Chapter  41:  Ayurvedic Interventions for Diabetes Mellitus: A Systematic Review

Overview

The objective of this evidence report was to conduct a search of the published literature on the use of Ayurvedic medicine/therapies for the treatment of health conditions and, on the basis of that search, to choose either a condition or a modality for a comprehensive review. A broad search of Ayurvedic medicine/ therapies showed that there was sufficient literature to support a systematic review of the use of Ayurvedic therapies for the treatment of diabetes. Diabetes is a common illness, and many traditional medical systems have developed strategies to treat this condition. The Ayurvedic therapy most commonly used to treat diabetes was herbal therapy, which therefore became the primary focus of this review. A small number of studies looking at diet therapy were also found.

Empirical evidence of efficacy for the Ayurvedic treatment of diabetes would be helpful to health care providers managing diabetic patients and would be useful in identifying areas for future research. The specific questions addressed in this project were:

1. What Ayurvedic therapies have been reported in the literature, for which conditions/body systems, and using what kinds of research designs?

2. What is the efficacy of Ayurvedic medicine/therapies for the treatment of diabetes?

Reporting the Evidence

An initial broad search of the literature found 2,565 titles, of which 1,214 were judged to represent Ayurvedic therapies that were neither veterinary nor agricultural in nature and thus potentially relevant to our review. To answer our first research question, these 1,214 titles were screened for subject, language, body system/disease state, study population, and study design. To answer our second research question, we further assessed the potentially relevant articles, including literature received from India. Fifty-four articles containing the results of 62 human clinical studies using Ayurvedic therapy for diabetes were identified. These studies were analyzed in detail to determine if evidence existed regarding the efficacy of Ayurvedic therapy for diabetes.

Methodology

A panel of technical experts representing diverse disciplines was established to advise us throughout the course of our research. A number of databases were searched: MEDLINE^®, HealthSTAR, EMBASE^®, Allied and Complementary Medicine^™, MANTIS^™, BIOSIS Previews^®, CAB HEALTH, and CINAHL^®. We used the MeSH terms “Ayurveda” or “Ayurvedic” combined with the botanical names of 16 herbs commonly used in Ayurvedic treatment. In addition, a strategy was developed to identify and retrieve literature from India. This involved using an abstracting service in India to identify potentially relevant literature.

There was no language restriction in the Western literature search. The Indian search was limited to studies published in English because we could not read studies in other languages and did not have the resources to have them translated. Additional articles were identified from supplemental searches that focused on the Ayurvedic herbs most often used for diabetes, on review articles, and on citations of articles. All titles, abstracts, and articles were reviewed by two reviewers, and all disagreements were resolved by consensus.

Data were collected using screening forms that we developed for this purpose. We analyzed the data regarding the general characteristics of the Ayurveda literature and used this information to select a topic for a focused review.

We then conducted a focused literature review using the articles we had identified from the Western literature and abstracts of articles published in India. We selected all articles identified as studying Ayurvedic therapies for diabetes in either the initial or the focused search. We identified 54 articles containing the results of 62 studies in this manner. Because of the heterogeneity of these studies, a meta-analysis was not possible. Approximately one-third of the studies were subjected to further analysis in which we calculated a common effect statistic. We also conducted a qualitative analysis on these studies.

Findings

• The most common conditions or body systems for which studies of Ayurvedic therapies have been published are:

  • Diabetes mellitus

  • Liver/hepatitis

  • Infectious diseases

  • Hypercholesterolemia

  • Central nervous system disorders (dementia/depression)

  • Cardiovascular diseases

• The Ayurvedic therapy that was the most common subject of published studies was herbal therapy. Almost no studies were found on any other Ayurvedic modalities.

• No studies were found that tested Ayurveda as a whole system or that tested multiple modalities for the same disease state at the same time.

• A significant body of literature in English exists in India; it can be identified, and a large portion of the studies can be obtained with effort. However, even after extensive efforts, a handful of English-language studies in India could not be found. Studies in non-English languages also exist but were not reviewed.

• Significant heterogeneity exists in the studies identified. More than 45 single herbs or combination herbal therapies were tested. The study designs likewise were varied. The 54 articles reported the results of 62 studies. Of these, 7 were randomized controlled trials (RCTs), and 10 were controlled clinical trials (CCTs). There were 38 case series, the most frequently used clinical design, and 7 cohort studies.

• The most common single herbs studied were Gymnema sylvestre, Coccinia indica, fenugreek (Trigonella foenum-graecum), and Eugenia jambolana. A number of herbal formulas were tested, but Ayush-82 and D-400 were the two most often studied.

• There is evidence to suggest that the single herbs Coccinia indica, holy basil, fenugreek, and Gymnema sylvestre and the herbal formulas Ayush-82 and D-400 have a glucose-lowering effect and deserve further study.

• Evidence of effectiveness of several other herbs is less extensive (C. tamala, Eugenia jambolana, and Momordica charantia).

Future Research

Our review has identified a number of interesting areas for future research. Basic scientific studies of Ayurvedic medicine have not been rigorously pursued. There are currently few RCTs and CCTs in the literature, which hinders the assessment of efficacy. Future trials need to enroll an adequate number of subjects. Interventions should be compared to placebo preparations, and care should be taken to construct placebos that cannot be distinguished from the trial drug.

The clinical trials of Ayurvedic therapies for diabetes need to be better reported. The method of patient selection and assignment to arms needs to be better described, and the reporting of results should follow good statistical practice. In addition, the trials need to be of sufficient length to determine a relevant clinical effect.

It would also be useful to investigate the efficacy of single-herb therapies versus the relatively complex Ayurvedic formulas used. It is not clear from the currently available literature if the formulas provide any additional benefit over single-herb therapies.

Field studies to determine how Ayurvedic medicine is used in real-life clinical practice should be conducted. The interaction between botanicals and other Ayurvedic modalities (yoga, for example) on diabetes could be assessed. The relationship between Ayurvedic diagnosis and Western diagnosis needs to be established.

Furthermore, trials incorporating more sophisticated diabetes research should be conducted. Studies evaluating the impact of Ayurveda using more current methods of assessing diabetes are also needed.

Purpose

This evidence report details the methodology, results, and conclusions of a literature review on the use of traditional Indian medical practices known as Ayurveda for the treatment of diabetes mellitus. The specific Ayurvedic modality examined was mainly herbal therapies, which are augmented at times with mineral compounds. The purpose of this work was to identify those Ayurvedic herbal therapies that have empirical support of efficacy for diabetes mellitus. Such information can be used to help health care providers counsel patients who use these therapies and to identify future research needs.

During the course of the project, we were informed that significant literature relevant to our search existed in the Indian literature. Thus, a secondary goal of the project developed: to assess the extent, nature, and accessibility of literature from India. We developed a methodology of identifying appropriate studies and of assessing literature from India. This allowed us to see if the study of Ayurveda in India was significantly different from study in the West.

Scope of Work

The work involved an initial survey of the medical literature on Ayurveda indexed in computerized databases in the West. From this initial review, we were able to identify the range of conditions that have been studied with Ayurveda, the modalities used, the probable study designs, and whether the studies were done on animals or humans. In deciding the scope of our review, we were confronted with the question of whether we should review literature that reports evaluations of Ayurveda as a whole system or only literature that reports evaluations on specific parts of Ayurveda. Very few, if any, of the articles identified in our preliminary searches looked at Ayurveda applied as a whole system. The majority of articles focused mainly on drug therapy, which was almost entirely herbal.

After discussions with our expert advisory panel and with the agencies involved with the project—the National Center for Complementary and Alternative Medicine (NCCAM) and the Agency for Healthcare Research and Quality (AHRQ)—we narrowed the focus of our study to the use of Ayurvedic therapies for the treatment of diabetes mellitus. This decision was based on the expectation that the quantity and quality of the evidence would be sufficient to support a systematic review. For the purpose of this review, our definition of Ayurveda encompasses not only the original Hindu form of this traditional medical system but also variations practiced in India, namely, Siddha, Unani Tibb, and yoga. The Siddha system, a variant of Ayurveda, is prevalent primarily in Tamil Nadu, a state in southern India, and it closely resembles the Ayurveda system. The Unani Tibb system, originally derived from Greek and Arabic medicine, also has much in common with Ayurveda except that its concepts of disease and diagnosis are similar to those of the early stages of modern medicine. Yoga is a distinct traditional system; the therapeutic aspects of yoga are merely one facet of Ayurveda holistic doctrine of the evolution of the human personality (Lodha and Bagga, 2000). In addition to Ayurveda, we also focused separately on the use of major Ayurvedic herbs whether they were studied by Indian researchers or not. Since herbal therapy is overwhelmingly the best studied aspect of Ayurveda, including these herbs was likely to improve the completeness of the search.

Additionally, we decided to search for literature that was published on the Indian subcontinent and thus not accessible in Western databases. Our expert panel felt that a substantial body of literature existed and that this literature was different quantitatively and qualitatively from Western literature and should be included in any systematic review of Ayurveda.

Ayurveda and the Ayurvedic View of the Individual and Health

Ayurveda (Sanskrit for “knowledge of life” or “knowledge of longevity”) is a comprehensive system of traditional health care that emphasizes the relationship among body, mind, and spirit. Originating in India roughly three thousand years ago, Ayurveda seeks to restore an individual's innate harmony. Primary Ayurvedic treatments include diet, exercise, meditation, herbs, massage, exposure to sunlight, controlled breathing, and detoxification treatments. This form of alternative medicine has spread beyond India's borders to include the rest of the Indian subcontinent, Sri Lanka, Malaysia, Mauritius, South Africa, Japan, Russia, Europe, and North America.

The historical roots of Ayurvedic teachings are shrouded as much in myth as in tradition. Custom alleges that the Vedas—four sacred books of mythico-religious hymns and medical lore that came from India's ancient Indus River civilization—are the original material from which Ayurveda developed. Diabetes is mentioned in one of these books, the Atharvaveda, a compilation of materials that date from around 1500 to 1000 B.C. Ayurveda itself reached a golden age roughly between 800 B.C. and A.D. 1000, during which time three important Ayurvedic treatises appeared: the Caraka Samhita on medicine, attributed to the physician Caraka; the Susruta Samhita on surgery, attributed to the physician Susruta; and the Ashtanga Hridaya Samhita, attributed to Vagbhata, who incorporated the teachings of the two earlier works. These classic texts are still printed today in India and are part of the training of Ayurvedic physicians.

Ayurveda is a rich and sophisticated form of medicine that is both like and unlike Western medicine. Appendix A gives a summary of Ayurveda's history, beliefs, and practices in more detail. Some of the basic Ayurvedic concepts are reviewed here in preparation for the discussion in the next section on the Ayurvedic diagnosis and treatment of diabetes.

Ayurveda has traditionally considered human beings to be a microcosm of nature. Humans and nature are both believed to consist of five basic elements: ether (space), air, fire, water, and earth. In humans these elements combine with each other and manifest themselves in the human body as three humors or doshas known as vata, pitta, and kapha. The doshas govern all biological, psychological, and pathophysiological functions in the body, mind, and consciousness. They are fundamental to human health, and an imbalance of the doshas brings on illness. Consequently, Ayurveda seeks to bring the doshas back into harmony. Balanced doshas, as well as good quality tissues (dhatus) and proper digestion and elimination of excretions (malas), are considered essential in Ayurveda for maintaining health.

Humans are endowed at birth with one of seven different body types, depending on which dosha or combination of the three basic doshas dominate. Body types can be dominated by a single humor (vata, pitta, or kapha), or they can be dominated by combinations: vata-kapha (when vata and kapha are present in almost equal amounts); vata-pitta; pitta-kapha; and vata-pitta-kapha (when all three doshas are present in almost equal amounts). If one considers that the two-dosha admixtures are different depending on which of the two doshas is dominant, then in fact there would be 10 different body types. A person's dosha body type is expressed both physically and emotionally; for example, a person with a vata-dominant body type will have a thin frame and an insecure temperament. The Ayurvedic physician takes body type and imbalances among the doshas into consideration when treating a patient.

Ayurvedic Diagnosis and Treatment of Diabetes

Etiology and Pathogenesis of Madhumeha

The etiology of madhumeha is multifactorial in the Ayurvedic system. Causes may be traced to tendencies inherited at birth or to derangements acquired afterwards. Specifically, if the three major doshas become imbalanced, this may lead to madhumeha. Likewise, disorder may arise directly from abnormalities in the tissues of the body, such as fat (medas), muscle (mansa), and muscle fat (vasa), or from the action of imbalanced doshas on these tissues. External causes felt to contribute excesses of the doshas include excessive sleep, excessive appetite (especially for sweet food), lack of physical exercise, excess sexual intercourse, suppression of natural urges, uneven body postures, and other behaviors.

Madhumeha is classified in the group of urination disorders known as prameha. Based on the main imbalance of the bodily humor (dosha) involved, these disorders are further classified as kaphaja, pittaja, and vataja (from the three doshas: kapha, pitta, and vata, respectively). There are at least 20 types of urinary disorders described in traditional texts; madhumeha falls under the category of vataja pramehas.

Even as early as Susruta's time, the varied presentation of diabetes was appreciated. Two types of vataja pramehas are described for diabetes. The first, called sahaja, is thought to be due to a defect in genetic substance, either in the mother or father. This form corresponds to juvenile-onset diabetes, and such patients are often described as thin and are thought to have more serious disease. The second type, apathyanimittaja, is believed to be acquired later in life due to excessive habits, such as overindulgence in food or sweets. This corresponds to adult-onset diabetes, and the patients are described as obese.

Caraka wrote that all pramehas (urinary disorders) start with a derangement of the bodily humor kapha. The vitiated kapha spreads throughout the body and mixes with fat (medas) because fat has properties similar to those of kapha. The affected body fluids are passed in the urine, but they block the openings of the urinary tubules coming out of the bladder. This is believed to be the cause of the frequent urination observed in madhumeha. Even though impaired kapha plays a dominant role in diabetes, the other two bodily humors—vata and pitta—are also important in the development of different types of the disease.

Symptoms

In their writings, Vagbhata and Susruta describe the symptoms of diabetes as a honey-like sweetness of urine as well as thirst, polyphagia, lassitude, tiredness, obesity, looseness of limbs, non-relishing of food, burning sensation of the skin, epileptic fits, insomnia, numbness of body, and constipation. Caraka wrote that chronic pramehas, of which madhumeha is one type, give rise to the following symptoms: oppressive feeling of the heart, anger, desire for foods of all different tastes, insomnia, and boils and abscesses, thus anticipating many of the sequelae of diabetes mellitus observed today.

Diagnosis

Table 1. Eight-point diagnosis for diabetes

Table 1. Eight-point diagnosis for diabetes

Ayurvedic term	Diagnostic technique	Characteristic in diabetes
Nadi pariksha	Pulse diagnosis	Depends on dominant dosha
Mutra pariksha	Urine examination	Sweet
Vata	Nervous system assessment	Vata usually depleted. If in excess, poor prognostic sign
Pitta	Assessment of digestive fire and metabolic secretions	May be elevated
Kapha	Mucous and mucoid secretions assessment	Generally increased; may be highly exaggerated
Mala pariksha	Stool examination	Depends on dominant dosha
Jihva pariksha	Tongue examination	Depends on dominant dosha
Shabda pariksha	Examination of body sounds	Depends on dominant dosha

In Ayurveda, diagnosis is based more on symptoms than on any laboratory results. The sweetness of urine and urine being assailed by a swarm of flies or ants is enough to make the diagnosis of madhumeha according to some authorities. Specifically, disease is diagnosed through a clinical examination called the Eight-Point Diagnosis or astha sthana pariksha. It includes an assessment of the state of a patient's doshas as well as various physical signs. This Eight-Point Diagnosis applied to diabetes is shown in Table 1. The observation of oversweet urine is correlated with the constitution of the affected individual and with the premonitory symptoms mentioned above. From this, the diagnosis of mahumeha of the kaphaja, pittaja, or vataja type is made.

Prognosis

Ayurveda holds that, if there is ulvanata (predominance) of pitta or kapha in persons suffering from madhumeha, their prognosis is better. If there is an excess of vata relative to kapha and pitta, then madhumeha is said to be incurable.

Concordance Between Ancient and Modern Descriptions of Madhumeha (Diabetes)

Several published studies from the Indian literature attempted to demonstrate a correlation between the classical Ayurvedic descriptions of the etiology, classification, pathogenesis, and treatment of madhumeha with more modern, scientific knowledge of diabetes. These studies, although interesting, have serious methodological flaws. The studies are summarized below.

Bharti and Singh (1995) surveyed 50 cases of diabetes and reported that there was a relatively higher incidence of patients with kapha-dominant and vata-kapha-dominant body types. Unfortunately, the authors did not show that the patients were representative of the general diabetic population, and they made no attempt to control for the subjective nature of the dosha assessments by having independent examiners reconcile their assessments.

Table 2. Body type relationship to blood chemistry (more...)

