Cornstarch: the thickener that is also a night-time medicine

The best tested thing about cornstarch is a failure. Across three randomised trials covering 725 hypoglycaemic events in awake people with diabetes, dietary sugars relieved symptoms at 15 minutes less often than glucose tablets (RR 0.89, 95% CI 0.83 to 0.95). Cornstarch hydrolysate was one of eight sugar forms pooled there, so that estimate describes the class rather than cornstarch alone. Hours before a fall is where the tested use lies, and every trial of it enrolled people with diabetes or glycogen storage disease. Eaten raw at bedtime it peaked in the blood after 4.3 hours instead of 2.0 hours for a wholemeal bread snack, and the peak was lower, 2.9 against 5.2 mM. That comparison pooled two overnight studies in 24 patients with non-insulin dependent diabetes, neither of them randomised. In 12 adults with intensively treated type 1 diabetes, 0.3 g of uncooked cornstarch per kg at 23.00 cut self-estimated hypoglycaemia at 03.00 by 70 percent over four weeks, with HbA1c and fasting lipids unchanged. Those episodes were counted by the patients themselves rather than confirmed in a laboratory, in a trial of twelve people at one centre. None of this is about cooking. USDA holds a single record for cornstarch: 91.27 g of carbohydrate per 100 g, with 0.05 mg of copper and 2.8 mcg of selenium beside it. It is an SR Legacy row with no sampled lots behind it to give a range.

The numbers, per 100 g

How much is actually in it

• USDA lists cornstarch at 91.27 g of carbohydrate per 100 g, with only 0.05 mg of copper and 2.8 mcg of selenium alongside it. — USDA reference entry for cornstarch, FDC ID 169698, SR Legacy, group Cereal Grains and Pasta (n = 1 record) Evidence E source
  Carbohydrate 91.27 g per 100 g (33% DV), copper 0.05 mg (6% DV), selenium 2.8 mcg (5% DV)

What a real portion looks like

• The amount of cornstarch a patient actually takes has nothing to do with cooking: the median night-time dose in this group was 98.0 g (IQR 83.0 to 101.8) at a median age of 13 years. — 22 patients with hepatic glycogen storage disease at one Korean centre, median age 13 years (IQR 11-14), median weight 47.1 kg, baseline characteristics before switching products (n = 22) Evidence C source
  Median night-time cornstarch intake 98.0 g per dose (IQR 83.0 to 101.8)

What your body absorbs

• Uncooked cornstarch is what a slowly digested starch looks like in blood: eaten at bedtime it peaked after 4.3 hours instead of 2.0 hours for a bread snack, and the peak was lower (2.9 versus 5.2 mM). — two pooled overnight studies in comparable groups of patients with non-insulin dependent diabetes mellitus (14 and 10 patients), cornstarch or wholemeal bread at 0.6 g of carbohydrate per kg body weight (n = 24) Evidence C source
  Blood glucose peak 2.9 +/- 0.5 vs 5.2 +/- 0.6 mM (P = 0.01), time to peak 4.3 +/- 0.6 vs 2.0 +/- 0.0 h (P < 0.01)

What it does to your health

• In 12 adults with intensively treated type 1 diabetes, uncooked cornstarch at bedtime cut self-estimated 03.00 hypoglycaemia by 70% over four weeks without moving HbA1c or fasting lipids. — 12 intensively treated type 1 (IDDM) patients, double-blind randomised 4-week cross-over, 0.3 g of cornstarch per kg body weight at 23.00 h (n = 12) Evidence B source
  70% fewer self-estimated hypoglycaemic episodes at 03.00 h (P < 0.05), HbA1c and fasting lipids unchanged
• A single bedtime dose of 1.76 g/kg uncooked cornstarch held plasma glucose at or above 3.9 mmol/L for 7 hours in five of seven young adults with type I glycogen storage disease. Three of the seven still needed treatment for hypoglycaemia after 7 to 9.5 hours. — 7 patients with type I glycogen storage disease, mean age 19.5 y (range 18.8-21.7 y), single bedtime dose versus divided doses in random order on consecutive nights (n = 7) Evidence B source
  Single dose held glucose at or above 3.9 mmol/L for 7 h in 5 of 7, divided doses for 9 h in 6 of 7
• Pooling three non-randomised controlled trials in 26 people with glycogen storage disease type 1a, overnight uncooked cornstarch raised blood glucose by 0.62 mmol/L (95% CI 0.23 to 1.00) against continuous nocturnal dextrose feeding, and raised total cholesterol by 0.68 mmol/L as well. — five controlled trials with 49 participants with GSD-1a, follow-up 2 days to 14 years; the glucose comparison pooled 26 participants from three non-randomised trials (n = 26 pooled of 49 total) Evidence C source
  Blood glucose +0.62 mmol/L (95% CI 0.23 to 1.00), serum insulin +62.37 pmol/L (95% CI 32.19 to 92.55), total cholesterol +0.68 mmol/L (95% CI 0.17 to 1.20)
• In 22 Korean patients with hepatic glycogen storage disease, replacing night-time uncooked cornstarch with an extended-release cornstarch product moved no nocturnal continuous glucose monitoring metric over one month. — 22 patients with hepatic GSD at one Korean centre, median age 13 years, prospective before-and-after observational study with continuous glucose monitoring in both periods (n = 22) Evidence C source
  No significant difference in nocturnal mean glucose, coefficient of variation, time in range 70-180 mg/dL or time below 70 and 54 mg/dL

