Costus and Blood Sugar: The Real Human Evidence

Most herb pages on this site have to report that the marketed blood-sugar claim has never been tested in a human being at all. Crepe ginger is the opposite case, and it is worth saying plainly up front because the main Costus page currently understates it: Costus speciosus has real human evidence behind a genuine glucose-lowering effect. It is not a clinical trial. It is three independent hospital surveys in Sri Lanka, conducted by different research groups between 2014 and 2025, all reporting the same finding — people who eat this plant develop hypoglycaemia at rates their doctors documented. That is evidence the plant does something real to blood glucose in people, not just in rats. It is also, read correctly, a safety warning rather than grounds for using it as a treatment, and the difference between those two readings is the entire subject of this page.

Table of Contents

1. The Claim, and Why This Herb Is Different
2. Four Independent Mechanisms, Each With Its Own Study
3. Reading the Rodent Models Properly
4. The Ceiling Set by Acarbose
5. The Human Surveys: Real Hypoglycaemia in Real Users
6. What These Surveys Can and Cannot Show
7. The “Insulin Plant” Confusion, Revisited
8. A Nanoparticle Study, and a New Substitution to Watch For
9. What This Means If You Have Diabetes
10. Key Research Papers
11. Connections

The Claim, and Why This Herb Is Different

Crepe ginger is sold across South and Southeast Asia, in tea, capsule and raw-leaf form, on a simple promise: it lowers blood sugar. The usual pattern on this site, for a herb with a claim like this, is to report a body of animal and cell-culture work, note that nobody has ever tested it properly in a person, and leave the claim in the “absent, not refuted” category — plausible mechanism, no human data either way.

Crepe ginger breaks that pattern, and it breaks it from an unexpected direction. There is human data. It did not come from anyone testing the plant as a treatment. It came from hospital researchers in Sri Lanka investigating why their diabetic patients kept turning up hypoglycaemic, and finding crepe ginger — eaten as a household food, not a medicine — among the leading causes. Three independent teams found this, in three different patient samples, over more than a decade. That is a genuinely unusual amount of real-world human signal for a plant on this site, and the honest response is not to ignore it because it did not arrive as a randomised trial. It is to read it for what it actually shows.

Four Independent Mechanisms, Each With Its Own Study

Before the human data, the mechanistic case, because it is what makes the human survey findings plausible rather than coincidental. At least four separate lines of evidence, from different groups using different methods, converge on the same conclusion: this plant contains something that lowers blood glucose.

1. Whole rhizome extract in alloxan-induced diabetic rats. Bavarva and Bhatt (2008) gave ethanolic Costus speciosus root extract to alloxan-diabetic rats and reported both antihyperglycaemic and hypolipidaemic effects.
2. Whole rhizome extract in streptozotocin-induced diabetic rats, with mechanism. Ali, Almaghrabi and Afifi (2014) used a different chemical diabetes model — streptozotocin (STZ), which destroys insulin-producing cells by a different route than alloxan — and reported changes in the expression of genes governing insulin synthesis and glucose-catabolising enzymes.
3. Two isolated sesquiterpene lactones, tested separately, in STZ rats. Eliza and colleagues (2009) isolated costunolide from the rhizome and showed a normo-glycaemic and hypolipidaemic effect in STZ-diabetic rats at 20 mg/kg/day; a companion paper the same year isolated eremanthin from the same plant and reported an antidiabetic and antilipidaemic effect at the same dose. Two different named compounds, two separate papers, same model, same research group — this is a genuine attempt to move past “crude extract did something” toward identifying which molecule is responsible.
4. Alpha-glucosidase and glycation inhibition, in vitro, from the leaf. Perera, Premadasa and Poongunran (2016) tested Costus speciosus leaf extracts directly against the enzyme alpha-glucosidase — the intestinal enzyme that releases glucose from starch — and against protein glycation, and reported inhibition of both.
5. Isolated raffinose, with a named signalling pathway. Muthukumaran and colleagues (2018) isolated the oligosaccharide raffinose from the rhizome and reported that it attenuates lipid synthesis through the PPAR/SREBP-1c pathway and improves insulin sensitivity through PI3K/AKT signalling in laboratory models — a genuine attempt at a named molecular mechanism rather than a black-box “extract lowered glucose” result.
6. Independent cross-confirmation from an unrelated group. Lee and colleagues (2015), screening eighteen commonly consumed edible plants for antioxidant and starch-hydrolase inhibitory activity as candidate functional foods, found Costus speciosus among the top five for inhibiting both alpha-amylase and alpha-glucosidase — a different laboratory, a different country, a different assay panel, arriving at the same enzyme-inhibition conclusion as Perera’s dedicated study.