Table 2. Body type relationship to blood chemistry in diabetic patients

	Vata-dominant	Pitta-dominant	Kapha-dominant
Fasting blood sugar (mg/dl)	256.35 ± 126.99	189.09 ± 94.49	173.77 ± 68.43
Post-prandial sugar (mg/dl)	393.45 ± 175.46	304.83 ± 121.47	299.37 ± 103.92
Plasma insulin (μU)	8.25 ± 5.76	9.15 ± 5.22	18.44 ± 10.70

Source: Chandola, Tripathi, Udupa (1994).

According to Ayurvedic texts, the process of wear and tear due to diabetes in patients with a kapha-dominant constitution is supposed to be slower, and the management of the disease is relatively easier. Conversely, patients with pitta-dominant and vata-dominant constitutions should show more rapid progress of the disease. In order to test this theory, Chandola, Tripathi, and Udupa (1994) conducted a case series of patients (n=40, age 40–60 years) with maturity-onset diabetes. These patients did not exhibit a kapha-dominant body type as was seen in the previous study. Instead, 15 patients had a vata-dominant constitution, 16 had a pitta-dominant constitution, and 9 had a kapha-dominant constitution. Body surface area, however, did relate to the expected body habitus with maximum area corresponding to a kapha-dominant constitution (1.625 ± 0.119 m²) and minimum area to a vata-dominant constitution (1.46 ± 0.15 m²). These are not unexpected results given the fact that body size is a component of the definition of the doshas. Vata-dominant patients also had the highest average levels of fasting blood sugar (FBS) and post-prandial blood sugar (PPBS), as well the maximum increase in blood sugar over the duration of illness. These results are summarized in Table 2.

The authors concluded that these results lend objective credence to the classical descriptions of madhumeha, especially regarding severity of the disease. However, the authors did not show that the patients were representative of the general diabetic population, nor did they have the dosha assessments conducted by at least a second independent examiner and any differences reconciled. This study did not include comparison arms of normal patients.

Table 3. Physical characteristics and blood chemistry (more...)

Table 3. Physical characteristics and blood chemistry of diabetic patients

Ayurvedic diagnostic group	Body build	Fasting glucose (mg/dl)	Post-prandial glucose (mg/dl)
Kapha dominant	Obese	100–120	180–225
Pitta dominant	Average	120–200	225–300
Vata dominat	Thin	>200	>300

Source: Kar, Upadhyay, and Ojha (1997).

Table 4. Glucose tolerance test results by Ayurvedic (more...)

Table 4. Glucose tolerance test results by Ayurvedic diagnostic group

Ayurvedic diagnostic group	Fasting blood sugar (mg/dl)	Post-prandial blood sugar (mg/dl)	Serum insulin levels (μU/ml)
Kapha dominant	123.38 ± 14.042	201.20 ± 27.95	49.21 ± 26.89
Pitta dominat	175.00 ± 33.02	275.41 ± 27.33	18.39 ± 3.42
Vata dominant	248.50 ± 25.66	332.60 ± 44.14	9.51 ± 1.672

Source: Kar, Upadhyay, and Ojha (1997).

The most interesting associations between Ayurvedic and Western descriptions of diabetes are seen in the study by Kar, Upadhyay, and Ojha (1997) of 40 type 2 diabetic patients who were classified into arms based on body habitus. Their physical characteristics and blood chemistry are summarized in Table 3. The results of a glucose tolerance test (GTT) performed on all subjects are shown in Table 4.

According to this study, the patients with kaphaja prameha showed relative hyperinsulinemia and would be the equivalent of the obese insulin-resistant diabetic patients seen in Western medicine. The patients with pittaja prameha had intermediate insulin levels, and the patients with vataja prameha had the least endogenous insulin. Ayurveda would predict a better prognosis for patients whose diabetes fell in the kaphaja prameha and pittaja prameha groups than for patients in the vataja prameha group. This prognosis is reflected in ancient texts that state that kaphaja is sadhya (curable), pittaja is yapya (palliatable), and vataja is asadhya (incurable).

There does seem to be some physiologic correlation with the various dosha types in diabetes that may support the observations made in the ancient Indian texts. It would be very interesting to have a scientifically rigorous study to see which associations remain once observer bias has been eliminated.

Herbal Therapies for Diabetes

Ethnobotanical studies of traditional herbal remedies used for diabetes around the world have identified more than 1,200 species of plants with hypoglycemic activity. These plants are broadly distributed throughout 725 different genera. The large number of traditional remedies dedicated to diabetes likely reflects the relative ease of diagnosing this disease—sugar in the urine can be determined even in technology-poor societies. This traditional knowledge, derived empirically, is supported by scientific testing. When traditional diabetic remedies have been tested for antidiabetic activity, plants with a traditional indication for diabetes are more likely than randomly selected plants to show activity in standard hypoglycemic assays (81 percent vs. 47 percent) (Marles and Farnsworth, 1995).

The pharmacopoeia of India is especially rich in herbal treatments for diabetes. Eighty-five percent of the 20 antidiabetic plants most widely used around the world are prescribed in India (Marles and Farnsworth, 1995). Basic scientific data now supplement traditional lore for the most commonly used Ayurvedic herbs.

Ayurvedic herbal drugs for diabetes are selected on the principles of rasa (taste), guna (physicochemical properties), veerya (potency), vipaka (post-digestive effect), and prabhava (unique action). These principles are described in Appendix A under “Properties of Ayurvedic Herbs.” Each of these principles is felt to have specific effects on the doshas and functions of the body, which is how they exert their therapeutic effects. Additional background information on Ayurvedic pharmacology is found in Mishra, Singh, and Dagenais (2001a).

Common Herbs Used in Ayurveda To Treat Diabetes

Table 5. Ayurvedic characteristics of herbs commonly (more...)

Table 5. Ayurvedic characteristics of herbs commonly used to treat diabetes

Herb	Taste (rasa)	Increases (aggravates)	Decreases (pacifies)
Gymnema	Kasaya (astringent)		Pitta, Kapha
Momordica	Tikta (bitter)		Kapha, Pitta
Trigonella (fenugreek)	Tikta (bitter) Madhura (sweet)	Pitta	Kapha, Vata
Coccinia indica	Kasaya (astringent) Tikta (bitter)	Vata, Pitta
Pterocarpus	Kasaya (astringent)	Vata

Source: Kapoor, 1990; Dash, 1987; Mishra, Singh, and Dagenais, 2001b.

A few of the herbs commonly used by Ayurveda practitioners to treat diabetes are described below. The Ayurvedic characteristics of these plants are summarized in Table 5.

Gymnema sylvestre R. Br. This plant, of the family Asclepiadaceae, is a woody, climbing vine common in central and southern India. It has been used to treat diabetes for more than two thousand years. Its Hindu common name, gurmara or gurmar, means “sugar destroyer.” Traditionally, the leaves are either chewed whole, taken as a powder, or drunk as a water decoction (Kapoor, 1990; Nadkarni and Nadkarni, 1976; Jain and DeFilipps, 1991; Dash, 1987).

A number of constituents have been isolated from this plant since the first chemical studies were done at the end of the nineteenth century. Most important for the treatment of diabetes are the gymnemic acids, which were reportedly first isolated by Hooper in 1889 (Nadkarni and Nadkarni, 1976). Regan began modern pharmacologic studies in 1930 (Nadkarni and Nadkarni,1976; Lawrence Review, 1993). The best studied extract of Gymnema, GS4, contains a group of at least 15 triterpene sapinoids (the gymnemic acids) plus a polypeptide, gurmarin (Alternative Med Rev, 1999; Bone, 1996).

The pharmacologic actions of Gymnema have been studied in a number of animal models. The plant has been able to normalize blood sugar in animals that were treated with agents that destroy beta-cell function but not in animals that have been pancreatectomized (Prakash, Mathur, and Mathur, 1986). Conversely, Gymnema has shown little effect on the blood sugar of normal animals (Prakash, Mathur, and Mathur, 1986). This observation has not been confirmed by other investigators. Chattopadhyay, Medda, Das, et al. (1993) have shown that Gymnema in a water based extraction increased the effect of exogenous insulin in normal rats, increased the tolerance for glucose in normal and hyperglycemic rats, and decreased the plasma glucose in mildly diabetic rats.

Shanmugasundaram and colleagues² observed that the powdered leaf and water-based extracts from the leaf stimulate insulin secretion in rats. Persaud, Al-Majed, Raman, et al. (1999) demonstrated increased insulin release in isolated pancreatic beta cells from rats using the GS4 extract. Further, they determined that gymnemic acid VIII was the most potent single constituent causing this effect.

Gymnema has also been demonstrated to have protective effects on the pancreas. A partial protective effect of this plant was seen in animals pretreated for 2 weeks with powdered Gymnema leaves before exposure to beryllium, a potent pancreatic toxin (Prakash, Mathur, and Mathur, 1986). Shanmugasundaram, Gopinath, Shanmugasundaram, et al. (1990a) and Baskaran, Ahamath, Shanmugasundaram, et al. (1990) further suggested that it may also promote regeneration of islet cells in streptozotocin-treated rats. In fact, Srivastava, Venkatakrishna-Bhatt, Jhala, et al. (1986) demonstrated that alloxan diabetic rats treated with Gymnema lived significantly longer than untreated rats.

Extra-pancreatic mechanisms of action have been demonstrated to explain the hypoglycemic activity of Gymnema. Increase of the activity of key enzymes of insulin-dependent glucose utilization pathways, such as phosphorylases and gluconeogenic enzymes, has been reported in alloxan-treated rabbits (Shanmugasundaram, Panneerselvam, Samudram, 1981). Shimizu, Iino, Nakajima, et al. (1997) demonstrated that gymnemic acids II, III, and IV decreased absorption of glucose from isolated rat intestine. Liver glycogen also was shown to be decreased in normal and hyperglycemic rats treated with Gymnema (Chattopadhyay, 1998).

It has been traditionally observed that chewing Gymnema leaves interferes with the perception of sweet taste, an effect that can last for 1 to 2 hours. Bitter taste is also obscured, but not other tastes such as salty, pungent, acidic, or astringent (Nadkarni and Nadkarni, 1976). The leaf constituents most responsible for this effect are the gymnemic acids and gurmar, via direct activity on the nerves of the sensory apparatus in the tongue (Frank, Mize, Kennedy, et al., 1992; Hellekant, Hagstrom, Kasahara, et al., 1974; Imoto, Miyasaka, Ishima, et al., 1991; Kamei, Takano, Miyasaka, et al., 1992; Yoshikawa, Nakagawa, Yamamoto, et al., 1992; Yoshikawa, Kondo, Arihara, et al., 1993). Human subjects who drank a solution prepared with Gymnema prior to eating decreased their calorie consumption; this was attributed to a reduced perception of sweet taste and a resulting decrease in appetite (Brala and Hagen, 1983).

In summary, a review of the in-vitro and animal data suggests three possible physiologic mechanisms of actions for Gymnema: (1) increased insulin secretion through action on the pancreas, (2) increased tissue sensitivity to insulin, and (3) decreased oral intake of calories due to an alteration in the sensation of taste. This herb is frequently included in Ayurvedic formulas for diabetes and is often used as a folk treatment for diabetes (Khajuria and Thomas, 1992).

Generally, no significant toxicities are reported with the use of this herb. However, there are observations of nausea occurring in patients taking more than 3 g of Gymnema (Gerson, 2000).

Momordica charantia L. An herbaceous climbing vine of the Cucurbitaceae family, this plant is the most widely used traditional remedy for diabetes. It is believed to alleviate kapha and pitta (Dash, 1987). The bitter, unripe fruit or its juice is used in India, Africa, China, the West Indies, and Central America (Marles and Farnsworth, 1995; Bhandari and Grover, 1998). Karela, the common Indian name for the gourd produced by this plant, is traditionally taken in the form of a fried vegetable or as expressed juice (Nadkarni and Nadkarni,1976; Dash, 1987; Jain and DeFilipps, 1991). The common English name for this plant is bitter gourd (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991).

Triterpenoid and peptide constituents with hypoglycemic activity have been isolated from Momordica and tested in both in-vitro and in-vivo models. Charantin, an alcoholic extract of the fruit, contains both β-sitosterol-D-glucoside and 5,25-stigmastaadien-3-B-ol-D-glucoside in a 1:1 mixture (Marles and Farnsworth, 1995; Zafar and Neerja, 1991). In addition, a 17-amino-acid polypeptide, referred to as vegetable insulin, has been isolated from Momordica fruit, seeds, and seedlings. This polypeptide does not crossreact in immune assays for bovine insulin (Marles and Farnsworth, 1995; Zafar and Neerja, 1991). It is assumed that the polypeptide would not survive exposure to stomach acid; in fact, the studies that use this extract deliver it by injection. Bailey and Day (1989) reported isolating at least two other active components from Momordica; one has a rapid-onset of action, and the other contains a slow-acting constituent present in an alkaloid-rich fraction. The two active chemicals have not yet been fully characterized (Bailey and Day, 1989). Research attention is also being given to the bitter components common to this and many other members of the Cucurbitaceae family. They exist as a series of triterpene glycosides classified as momoridicosides; 11 have been identified so far (Zafar and Neerja, 1991).

Both pancreatic and extra-pancreatic mechanisms have been postulated for the hypoglycemic effects of the karela (bitter gourd), based on in-vitro and animal data. Hypoglycemic activity of the raw unripened fruit and aqueous extracts of the fruit are cited (Bailey and Day, 1989; Ivorra, Paya, and Villar, 1989). Charantin has been reported to have glucose-lowering activity in alloxan-treated rabbits, normal rabbits, rats, and cats (Marles and Farnsworth, 1995). However, other investigators report difficulty documenting the hypoglycemic effects of this constituent (Bailey and Day, 1989). In pancreatectomized animals, the effect of Momordica was equivocal (Marles and Farnsworth, 1995), but Raman and Lau (1996) noted efficacy of karela, also known as bitter melon, in lowering blood sugar in animals with very little pancreatic reserve. Decreased glucose uptake from intestine has been reported, as well as increased glucose uptake by muscle but not adipose tissue (Marles and Farnsworth, 1995; Bailey and Day, 1989). Bitter melon reportedly does not increase insulin levels in animals, despite the fact that insulin stimulation from isolated pancreatic islet cells has been reported (Raman and Skett, 1998). A purified protein called peptide V isolated from bitter melon has demonstrated hypoglycemic activity in animals and humans when injected (Bhandari and Grover, 1998; Marles and Farnsworth, 1995; Bailey and Day, 1989).

There are several reports of toxicity with Momordica. Bhandari and Grover (1998) noted uterine hemorrhage in pregnant rabbits given a crude extract of this plant. Cases of vomiting and diarrhea in humans have also been reported (Lewis and Elvin-Lewis, 1977). Marles and Farnsworth (1995) identified a mildly toxic lectin in the seeds and outer rind of the fruit. As a food in the Indian diet, no toxicity has been reported (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991).

Trigonella foenum-graecum Linn. Commonly known as fenugreek, this annual herb is widely cultivated throughout the world as a spice. It is in the family Leguminosae (also called Fabaceae). The herb is known in Sanskrit and Hindi as medhika or methi (Kapoor, 1990; Jain and DeFilipps, 1991; Nadkarni and Nadkarni, 1976). In addition to its role in Ayurveda, fenugreek was included in the pharmacopoeia of many ancient cultures, including Egyptian and Greek. Volatile and essential oils give this plant its characteristic odor, and a small amount (5 percent) of bitter fixed oil contributes to its taste (Lawrence Review, 1996). The leaves are eaten in India as a vegetable. The seed, on the other hand, is used to treat diabetes. It is prepared as a gruel or a drink, or it is baked into bread or mixed into curry (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991; Kapoor, 1990).

Fenugreek seeds have a high fiber content, up to 50 percent; it is mucilaginous and rich in galactomannans (Lawrence Review, 1996; Blumenthal, Goldberg, and Brinkman, 2000; Jain, Agrawal, and Sharma, 1996; Jellin, Batz, and Hitchens, 1999). Trigonelline, an alkaloid derived from the metabolism of nicotinic acid, has been isolated from the seed and shown to have hypoglycemic effects (Bailey and Day, 1989; Jain, Agrawal, and Sharma, 1996; Marles and Farnsworth, 1995).

Whole seeds have been shown to be hypoglycemic in normal and mildly diabetic animals but not in those with severe disease (Bailey and Day, 1989). Bailey and Day postulate that the high fiber content of fenugreek seeds decreases absorption of glucose by slowing transit time in the gut. Defatted seeds lowered blood glucose as well as glucagon in dogs, both normal and diabetic (Lawrence Review, 1996). Trigonelline showed a weak and transitory hypoglycemic effect when given orally to diabetic patients, presumably by slowing the metabolism of nicotinic acid, a hyperglycemic constituent (Marles and Farnsworth, 1995). In addition to its effects on glucose, fenugreek seed, especially the fiber component, lowers cholesterol and triglyceride levels in normal and diabetic animals and patients (Indian Council for Medical Research, 1987).