Safety, limits and interactions

• A 17-year-old who ate one box of cornstarch a day for six months reached intensive care with multisystem organ failure from propylene glycol toxicity, and needed fomepizole and continuous renal replacement therapy. — single case, 17-year-old female, one box of cornstarch daily for the previous 6 months (n = 1) Evidence D source
  Anion gap 18 mEq/L with no osmolar gap, negative toxicology screen, ICU admission with multisystem organ failure
• A non-randomised drug challenge in one 37-year-old woman showed that the cornstarch used as a tablet excipient, and not the antihistamine, caused a skin reaction that had been recorded as a desloratadine allergy. — single patient, 37-year-old woman with corn and maize allergy, open challenge with two desloratadine products, one containing cornstarch and one not (n = 1) Evidence C source
  Reaction reproduced with the cornstarch-containing tablet and absent with the cornstarch-free one, so the diagnosis changed from drug allergy to food allergy

What people believe that the data does not support

• For first-aid treatment of symptomatic hypoglycaemia in awake people with diabetes, dietary sugars including cornstarch hydrolysate relieved symptoms at 15 minutes less often than glucose tablets (RR 0.89, 95% CI 0.83 to 0.95). — awake patients with diabetes; three randomised trials totalling 502 hypoglycaemic events treated with dietary sugars and 223 with glucose tablets, plus one small non-randomised study (n = 725 hypoglycaemic events in the randomised trials) Evidence A source
  RR 0.89 (95% CI 0.83 to 0.95) for resolution of symptoms 15 min after treatment, favouring glucose tablets

Who this works differently for

• In 14 pre-school children with type 1 diabetes, 0.3 g/kg of cornstarch at bedtime raised mean 02.00 blood glucose by 2.2 mmol/L and cut readings below 5 mmol/L by 64%. — 14 pre-school children with type 1 diabetes, randomised double-blind placebo-controlled, cornstarch or placebo on five occasions each (n = 14) Evidence B source
  Mean nocturnal glucose at 02.00 h higher by 2.2 mmol/L, 64% fewer readings below 5 mmol/L
• In a planned pregnancy with glycogen storage disease type IIIa, nocturnal cornstarch covered basal carbohydrate needs, yet labour and the first 24 hours after birth still needed intravenous glucose at 3.5 mg/kg/min with hourly glucose checks. — single case, 25-year-old woman with GSD-IIIa diagnosed at 5 months of age, induction at 38 weeks and emergency caesarean section (n = 1) Evidence D source
  Nocturnal cornstarch plus a high protein daytime diet kept her well antenatally, while the peripartum period required 10% glucose intravenously at 3.5 mg/kg/min

The bottom line

The doses in this research are nothing like a recipe. Among 22 patients with hepatic glycogen storage disease at one Korean centre, median age 13 years, the median night-time dose was 98.0 g (IQR 83.0 to 101.8). That is a therapeutic amount in a rare disease. In seven young adults with type I glycogen storage disease, a single bedtime dose of 1.76 g/kg held plasma glucose at or above 3.9 mmol/L for 7 hours in five of them. Three of the seven needed treatment for hypoglycaemia after 7 to 9.5 hours. Pooling three non-randomised trials in 26 people with glycogen storage disease type 1a, overnight uncooked cornstarch raised blood glucose by 0.62 mmol/L (95% CI 0.23 to 1.00) against continuous nocturnal dextrose feeding. It raised total cholesterol by 0.68 mmol/L (95% CI 0.17 to 1.20) along with it, and because none of those three trials was randomised the pooled figure carries less weight than a meta-analysis usually suggests. Swapping the raw powder for an extended-release cornstarch product in the same 22 patients moved no nocturnal continuous glucose measure over one month, in a before-and-after study with no control period. Children and pregnancy appear here only inside the treatment literature. In 14 pre-school children with type 1 diabetes, the same bedtime 0.3 g/kg of uncooked cornstarch raised mean glucose at 02.00 by 2.2 mmol/L and cut readings below 5 mmol/L by 64 percent. That was a randomised placebo-controlled trial of fourteen children with no confidence intervals reported, and a reading under 5 mmol/L is a safety margin rather than a low. In one planned pregnancy with glycogen storage disease type IIIa, nocturnal cornstarch covered basal carbohydrate needs antenatally. Labour and the first 24 hours after birth still needed intravenous glucose at 3.5 mg/kg/min with hourly checks, in a single case describing what one team did. Two human reports mark the limits. A 17-year-old who ate one box of cornstarch a day for six months reached intensive care with multisystem organ failure from propylene glycol toxicity, and needed fomepizole and continuous renal replacement therapy. The report names propylene glycol toxicity and does not say how a daily box of cornstarch delivered it, so one case leaves the route of exposure unexplained. A drug challenge in one 37-year-old woman with corn allergy showed that the cornstarch used as a tablet excipient was causing a reaction filed as desloratadine allergy. The same drug in a cornstarch-free tablet caused nothing, so her diagnosis changed to food allergy. The same report notes cornstarch on medical gloves as another route of exposure. Corn allergy is rare, and each of these reports describes one patient. No source on this page sets an upper amount for the raw powder. Every bedtime dose above comes from a trial in people being treated for type 1 diabetes or for glycogen storage disease, and none of them was tested in anybody else.

