Does a low-carb diet cause constipation?

In ordinary weight-loss diets, the trials do not show it clearly. A 2022 Cochrane review found constipation about as common on low-carbohydrate as on balanced diets in adults without diabetes, a risk ratio of 1.06 across 4 trials and 564 people over three to six months, on very low certainty evidence. Medical ketogenic diets are a different story. In people with drug-resistant epilepsy on the modified Atkins diet, constipation was the most common side effect, reported in 31 percent.

Unsettled. The weight-loss trials cannot settle it. Only a handful reported constipation at all, people reported it themselves, and the reviewers rated the evidence very low certainty. In adults with type 2 diabetes the estimate was 37 percent higher risk, a ratio of 1.37 in 177 people, and the interval runs from 0.86 to 2.18, so it does not show an effect either way.

What the trials found

The Cochrane review behind the lead pooled 61 trials of low-carbohydrate against balanced-carbohydrate diets for weight loss. Four of them, in 564 overweight adults without diabetes, counted constipation at three to six months. The ratio was 1.06, with an interval from 0.81 to 1.38. A single trial of 148 people that ran for a year gave 0.63, with an interval that also crosses 1.

Where the gut changes is fiber. In an 8-week randomized trial of 91 adults with obesity, a very-low-carbohydrate high-fat diet lowered stool output by 61 g while the high-carbohydrate high-fiber diet raised it by 21 g. Defecation frequency, butyrate and bifidobacteria also fell on the low-carb side. Gut symptoms did not differ between the groups, and stool form did not change. The comparison diet was rich in fiber, so the trial cannot pull apart less carbohydrate from less fiber.

The cohort data point the same way. In three US cohorts of 95,917 middle-aged and older adults, the fifth eating the most low-carb way had a constipation risk ratio of 1.03 against the fifth eating the least, which was not significant. After cereal fiber was taken into account it became 1.10, with an interval of 1.02 to 1.19. That is observational, and the score describes habitual eating rather than a ketogenic diet.

Fiber itself has trial evidence. Across 16 trials in 1,251 adults with chronic constipation, 66 per cent responded to fiber supplements against 41 percent on control, a risk ratio of 1.48. Psyllium and pectin worked, and the benefit appeared at doses above 10 g a day taken for four weeks or more.

A low-carb diet can be built with fiber in it. In a 4-week trial of 294 adults with irritable bowel syndrome, a fiber-optimized low-carbohydrate diet cut symptom scores by 50 points or more in 71 percent of people, against 58 percent on optimized medicine and 76 percent on a low FODMAP diet with traditional advice. The groups mixed IBS subtypes, so this does not test constipation on its own.

Established. Low fiber intake is a recognized cause of constipation. NIDDK lists not eating enough fiber among its causes, and a meta-analysis of 16 randomized trials found fiber supplements raised the response rate in chronic constipation.

Who should be careful

People on a medical ketogenic diet, many of them children with epilepsy. A 2020 Cochrane review of 13 trials in 932 people with drug-resistant epilepsy, 711 of them children, found vomiting, constipation and diarrhea were the most commonly reported side effects, in both the diet and usual-care groups, on low-certainty evidence. It gave no pooled ratio for constipation. A 2025 meta-analysis found constipation the most common side effect of the modified Atkins diet, at 31 percent, though the size of that pool was not available to us.

Not for everyone. Across 36 prospective studies of ketogenic diets of every kind, constipation was the single most reported adverse event, 23.9 percent of 743 recorded events, and stricter diets brought more events. Most of those studies had no control group, so they show how often it happens, not how much the diet adds. If you or your child are on a medical ketogenic diet, constipation is expected and is worth raising with the team running it.

For adults with type 2 diabetes the weight-loss estimate leaned toward more constipation, with an interval that includes no difference. That is a reason to watch for it, not a measured risk.

What expert bodies say

NIDDK, part of the US National Institutes of Health, lists not eating enough fiber among the causes of constipation, alongside not drinking enough liquids, low physical activity and changes in diet. Its patient page does not mention low-carb diets, and it was last reviewed in May 2018. We found no guideline that addresses constipation on low-carb diets directly.

For which foods carry fiber and how much, see our guide to fiber.