Fenugreek is also used to treat dysentery, dyspepsia, rheumatism, and chronic cough and to increase milk production. The leaves, made into a poultice, are applied topically to reduce swelling and treat abscesses or other wounds.

Generally, no significant toxicities are reported with the use of this herb.

Coccinia indica Wight and Arun. Also known as C. grandis and C. cordifolia Cogn., this leafy creeping plant of the Cucurbitaceae family grows wild over much of India and elsewhere (e.g., Hawaii) where it is a weed. The common English name is ivy gourd; in Hindi it is kanduri, and in Sanskrit it is bimba (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991; Kapoor, 1990). Traditionally, the leaves, root, fruit, and bark have been used medicinally. However, more recent scientific studies have focused primarily on the leaf (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991; Kapoor, 1990; Azad Kahn, Akhtar, and Mahtab, 1979).

Coccinia indica has not been studied as much as some of the other traditional remedies for diabetes, but the available data on this plant are chemically and pharmacologically interesting. The juice contains an amylase as well as β-sitosterol and a cucurbitacin, B-glycoside (Kapoor, 1990). Water-soluble fractions of the leaf that test positive for alkaloids show hypoglycemic activity (Hossain, Shibib, and Rahman, 1991). A novel saponin has also been identified from an alcohol extract of the leaf (Vaishnav and Gupta, 1995). The mechanism of action of Coccinia indica is not clearly understood, but studies with rats demonstrate involvement in the repression of a key gluconeogenic enzyme, glucose-6-phosphatase (Hossain, Shibib, and Rahman, 1991). Other studies showed that a suspension of the powdered leaf of C. indica lowers blood sugar in alloxan-treated dogs but not in normal animals (Ivorra, Paya, and Villar, 1989; Singh, Singh, Vrat, et al., 1985). This suspension also reduced blood glucose in both normal and diabetic dogs during a glucose tolerance test (Ivorra, Paya, and Villar, 1989; Singh, Singh, Vrat, et al., 1985). Both alcohol and water extracts of the root show antidiabetic activity in healthy rabbits (Bailey and Day, 1989; Ajgaonkar, 1979; Brahmchari and Augusti, 1963). Researchers have variously reported a rapid onset of action for C. indica or a delayed onset of at least 3 weeks (Azah Khan, Akhtar, and Mahtab, 1979).

Other traditional uses for this plant include the treatment of jaundice, wounds (when applied topically), bronchial complaints, psoriasis, ringworm, and sexually transmitted diseases such as syphilis and gonorrhea (Nadkarni and Nadkarni, 1976; Dash, 1987; Jain and DeFilipps, 1991; Kapoor, 1990).

No toxic side effects of this plant have been reported in the literature.

Pterocarpus marsupium Roxb. and Pterocarpus santalinus Linn. These two closely related species are used as folk remedies for diabetes in southern India. P. marsupium is a large deciduous tree common in central India. Its English name is false teak, in Sanskrit it is pitasala, in Hindi it is bijasal, and it is also known as vijayasara. P. santalinus, or red sandalwood, is a smaller tree found in the deciduous forests of southern India. It is known as Rakta chandana in Sanskrit and as Lal chandan in Hindi (Kapoor, 1990; Jain and DeFilipps, 1991; Dash, 1987; Nadkarni and Nadkarni, 1976). The wood and bark of the trees are most commonly used, often as a decoction. In addition, a novel method of using P. marsupium medicinally has been reported. Heartwood is carved into a cup, which is filled with water that is allowed to steep overnight. Diabetic patients then drink the water, called beeja water, the following day. It is not clear from the sources if the cup can be used more than once (Bhandari and Grover, 1998).

The sap from P. marsupium yields a reddish gum, called kino, which is high in tannic acids. Not surprisingly, epicatechins have been isolated from bark and heartwood (Kapoor, 1990; Marles and Farnsworth, 1995; Hii and Howell, 1984). These flavonoids have demonstrated an ATP-dependent enhancement of glucose-stimulated insulin release from isolated pancreatic islet cells in vitro (Marles and Farnsworth, 1995; Hii and Howell, 1984). Less substantiated claims have been made that the flavonoids contribute to the regeneration of beta cells in the pancreas (Bhandari and Grover, 1998; Chakravarthy, Gupta, Gambhir, et al., 1980). Phenolic constituents isolated from P. marsupium, identified as marsupin, ptersupin, and pterostilbene, have demonstrated hypoglycemic activity in rats (Manickam, Ramanathan, Jahromi, et al., 1997). Contradictory data have been reported about the efficacy of P. marsupium to lower blood sugar in animal models (Bhandari and Grover, 1998; Marles and Farnsworth, 1995).

A series of experiments in rats using a 95 percent alcohol extract of wood powder from P. santalinus showed a hypoglycemic effect in healthy albino rats and in streptozotocin-treated rats (Nagaraju, Prasad, Gopalakrishna, et al., 1991).

Additional traditional uses for these plants include the treatment of diarrhea, toothache, and skin wounds and infections (when applied topically). The gum from P. santalinus is often included in other medicinal salves.

There are no reports in the literature of significant toxic effects of these plants.

Scope of Work

Our literature review process consisted of the following steps:

• Identify sources in the literature reporting evidence for Ayurvedic therapy.

• Develop a strategy to maximize retrieval of Ayurvedic literature.

• Conduct a search of the Ayurvedic (broadly defined) literature to identify topic areas with sufficient publications to support a detailed review.

• Develop a strategy to identify English-language literature published in India but not normally cited in the Western indexes.

• Conduct a focused literature search of therapies for the treatment of diabetes in the Ayurvedic literature from both Indian and Western sources.

• Assess the strategies for completeness.

• Evaluate articles for methodological quality and relevance.

• Extract characteristics and results from studies meeting methodologic and clinical criteria

• Synthesize the results.

• Submit the results to technical experts for review.

• Incorporate the reviewers' comments into a final report for submission to AHRQ.

Objectives

We conducted a literature search of the field of Ayurveda to establish the distribution of studies using Ayurvedic interventions. These studies were then evaluated to determine if there was a sufficient body of literature in any one combination of disease and/or Ayurvedic modality to enable a comprehensive systematic review.

The initial search was guided by the following research questions:

• Is Ayurveda studied as a whole system?

• What disease states or modalities are the major focus of the Ayurvedic studies?

• Is there a sufficient body of literature in any one combination of disease and/or modality to do a traditional systematic review?

• If not, what type of review is possible?

• Is it possible to access the literature in India in a manner that is both cost-effective and comprehensive?

• Is the Indian body of literature qualitatively different from Western literature?

Based on the results of this search and on further discussions with our technical experts and the sponsoring agencies, we chose Ayurvedic therapies for diabetes mellitus as the focus of the comprehensive review presented in this report.

Ayurvedic Literature Search Design

Preliminary Search of the Ayurvedic Literature

We conducted a preliminary and nonsystematic search of Western databases to quickly survey the Ayurvedic literature and to help us define our search strategies. To determine what might be available in the Ayurvedic literature, we performed a preliminary search in PubMed, an online database encompassing the Cochrane and MEDLINE^® databases from 1966 to 1999. This simple review yielded only 708 citations. Because of the low yield, we then tested the strategy of surveying all the Asian Indian journals indexed in the Index Medicus from 1966 to see how many additional published studies these sources yielded. There were fewer than 1,200 citations published in these journals, and the majority of them covered conventional Western medical topics. Most of the studies that appeared to focus on Ayurvedic medicine were descriptive or historical in nature (as opposed to research studies). Of the research studies identified in this preliminary survey, the large majority examined botanical therapies. Consulting additional sources of information on Ayurveda, we also examined texts in the field of Ayurveda, both lay and professional.

Table 6. Results of preliminary search in PubMed

Table 6. Results of preliminary search in PubMed

Search terms	Number
Keyword: Ayurved-*	708
Limited to human studies	373
Limited to clinical trials	35
Limited to randomized controlled trials	20
Limited to reviews	41

*

A dash after a term indicates that the term was truncated (e.g. “AYURVED-” will search for “AYURVEDA” and AYURVEDIC."

The consensus of the experts we consulted and the results of the preliminary work outlined above suggested that the most useful and effective way to maximize the yield of Ayurvedic studies from a literature survey was to focus on herbs. Our preliminary work also identified which disease states were most common in the Western literature on Ayurvedic therapy. This showed that these diseases are described in Western diagnostic terminology and not in terms unique to Ayurveda. The results of the PubMed search are summarized in Table 6.

Table 7. Preliminary search, clinical trials by disease (more...)

Table 7. Preliminary search, clinical trials by disease state

MeSH heading	Number of articles
Total*	37
Acne	1
Bronchial asthma	1
Constipation	1
Coronary artery disease	1
Diabetes	4
Dyspepsia	1
Hyperlipidemia/cholesterol	1
Irritable bowel syndrome	1
Infectious disease	1
Lactational inadequacy	1
Liver disease	4
Tobacco dependence	1
Obesity	1
Ocular problems	4
Parasites	2
Parkinson's disease	1
Rectal fistula	1
Rheumatologic disease	2
Varicose veins	1
Unknown	7

*

Total is greater than 35 because an article may report on more than one disease state.

Table 7 presents the disease states described by the MeSH³ headings in the 35 articles identified as clinical trials in Ayurveda from the PubMed search. Diabetes, liver disease, and ocular problems were the conditions with the largest number of articles.

On the basis of this preliminary nonsystematic search, we decided that a more comprehensive and systematic survey of the Ayurvedic literature was justified. We then moved forward with our initial systematic search.

Initial Systematic Search of the Ayurvedic Literature

Guided by the results of the preliminary search outlined above and with input from our panel of technical advisors, we used the search term Ayurveda plus the names of 16 major botanicals characteristically used in Ayurveda for the initial systematic search of the Western literature. We added terms for 16 herbs that we identified in published articles, including ones recognized as important by experts. The herbal terms were added to the search to increase its sensitivity, making it possible to find studies that used Ayurvedic herbal therapy without necessarily being directly identified as Ayurvedic studies.

The search was done hierarchically and initially looked for articles published in English. The focus was Ayurvedic medicine for common Western medical diseases. The search terms were Ayurveda or Ayurvedic or the scientific names of the most common botanicals: Adhatoda vasica or Albizzia lebbeck or Andrographis paniculata or Bacopa monniera or Coleus forskohlii or Commiphora mukul or Crataeva nurvala or Gymnema sylvestre or Hemidesmus indicus or Inula racemosa or Phyllanthus amarus or Picrorrhiza kurroa or Terminalia arjuna or Tylophora indica or Withania somnifera.

Table 8. Database search characteristics

Table 8. Database search characteristics

Database	Time period
MEDLINE®MEDLARS onLINE: database for biomedical literature	January 1966-December 1999
HealthSTAR Health Services Technology, Administration, and Research: database on health planning and administration	January 1975-December 1999
EMBASE®Comprehensive literature index on human medicine and related disciplines	January 1974-November 1999
Allied and Complementary Medicine™ Database on complementary or alternative medicine and allied health	January 1984-January 1999
MANTIS™ Manual, Alternative and Natural Therapy: database for health care disciplines not covered by major biomedical databases	January 1880-November 1999
CAB HEALTH Database on human health and communicable diseases	1983-October 1999
Biosis Previews®Database on research in the biological and biomedical sciences	1993-September 1999
CINAHL®Cumulative Index to Nursing and Allied Health Literature	1982-August 2000

Table 8 shows the databases we used, and the time periods covered, for the initial systematic search of the Ayurvedic literature. The exact search terms used are listed in Appendix C.

Ayurvedic Review Strategy

Results of the Initial Systematic Search

We downloaded 2,565 citations from the online searches into a Microsoft Access database. Of these citations, abstracts were available for 1,562, and the remainder (1,003 citations) were obtained as titles. We initially only screened those articles for which we had abstracts. Thus, in a sense, the lack of an abstract was an exclusion criterion for the initial review.⁴ Of the 1,562 articles for which we had abstracts, 1,214 were related to Ayurvedic medicine and met the inclusion criteria. Each of these abstracts was independently evaluated by two reviewers using the screening form.

Studies were excluded from further review if they were:

• Treatment studies having to do with veterinary treatment of animal disease.

• Botanical studies having to do with the growth or botanical identity of a plant.

• Studies in which the plant was used as a pesticide or fungicide.

• Studies not using Ayurvedic techniques, philosophies, or materials.

• Articles in which Ayurveda was not the main or a major therapeutic focus (i.e., survey articles of complementary and alternative medicine in general).

Diagram 1. Ayurveda Literature Search and Review Strategy: (more...)

   Diagram 1. Ayurveda Literature Search and Review Strategy: Initial Systematic Search

See Diagram 1 at the end of this chapter for a summary of the steps taken in the initial systematic search.

Table 9. Study designs identified in initial systematic (more...)

Table 9. Study designs identified in initial systematic search

Type of study	Frequency	Percentage
Historical/descriptive	114	9
Pharmacological	569	47
Ethnobotanical	36	3
Clinical study	394	33
Type of clinical study:
Potential controlled trial	247	63*
Case series	41	11*
Case control	1	0*
Unclear	105	26*
Review	62	5
Unclear	26	2
Other	13	1
Total	1214	100

*

Percent of the 394 clinical studies.

We classified 47 percent of the 1,214 abstracts as pharmacological studies. Clinical studies represented 33 percent of our abstracts; and of those, 63 percent were thought to represent controlled trials. The study designs were broken down into the categories shown in Table 9.

The articles identified as clinical studies were of special interest, since these were the types of studies we would focus on for an in-depth review. We identified the following clinical study types in our initial abstract screening:

• Controlled trials—These include randomized controlled trials (RCTs), where subjects are randomly assigned to an intervention. We were particularly interested in RCTs. However, given their probable scarcity, we broadened this design category to include controlled clinical trials (CCTs), where the allocation is not done randomly, as well as any study that might have a comparison arm. Examples might be a study that compares patients treated with an Ayurvedic therapy to patients treated in some other way or a study that has two groups treated with different therapies.

• Case control study—This is a study in which subjects are chosen because they have a disease, and they are compared to a comparison arm that does not. None of the studies we found fell into this category.

• Case series—This category includes both series and single case studies without a comparison arm. Such studies involve a simple series of sequential cases (or a single case). This category accounted for 11 percent of the clinical studies. Distinctions between descriptive and experimental case series cannot be made at this stage.

• Unclear clinical design—This category includes those studies where the two reviewers were unable to determine the study design based on the abstract alone.

Table 10. Body systems or conditions identified in (more...)

Table 10. Body systems or conditions identified in initial systematic search

Body system or condition	All articles	Potential controlled trials
Total number of articles1	1214	247
Diabetes	120	44 (17.8%)
Liver/hepatitis	112	42 (17.0%)
Musculoskeletal	23	8 (3.2%)
Hyperlipidemia/cholesterol	46	23 (9.3%)
Central nervous system disorders*	54	23 (9.3%)
Gastrointestinal (non-hepatitis)	49	17 (6.9%)
Psychiatric conditions (depression)	22	6 (2.4%)
Cardiovascular diseases	63	18 (7.3%)
Pulmonary diseases	20	5 (2.0%)
Infectious diseases	103	15 (6.1%)
Allergy	17	2 (0.8%)
Cancer	19	9 (3.6%)
Dermatology	17	6 (2.4%)
Toxicology	36	6 (2.4%)
Hematology	16	1 (0.4%)
Other	264	64 (25.9%)
Unclear	338	1 (0.4%)

1

The total number of articles will not equal the column totals since articles can report on more than one body system or condition.

*

Dementia and cerebrovascular accident (stroke).

Table 10 shows the distribution of target body systems or conditions identified in the 1,214 articles. One column shows the distribution for all articles, and the other lists the distribution for the 247 potential controlled trials that are of primary interest to this report. This breakdown was used to identify the potential focus areas for our comprehensive review. A single article can focus on more than one body system or condition; therefore, it is not useful to total the columns in Table 10.

The most common body systems or conditions we identified in our Ayurveda literature search were diabetes, liver/hepatitis, infectious disease, and cardiovascular disease; there were 120, 112, 103, and 63 articles, respectively. Diabetes and hepatitis had the largest number of potential controlled trials, 44 and 42, respectively, followed by central nervous system disorders and hypercholesterolemia with 23 articles each.

Table 11. Ayurvedic modalities identified in initial (more...)

Table 11. Ayurvedic modalities identified in initial systematic search

Modality	All articles	Potential controlled trials
Total number of articles1	1214	247
Botanical	1091 (89.9%)	246 (99.6%)
Mind-body	4 (0.3%)	0 (0.0%)
Panchakarma	12 (1.0%)	0 (0.0%)
Other	69 (5.7%)	10 (4.0%)
Unclear	64 (5.3%)	1 (0.4%)

1

The total numbers of articles will not equal the column totals since articles can report on more than one body system or condition.

Table 11 identifies the various Ayurvedic modalities and therapies represented in the literature. We show the distribution of modalities for all articles and also for those articles containing potential controlled trials. Botanical modalities were the most common, accounting for 88 percent of all the studies and 95 percent of the possible comparative studies that were reviewed. This is not unexpected since the search was constructed using specific botanical names to increase the yield of Ayurvedic articles. No other modalities had significant numbers of articles, including some that are considered important in treatment of Ayurvedic patients, such as yoga or panchakarma. Some articles may have reported the use of more than one modality.