Six studies, at least five independent research groups, two different diabetes-induction chemistries, two different named compounds isolated and tested separately, and one outside cross-confirmation. For a plant with no controlled human trial, this is a genuinely substantial and convergent preclinical case — considerably more so than the single-study literatures this site more often reports.

Back to Table of Contents

Reading the Rodent Models Properly

Convergence is not the same as relevance, and the design of these studies needs the same scrutiny this site applies to any preclinical diabetes literature.

Streptozotocin and alloxan are both beta-cell toxins. They work by chemically destroying the insulin-producing cells of the pancreas, which models insulin-deficient diabetes — closer to type 1 diabetes, or to the late, beta-cell-exhausted stage of type 2, than to the insulin-resistant state most readers researching this page actually have. A compound that helps a rat with almost no functioning beta cells is not automatically doing anything for a person whose pancreas makes plenty of insulin that their tissues are not responding to properly. This is the same critique this site has applied to bael’s antidiabetic literature, and it applies with equal force here, to the same two model chemistries.

The dosing route rarely matches how anyone actually takes this plant. These studies administer ethanolic extract, isolated costunolide, or isolated eremanthin by gavage at fixed milligram-per-kilogram doses. Nobody eating crepe ginger as a household vegetable or drinking it as tea is dosing themselves this precisely, and the studies do not report what fraction of the isolated compound a realistic preparation would actually deliver.

None of the six studies used a genetic or diet-induced model of insulin resistance — the db/db mouse, the high-fat-diet mouse, or any comparable model that would speak more directly to type 2 diabetes as most patients experience it. This is a gap worth naming rather than glossing over: the mechanistic case here is real, but it has been built almost entirely in a model that answers a different physiological question than the one most readers are asking.

Back to Table of Contents

The Ceiling Set by Acarbose

Where a claim rests on a named enzyme, the strongest honest move is to find where that same mechanism was already developed into a licensed drug and see how well it performs there. Two of the six mechanistic studies above — Perera’s dedicated assay and Lee’s independent 18-plant screen — converge on alpha-glucosidase inhibition. That is not a novel mechanism. It is acarbose’s mechanism, a drug class in clinical use since the 1990s with a well-characterised, modest effect on HbA1c, and a well-known side-effect profile (flatulence, bloating, diarrhoea) that comes directly from undigested carbohydrate reaching the colon.

This matters for calibrating expectations. Even in the best case — even if crepe ginger’s alpha-glucosidase inhibition translated perfectly from a test tube into a person’s gut at a realistic dose, which has not been demonstrated — the outcome it could plausibly produce is already known, because a purified drug built on exactly this mechanism has been tested at scale. It is a real but modest HbA1c reduction, not a cure and not a substitute for metformin or insulin in anyone who needs them.

Back to Table of Contents

The Human Surveys: Real Hypoglycaemia in Real Users

This is the section that changes the picture, and it did not exist when the main Costus page was written to say “no human data.” It should have said something closer to “no controlled human trial, but a real and growing human safety signal.”

Medagama and colleagues, 2014 (n=254). A cross-sectional survey at a Sri Lankan university teaching hospital interviewed 254 type 2 diabetic patients about complementary and alternative medicine use and hypoglycaemic episodes. Crepe ginger was used by 92 patients (36% of the sample) — the third most common agent, behind bitter gourd and ivy gourd. The headline finding: “Ingestion of Costus speciosus (crepe ginger) was associated with a higher incidence of hypoglycaemia (P = 0.01).” That is a statistically significant association, specifically named to this species, in a sample of nearly a hundred users.

Dissanayake and colleagues, 2018 (n=1,000). A much larger observational study at a Sri Lankan out-patient diabetes clinic interviewed 1,000 patients about hypoglycaemic episodes in the preceding month. Of those, 16.9% attributed an episode to a non-prescribed native food item. Within that group, Costus speciosus accounted for 52.3% of the implicated cases — second only to bitter gourd (Momordica charantia, 54.5%) and ahead of Salacia prinoides, ivy gourd and Adenanthera pavonina. In a sample four times larger than the 2014 study, crepe ginger again lands near the top of the list of foods causing real hypoglycaemic episodes.