How we searched

What would change this answer

• A trial that randomizes people prone to constipation to a low-carb diet, with constipation as the main outcome. We did not find one, and no systematic review of low-carb diets takes constipation as its primary outcome.
• More weight-loss trials that record bowel habits in a standard way. The current answer rests on six trials that happened to report constipation, and people reported it themselves.
• A trial of a low-carb diet with and without added fiber in people without IBS. That would show whether the fiber gap explains the change in stool output.
• We found no registered trial on ClinicalTrials.gov looking at ketogenic diets and constipation.

Sources

1. ev-lccon-01 · Meta-analysis or systematic review · Cochrane systematic review and meta-analysis of RCTs (61 RCTs, 6925 participants overall) · n = 564
   The meta‐analysis of the risk ratio of the number of participants per group reporting constipation at three to six months was 1.06 (95% CI 0.81 to 1.38, I2 = 0%, 564 participants, 4 RCTs, very low‐certainty evidence, Analysis 1.24) when comparing the two diet groups.
   
   Who: overweight and obese adults without type 2 diabetes, weight-reducing phase only
   Effect: constipation RR 1.06 (0.81 to 1.38), I2 0%, 4 RCTs, 3-6 months; at 1 year RR 0.63 (0.32 to 1.25), 1 RCT, 148 people; very low certainty
   Certainty: Very low certainty: constipation was participant-reported and judged high risk of bias; few trials reported it.
   Cochrane Database Syst Rev, 2022 · checked 2026-10-07 · we read the fulltext
2. ev-lccon-02 · Meta-analysis or systematic review · Cochrane systematic review and meta-analysis of RCTs · n = 177
   The meta‐analysis of the risk ratio of the number of participants per group reporting constipation at six months was 1.37 (95% CI 0.86 to 2.18, I2 = 0%, 177 participants, 2 RCTs, very low‐certainty evidence, Analysis 3.19) when comparing low‐carbohydrate and balanced‐carbohydrate diet groups.
   
   Who: overweight and obese adults with type 2 diabetes, weight-reducing phase only
   Effect: constipation RR 1.37 (0.86 to 2.18), I2 0%, 2 RCTs, 6 months; very low certainty
   Certainty: Interval includes no effect; downgraded for risk of bias and twice for imprecision.
   Cochrane Database Syst Rev, 2022 · checked 2026-10-07 · we read the fulltext
3. ev-lccon-03 · Meta-analysis or systematic review · systematic review and meta-analysis of RCTs and prospective cohort studies · n = —
   Adverse events varied: constipation was most common with MAD (30.97%), whereas KD had higher rates of respiratory infections (42.77%, P < 0.0001 vs. MAD) and diarrhea (13.75% vs. 8.11%, P = 0.0017).
   
   Who: children and adults with drug-resistant epilepsy on ketogenic diet, modified Atkins diet or low glycemic index treatment
   Effect: constipation 30.97% with MAD; KD had more diarrhea (13.75% vs 8.11%, P=0.0017) and respiratory infections (42.77%)
   Certainty: Pooled RCT and cohort data, I2 above 75% in most analyses; full text no_pmc, so N for the adverse-event pool not seen.
   World J Pediatr, 2025 · checked 2026-10-07 · we read the abstract
4. ev-lccon-04 · Meta-analysis or systematic review · Cochrane systematic review and meta-analysis of RCTs and quasi-RCTs · n = 932
   The most commonly reported adverse effects were vomiting, constipation and diarrhoea for both the intervention and usual care group, but the true effect could be substantially different (low-certainty evidence).
   
   Who: 711 children (4 months to 18 years) and 221 adults with drug-resistant epilepsy
   Effect: vomiting, constipation and diarrhea most common in both diet and usual-care arms; low certainty; no pooled constipation figure
   Certainty: Adverse effects reported in both arms; no pooled risk ratio for constipation; full text no_fulltext.
   Cochrane Database Syst Rev, 2020 · checked 2026-10-07 · we read the abstract
5. ev-lccon-05 · Observational data · systematic review of prospective interventional studies, mostly single-arm, no meta-analysis · n = —
   Constipation was the most frequently reported individual AE within this category, accounting for 23.9% of all recorded AEs.
   