Based on the results of our detailed analysis of the data extracted from the 1,214 abstracts, and in consultation with the funding agencies, we chose diabetes as the focus of our comprehensive review. This decision was based on the fact that diabetes was the focus of the largest number of articles and had the most potential controlled trials. In addition, it has well-established diagnostic criteria and measurable outcomes that could be extracted from the Ayurvedic clinical studies. Further, diabetes is well known in Ayurvedic medicine, and the concordance of the Ayurvedic diagnosis with the Western diagnosis was close (see “Ayurvedic Diagnosis and Treatment of Diabetes” in Chapter 1).

Search of the Indian Literature

Several of the experts we consulted were of the opinion that a large body of literature existed in India that could not be obtained through the conventional search strategy outlined above. Hence, we decided after consultation with the sponsoring agencies to conduct a search for English-language Ayurvedic literature from the Indian subcontinent in December 1999-January 2000. Our goal was to ascertain how much research literature exists in Indian publications and at Indian institutions, and whether it is readily available and can be obtained in a cost-effective manner. An SCEPC staff member, a physician fluent in English and Hindi who had trained in India, went to India with the objectives described below.

Focused Search of the Ayurvedic Literature

After choosing diabetes as our focus topic, we conducted a focused literature search to identify articles we might have missed in our initial systematic search and Indian literature search.

Reasons for Exclusion

Of the 73 reviewed articles, 19 were deemed inappropriate for inclusion in the final data set for analysis. The reasons for exclusion were:

• Articles containing only information about Ayurveda or treatment of diabetes, instead of original research, were considered background articles or review articles and rejected for further analysis.

• Articles describing pharmacological, animal, and in-vitro studies were excluded.

• Duplicate studies (those entered twice in the database) were excluded, as well as studies that appeared in different journals but contained the same or similar data.

• Articles that had no relevant outcomes, were unobtainable, or were case reports of a single patient, were excluded from subsequent analysis.

Diagram 2. Ayurveda Literature Review Strategy of Focused (more...)

   Diagram 2. Ayurveda Literature Review Strategy of Focused Search for Human Diabetes Articles

See Diagram 2 at the end of this chapter for a description of the filtering strategy used to identify articles focusing on the Ayurvedic treatment of human diabetes.

Selection of Studies for Further Analysis

We felt that further analysis could be done on a number of studies to allow a more direct comparison of the effects of different therapies. From the studies that had a complete review, we selected ones for further analysis if they met the following criteria:

• The study had to be an RCT or a CCT with a comparison arm that did not receive an herb. Where there was more than one comparison arm, a diabetic comparison was preferred. RCTs of any size were included. For CCTs, the study had to have at least one treatment arm that contained at least 10 patients.

• The study (generally case series or cohorts) had to have arms with at least 10 patients, if it was not an RCT or a CCT.

• The study had to test a single agent, a formula acting as a single agent, or a limited combination of products (no more than three) acting as a single agent. The agent had to be dispensed more than one time to the study patients.

• The study had to report on at least one of three outcome measures at 30 days minimum following the start of the study: hemoglobin A_1c (glycosylated hemoglobin), fasting blood glucose, or post-prandial blood glucose at either 2 hours (preferred) or 1 hour (acceptable).

• The study had to provide sample sizes, pre- and post-intervention means, and standard deviations or standard errors separately for each arm that we included in our analysis.

For studies that reported on both 1-hour and 2-hour post-prandial blood glucose levels, we chose the 2-hour measurement for the analysis. For studies that reported more than one followup time, we selected the followup time that was closest to 3 months.

Table 12. Summary of studies included for further analysis (more...)

Table 12. Summary of studies included for further analysis

Study design	#Studies	#Arms
RCT with nonherbal arm	2	4
CCT with nonherbal arm	5	10
Pre/post data available	15	19

Study = discrete clinical investigation; arm = a comparison group within a study.

Twenty-two studies were selected for further analysis based on the criteria described above. These studies represented either RCTs (2) or CCTs (5) with a comparison arm that did not receive an herb, or studies that had pre/post⁵ comparison data available (15). The study types selected for further analysis are summarized in Table 12

Given the clinical heterogeneity of the interventions and the populations across the studies, we decided not to pool data across studies. Instead, separately for each of the three clinically relevant outcome measures (fasting blood glucose, post-prandial blood glucose, and hemoglobin A_1c), we estimated a common effect size and its confidence interval for each study.

We considered two subgroups of eligible studies separately: the RCTs and CCTs that had nonherbal comparison arm(s) were considered one subgroup, and all other eligible (pre/post data available) studies made up the second subgroup. The RCT/CCT subgroup allowed us to do a comparison between herbal and nonherbal arms within a study, providing a stronger level of evidence than pre/post comparisons. For these studies, we were able to compare the effects of different treatments after the placebo or nonherbal arm results were taken into account. The pre/post studies subgroup did not allow a concurrent comparison, but the difference between pre- and post-treatment values could be estimated.

The results of this analysis are displayed graphically for ease of comparison across studies and are presented in Chapter 3. Appendix K summarizes the data and methods used to calculate the common effect sizes.

Description of the Evidence

Fifty-four articles reported on the results of 62 studies in diabetes during the search and review process for this evidence report, and they were then reviewed in depth. Thirty-five studies came from the Western literature, and 27 came from the Indian literature. The designs of the 62 studies were varied. There were 7 randomized controlled trials and 10 controlled clinical trials. There were 38 case series, the most frequently used clinical design, and 7 cohort studies.

Figure 1. Difference of Differences for RCTs and CCTs (more...)

   Figure 1. Difference of Differences for RCTs and CCTs

Figure 1

Figure 2. Pre/Post Comparison Group

   Figure 2. Pre/Post Comparison Group

Figure 2

In terms of interventions, 45 unique herbs or herbal formulas were used 142 times in the 62 studies for which we did detailed analysis. (A study could have more than one intervention; thus the total number of interventions exceeded the number of studies). The two most commonly tested single herbs were Gymnema sylvestre (used 16 times in 12 studies) and fenugreek (used 18 times in 11 studies). The Ayurvedic formulas D-400 (used seven times in four studies) and Ayush-82 (used six times in five studies) were the two most commonly tested herbal formulas. These four interventions plus guar gum and Eugenia jambolana accounted for 50 percent of all the herbal interventions tested. Mineral substances were added to herbal formulas only six times. Diet was the focus of two studies but was used as an intervention or co-intervention 9 and 35 times, respectively.

Quality of the Evidence

Only the RCTs and CCTs were given a Jadad score, which rates studies on a 0 to 5 scale (Jadad, Moor, Carroll, et al., 1996). A score is based on the answer to three questions: Was the study randomized? Was the study described as double blind? Was there a description of withdrawals and dropouts? One point is awarded for each “yes” answer, and no points are given for a “no” answer. An additional point is given if the randomization method was described and was appropriate. A point is deducted if the method is described but is not appropriate. In response to the second question, a point is awarded if the method of blinding is appropriate and described, and one point is deducted if the described method is inappropriate. Given the nature of some of the interventions used in Ayurvedic medicine, a true double-blind study would be difficult to perform, and the maximum Jadad score expected would then be 3 or 4. Empirical evidence has shown that studies scoring 2 or less report exaggerated results compared with studies scoring 3 or more (Moher, Pham, Jones, et al., 1998). Studies with a Jadad score of 3 or more are referred to as “high” quality, and studies scoring 2 or less are referred to as “poor” quality.

Of the seven RCT studies, four received a Jadad score of 1, two studies scored a 2, and only one study received a 4 on this scale. This is the only high quality study in our analysis. Seven of the CCT studies had a Jadad score of 0, and the remaining three studies scored only 1.

A methodological problem common to many of the studies was lack of statistical power—32 arms or studies had fewer than 10 subjects. This was especially true of the studies in Appendix L. Cohen (1988) defines an effect size as the absolute value of the difference in the treatment and control group means divided by the standard deviation. He further defines a “small” effect size as 0.2, a “medium” effect size as 0.5, and a “large” effect size as 0.8. Therefore, to achieve 80 percent power to detect a small effect size at a level of significance of p<0.05, at least 393 subjects would be needed in both treatment and control groups. To detect a medium effect size, 63 patients would be required in each group, and to detect a large effect size, 25 patients would be required. None of our studies had sufficient numbers to detect a small effect size, and most would not be able to detect even a medium effect size. Thus, the studies that failed to demonstrate an effect of herbal therapy could have been either truly ineffective or underpowered. The majority of the studies did show statistically significant results. Given the preceding discussion, we can postulate that either the effect sizes were large for these studies, the lack of random design exaggerated the beneficial effects of the interventions, or there exists a publication bias for studies reporting a beneficial effect.

Synthesis of the Evidence

Controlled Trials With Nonherbal Comparison Arm(s)

The seven trials that met the inclusion criteria (see “Quality of the Evidence,” above) were either RCTs (2) or CCTs (5); these studies are summarized in Evidence Table 1. All of them tested herbs as single agents, not in formulas or in combinations. Detailed descriptions of these studies follow.

Randomized Controlled Trials

Azad Khan, Akhtar, and Mahtab (1979) performed a randomized controlled trial to assess the hypoglycemic properties of a preparation of the herb Coccinia indica on uncontrolled and untreated type 2 diabetic patients. Thirty-eight patients with type 2 diabetes who were attending the outpatient department of the Bangladesh Institute of Research and Rehabilitation in Diabetes in Dhaka, Bangladesh, were recruited for the study. Only uncontrolled or newly diagnosed patients were enlisted. Patients who had evidence of renal disease, hepatic disease, or neuropathy or who had ketones in their urine were excluded. There were six dropouts from the study, and 32 (84 percent) patients completed the study. The patients were randomly allotted to receive either a plant preparation (Arm II, n=16, 23 percent female, age 51 ± 10 years) or placebo (Arm I, n= 16, 14 percent female, age 44 ± 5 years). Four of the dropouts were from Arm I, and two were from Arm II. The patients were reported to be similar with respect to diagnosis and therapy. No mention of socioeconomic status was made. Arm I patients received placebo twice a day for 6 weeks. The intervention consisted of a freeze-dried powder of crushed leaves of Coccinia indica, given in a dose of 900 mg twice a day for 6 weeks for Arm II.

Table 13. Study results for Azad Khan, Akhtar, and (more...)

Table 13. Study results for Azad Khan, Akhtar, and Mahtab, 1979

	Arm I (placebo)		Arm II (Coccinia)
Variable	Pre-treatment	Post-treatment	Pre-treatment	Post-treatment
Fasting blood sugar (mg/dl)	195.5	181.3	178.8	122.1 (p<0.01)
1-hour post-prandial blood sugar (mg/dl)	281.1	267.6	268.1	224.9 (p<0.05)
2-hour post-prandial blood sugar (mg/dl)	255.1	252.0	245.4	186.9 (p<0.01)

Study results are shown in Table 13. Of the 16 patients in receiving Coccinia indica, 10 patients showed a marked improvement in the glucose tolerance tests to values closer to normal, while none in the placebo arm showed marked improvement. Limitations of this trial are its short duration⁷ and the small number of patients. The Jadad score for the trial was 4.

Agrawal, Rai, and Singh (1996) reported the results of a randomized, placebo controlled crossover trial of the effects of a leaf extract of holy basil (Ocimum sanctum) on type 2 diabetic patients. The patients (n=62) were recruited from community advertisements in Kanpur, India, and through visits to the offices of local dieticians. The diagnosis of diabetes was made through a glucose tolerance test. Patients were excluded from the trial if they used holy basil leaves frequently; had a blood urea level > 40 mg/dl; had diarrhea or dysentery; had cancer; had 2-hour PPBS > 350 mg/dl; did not like to eat holy basil; had type 1 diabetes; or did not have diabetes. Of the 62 patients selected, 22 were excluded. The remaining patients (n=40, 37.5 percent female, age 41–65 years) were entered into the trial. The mean duration of diabetes was 8.6 years (1.5–13 years), body weight ranged from 55–75 kg, mean systolic blood pressure was 137 mm Hg (122–170 mm Hg) and mean diastolic blood pressure was 89 mm Hg (80–110 mm Hg).

The crossover design was performed as follows. All subjects had a 5-day run-in period during which they all consumed a tea made from holy basil leaves. This was followed by an 8-week experimental period. The patients were randomly assigned to one of two arms. The first arm drank the holy basil leaf tea for 4 weeks followed by placebo leaves for 4 weeks. The second arm had the placebo leaves first, followed by the holy basil leaf tea. All the patients were advised not to change their physical activity and diets and to stop taking their oral hypoglycemic agents 7 days prior to the start of the trial. The active medication was dispensed in a sachet containing dried leaf powder made from 2.5 g of fresh leaves of holy basil. The placebo was identical except that it contained spinach leaf powder. Subjects were instructed to mix the contents of the sachet in 200 ml of water and to drink the mixture on an empty stomach in the morning. Compliance was monitored by counting the number of unused sachets returned by the patients at each visit.

Table 14. Study results for Agrawal, Rai, and Singh, (more...)

Table 14. Study results for Agrawal, Rai, and Singh, 1996

Variable	Holy basil treatment first			Placebo treatment first
	Baseline	First 4 weeks	Second 4 weeks	Baseline	First 4 weeks	Second 4 weeks
Fasting blood sugar (mg/dl)	134.5	99.7 (p<0.1)	115.6	132.4	123.2	97.2 (p<0.1)
Post-prandial blood sugar (mg/dl)	223.9	204.0 (p<0.1)	217.1	221.6	215.1	197.1 (p<0.1)
Total cholesterol (mg/dl)	238.2	221.5	236.0	233.0	236.2	220.1

Table 15. Laboratory data at end of each treatment (more...)

Table 15. Laboratory data at end of each treatment period, Agrawal, Rai, and Singh, 1996

Variable	Placebo treatment	Holy basil treatment	Difference in effects of treatment	Change (%)
Fasting blood glucose (mg/dl)	119.4	98.6	-31.4 to -11.2 (p<0.001)	-17.6
Post-prandial blood sugar (mg/dl)	216.1	200.5	-27.0 to -5.6 (p<0.02)	-7.3
Total cholesterol (mg/dl)	236.1	220.8	-26.5 to 8.7 (p<0.02)	-6.5

Study results (Table 14) and laboratory data at the end of each treatment period (Table 15) are summarized below.

The authors concluded that holy basil leaf powder caused significant reduction in the level of fasting blood sugar and post-prandial blood sugar, and a moderate reduction in cholesterol. They suggested that holy basil leaf powder may be a useful adjunct to other treatments for type 2 diabetes. The small numbers of patients and the fact that the patients were not blinded to the sequence of treatment given (holy basil powder has a distinctive taste and does not taste like spinach powder) are the significant drawbacks of the study as well as its short duration. The Jadad score for this trial is 1.

Controlled Clinical Trials

Chandola, Tripathi, and Udupa (1980a) performed a series of three studies with type 2 diabetic patients to study the hypoglycemic effects of Cinnamonum tamala. Only one study is a CCT, but all three studies are discussed together here.

In the first study, a group of type 2 diabetic patients (Arm I, n=32, 5 percent female, ages 35–70) were recruited from the outpatient departments of the Institute of Medical Sciences, Varanasi, India. Another group of type 2 diabetic patients (Arm II, n=8, 0 percent females, ages 39–50) served as controls. The socioeconomic characteristics of the patients were not noted. The authors did not report if the two arms of patients were statistically equivalent on important prognostic variables at the start of the trial. The intervention consisted of leaves of Cinnamonum tamala (tejpatra) that were dried, pulverized, and filtered through a fine mesh. The powder was packed into 100 g packets. The patients in Arm I were given 2 heaped teaspoons of the powder four times a day before meals and tea for a total of 1 month. All patients were advised to follow an 1,800-calorie diet with restriction of sugar, potato, and rice. The patients in Arm II were only instructed to follow the above diet with no other medications. Fasting blood sugars were assessed in both arms of patients before the trial and at the end of 1 month.

The results of the first study are as follows: In Arm I, mean fasting blood sugar improved from 153.44 mg/dl to 112.65 mg/dl (p<0.001) at the end of 1 month. In Arm II, the mean fasting blood sugar increased from 156.37 mg/dl to 164.12 mg/dl (p<0.01) at the end of 1 month. The Jadad score for this CCT was 0.

Table 16. Study results of Chandola, Tripathi, and (more...)

Table 16. Study results of Chandola, Tripathi, and Udupa, 1980a, second study

	Fasting blood sugar (mg/dl)	1-hour post-prandial blood sugar (mg/dl)	2-hour post-prandial blood sugar (mg/dl)
Pre-trial	144.84	258.64	236.15
Post-trial	103.51 (p<0.001)	206.89 (p<0.001)	170.59 (p<0.001)

The second study was a case series with pre/post data available in which a series of type 2 diabetic patients (n=25, 4 percent female, ages 36–70), diagnosed by glucose tolerance tests, were recruited from the outpatient departments of the Institute of Medical Sciences, Varanasi, India. The patients were given Cinnamonum tamala for 1 month. It was prepared in a similar manner and dose as described for the first study. Glucose tolerance tests were done at the start and at the end of 1 month. The results for the second study are reported in Table 16.