Medagama and Senadhira, 2015 (n=220). A separate cross-sectional survey of 220 diabetic patients using herbal remedies found crepe ginger the third most commonly used (25%), behind ivy gourd (32%) and bitter gourd (20%) — confirming it as a mainstream, everyday household remedy in this population rather than a fringe practice.

Thilakarathna and colleagues, 2025 (n=280). The most recent survey, at a different Sri Lankan tertiary centre, again found “canereed leaves (Costus speciosus)” among the most commonly used herbs for diabetes, alongside bitter gourd and fenugreek, more than a decade after the first survey — this is not a fading practice.

The evidence changed between 2014 and 2018, and it is worth showing that change explicitly. Medagama and Bandara’s own 2014 review, published the same year as their survey, concluded that “crepe ginger has not been studied adequately in humans to draw conclusions.” That was an accurate statement in 2014, before the larger Dissanayake study existed. Repeating that sentence today, as the main Costus page currently does in substance, is now out of date — the same research community answered its own question four years later, in a sample four times larger, and found the same signal again.

Back to Table of Contents

What These Surveys Can and Cannot Show

Read this evidence for exactly what it is, in both directions.

What it can show, and does: a real, replicated, statistically supported association between eating this specific plant and experiencing hypoglycaemia, across three independent research teams, three separate patient samples totalling nearly 1,500 people, over an eleven-year span. That consistency is the kind of thing that is hard to produce by chance or by a single group’s bias, and it lines up with the mechanistic literature above. Taken together, this is about as close to “this plant genuinely lowers blood glucose in people” as an uncontrolled observational literature can get.

What it cannot show:

• Dose. None of the surveys recorded how much crepe ginger each patient ate, in what preparation, or how often. “Associated with hypoglycaemia” does not translate into a safe amount or an effective amount.
• Causation in any individual case. Cross-sectional surveys report association, not mechanism-confirmed causation in a given patient. Medagama’s 2014 sample shows many patients used two, three or more complementary agents simultaneously (34% used two, 21% used three, 2.4% used more than three) alongside their prescribed diabetes medication — so a reported episode could reflect crepe ginger, another agent, a medication interaction, or a combination, and the survey design cannot fully separate them for any single patient.
• Efficacy as a treatment. Causing hypoglycaemia in some real-world users is evidence of a pharmacological effect. It is not the same as evidence that controlled, dosed use safely and reliably improves glycaemic control over time — that would require the randomised trial that has never been run, with a measured dose, a measured HbA1c outcome, and safety monitoring built in.
• Generalisability of the exact preparation. These are Sri Lankan populations using it as a household “native food,” not a standardised capsule or extract. A commercial supplement standardised to a diosgenin or costunolide percentage is a different product from the leaf someone’s grandmother added to a curry.

The correct one-sentence summary: this is strong evidence of a real effect and weak evidence of a safe or useful one — which is precisely the asymmetry a reader needs before deciding what to do with the information.

Back to Table of Contents

The “Insulin Plant” Confusion, Revisited

The main Costus page already documents the four-way “costus” naming problem in detail, and specifically flags that Chamaecostus cuspidatus (Costus igneus), the “insulin plant,” has its own separate human case reports of hypoglycaemia. It is worth stating precisely how this page’s findings relate to that one, because they are not the same finding wearing two names.

All three Sri Lankan surveys above name their agent specifically as Costus speciosus or its English/vernacular equivalents (crepe ginger, canereed) — not Costus igneus or “insulin plant,” which is more strongly associated with South Indian home cultivation than with Sri Lankan household use. So this is genuine, separate, species-specific human evidence for Costus speciosus itself, not a borrowed finding from its more famous cousin. That is a meaningful correction to the main page’s current framing, which implies the human hypoglycaemia signal belongs only to the “insulin plant.” It does not. Crepe ginger has its own.

One honest limit remains, and it sits a layer below what any citation check can verify: these are self-reported ethnobotanical surveys, where a patient told an interviewer they ate “crepe ginger” or its Sinhala equivalent. Whether every patient’s home garden plant was botanically the species the researchers assumed is not something a survey of this kind can confirm. This is a smaller and different uncertainty than the marketplace mislabeling the main page describes for internationally traded “costus” products, but it is not zero.