   Who: prospective interventional studies 2019-2024, all ages and conditions, classic KD, MAD, MCT KD, LGIT
   Effect: at least one adverse event in 43% of participants; gastrointestinal 40% of events; constipation 23.9% of 743 recorded events; gastrointestinal events in 36 of 42 interventions
   Certainty: Descriptive pooling without control groups, so it cannot say how much a ketogenic diet adds over baseline; level C for uncontrolled designs.
   BMC Nutr, 2026 · checked 2026-10-07 · we read the fulltext
6. ev-lccon-06 · Randomized controlled trial(s) · parallel randomized controlled trial, 8 weeks · n = 91
   Compared with the HC group, there were significant reductions in the LC group for faecal output (21 (sd 145) v. - 61 (sd 147) g), defecation frequency, faecal excretion and concentrations of butyrate ( - 0.5 (sd 10.4) v. - 3.9 (sd 9.7) mmol/l) and total SCFA (1.4 (sd 40.5) v. - 15.8 (sd 43.6) mmol/l) and counts of bifidobacteria (P < 0.05 time x diet interaction, for all).
   
   Who: overweight and obese adults, age 50.6 y, BMI 33.7; energy-restricted (about 30% deficit) isoenergetic diets
   Effect: fecal output -61 g vs +21 g; lower defecation frequency, butyrate, total SCFA and bifidobacteria (P<0.05); no difference in adverse GI symptoms; fecal form unchanged
   Certainty: Short trial; the control arm was also high-fiber, so the gap mixes carbohydrate and fiber; full text no_pmc.
   Br J Nutr, 2009 · checked 2026-10-07 · we read the abstract
7. ev-lccon-07 · Meta-analysis or systematic review · systematic review and meta-analysis of RCTs · n = 1251
   Overall, 311 of 473 (66%) participants responded to fiber treatment and 134 of 329 (41%) responded to control treatment [RR: 1.48 (95% CI: 1.17, 1.88; P = 0.001); I2 = 57% (P = 0.007)], with psyllium and pectin having significant effects.
   
   Who: adults with chronic constipation
   Effect: response RR 1.48 (1.17 to 1.88), I2 57%; stool frequency SMD 0.72 (0.36 to 1.08); psyllium and pectin effective; doses over 10 g/d and 4 weeks or more
   Certainty: Indirect for low-carb diets: shows fiber matters, which is what carbohydrate restriction often removes.
   Am J Clin Nutr, 2022 · checked 2026-10-07 · we read the abstract
8. ev-lccon-08 · Observational data · three prospective cohorts, food frequency questionnaires · n = 95917
   After additional adjustment for cereal fiber, the highest quintile of the LCD score was associated with a 10% (RR: 1.10, 95%CI: 1.02-1.19) increased risk for constipation, compared to the lowest quintile.
   
   Who: NHS, NHSII and HPFS participants, mean age 60-79 y; 7519 incident cases over 2-4 years
   Effect: top vs bottom quintile of low-carb score RR 1.03 (-3% to +11%) after confounders; RR 1.10 (1.02 to 1.19) after adjusting for cereal fiber
   Certainty: Observational; the low-carb score measures habitual eating, not a ketogenic diet.
   Gastroenterology, 2025 · checked 2026-10-07 · we read the fulltext
9. ev-lccon-09 · Randomized controlled trial(s) · single-center single-blind randomized controlled trial (CARIBS) · n = 294
   After 4 weeks, 73 (76%) of 96 participants in the LFTD diet group, 69 (71%) of 97 participants in the low-carbohydrate diet group, and 59 (58%) of 101 participants in the optimised medical treatment group had a reduction of 50 or more in IBS-SSS compared with baseline, with a significant difference between the groups (p=0·023).
   
   Who: adults with moderate-to-severe IBS (Rome IV), 82% women, mean age 38
   Effect: IBS-SSS reduction of 50 or more: low-carb 69/97 (71%), low FODMAP plus traditional advice 73/96 (76%), medical treatment 59/101 (58%); p=0.023
   Certainty: The low-carb arm was deliberately fiber-optimized, and IBS subtypes were mixed; it does not test constipation alone.
   Lancet Gastroenterol Hepatol, 2024 · checked 2026-10-07 · we read the abstract
10. ev-lccon-10 · Position of an expert body · agency patient guidance · n = —
    not eating enough fiber
    
    Who: people with constipation (US patient guidance)
    Effect: causes listed include not eating enough fiber, not drinking enough liquids, low physical activity, and changes in diet
    Certainty: Patient advice without its own analysis; page last reviewed May 2018; does not name low-carb diets.
    NIDDK, Symptoms & Causes of Constipation (NIH) · checked 2026-10-07 · we read the section