Sources

1. cstr-r3-12 · food composition dataset
   Cornstarch, Carbohydrate, by difference, 91.27 g per 100 g, FDC ID 169698, SR Legacy
   USDA FoodData Central, SR Legacy, FDC ID 169698 · checked 2026-09-24
2. cstr-r3-11 · prospective observational study
   The median age was 13 (IQR: 11–14) years, and 14 patients (63.6%) were male...Prior to Glycosade initiation, the median nighttime cornstarch intake was 98.0 (IQR: 83.0–101.8) g per dose
   Nutrients, 2026 · checked 2026-09-24
3. cstr-r3-03 · pooled non-randomised comparison in patients
   Bedtime uncooked cornstarch ingestion led to a lower (2.9 +/- 0.5 vs. 5.2 +/- 0.6 mM, P = 0.01) and delayed (4.3 +/- 0.6 vs. 2.0 +/- 0.0 h, P < 0.01) BG peak, compared with a conventional snack, in NIDDM patients.
   J Intern Med, 1999 · checked 2026-09-24
4. cstr-r3-02 · randomised cross-over trial
   Four weeks of bedtime uncooked cornstarch supplement, as compared with placebo, led to a 70% reduction in the frequency of self-estimated hypoglycaemia at 03.00 h (P < 0.05), without affecting HbA1c or fasting lipids in IDDM patients.
   J Intern Med, 1999 · checked 2026-09-24
5. cstr-r3-05 · randomised cross-over clinical trial
   After a single dose, plasma glucose concentrations were > or = 3.9 mmol/L for 7 h in five of seven subjects; three subjects were treated for hypoglycemia after 7-9.5 h.
   Am J Clin Nutr, 1997 · checked 2026-09-24
6. cstr-r3-06 · meta-analysis of nonrandomised controlled trials
   Results from three nonrandomised controlled trials comparing uncooked cornstarch with continuous nocturnal feeding of dextrose were pooled in a meta-analysis based on a fixed-effect model...Short- to long-term overnight intermittent administration of uncooked cornstarch prevents nocturnal hypoglycaemia in GSD-1a children more effectively than continuous nocturnal feeding of dextrose
   J Hum Nutr Diet, 2013 · checked 2026-09-24
7. cstr-r3-10 · prospective observational study
   No significant differences were observed in the nocturnal CGM metrics between the treatment periods.
   Nutrients, 2026 · checked 2026-09-24
8. cstr-r3-07 · case report
   Twelve hours after ED arrival, she admitted to consuming one box of cornstarch daily for the past 6 months. She was admitted to the intensive care unit (ICU) with multisystem organ failure due to propylene glycol toxicity.
   J Am Coll Emerg Physicians Open, 2024 · checked 2026-09-24
9. cstr-r3-08 · non-randomised challenge in a single patient
   Finally, based on the results of a drug challenge test performed with two desloratadine-containing medications-desloratadine/aerius containing cornstarch as an excipient and desloratadine/lordestinenot containing cornstarch, the causative significance of corn was confirmed.
   Iran J Allergy Asthma Immunol, 2020 · checked 2026-09-24
10. cstr-r3-01 · meta-analysis of randomised controlled trials
    The dietary forms of sugar included sucrose, fructose, orange juice, jelly beans, Mentos, cornstarch hydrolysate, Skittles and milk. In the pooled analysis, patients treated with dietary sugars had a lower resolution of symptoms 15 min after treatment compared with glucose tablets (RR 0.89, 95% CI 0.83 to 0.95).
    Emerg Med J, 2017 · checked 2026-09-24
11. cstr-r3-04 · randomised placebo-controlled trial
    The mean nocturnal blood glucose concentration (at 02.00 h) was 2.2 mmol/L higher and the number of blood glucose concentrations below 5 mmol/L were reduced by 64% when cornstarch had been ingested at bedtime.
    Acta Paediatr, 1999 · checked 2026-09-24
12. cstr-r3-09 · case report
    Nocturnal cornstarch was provided to meet basal carbohydrate requirements and a high protein diet with regular carbohydrates was consumed throughout the day...She was managed in labour with glucose 10% IV at 3.5 mg/kg/min, hourly blood sugar level (BSL) monitoring and early epidural anaesthetic
    JIMD Rep, 2022 · checked 2026-09-24