Back to Table of Contents

A Nanoparticle Study, and a New Substitution to Watch For

One more 2022 study deserves inclusion, with a caveat precise enough to be worth naming as its own category. Bakhshwin and colleagues tested gold nanoparticles synthesised using Costus speciosus extract in streptozotocin-diabetic rats, reporting decreased blood glucose, increased insulin and testosterone, reduced pro-inflammatory cytokine expression (IL-6, IL-1β, TNF-α) in prostate tissue, and preserved prostatic histology.

Read the method, not just the result. This is not a test of crepe ginger extract administered on its own. It is a test of a gold nanoparticle, for which the plant extract served as the “green synthesis” reducing and capping agent during nanoparticle manufacture — a standard nanotechnology technique with a fast-growing literature across many plants. Gold nanoparticles have their own documented biological activity, independent of whatever reducing agent was used to make them. Crediting the observed effect to “Costus speciosus” without separating the nanoparticle’s own contribution is a real and easy-to-miss error, worth naming precisely because this site’s doctrine has not previously had occasion to catalogue it: call it nanoparticle-vehicle substitution — crediting a plant for a biological effect when the plant extract’s documented role in the experiment was as a chemical reagent in manufacturing a different active agent, not as the active agent itself.

The testosterone increase in this study is not evidence for the diosgenin hormone-precursor myth the main page and the diosgenin page both address. The paper’s own proposed mechanism is indirect — better glycaemic control and reduced inflammatory cytokine damage to testicular and prostatic tissue, not a direct steroidogenic action — and it is further confounded by the unresolved nanoparticle-versus-phytochemical question above. Two layers of uncertainty sit between this result and any claim about crepe ginger boosting hormones, and neither should be quietly dropped when the finding gets summarised.

Back to Table of Contents

What This Means If You Have Diabetes

• If you have type 2 diabetes and eat crepe ginger regularly — as a vegetable, tea or household remedy — treat it as a real, active contributor to your glucose control, not a neutral food. The human survey evidence for that is stronger here than for almost any other herb on this site.
• If you also take insulin, a sulfonylurea (glibenclamide, gliclazide, glipizide) or a meglitinide, this is a genuine, underappreciated hypoglycaemia risk stacking on top of your prescribed treatment. Monitor more closely if you use both, and tell your prescriber that you eat it — the Sri Lankan literature suggests many patients do not volunteer this information.
• None of this supports using crepe ginger as a substitute for prescribed diabetes treatment. There is no dose-finding study, no controlled efficacy trial, and no long-term safety data. The evidence here establishes that the plant does something to blood glucose, not that using it deliberately is a safe or effective way to manage diabetes.
• Watch for hypoglycaemia symptoms — sweating, shakiness, confusion, palpitations — especially after a meal containing crepe ginger, and know that “I only ate a vegetable” is not a reason to dismiss them in this specific case.

Back to Table of Contents

Key Research Papers

Every citation below was checked against the live PubMed record before being written onto this page, using a title/abstract-scoped, species-locked search to confirm the paper concerns this plant specifically.