In the third study, another case series with pre/post data available, the immediate hypoglycemic effect of Cinnamonum tamala was evaluated in a series of seven patients (0 percent female, ages 35–68 years). No other demographic characteristics of the patients were reported. All patients had blood drawn for a fasting blood sugar after which they were given 20 g of Cinnamonum tamala powder. Blood sugar levels were assessed at 1 hour and 2 hours after administration of the herb. The authors reported that mean blood sugar decreased from 166.28 mg/dl to 145.57 mg/dl at 1 hour (p<0.02) and to 122.28 (p<0.01) mg/dl at 2 hours.

Based on these three studies, the authors concluded that Cinnamonum tamala has a definite hypoglycemic effect and its role in the management of diabetes is established. However, the short duration of all the studies, the lack of randomized control arms, and the small numbers of patients prevent definitive conclusions being drawn.

Kamble, Jyotishi, Kamalakar, et al. (1996) studied the effect of fresh leaves of Coccinia indica on blood sugar levels and on hyperlipidemia in non-insulin-dependent diabetes mellitus (NIDDM, type 2) patients at an Ayurvedic clinic in Nagpur, India. Patients were classified into four arms: a healthy comparison arm (Arm I, 15 patients); a NIDDM comparison arm (Arm II, 30 patients); NIDDM patients treated with Coccinia (Arm III, 25 patients); and a NIDDM arm treated with Chlorpropamide (Arm IV, 15 patients). Patients received a glucose tolerance test and were evaluated for existing diabetic complications. Patients ranged in age from 21 to 55 years; no other demographic information was provided. Patients were given the Coccinia (decocted from fresh leaves, dried into a paste, and made into a 3 g tablet) twice a day for 12 weeks.

For patients in Arm III, the study showed a statistically significant reduction in mean fasting blood sugar, from 160 to 110 mg/dl (p<0.001), and a significant decrease in blood sugar levels after glucose tolerance tests: the 2-hour level dropped from 308 mg/dl to 142 mg/dl (p<0.001). These results were roughly equivalent to those found in the Chlorpropamide arm. Additionally, after 12 weeks of treatment, the Coccinia treated patients showed significant reductions in levels of cholesterol (p<0.001), phospholipid (p<0.05), triglyceride (p<0.001), and free fatty acid (p<0.001) compared to untreated diabetics. These levels approach those of the healthy comparison arm. Within the treatment group, no pre/post data are reported for the changes in the lipid levels. No adverse effects were reported from this trial.

The researchers concluded that Coccinia indica significantly reduces hyperlipidemia and serum glucose levels in NIDDM (type 2) patients. The authors postulate that the hypocholesterolemic effect may be due to the presence of beta-sito-sterol containing an ethyl group that may interfere with the absorption of cholesterol in the gut. They also postulate that the hypoglycemic effect, which was similar to that of Chlorpropamide, may be a result of the anabolic insulin-like properties of the plant. The authors' conclusions would be further supported if the study had been done in a randomized and blinded manner. The Jadad score was 0.

Kohli and Singh (1993) studied the effects of Jamun beej (Eugenia jambolana) on fasting blood sugar, glucose tolerance test, and symptomatic relief of symptoms associated with diabetes. Thirty patients identified from the outpatient unit of the Sunderlal Hospital in Varanasi, India, who had NIDDM confirmed by glucose tolerance test, were given the herb. Seeds of Eugenia jambolana were obtained from the local market and crushed into a fine powder. The patients were instructed to take 4 g three times a day for 3 months. No dietary restrictions were included in the study. A second arm of six patients with confirmed NIDDM was given Chlorpropamide (250 mg per day). No demographic information was provided on these patients. The Jadad total is 1.

Table 17. Study results for herb-treated arm, Kohli (more...)

Table 17. Study results for herb-treated arm, Kohli and Singh, 1993

	Pre-study	1 month	2 months	3 months
Fasting blood sugar (mg/dl)	163	129 (p<0.001)	99.64 (p<0.001)	130.1 (NS)
1-hour post-prandial blood sugar (mg/dl)	279	222.68 (p<0.001)	183.55 (p<0.001)	220.33 (p<0.001)
2-hour post-prandial blood sugar (mg/dl)	304.67	249.00 (p<0.001)	192.62 (p<0.001)	226.00 (p<0.001)

NS=nonsignificant

Results for the herb-treated arm are shown in Table 17. The study showed a statistically significant reduction in mean fasting blood sugar of 51.86 mg/dl at 2 months (p<0.001) but not at 3 months. The study also demonstrated a significant reduction in the response to the glucose tolerance test at both 2 months (p<0.001) and at 3 months (p<0.01). The Eugenia jambolana arm was compared with the Chlorpropamide arm at 1 month. The Chlorpropamide arm did not show a significant reduction in fasting blood sugar as did the Eugenia jambolana arm. Both the Chlorpropamide and the Eugenia jambolana arms showed a significant reduction in serum blood sugar 2 hours after the glucose tolerance test (p<0.05 and p<0.001, respectively). The study also revealed statistically significant reductions in all of the symptoms associated with diabetes (polyurea, polydipsia, polyphagia, weight loss, weakness, leg cramps, and joint pain) at 1 month, 2 months, and 3 months. The authors noted that of the original 30 patients enrolled in the study, only 9 were still in the study at 3 months. No adverse effects were reported from this trial.

The researchers concluded that Eugenia jambolana produces symptomatic relief of diabetes and a simultaneous reduction in fasting glucose levels. Given the widespread availability, inexpensive price, and lack of side effects of this herb, the authors recommend making Eugenia jambolana part of the regimen of all NIDDM (type 2) patients. The study's significant design flaws make it difficult to draw the same conclusions.

Baskaran, Ahamath, Shanmugasundaram, et al. (1990) studied the effects of GS4, a specific isolate of Gymnema sylvestre, on the blood sugar and cholesterol levels of type 2 diabetic patients. Forty-seven patients (23 percent female, age 40–63) were recruited. The patients were classified into two arms. Arm I consisted of 22 diabetic patients on oral hypoglycemic therapy, which was continued during the trial, plus the herbal therapy with gymnema. The duration of diabetes in this arm varied from 1 to 12 years (mean = 4.6 years), and the blood glucose control was generally noted to be poor. Arm II consisted of 25 type 2 diabetic patients on conventional medications alone. The duration of diabetes in this arm ranged from 1–5 years (mean = 2.7 years). No socioeconomic characteristics were reported for the patients. Although the arms were reasonably matched for age and Broca index (percent of ideal body weight), they were not clinically equivalent at the start of the trial. Importantly, based on analysis of individual patient data in the study, the two arms were statistically different at baseline with respect to fasting blood sugar.

The intervention consisted of GS4, which is a water-soluble acidic fraction of an ethanol extract of the leaves of Gymnema sylvestre. The dose was 400 mg/day given to the patients in Arm I for a period of 18–20 months in addition to their conventional oral hypoglycemics. Patients in Arm II continued to take their conventional medicines only.

The study results were as follows. In Arm I after 12 months of followup, mean FBS improved from 174 mg/dl to 124 mg/dl (p<0.001); hemoglobin A_1c improved from 11.9 percent to 8.48 percent (p<0.001); cholesterol improved from 260 mg/dl to 231 mg/dl (p<0.001); and triglycerides improved from 170 mg/dl to 142 mg/dl (p<0.001). In the Arm II comparison group after 12 months of followup, mean FBS changed from 150 mg/dl to 157 mg/dl; mean hemoglobin A_1c went from 10.2 percent to 10.47 percent; mean cholesterol worsened from 252 mg/dl to 261 mg/dl (p<0.05); and mean triglycerides changed from 148 mg/dl to 164 mg/dl (p<0.001). Virtually all patients in Arm I developed hypoglycemic symptoms, and the dose of their normal oral hypoglycemic agent needed to be changed or stopped. Most of the patients in Arm I reported a sense of well being, less exhaustion, and less pain. Five of the 22 patients in Arm I were able to discontinue their oral hypoglycemic agent.

The authors concluded that GS4 supplementation in type 2 diabetic patients has an advantage over conventional therapy alone in reducing hyperglycemia and hyperlipidemia. The lack of a placebo comparison arm, lack of random allocation, the small number of patients, and baseline differences between arms are the drawbacks of this study. The Jadad score is 0.

Shanmugasundaram, Rajeswari, Baskaran, et al. (1990b) reported the effect of an herbal compound on blood glucose, insulin requirements, and cholesterol in type 1 and type 2 diabetic patients. The patients (n=64) were recruited from the outpatient departments of Ambedkar Institute for Diabetes in Madras, India. All the patients were free from diabetic complications. No socioeconomic characteristics were reported. The patients were studied in two arms. Arm I consisted of 37 type 1 diabetic patients (ages 8–30 years, percent female not reported) who were on insulin therapy and served as controls. Arm II consisted of 23 type 1 diabetic patients (ages 10–31 years, 39 percent female) and 4 type 2 diabetic patients (ages 44–50 years, 25 percent female) who continued their insulin in addition to taking the herbal intervention. The clinical equivalence of the two arms at baseline was not mentioned. The intervention drug was GS4, a re-crystallized precipitate of the alcoholic extract of the acidic fraction of Gymnema sylvestre; a dose of 400 mg/day was given to patients in Arm II alone. Patients in Arm I served as controls and continued to take their insulin. They were given the same care and education as Arm II patients.

Table 18. Study results for Shanmugasundaram, Rajeswari, (more...)

Table 18. Study results for Shanmugasundaram, Rajeswari, Baskaran, et al., 1990b

	Insulin required (units/day)	Fasting glucose (mg/dl)	Hemo-globin A1c (%)	Choles-terol (mg/dl)	Triglyce-rides (mg/dl)	Free fatty acid (mg/dl)
Arm I
Initial	55	233	12.7	225	124	84
At 10–12 months	55	224	11.8	209	112	80
Arm II	60	232	12.8	206	134	84
Pre GS4
At 6–8 months	45	177	9.5 (p<0.01)	208	121	77

Study results are summarized in Table 18. (Data are reported for the longest followup interval for all patients, 6–8 months.) Some patients continued in the trial for a total of 10–12 months.

There were 11 dropouts (40 percent) in the treatment arm (5 left the city, 1 had frequent episodes of hypoglycemia, and 5 did not get continued support from their families). All patients in Arm II developed hypoglycemic episodes, and their insulin doses were reduced by 10 units at a time. Based on these results, the authors concluded that, compared to the insulin-only arm, those patients who received GS4 supplementation had a reduction in insulin requirement, improved blood glucose homeostasis, and better control of hyperlipidemia. The drawbacks of this study were the lack of a placebo comparison arm, lack of random allocation and blinding, as well as a high dropout rate. The Jadad score was 1.

Effect Size Analysis: Difference of Differences for RCTs and CCTs

As described in Chapter 2, the studies selected for inclusion in Evidence Table 1 were subjected to further analysis. For each of the three outcomes, a common effect size known as a “difference of differences” was estimated for each RCT or CCT that met the inclusion criteria and had data available for that outcome. This statistic is calculated in three steps:

• Step 1—Calculate the difference in the outcome associated with the intervention for the treatment arm (the post-intervention mean value minus the pre-intervention mean value).

• Step 2—Calculate the same difference for the nonherbal arm.

• Step 3—Subtract the results of Step 2 from those of Step 1.

We then constructed a confidence interval for this estimate as described in Appendix K. Given the marked heterogeneity of the studies selected, these effect sizes are only compared graphically across studies in Figure 1; they are not pooled together into an overall effect size.

To facilitate the comparison of these study outcomes, we constructed side-by-side forest plots for the three outcomes: fasting glucose (left), post-prandial glucose (center), and hemoglobin A_1c (right). Each effect size is plotted as a point with its 95-percent confidence interval shown as a horizontal line. Placing the plots side-by-side allows the reader to compare outcomes within a study (across the same horizontal line) and across studies (vertically). We added a vertical reference line at zero (no difference between pre-intervention and post-intervention mean values) for ease of reference. To further help the reader, we state “favors treatment” at the bottom of the forest plots to indicate values of the effect size that are associated with a beneficial effect of the intervention.

The two RCTs are presented first, in descending alphabetical order according to the herbal agent used. These are followed by the five CCTs. The herbal agents are designated by their common English name, if one exists, or by their botanical name. All the agents reported in Figure 1 were tested as single agents.

Figure 1 demonstrates the following results: five studies favor treatment to reduce fasting blood sugar; the remaining two studies, which tested Coccinia indica and Eugenia jambolana respectively, had statistically nonsignificant results. Only four of the seven studies measured post-prandial blood sugar. Of those four studies, only two favored treatment; two did not clearly support treatment. Two of the seven studies measured hemoglobin A_1c, and both favored treatment.

The confidence intervals for some effect sizes are very narrow, which concerned us given the small sample sizes of the studies. We would not expect such precision based on these numbers of patients. We hypothesized that some studies that reported a statistic as the “standard deviation” meant the standard deviation of a mean (the standard error), not the standard deviation of the underlying population. If we mistakenly use a standard error in the calculations discussed in Appendix K, we would construct an overly narrow confidence interval. We attempted to check the accuracy of the data, for example, by inferring from a reported test statistic or p value what the standard error was; but we were unsuccessful at providing adequate evidence to make any changes to the data. We alert the reader to the fact that some of the confidence intervals seem unnaturally narrow. We note that Chandola, Tripathi, and Udupa (1980a), Baskaran, Ahamath, Shanmugasundaram, et al. (1990), and Shanmugasundaram, Rajeswari, Baskaran, et al. (1990b) all reported patient-level data from which we could calculate our results directly; therefore, the confidence intervals are realistic for the effect sizes of these studies.

Pre/Post Comparison Studies

Single Herbs

Chandola, Tripathi, and Udupa (1980a) performed a series of three studies with type 2 diabetic patients to study the hypoglycemic effects of Cinnamonum tamala. All three studies are discussed together in the synthesis of evidence under the heading “Controlled Clinical Trials” above. Only one of the studies was a CCT; the other two were case series of a single herb that had pre/post data available. Of these latter two studies, only the first had an adequate sample size to be included in Evidence Table 2 and the associated Figure 2.

Kamble, Kamlakar, Vaidya, et al. (1998) studied the effects of Coccinia indica on enzymes in the glycolytic and lipolytic pathway in diabetes.⁸ Thirty patients had documented NIDDM (type 2), of which 15 had mild diabetes (fasting blood glucose levels of 120 to 140 mg/dl) and 15 had severe diabetes (fasting blood glucose >200 mg/dl). There were 30 normal healthy controls. The age, gender, and other demographic and socioeconomic characteristics of the patients were not provided.

Patients with severe diabetes were given 50 mg/kg of body weight of a dried extract of Coccinia Indica, which was taken with water twice a day before meals. It was not noted whether these patients were taking any concurrent allopathic medications.

The researchers reported that the mean fasting blood sugar level of the severe diabetic patients treated with Coccinia decreased from 365 mg/dl to 112 mg/dl (p not reported). There were also statistically significant decreases reported in levels of glucose-6-phosphatase, lactate dehydrogenase, and lipoprotein lipase (p<0.001).

The short duration of the study and small number of patients are weaknesses of this study. Although these weaknesses limit the generalizability of the conclusions, the results support the argument for further research on Coccinia indica.

Kuppurajan, Seshadri, Revathi, et al. (1986), at the Central Council for Research in Ayurveda and Siddha, studied the hypoglycemic effect of Coccina indica. A series of patients with adult onset diabetes were recruited (source not described). Although gender was not described, the patients were between 30 and 60 years of age. Patients were included if they were recently diagnosed NIDDM (type 2) confirmed by GTT. Patients were excluded for conditions such as severe hypertension and tuberculosis. Patients were given Coccina indica in the form of a powder made from fresh stems of the plant, which had been pharmacognostically identified prior to being harvested. The powder was placed in capsules and administered to patients at doses of 3 g/day (for mild diabetes) and 4 g/day (for moderate diabetes), in divided doses for 30 days. Patients' urine was tested for sugar on day 11 and day 21. A glucose tolerance test was done on day 31. A total of 16 patients completed the study (10 with mild diabetes and 6 with moderate diabetes). Out of the 10 mild diabetic patients, 8 showed good response to treatment and 2 did not respond; out of the 6 moderate diabetic patients, 4 showed good response to treatment and 2 had no response. Patients who responded showed significant benefit from the treatment in their fasting and 2-hour post-prandial glucose values.

Table 19. Blood glucose levels from Kuppurajan, Seshadri, (more...)

