baseline ranged from a 9 mmHg reduction to a 4 mmHg increase with low- carbohydrate diets and from an 11 mmHg reduction to 2.9 mmHg increase with balanced-carbohydrate diets across the 11 trials that reported this. The meta-analysis of the mean difference in change in DBP between the low-carbohydrate and balanced-carbohydrate diets demonstrated little to no difference at one to two years (MD -0.09 mmHg, 95% CI -1.29 to 1.12, I2 = 44%, 1419 participants, 11 RCTs, moderate-certainty evidence, Analysis 1.17). Seven of the 11 trials had follow-up at one year, one at 15 months, one at 18 months and the other two trials at two years.
Change in SBP (mmHg) at ≥ 12 months
Average change in SBP from baseline in the 11 trials reporting this outcome ranged from reductions of 13.8 mmHg to 0.2 mmHg in low-carbohydrate diet groups and from a reduction of 14.6 mmHg to an increase of 1.7 mmHg in balanced-carbohydrate diet groups over one to two years of follow-up. For this comparison, the mean difference in change in SBP at one to two years was -1.37 mmHg (95% CI -2.99 to 0.24, I2 = 34%, 1397 participants, 11 RCTs, Analysis 1.18). Seven of the 11 trials had follow-up at one year, one at 15 months, one at 18 months and the other two trials at two years.
Change in LDL cholesterol (mmol/L) at ≥ 12 months
GRADE assessment suggests that low-carbohydrate weight- reducing diets probably result in little to no difference in change in LDL cholesterol over one to two years compared to balanced-carbohydrate weight-reducing diets in overweight and obese participants without T2DM (moderate-certainty evidence, downgraded once for risk of bias).
Over one to two years, average change in LDL cholesterol from baseline across the 13 trials that reported this outcome ranged from reduction of 0.3 mmol/L to an increase of 0.6 mmol/L with low-carbohydrate diets and from a reduction of 0.31 mmol/L to an increase of 0.1 mmol/L with balanced-carbohydrate diets. The meta-analysis of the mean difference in change in LDL cholesterol between the low-carbohydrate and balanced-carbohydrate diets demonstrated little to no difference at one to two years (MD 0.04 mmol/L 95% CI -0.05 to 0.12, I2 = 33%, 1494 participants, 13 RCTs, moderate-certainty evidence, Analysis 1.19). Nine of the 13 trials had follow-up at one year, one at 15 months, one at 18 months and the other two trials at two years. One of the included trials measured this outcome but it was not reported in a usable format (Table 4).
Change in HDL cholesterol (mmol/L) at ≥ 12 months
Average change in HDL cholesterol from baseline over one to two years ranged from a reduction of 0.1 mmol/L to an increase of 0.3 mmol/L in low-carbohydrate diet groups and from a reduction of 0.21 mmol/L to an increase of 0.18 mmol/L in balanced-carbohydrate diet groups in the 13 trials that reported this outcome. When low-carbohydrate weight-reducing diets were compared to balanced-carbohydrate weight-reducing diets, the mean difference in change in HDL cholesterol at one to two years was 0.06 mmol/L (95% CI 0.02 to 0.10, I2 = 48%, 1519 participants, 13 RCTs, Analysis 1.20). Nine of the 13 trials had follow-up at one year, one at 15 months, one at 18 months and the other two trials at two years.
Change in non-HDL cholesterol (mmol/L) at ≥ 12 months
Two small studies reported on this outcome at one year and the mean difference in non-HDL cholesterol was 0.06 mmol/L (95%
CI -0.2 to 0.33, I2 = 47%, 225 participants, 2 RCTs, Analysis 1.21) with average change in non-HDL ranging from a reduction of 0.13 mmol/L to an increase of 0.24 mmol/L in low-carbohydrate weight- reducing groups and average reductions ranging from 0.1 mmol/L to 0.02 mmol/L in balanced-carbohydrate weight-reducing groups.
Change in total cholesterol (mmol/L) at ≥ 12 months
Over one to two years, average change in total cholesterol from baseline ranged from a reduction of 0.4 mmol/L to an increase of 0.7 mmol/L with low-carbohydrate diets and from a reduction of 0.42 mmol/L to an increase of 0.13 mmol/L with balanced-carbohydrate diets across the 11 trials that reported this. On average, there was no difference in change in total cholesterol at one to two years between the two types of diet (MD 0.01 mmol/L, 95% CI -0.1 to 0.12, I2 = 41%, 1291 participants, 11 RCTs, Analysis 1.22). Eight of the 11 trials had follow-up at one year, one at 15 months and the other two trials at two years.
Change in triglycerides (mmol/L) at ≥ 12 months
Across the 13 trials that reported on triglycerides over one to two years, average change from baseline ranged from reductions of 0.58 mmol/L to 0.1 mmol/L with low-carbohydrate diets and from a reduction of 0.44 mmol/L to an increase of 0.02 mmol/L with balanced-carbohydrate diets. For this comparison, the mean difference in change in triglycerides at one to two years was -0.11 mmol/L (95% CI -0.16 to -0.06, I2 = 0%, 1559 participants, 13 RCTs, Analysis 1.23). Nine of the 13 trials had follow-up at one year,
one at 15 months, one at 18 months and the other two trials at two years.
Participant-reported adverse effects
Participant-reported adverse effects were scantily reported by six included trials for comparison 1, with all outcomes except short- term constipation being reported only by single trials, and one of the trials reporting on most of these outcomes (Bazzano 2014).
All 19 dichotomous outcomes reported very few events, had small total sample sizes and had wide 95% CIs. The four continuous outcomes all had small total samples sizes, mean differences close to 0 and wide 95% CIs.
Gastrointestinal
GRADE assessment suggests that the evidence is very uncertain about whether there is a difference in the number of participants per group with constipation at three to six months when comparing low-carbohydrate weight-reducing diets with