1. Bavarva JH, Bhatt HB. Antihyperglycemic and hypolipidemic effects of Costus speciosus in alloxan induced diabetic rats. Phytotherapy Research, 2008. — PubMed search
2. Ali HA, Almaghrabi OA, Afifi ME. Molecular mechanisms of anti-hyperglycemic effects of Costus speciosus extract in streptozotocin-induced diabetic rats. Saudi Medical Journal, 2014. — PubMed search
3. Eliza J, Daisy P, Ignacimuthu S, Duraipandiyan V. Normo-glycemic and hypolipidemic effect of costunolide isolated from Costus speciosus (Koen ex. Retz.) Sm. in streptozotocin-induced diabetic rats. Chemico-Biological Interactions, 2009. — PubMed search
4. Eliza J, Daisy P, Ignacimuthu S, Duraipandiyan V. Antidiabetic and antilipidemic effect of eremanthin from Costus speciosus (Koen.) Sm., in STZ-induced diabetic rats. Chemico-Biological Interactions, 2009. — PubMed search
5. Perera HK, Premadasa WKVK, Poongunran J. α-Glucosidase and glycation inhibitory effects of Costus speciosus leaves. BMC Complementary and Alternative Medicine, 2016. — PubMed search
6. Muthukumaran P, Thiyagarajan G, Arun Babu R, Lakshmi BS. Raffinose from Costus speciosus attenuates lipid synthesis through modulation of PPARs/SREBP1c and improves insulin sensitivity through PI3K/AKT. Chemico-Biological Interactions, 2018. — PubMed search
7. Lee YH, Choo C, Watawana MI, Jayawardena N, Waisundara VY. An appraisal of eighteen commonly consumed edible plants as functional food based on their antioxidant and starch hydrolase inhibitory activities. Journal of the Science of Food and Agriculture, 2015. — PubMed search
8. Gavillán-Suárez J, et al. Chemical profile and in vivo hypoglycemic effects of Syzygium jambos, Costus speciosus and Tapeinochilos ananassae plant extracts used as diabetes adjuvants in Puerto Rico. BMC Complementary and Alternative Medicine, 2015. — PubMed search
9. Shediwah FMH, et al. Antioxidant and antihyperlipidemic activity of Costus speciosus against atherogenic diet-induced hyperlipidemia in rabbits. Journal of Integrative Medicine, 2019. — PubMed search
10. Medagama AB, Bandara R, Abeysekera RA, Imbulpitiya B, Pushpakumari T. Use of Complementary and Alternative Medicines (CAMs) among type 2 diabetes patients in Sri Lanka: a cross sectional survey. BMC Complementary and Alternative Medicine, 2014. The n=254 survey finding a statistically significant hypoglycaemia association specifically with Costus speciosus. — PubMed search
11. Dissanayake HA, et al. Hypoglycaemia in diabetes: do we think enough of the cause? An observational study on prevalence and causes of hypoglycaemia among patients with type 2 diabetes in an out-patient setting in Sri Lanka. BMC Endocrine Disorders, 2018. The n=1,000 survey where Costus speciosus accounted for 52.3% of native-food-attributed hypoglycaemia. — PubMed search
12. Medagama AB, Senadhira D. Use of household ingredients as complementary medicines for perceived hypoglycemic benefit among Sri Lankan diabetic patients; a cross-sectional survey. Journal of Intercultural Ethnopharmacology, 2015. — PubMed search
13. Medagama AB, Bandara R. The use of complementary and alternative medicines (CAMs) in the treatment of diabetes mellitus: is continued use safe and effective? Nutrition Journal, 2014. The review whose “not studied adequately” conclusion the later surveys superseded. — PubMed search
14. Thilakarathna M, Appuhami K, Darshana N, Perera J. Complementary and alternative medicine use among patients with type-2 diabetes mellitus attending a suburban tertiary healthcare centre in Sri Lanka. BMC Complementary Medicine and Therapies, 2025. — PubMed search
15. Bakhshwin D, Faddladdeen KAJ, Ali SS, Alsaggaf SM, Ayuob NN. Nanoparticles of Costus speciosus Ameliorate Diabetes-Induced Structural Changes in Rat Prostate through Mediating the Pro-Inflammatory Cytokines IL 6, IL1β and TNF-α. Molecules, 2022. Read with the nanoparticle-vehicle caveat above. — PubMed search
16. Holman RR, et al. Acarbose and the ceiling on alpha-glucosidase-inhibitor efficacy in type 2 diabetes — topic search. — PubMed search

Back to Table of Contents

Connections

• All Herbs
• Costus Benefits Hub
• Diosgenin, Costunolide and the Industrial Chemistry — why the testosterone finding above is not a hormone-precursor claim.
• Anti-Inflammatory and Analgesic Evidence — costunolide and eremanthin’s other documented activity.
• Antimicrobial and Antiviral Evidence
• Costus (Cheilocostus speciosus) — the main page, with the “insulin plant” naming problem and the diosgenin myth in full.
• Type 2 Diabetes
• Hypoglycemia Awareness and Prevention — the actual documented harm from this plant.
• Bael: Blood Sugar and Metabolic Research — the same streptozotocin/alloxan model critique applied to a different plant.
• Fenugreek — the other major diosgenin-bearing plant with a genuine, separately documented human antidiabetic literature.
• Hemoglobin A1C — the outcome measure no crepe-ginger study has ever reported.
• Continuous Glucose Monitor (CGM)

Back to Table of Contents