Table 19. Blood glucose levels from Kuppurajan, Seshadri, Revathi, et al., 1986

	Fasting blood sugar (mg/dl) (SD)		2-hour post-prandial blood sugar (mg/dl) (SD)
	Initial	Final	Initial	Final
Responders (n=12)	127.1 (11.2)	95.5 (4.9)	242.4 (32)	168.4 (14.8)
Non-responders (n=4)	123.0 (29.7)	182.5 (80)	205.0 (46.4)	285.5 (80.2)

SD=standard deviation

Table 20. Effects on mean glucose tolerance test in (more...)

Table 20. Effects on mean glucose tolerance test in responder cases (n=12), Kuppurajan, Seshadri, Revathi, et al., 1986

	Fasting	30 min	60 min	90 min	120 min
Mean difference ± SE	-31.6 ± 8.14	20.9 ± 11.9	-58.28 ± 13.3	-59.0 ± 19.4	-74.3 ± 21.5
P value	<0.01	NS	<0.01	<0.02	<0.01

SE=standard error; NS=nonsignificant; min=minute

Table 21. Effects on mean glucose tolerance test in (more...)

Table 21. Effects on mean glucose tolerance test in nonresponder cases (n=4), Kuppurajan, Seshadri, Revathi, et al., 1986

	Fasting	30 min	60 min	90 min	120 min
Mean difference ± SE	59.5 ± 50.9	44.0 ± 19.8	-17.75 ± 25.1	-40.5 ± 26.5	-80.2 ± 46.1
P value	NS	Borderline	NS	NS	NS

SE=standard error; NS=nonsignificant; min=minute

Data were presented separately for the responder and nonresponder arms (Tables 19-21). However, there was no description of differences between responders and nonresponders. Thus no data are available to select patients who would be most likely to respond to the treatment. The authors did demonstrate that some patients will respond favorably to the drug at the stated dosage levels. Lipids were not significantly affected in any patients. The short duration of followup is a significant weakness of this study.

Kuppu Rajan, Srivatsa, Krishnaswami, et al. (1998) conducted a case series study with type 2 diabetic patients using fenugreek (Trigonella foenum-graecum), also known as methica churna in Hindi.

Table 22. Study results of Kuppu Rajan, Srivatsa, Krishnaswami, (more...)

Table 22. Study results of Kuppu Rajan, Srivatsa, Krishnaswami, et al., 1998

Parameter	Initial mean value (SD)	Final mean value (SD)	Level of significance
Fasting glucose (mg/dl)	148.00 (64.4)	128.27 (12.70)	NS
2-hour post-prandial blood sugar (mg/dl)	314.27 (225.19)	264.80 (21.31)	P< 0.02
Cholesterol (mg/dl)	205.33 (9.58)	187.93 (7.30)	P < 0.001
Triglycerides (mg/dl)	142.07 (13.40)	129.42 (8.98)	P < 0.05

SD=standard deviation; NS=nonsignificant; h=hour

Patients, referred from a Western mainstream practice, were studied in collaboration with a research unit in indigenous medicine. Fifteen patients (ages 30–60 years, 9 males, 6 females) with type 2 diabetes confirmed by GTT were enrolled. No inclusion/exclusion criteria were described. All patients were given encapsulated fenugreek powder; each capsule contained 0.5 g of the powder. The total dose was 9 g/day given in three divided doses for 3 months. At the end of 90 days, the glucose tolerance test and tests for blood urea nitrogen, serum creatinine, cholesterol, uric acid, and triglycerides were repeated. Statistically significant reductions between initial and final blood sugar values at 2 hours were observed. The most significant reductions between initial and final measures for the other values were seen in serum cholesterol (mean difference –17.40 ± 4.22, p<0.001) and in serum triglycerides (mean difference -13.25 ± 6.51, p<0.05). Study results are shown in Table 22.

The authors concluded from this study that the traditional use of methica churna (fenugreek) for diabetes is justified by the evidence at the dosage and duration used in this study. It is further noted that the effect on lipids is apparently greater than the effect on blood sugar.

Kumar, Kumar, and Sharma (1999), in a CCT, reported the effects of three different herbal preparations—two combinations and one single herb respectively—on three arms involving type 2 diabetic patients. A total of 111 patients (55 percent female) were recruited from the outpatient department of the Regional Research Center for Ayurveda, Jammu, India. The patients had a fasting blood sugar between 120 mg/dl and 300 mg/dl; had the disease for less than 10 years; and did not have any diabetic complications. Socioeconomic status was reported for class (high class, 30 percent; middle class, 49 percent; low class, 21 percent). The authors did not report if the arms were clinically equivalent prior to the trial.

Patients were classified into three arms. Arm I (n=30) received Ayush-82, a combination of seeds from Mangifera indica, Syzygium cuminii, and Momordica charantia as well as leaves of Gymnema sylvestre and the mineral preparation known as Shuddha Shilajit, which is black bitumen purified in Triphala water. (Triphala is a combination of Terminalia chebula, Terminalia bellerica, and Emblica officinalis.) One gram of Ayush-82 was given three times daily for 6 weeks. Arm II (n=30) received Chandraprabhavati, 1 g; Trivanga bhasma, a mineral preparation, 250 mg; and Vijayasara Kwatha (Pterocarpus marsupium), 25 ml. This treatment was taken daily for 6 weeks. Arm III (n=51) received 3 g of powdered Trigonella foenum graecum (fenugreek) seed taken three times daily for 6 weeks.

The study results were as follows. In Arm I patients, fasting blood sugar improved from 169.17 mg/dl to 120.9 mg/dl (p<0.001), and post-prandial blood sugar fell from 218.4 mg/dl to 172.9 mg/dl (p< 0.001) after treatment; cholesterol levels decreased from 251.6 to 225.33 (p< 0.001) after treatment. In Arm II patients, fasting blood sugar decreased from 168.37 mg/dl to 119.2 mg/dl (p<0.001), and post-prandial blood sugar fell from 234.07 to 181.83 mg/dl (p<0.001); cholesterol levels decreased from 239.1 to 223.63 after treatment (p<0.001). In Arm III patients, fasting blood sugar decreased from 174.63 to 121.06 mg/dl, and post-prandial blood sugar decreased from 236.53 to 183.83 mg/dl (p<0.001); cholesterol levels also decreased from 218.14 mg/dl to 198.29 (p<0.001) after treatment. In all three arms there was significant improvement in the symptoms of polyuria, polyphagia, polydipsia, tiredness, palpitations, anxiety, and pruritis.

The authors concluded that the preparations tested had a highly significant hypoglycemic effect and were effective in symptomatic relief as well. They noted that there were no hypoglycemic episodes despite the significant decrease in blood sugar. The Jadad score for this trial was 0.

Sharma, Sarkar, Hazra, et al. (1996a), in a CCT, studied the effects of fenugreek seeds in 60 NIDDM (type 2) patients (25 percent female, ages 30–70 years) at the S.N. Medical College in Agra, India. The patients were selected from the outpatient clinic, and all had uncontrolled blood sugar levels. Ten normal subjects were also studied; however the article did not report data sufficient for a statistical comparison of these two arms. Of the 60 diabetic patients, 22 had mild disease, 35 had moderate disease, and 7 had severe disease.⁹ Forty of the 60 patients were taking oral hypoglycemic agents, which were continued throughout the study. Patients were instructed to begin a 300 g carbohydrate diet and continue it for a 7-day control period after which a baseline glucose tolerance test was performed. The study also included 10 control patients (30 percent female, ages 30–70 years) chosen from healthy volunteers at the medical center. There was no blinding of either the patients or the researchers.

Patients were given powdered fenugreek seeds, 12.5 g twice a day, for 24 weeks in addition to their carbohydrate-restricted diet. Some patients complained of diarrhea and excess flatus, all of which resolved within 3 to 4 days. No other adverse effects from the fenugreek seeds were observed.

Table 23. Study results for diabetic arm of Sharma, (more...)

Table 23. Study results for diabetic arm of Sharma, Sarkar, Hazra, et al., 1996a

Time (hours)	Blood glucose (mg/dl)		Serum insulin (mU/L)
	Initial	24 weeks	Initial	24 weeks
0	152.45	112.52 (p<0.001)	16.2	17.3
2	259.53	172.41 (p<0.001)	38.2	25.8 (p<0.05)

Diabetic symptoms, such as polyuria, polyphagia, and polydipsia, improved in a majority of patients. The quantitative results are reported in Table 23.

In addition, mean urinary glucose excretion was reduced by 13 percent (p<0.001), and there was a significant reduction in glycosylated hemoglobin of 1.2 percent (p<0.001).

The researchers concluded that the addition of 25 g of fenugreek seed to the daily diet of NIDDM (type 2) patients could be an effective supportive therapy in the prevention and management of long-term complications of diabetes. The results of this study are encouraging, and the use of fenugreek seed in the management of NIDDM (type 2) patients merits further investigation. The Jadad score for this trial was 0.

The Indian Council for Medical Research (1998) and its collaborating centers studied the effects of the bark of Pterocarpus marsupium on 223 mild type 2 diabetic patients. Only patients with newly diagnosed or untreated diabetes were admitted to the study. After 1 month of diet therapy, 124 of these patients had fasting blood glucose between 120 mg/dl and 180 mg/dl and post-prandial blood glucose levels from 180 to 250 mg/dl and were admitted to the herbal study. Of 223 patients enrolled in the diet phase of the study, 15 dropped out during diet therapy; 49 were controlled by diet alone; 35 had blood sugars too high after the diet phase to enroll in the herb study; and 24 dropped out after the herbal study began. In addition, one patient was excluded from the analysis and two withdrew from the study for consistently high blood sugars; this left 97 patients available for analysis. Patients who had any systemic diseases, body mass index < 19, or who were pregnant were excluded from entering the study. The ages of the patients ranged from 35 to 60 years; gender and socioeconomic status were not reported.

All patients were given an extract (aqueous decoction subsequently dried) of the bark of Pterocarpus marsupium. Initially, 1 g of drug was given twice a day. This was increased to 1.5 g twice a day and then to 2 g twice a day, as required to control blood sugar. All patients were also given a standard diet, and they were discouraged from taking drugs for any other ailment.

Results were reported as follows. Four of the 97 patients had to be withdrawn because of post-prandial blood sugar > 300 mg/dl on two occasions. Mean fasting blood sugar of the remaining patients fell from 151 mg/dl to 119 mg/dl (p< 0.001). Mean post-prandial blood sugar fell from 216 mg/dl to 171 mg/dl (p< 0.001), and mean hemoglobin A_1c (reported only for 67 patients) fell from 9.8 to 9.4 (p<0.001). Control of glucose was achieved in 67 of 93 patients, and no hypoglycemic episodes were noted. There was also a reduction in polyuria in all patients who had it, as well as a modest decrease in polyphagia and polydipsia. On a three-point quality of life scale (fatigue, sense of well being, and optimization of weight), there was mild improvement.

The authors concluded that Pterocarpus marsupium is useful in the treatment of newly diagnosed and untreated diabetic patients. The short duration of the study is a limitation. The variation in the dosage is not adequately controlled for, nor are the results reported in relation to the dosage required for control. This lack of detail is another weakness in the study.

Goyal and Tiwari (1999) reported the effects of Vinca rosea (now Catharanthus roseus) on blood sugar levels in a CCT consisting of 40 diabetic and 10 normal patients. The source, duration of disease, age, and sex of the patients were not mentioned. Socioeconomic status of patients, inclusion and exclusion criteria, and the presence of complications were not reported.

The trial drug was prepared using the leaves of Vinca rosea made into a paste. The patients were classified into four arms. Arm I (n=25) consisted of diabetic patients with fasting blood sugar ranging from 100 mg/dl to 200 mg/dl. They received 5 g of the trial drug twice a day for 30 days. Arm II (n=10) consisted of diabetic patients with fasting blood sugar levels <200 mg/dl. They were given 5 g of the trial drug twice a day along with the oral hypoglycemic agent glipizide given at 2.5 mg once a day for 20 days. Arm III (n=10) consisted of subjects who were of normal health and had normal fasting blood sugars. They were given 5 g of the trial drug twice a day for 10 days. Arm IV (n=5) consisted of diabetic patients who were given a single 10 g dose of the trial drug.

Results were as follows. In Arm I patients, the average fasting blood sugar decreased from 131.37 mg/dl to 98.98 mg/dl (p< 0.001) after 1 month of treatment. Post-prandial blood sugar fell from 224.90 mg/dl to 144.24 mg/dl after 30 days of treatment (p<0.001). In Arm II patients, the average fasting blood sugar fell from 217.45 mg/dl to 91.55 mg/dl (p<0.001) after 20 days of treatment. Post-prandial blood sugar fell from 314.50 mg/dl to 140 mg/dl (p<0.001). In Arm III and Arm IV there was no significant decrease in fasting or post-prandial blood sugars. Symptomatic improvement was assessed in Arm I and Arm II patients only. These patients showed statistically significant (p<0.001) improvement in polyuria, polydipsia, and weakness, while there was no significant improvement in polyphagia, cramps, libido, joint pain, and weight loss.

The authors concluded that treatment with the leaves of Vinca rosea caused a significant improvement in blood sugar levels of diabetic patients and also improved the symptoms of polyuria, polydipsia, and weakness. The authors did not describe how they assigned patients to the study arms. The Jadad score for this trial was 0.

Combinations and Formulas

Chowdhary, Dua, Bharti, et al. (1998) studied the effect of Ayush-82, an Ayurvedic formula, on 100 non-insulin dependent diabetic patients. The patients (48 percent female) were randomly selected from both the inpatient and outpatient departments of the Central Research Institute of New Delhi to participate in the study. Patients were included if they were diabetic for less than 3 years and had no evidence of diabetic complications. In this study, 52 percent of the patients were between 40 and 55 years old, and 48 percent were between 55 and 70 years old. Socioeconomic status was reported for both education level and income (22 percent high income, 50 percent middle income, 28 percent low income; 30 percent no education, 41 percent school level, 28 percent college level).

The drug treatment, Ayush-82, contained four herbs: the seeds of Mangifera indica, Syzygium cuminii ¹⁰, and Momordica charantia, and the leaves of Gymnema sylvestre. Five grams of the powder was given three times per day for 6 weeks. This preparation was given in conjunction with ShuddhaShilajit, a mineral preparation of black bitumen purified in Triphalawater (Triphala is a combination of Terminalia chebula, Terminalia bellerica, and Emblica officinalis.). The SudhaSilajitu was given in a 500 mg dose twice a day for the 6-week period. All patients were advised to consume a 1,200-calorie diet daily. Patients who were taking oral hypoglycemic medication at the start of the study were withdrawn from their drugs after 2 weeks of treatment with the Ayurvedic formula. No comparison arm was included in this study.

The results were as follows. Average blood sugar prior to treatment was 169.33 mg/dl, and after 6 weeks of treatment it was 144.20 mg/dl. The mean fall in fasting blood sugar was 25.21 mg/dl, and this was significant (p<0.001). Likewise, post-prandial blood sugar fell from an average of 249.6 mg/dl pre-treatment to 219.2 mg/dl post-treatment for an average fall of 30.43 mg/dl (p<0.001). Despite the significant decline in blood sugar, only 25 percent of patients were reported to have achieved good or complete control of their diabetes, 17 percent of patients were mildly controlled, and 47 percent did not have their diabetes controlled by the end of the study. During the treatment period, there was an improvement in the patient reporting rate of polyuria (40 percent to 22 percent), polydipsia (38 percent to 18 percent), polyphagia (24 percent to 16 percent), and weakness (47 percent to 23 percent).

The researchers concluded that Ayush-82 exerted a beneficial effect in NIDDM (type 2) patients, making it possible for them to reduce their use of pharmaceuticals and to improve their physical symptoms. The ambiguity of the process of withdrawal of oral medication is a design problem of this study. Further, the clinical response was not great since half the patients still remained uncontrolled.

Kumar, Kumar, and Sharma (1999) is described above under “Single Herbs.” It is listed in this section as well because combinations of herbs were used in two of the three study arms.

Pandey, Rajagopalan, and Chowdhary (1995) studied the effects of Ayush-82, an Ayurvedic herbal formulation, on diabetic patients. A total of 350 type 2 diabetic patients attending the outpatient department of the Central Research Institute of Ayurveda in New Delhi were screened. Out of these, 80 patients who were older than 40 years and who had the disease less than 3 years, with no neuropathy, nephropathy, retinopathy, or current infections and who were compliant with medications were selected. Of these 80 patients, 30 (46 percent female) were between 40 and 49 years old; 38 patients (55 percent female) were between 50 and 59 years old; and 12 patients (25 percent female) were more than 60 years old. Among the patients, 8.7 percent belonged to low-income groups, 76 percent belonged to middle-income groups, and 15 percent belonged to high-income groups.

The intervention consisted of the herbal formula Ayush-82. This contains four herbs: the seeds of Mangifera indica, Syzygium cuminii, and Mormodica charantia and the leaves of Gymnema sylvestre. These herbs were powdered together, and 5 g of the powder was taken three times a day for 24 weeks. This preparation was given in conjunction with Shuddha Shilajit, a mineral preparation of black bitumen purified in Triphala water. A 500 mg dose of this formulation was given twice a day for the 24-week duration. All patients were advised to consume a 1,200-calorie diet. Additionally, patients who were taking an oral hypoglycemic agent were withdrawn from this drug over a 15-day period prior to commencement of the study.

Additionally, the researchers reported that on a physician rating scale, 61 percent of the patients had a good response to therapy, 12.9 percent had a fair response, and 25.9 percent had a poor response.

Table 24. Study results for Pandey, Rajagopalan, and (more...)

Table 24. Study results for Pandey, Rajagopalan, and Chowdhary, 1995

Values	Blood sugar before treatment (mg/dl)	Blood sugar after treatment—24 weeks (mg/dl)
Fasting (males, n=42)	195 ± 48	131 ± 64 (p<0.001)
Fasting (females, n=38)	191 ± 48	141 ± 64 (p<0.001)
Post-prandial (males, n=42)	288 ± 74	204 ± 89 (p<0.001)
Post-prandial (females, n=38)	279 ± 105	201 ± 83 (p<0.001)

Note: The results were not reported for males and females as a whole group.

Study results are shown in Table 24. The authors concluded that the Ayurvedic formulation Ayush-82 lowers both fasting and post-prandial blood sugar levels in both sexes. The lack of a control arm and the ambiguity of the process of withdrawal of oral hypoglycemic agent are weaknesses of this study.

Shankar and Singhal (1995) studied the effects of Abraga Chendooram, an Ayurvedic formulation, on 130 patients with NIDDM (type 2). Patients were selected from the ambulatory clinic at Safdarjang Hospital in New Delhi. All patients (n=130, 43 percent female, ages 20–70 years) chosen for this study were given a glucose tolerance test and failed dietary control of their blood sugar for 15 days. The socioeconomic characteristics of the patients were not mentioned.

Patients were given a formula called Abraga Chendooram, which consists of Abragam (purified black mica, 80 g); Vengaram (dehydrated borax, 0.5 g); Saranaiver charu (juice of the root of Trianthema decandra Linn.); Adathodaielai charu (juice from the leaves of Adhatoda zeylanica Linn.); and Alam vizhuthu kudineer (root of Ficus benghalensis Linn.). Patients were given 200 mg of the drug (in gelatin capsules) twice daily for 45 days. During this period, caloric intake was restricted to 25 calories per kg of ideal body weight. No mention was made of whether or not any patients were taking any concurrent allopathic medications for diabetes.

The study showed a statistically significant reduction in mean FBS from 172.3 mg/dl to 110 mg/dl, which was a difference of 62.30 mg (p<0.005), and a reduction in the PPBS from 267.7 mg/dl to 183.0 mg/dl, a difference of 84.70 mg/dl (p<0.005). The authors created a secondary classification system consisting of “good response” (FBS and PPBS at or below normal levels with disappearance of all symptoms); “moderate response” (FBS 120–140 mg/dl and PPBS 180–200 mg/dl with disappearance of major symptoms); and “no response” (no change in blood sugar level or symptoms). The study showed that 57.69 percent of all patients had a good response, 26.93 percent showed a moderate response, and 15.38 percent showed no response. No adverse effects were reported from this study.

The researchers concluded that Abraga Chendooram reduces the blood sugar level in patients with mild to moderate NIDDM (type 2) when combined with dietary control; in 58 percent of cases, it brings blood sugar levels into the normal range. The short duration of followup is a design flaw of this study.

Maji and Singh (1995) reported the effects of an herbo-mineral preparation, D-400, on the blood sugar and cholesterol levels of diabetic patients. The patients (n=38, ages 35–76 years, percent female not reported) were classified into three arms. Arm I (n=19), consisted of patients who had failed therapy with oral hypoglycemic agents; Arm II (n=8) consisted of newly diagnosed diabetic patients who failed diet and exercise therapy; and Arm III (n=6) consisted of patients who were dependent on insulin therapy. The socioeconomic status of the patients and the clinical equivalence of the three arms were not reported.

The intervention consisted of an herbo-mineral compound, D-400, consisting of: Eugenia jambolana,¹¹ Pterocarpus marsupium, Ficus glomerulata, Gymnema sylvestre, Momordica charantia, Ocimumsanctum, and Shilajit. The concentrations of each of these ingredients and the method of preparation were not mentioned. All three arms of patients received two tablets of the drug three times a day for 6 months. All the patients were given a standard diet with the number of calories determined by the patient's body mass index.

Table 25. Study results for Maji and Singh, 1995

Table 25. Study results for Maji and Singh, 1995

	Arm I	Arm II	Arm III
Pre-treatment FBS (mg/dl)	208.47	219.88	267.33
Post-treatment FBS (mg/dl)	113.05 (p<0.01)	109.00 (p<0.01)	215.00
Pre-treatment PPBS (mg/dl)	262.00	284.13	311.17
Post-treatment PPBS (mg/dl)	141.16 (p<0.01)	131.00	302.17
Pre-treatment cholesterol (mg/dl)	194.82	182.57	182.83
Post-treatment cholesterol (mg/dl)	172.00 (p<0.01)	162.29 (p<0.01)	166.50 (p<0.01)
Pre-treatment HDL (mg/dl)	71.15	54.00	51.60
Post-treatment HDL (mg/dl)	80.85 (p<0.05)		54.60
Pre-treatment LDL (mg/dl)	170.73	118.25	168.40
Post-treatment LDL (mg/dl)	149.64 (p<0.01)	101.75	154.20

FBS=fasting blood sugar; PPBS=post-prandial blood sugar; HDL=high-density lipoprotein; LDL=low-density lipoprotein

Study results are provided in Table 25. The authors concluded that D-400 was effective in lowering blood sugar as well as cholesterol in patients who had failed oral hypoglycemic therapy and in newly diagnosed diabetic patients. It was less effective in patients who were already on insulin.

Sircar, Ahuja, Natu, et al. (1996) reported the effects of the Ayurvedic formulation MA-471 on blood sugar levels, on hemoglobin A_1c, and on cholesterol in 69 type 2 diabetic patients. Patients were recruited from the diabetic clinics of Gandhi Memorial Hospital and associated hospitals in Lucknow, India. Patients who were obese and had diabetic complications were excluded. No mention is made of the age and gender or the socioeconomic characteristics of these patients. Of the 69 patients, 9 were excluded (6 due to noncompliance and 3 due to other illnesses), leaving 60 patients completing the study.

The patients were classified into three arms. Arm I (n=15) consisted of patients who had never been on an anti-diabetic agent and were uncontrolled despite diet and exercise. Arm II (n=30) consisted of patients who were well controlled on an oral hypoglycemic agent. Arm III (n=15) consisted of patients who were not controlled despite maximum dose of oral hypoglycemic agents.

The Ayurvedic formulation MA-471 contained the herbs Enicostema littorale, Phyllanthus niruri, Eugenia jambolana, Melia azadirachta,¹² Terminalia arjuna, and Aegle marmelos as well as a mineral preparation of black bitumen called Shilajit. The herbs were processed in an aqueous extract made from the leaves of the Aegle marmelos and from the fruits of the Momordica charantia. All patients received MA-471, two 500 mg tablets twice a day for an average of 9 months. Arm I patients received only MA-471; Arm II patients had their oral hypoglycemic agent withdrawn in addition to receiving MA-471; and Arm III patients continued to take both the oral hypoglycemic agent and MA-471. All patients were prescribed a standard diet.

Table 26. Study results for Sircar, Ahuja, Natu, et (more...)

Table 26. Study results for Sircar, Ahuja, Natu, et al., 1996

Values	Pre/post treatment	Arm I	Arm II	Arm III
Fasting blood sugar (mg/dl)	pre	168.72	116.91	151.64
	post	124 (p<0.001)	103.64 (NS)	102.90 (p<0.001)
Post-prandial blood sugar (mg/dl)	pre	292.73	173.64	249.64
	post	221.09 (p<0.001)	171.82 (NS)	171.27 (p<0.001)
Hb A1c (%)	pre	10.3	7.0	10.1
	post	7.3 (p<0.05)	7.6 (NS)	6.4 (p<0.05)
Cholesterol (mg/dl)	pre	236.4	242.3	227.6
	post	182.3 (p< 0.01)	171.5 (p<0.001)	170.4 (p<0.01)
Triglycerides (mg/dl)	pre	200.3	180.3	210.6
	post	149.3 (p<0.05)	148.7 (p< 0.01)	159.2 (p<0.05)
HDL (mg/dl)	pre	33.8	34.8	31.6
	post	37.5 ( NS)	39.8 (NS)	35.8 (NS)

Hb=hemoglobin; HDL=high-density lipoprotein; NS=nonsignificant

In addition to the results shown in Table 26, significant improvement was also noted in the symptoms of polyuria, fatigue, and constipation; moderate improvement was noted in weakness, polydipsia, giddiness, muscle pain, palpitations, and anorexia.

The authors concluded that MA-471 has significant hypoglycemic properties as well as hypolipidemic properties. Additionally, they noted that there was great improvement in symptoms, which translated into better quality of life.

Shankar and Singhal (1994) reported the effects of a Siddha herbal preparation, Sandana Podi-a, on the blood sugar and symptoms of an arm of type 2 diabetic patients. The patients (n=20, 40 percent female, ages 20–70 years) were recruited from the diabetic clinic of Safdarjang Hospital, New Delhi, India. They had mild to moderately severe disease based on their blood sugar levels. Patients with diabetic complications were excluded from the study. No data about the socioeconomic status of the patients were reported.

The intervention consisted of Sandana Podi-a, an herbo-mineral preparation containing: 1 part Santalum album (sandalwood) saw dust; 1 part Andropogon citratus (lemongrass) root; 1 part Vetiveria zizanioides (vetiver) root; 1 part Syzygium aromaticum (clove) flower bud; 1 part Anacyclus pyrethrum (pyrethrum) root; 1 part of purified Shilajit, and 6 parts of Tinospora cordifolia sugar. This sugar was prepared by using whole chopped and crushed plant, which sat in water overnight. Then the water was decanted and placed in the sun, and the dried material was then re-suspended in water overnight. This process was repeated three times. Finally, the starch from this water was collected and dried. All ingredients for the preparation were ground into a fine powder, which was placed into capsules. All patients were given 300 mg of powdered drug twice a day for 45 days.

The results were reported as follows. The mean fasting blood sugar before treatment was 164.5 mg/dl, which decreased to 114.7 mg/dl after treatment. Mean post-prandial blood sugar before treatment was 281.7 mg/dl, and this decreased to 171.25 mg/dl after treatment. Both differences were statistically significant at p<0.01. Similarly, there was improvement in all symptoms, such as polyuria, excessive appetite, weight, giddiness, polydipsia, indigestion, vomiting, and abdominal pain.

The authors concluded that Sandana Podi-a reduces blood sugar and improves symptoms in mild-to-moderate type 2 diabetic patients. The short duration of the study is a major drawback.

Sivaprakasam, Rao, Yasodha, et al. (1984) studied the effects of two Siddha herbal preparations (Kadal Azhinjil choornam and Triphala tablets) on the blood sugar and symptoms of 25 type 2 diabetic patients. These patients were between the ages of 30 and 65 years (24 percent female) and were recruited from the outpatient department of the Central Research Institute for Siddha, Madras, India. The diagnosis of diabetes was made using Siddha as well as modern clinical and biochemical methods. There was no mention of the socioeconomic status or demographics of the patients.

All patients were asked to stop taking any medications that they had been on previously. The patients were then given two preparations. The first was Kadal Azhinjil choornam, botanically identified as Salacia chinensis. The roots and bark of this plant were used in a dose of 500 mg taken twice a day with water. The Triphala tablets contained three components: Terminalia chebula, Terminalia bellerica, and Phyllanthus emblica (also Emblica officinalis) in equal proportions. The 2.5 g Triphala tablets were taken three times a day with water. The total duration of the study was 120 days.

Table 27. Study results for Sivaprakasam, Rao, Yasodha, (more...)

Table 27. Study results for Sivaprakasam, Rao, Yasodha, et al., 1984

Clinical symptoms	Patients with symptom before treatment (%)	Patients with symptom after treatment (%)
Polyuria	24	20
Polyphagia	44	36
Polydipsia	52	44
Nocturia	52	40
Tiredness and general weakness	48	40
Giddiness	44	32
Pruritis	32	24
Pruritis vulva	12	8
Peripheral neuritis	24	16
Blurring of vision	48	16
Constipation	52	40

Results are shown in Table 27. No statistical tests of significance were reported for any of these percentages. Improvements in FBS values were reported before and after treatment. The mean blood sugar for the arm improved from 303.88 mg/dl to 184.68 mg/dl. Additionally, the authors reported that 4 patients showed a reduction in blood sugar of more than 50 percent; 11 showed a reduction of 40–50 percent; 6 showed less than a 40-percent reduction; and 4 showed no significant reduction.

The authors concluded that that the preparations have a definite effect on the blood sugar level. The ambiguity of reporting the results is a limitation of this study.

Effect Size Analysis: Difference in Mean Values for Pre/Post Comparison Studies

As described in Chapter 2, the studies selected for inclusion in Evidence Table 2 were subjected to further analysis. All studies included had treatment arms with measurements of blood glucose, hemoglobin A_1c, or post-prandial blood sugar before and after an intervention of either single herbs or groups of herbs. This subgroup of studies included CCTs, cohorts, and case series. For each of the three outcomes, a common effect size (a difference of means) was estimated for each arm of each study that met the inclusion criteria and had data available for that outcome. For a particular arm in a study, this statistic is the difference in the outcome associated with the intervention; i.e., the post-intervention mean value minus the pre-intervention mean value. We then constructed a confidence interval for this estimate as described in Appendix K. Given the marked heterogeneity of the studies selected, these effect sizes are only compared graphically across studies in Figure 2; they are not pooled together into an overall effect size. (Figure 2 is constructed similarly to Figure 1.)¹³

The studies in Figure 2 are in the following order: Studies or arms of studies using a single herbal preparation are listed first. Within that group, the studies are sorted alphabetically by the name of the herbal preparation used. Next, the group of studies that use a formula or a group of herbs and minerals is displayed. This group is sorted alphabetically in descending order by the name of the formula. Note that some studies with multiple arms will have listings in more than one place on the graph. Otherwise, the construction of Figure 2 is the same as for Figure 1.

As Figure 2 demonstrates, the majority of pre/post comparison studies using single herbs and formulas favor treatment. Hemoglobin A_1c is measured in only five of the studies. Two of those studies seem to strongly favor treatment; two are almost equivocal; and one does not favor treatment. As with Figure 1, we remind the reader that we relied on the data as reported in the individual studies, even though some of the resulting confidence intervals seem unnaturally narrow. We also note the Chandola, Tripathi, and Udupa (1980a) and Sivaprakasam, Rao, Yasodha, et al. (1984) reported patient-level data from which we could calculate our results directly; therefore, the confidence intervals are realistic for the effect sizes of these studies.

Studies Not Included in Further Analysis

Limitations of the Review

A number of factors prevent drawing any stronger conclusions from the data other than the ones presented above.

• Our searches showed that the Ayurvedic literature emphasizes botanical therapies. However, the structure of our search may have exaggerated this emphasis. Therefore, no conclusions can be drawn about the use of Ayurveda as a whole system for the treatment of diabetes.

• Articles without abstracts were excluded from the initial review by default. In addition, it was difficult to accurately complete the screening form based on titles alone, and the number of titles was too large to order all of the articles. It is possible that the inclusion of all of these articles might have changed the frequency of disease states in the initial review and, hence, the choice of our topic for in-depth review.

• Although extensive efforts were made to obtain the relevant Indian literature, studies not published in English were not included, and a handful of studies identified in India could not be retrieved. This means that there is information that was not included in the review.

• There were significant methodological difficulties with the literature analyzed. In general, random assignment to treatment was used infrequently; comparison arms were not included; and many of the studies had insufficient subjects to demonstrate effect sizes other than very large ones. In addition, we could not calculate a common statistical measure in all studies. Thus, we were limited in the comparisons we could make between different therapeutic interventions, and we had no way to check for publication bias in either the Western or Indian literature.

• A number of the studies tested immediate or very rapid effects of the herbs used in the intervention. Although this might be interesting to elucidate the mechanism of action of the herb, it did not replicate normal clinical practice, and it limited our ability to generalize based on these data.

• All of the studies that we identified and analyzed were not of equal methodological quality. In order to avoid giving all studies equal weight (and perhaps thereby giving undue weight to studies of poorer quality), we tried to emphasize studies of better quality by subjecting them to further analysis.

Introduction

Ayurveda is a Sanskrit word that translates into knowledge (veda) of life (ayur). Ayurveda refers to the traditional medicine that developed in India and has been practiced there for roughly three thousand years. Ayurveda is the sum of early magico-religious beliefs, which matured into an empirical or rational science and which incorporated influences from other regional medical practices.

The historical roots of Ayurvedic teachings are shrouded as much in myth as in tradition. Our understanding of its origin and evolution is limited in several respects: There is uncertainly about dates associated with the major texts; biographical material on the important medical writers is sketchy; and English translations of texts and related documents are often incomplete and inaccurate. Furthermore, available Ayurvedic texts often contain a mixture of legend and facts that are difficult to separate.

This appendix is a best effort at summarizing the history, beliefs, and practices of Ayurveda, but in no respect can it be considered complete. Ayurveda is a rich and sophisticated form of medicine that is unlike Western medicine. Any errors or omissions made in this summary are unintentional. The references consulted for the historical background on Ayurveda include: Ackerknecht, 1982; Barnett and Barone, 1996; Castiglioni, 1947; Frawley and Lad, 1988; Kutumbiah, 1962; Lad, 1999; Lad, 1995; Lad, 1985; Magner, 1992; Mishra, Singh, and Dagenais, 2001a, 2001b; International Society of Alternative Medicine, 1987; Porter, 1997; Sanyal, 1964; Sharma and Clark, 1996, 1997; Sigerist, 1961; Sodhi, 1992; Shah, 1995; Thorwald, 1963; Udupa, 1977; Udupa, 1985; Zysk, 1996.

History of Ayurveda

Properties of Ayurvedic Herbs

Ayurvedic herbs are described and classified according to five major properties: rasa (taste), guna (physicochemical properties), veerya (potency), vipaka (post-digestive effect), and prabhava (unique effect of the drug) (Frawley and Lad, 1988; Sanyal, 1964). (Note that the term rasa, used here to mean taste, can also mean dhatus or tissues. The two terms are distinguished based on context.)

Rasa is divided into six major tastes: madhura (sweet), amla (sour/acid), lavana (salty), katu (pungent), tikta (bitter), and kashaya (astringent). In Ayurveda, each taste is made of a combination of two of the five basic elements or mahabhutas (earth, water, fire, air, and ether). Each taste has corresponding effects on the three bodily doshas (pitta, vata, and kapha). Appendix Table 6 summarizes the characteristics of the major tastes (Dash, 1987; Frawley and Lad, 1988; Sanyal, 1964).

Appendix Table 6. Characteristics of the major tastes (rasa)

Rasa (taste)	Elements	Action	Increases	Decreases	Restorative property
Madhura (sweet)	Earth + water	Soothing	K	P
Amla (sour/acid)	Earth + fire	Soothing	P/K	V
Lavana (salty)	Water + fire	Soothing	P/K	V
Katu (pungent)	Fire + air	Exciting	P/V	K
Tikta (bitter)	Air + ether	Exciting	V	K/P
Kashaya (astringent)	Air + earth	Exciting	V	K/P

P = pitta; V = vata; K = kapha

Source: Dash, 1987; Frawley and Lad, 1988, Sanyal, 1964.

Guna represents the more physical aspects of a medicinal substance. There are five major classes of guna, and each class corresponds to one of the major elements: heaviness corresponds with earth; unctuousness with water; keenness and sharpness with fire; dryness with air; and light with ether. Gunas are generally considered in pairs: light/heavy, wet/dry, etc. There is an extensive subdivision of guna based on combinations with the elements, but its description is beyond the scope of this report (Sanyal, 1964; Udupa, 1977).

Veerya represents the active principle or potency of a drug. Factors such as growth conditions, harvesting technique, and storage affect an herb's veerya. Various authorities have different classification systems for this attribute. The simplest system classifies veerya as having properties that range between hot and cold. Substances that heat are pitta dominant; cooling ones are kapha and vata dominant.

Vipaka is the quality a substance takes on after it has been acted on by the body. After digestion, which is an important bodily function in Ayurveda, the rasas are simplified into three groups corresponding to three vipakas. The attributes of the vipaka system are summarized in Appendix Table 7 (Frawley and Lad, 1988; Nadkarni and Nadkarni, 1976; Sanyal, 1964).

Appendix Table 7. Attributes of Vipaka

Vipaka (post-digestion taste)	Rasa (pre-Digestion)	Increases dosha	Corresponding food
Madhura (sweet)	Sweet or salty	Kapha	Carbohydrate
Amla (sour)	Sour (unchanged)	Pitta	Protein
Katu (bitter)	Bitter or pungent or astringent	Vata	Fat

Source: Frawley and Lad, 1988; Nadkarni and Nadkarni, 1976; Sanyal, 1964.

Prabhava refers to a drug's unique influence on the body. Ayurvedic practitioners have observed that drugs may have the same rasa, guna, veerya, and vipaka and yet have different actions in the body. The drug's prabhava accounts for these differences (Udupa, 1977).

Reviewers

We gratefully acknowledge the following individuals² who reviewed the initial draft of this report and provided us with constructive feedback. Acknowledgments are made with the explicit statement that this does not constitute endorsement of the report.

Mayer B. Davidson, M.D.

Director, Clinical Trials Unit

Professor of Medicine

UCLA School of Medicine

Charles R. Drew University of Medicine & Science

Los Angeles, CA

Michael S. Dick, MA

The Ayurvedic Institute

Albuquerque, NM

Dr. James Duke

Herbal Vineyard

Fulton, MD

David Frawley, OMD

Director

American Institute of Vedic Studies

Santa Fe, NM

Scott Gerson, M.D.

Ayurvedic Medicine of New York

New York, NY

David Heber, M.D., Ph.D.

Professor of Medicine

Director, UCLA Center for Human Nutrition

Los Angeles, CA

Simon Mills,* MA, MCPP, FNIMH

Director for the Center of Complementary Medicine

University of Exeter

Exeter, England

Lakshmi C. Mishra,* BIMS, M Pharm, Ph.D.

Professor of Research

Southern California University of Health Science

Whittier, CA

Shri K. Mishra,* M.D., MS, Doctor of Ayurveda

Professor of Neurology and Coordinator

Integrative Medicine

USC School of Medicine

USC Keck School of Medicine

Los Angeles, CA

Robert Schneider, M.D.

Director

Center for Natural Medicine & Prevention

Dean, College of Maharishi Vedic Medicine

Maharishi University of Management

Fairfield, IA

Betsy B. Singh,* PhD

Dean of Research

Southern California University of Health Sciences

Whittier, CA

Robert Svoboda, BS, BAMS, Ayurvedacharya

The Ayurvedic Institute

Albuquerque, NM

Technical Expert Panel

We also wish to acknowledge the work of our technical expert panel. In addition to the four members who also acted as reviewers (see preceding page), the panel included the following individuals:

Betty L. Chang, DNSc, FNP-C, FAAN

Professor, School of Nursing, University of California, Los Angeles

Los Angeles, CA

Francesco Chiappelli, Ph.D.

Laboratory of Human Oral & Molecular

Immunology, School of Dentistry

University of California, Los Angeles

Los Angeles, CA

David Diehl, M.D.

Associate Clinical Professor of Medicine

Western Gastroenterologists, Inc

Irvine, CA

Seigward-Markus Elsas, M.D.

Clinical Fellow in Neurophysiology, University of California, Los Angeles

Los Angeles, CA

Glenn Clark, DDS

School of Dentistry

University of California, Los Angeles

Los Angeles, CA

Deborah Glik, Ph.D.

Associate Professor, School of Public Health

University of California, Los Angeles

Los Angeles, CA

Michael Goldstein, Ph.D.

Professor, School of Public Health

University of California, Los Angeles

Los Angeles, CA

Eric Hurwitz, D.C., Ph.D.

Assistant Professor, Department of Epidemiology, School of Public Health, University of California

Los Angeles, CA

Ka Kit Hui, M.D., FACP

Director, UCLA Center for East-West Medicine, University of California

Los Angeles, CA

Lucy Postolov, LAc

Postolova Acupuncture Group

Los Angeles, CA

David Riley, M.D.

Clinical Associate Professor, University of North West Medical School, Director, Integrative Medicine Research

Santa Fe, NM

George Solomon, M.D.

Professor Emeritus, UCLA School of Medicine, Dept. Psychiatry and Biobehavioral Medicine

University of California

Los Angeles, CA

Hitoshi Tomizawa, M.D.

Director, Japanese Executive Medical Services, Cedars-Sinai Medical Center

Los Angeles, CA

Xiao-Ping Xu, LAc

Burns and Allen Research Institute

Cedars Sinai Medical Center

Los Angeles CA

Contacts in India

Dr. Swamy Venturupalli would also like to acknowledge and thank the following individuals:

M.S.Baghel, M.D. (Ay) Ph.D., WHO Fellow, I/C Reader- Kayachikitsa

Institute of Post Graduate Teaching and Research in Ayurveda

Gujarat Ayurveda University, Jamnagar, Gujarat, India

Dr. Chandola, M.D. (Ay)

I/C Reader- Kayachikitsa

Institute of Post Graduate Teaching and Research in Ayurveda

Gujarat Ayurveda University, Jamnagar, Gujarat, India

Dr. Kohli, M.D. (Ay)

Dept. of Kaya-Chikitsa

Podar Ayurvedic Medical College

Mumbai, India

Adarsh Kumar, M.D. (Ay)

Assistant Research Officer

Central Council for Research in Ayurveda and Siddha

New Delhi, India

D.K. Mishra, BAMS, M.D. (Ay)

Assistant Director (Ay)

Central Council for Research in Ayurveda and Siddha

New Delhi, India

Bibhas Ray, MS, Ph.D.

Consultant - Coordination and Liaison- Drug Development

Dr. Reddy's Research Foundation

New Delhi, India

V. Ravi Shankar, M.D. (Ay)

Assistant Director (Siddha)

Central Council for Research in Ayurveda and Siddha

New Delhi, India

Dr. Gurudeep Singh

Dean, Institute of Post Graduate teaching and Research in Ayurveda

Gujarat Ayurveda University, Jamnagar, Gujarat, India

Dr. Kulwant Singh

Department of Surgery

Institute of Post Graduate Teaching and Research in Ayurveda

Gujarat Ayurveda University, Jamnagar, Gujarat, India

G. Veluchamy, M.D. (Ay)

Director

Central Council for Research in Ayurveda and Siddha

New Delhi, India

A. Venkateshwarulu, Ph.D.

President- Research Division.

Dr. Reddy's Laboratories, Hyderabad, India.

R.S. Yadava

Assistant Director (Library)

Central Council for Research in Ayurveda and Siddha

New Delhi, India

Others

Dr. Mary Hardy would also like to acknowledge and thank the following individuals:

Ruchi Mathur, M.D.

Endocronologist Fellow

Cedars Sinai Medical Center

Los Angeles, CA

Shannon Rhodes, MSA

Staff Assistant

Evidence-Based Practice Center

RAND

Santa Monica, CA

Nadia Khatib

Administrative Assistant

Evidence-based Practice Center

RAND

Santa Monica, CA

Jay Udani, M.D.

Los Angeles, CA

Initial Searches

Final Search

Single Herbs

The effect on the patients' tridosha was also reported. Highly significant (p<0.001) relief was found in the kapha- and pitta-dominant patients, with reductions of 89 percent and 73 percent of an unvalidated symptom score described by the authors. In their paper, the authors describe the action of Cassia auriculatain terms of Ayurvedic causality, but they do recommend its use in diabetic patients generally. Interpretation of these results is limited by the small numbers of patients and the subjective nature of the dosha assessment. The scale used by the authors could be useful if validated in larger groups of patients or if the dosha evaluations were performed by two independent examiners who reconciled any differences.

Sivaprakasam, Rao, Anandan, et al. (1985) studied the effects of Avarai elai (Cassia auriculata) and Koiiya elai (Psidium guajava, guava in English) on confirmed NIDDM (type 2) patients from the outpatient clinics of the Central Research Institute for Siddha and the Drug Research Scheme in India. Twenty-five patients were treated with Avarai elai (1 g, three times a day), and 20 patients were treated with Koiyya elai (1g, three times a day). Five patients dropped out of the Koiyya elai arm. Patients were instructed to follow the World Health Organization diabetic diet regimen and were instructed not to take any other hypoglycemic agents. Patients were treated for 24 months; however, there was no intent to treat analysis on patients who did not take the preparations for the full 24 months.

The researchers concluded that the formulations of Koiyya elai and Avarai elai at a dose of 1 g three times a day failed to show any significant reduction in blood glucose levels. There are numerous design flaws as noted above. The most significant of these was sample attrition, given the fact that less than 4 percent of the Avarai arm and 0 percent of the Koiyya arm remained in the study for the full 24 months. No adverse effects were reported from this study.

Chandola, Tripathi, and Udupa (1980b) reported the effects of Cinnamonum tamala on serum insulin levels and on fasting blood sugar in NIDDM (type 2) patients. The researchers identified a total of nine patients from the outpatient clinic at the Sunderlal Hospital in Varanasi, India, who had NIDDM confirmed with a glucose tolerance test. The researchers performed two separate studies measuring the effect of Cinnamonum tamala on fasting blood sugar and serum insulin levels. The first study involved five patients (20 percent female) who were given Cinnamonum tamala at a dose of 2 teaspoons four times a day for 15 days. The second study involved four patients who were given a single dose of Cinnamonum tamala (20 g) and were then followed for 4 hours. There were no comparison arms. Patients were asked to follow an 1,800-calorie diabetic diet during the study.

The preparation of Cinnamonum tamala leaves used in this study was a commercially available kitchen spice, called “Tej-patra,” which the researchers dried, pulverized, and filtered before giving it to the patients. There was no mention as to whether or not any patients were taking any concurrent allopathic medications for diabetes.

Patients in the 15-day study showed a statistically significant reduction in mean fasting blood sugar (56.6 mg/dl p<0.05) and a nonsignificant increase in plasma insulin levels (9.74 μU/ml p>0.05). Patients in the single-dose study (C. tamala given up to 2 hours before a GTT was done) also showed a significant reduction in mean fasting blood sugar (56.5 mg/dl p<0.05) as well as a significant increase in serum insulin levels (23 μU/ml p<0.05). No adverse effects were reported from this study.

The researchers concluded that Cinnamonum tamala has a positive effect on the release of insulin from the beta-cells of the islets of Langerhans. This conclusion was drawn solely from the results of the 2-hour study and was not corroborated by the15-day study performed by the same authors.

Chandola, Tripathi, and Udupa (1980a) performed a series of three studies with type 2 diabetic patients to study the hypoglycemic effects of Cinnamonum tamala. The article is discussed in Chapter 3 under the heading “Controlled Clinical Trials” because the first of the three studies is a CCT.

Chaturvedi, Subramaniyam, Tiwari, et al. (1983) conducted a series of case studies evaluating the oral hypoglycemic effect of Arani, an Ayurvedic drug prepared from the plant Clerodendron philomidis. The patients were recruited from the SS Hospital, Varanasi, India. Demographics, including age, gender, and inclusion/exclusion criteria were not described for this study. The diagnostic criteria were fasting blood sugar and urine glucose.

Thirteen patients were treated with Arani as a crude drug decoction (1:4 concentration) of Clerodendron phlomidis; 15 to 30 g daily was given in divided doses for an average of 5.2 weeks. Tolbutamide (500 mg/day) was given to three diabetic patients for 2 weeks. Three patients were given 30 units of insulin for 2.3 weeks. Four patients on diet only were followed for an unknown period. A marked improvement in symptoms was observed in the Arani-treated arm. Urine became free of sugar in 46 percent of patients (no p value given). Also a marked reduction in blood sugar values, from 202 mg/dl to 149 mg/dl was reported. A drop of urine sugar, from 2.2 mg/dl to 0.8 mg/dl, was also noted. No statistical tests were performed and the article does not report sufficient data for re-analysis.

Observations in this study showed a fall in blood sugar and urine sugar along with symptomatic improvement, but the lack of specific data (no means, no standard deviations) and the small number of patients in the multiple arms of the study make generalization difficult.

Nande, Kale, and Wagh (1983) studied of the effects of the Jambu fruit (Eugenia jambolana) on the blood sugar levels of seven healthy volunteers and five diabetic patients. The seven healthy volunteers were in Arm I (23–34 years old, 0 percent female). The five diabetic patients (25–70 years old, 60 percent female) were in Arm II. All the subjects were given 150 g of fresh Jambu fruit and asked to eat only the pulp. The remaining seeds were collected and weighed to determine the quantity of pulp consumed, which averaged 114 g (range 107–120 g) per person. One patient in the non-diabetic arm consumed 184 g of pulp and was excluded from analysis. Blood sugar was measured at baseline and at 1 hour, 2 hours, and 3 hours.

The authors concluded that Jambu fruit does not seem to have a significant role in the management of diabetes. The small number of patients, the use of just one dose of the fruit, and the nonreporting of actual data are the limitations of this study.

Sepaha and Bose (1956) studied the effects of Pterocarpus marsupium and Eugenia jambolana on 21 type 1 diabetic patients (9.5 percent female, 20 to 60 years old). Patients were selected from those admitted to the medical ward at the Medical College of Indore, India, and did not have end-stage diabetes or diabetic complications. Patients were put on an 1,800-calorie American Diabetes Association (ADA)-recommended diet and observed for 5 days without insulin before beginning treatment.

Arm I (n=17) patients were given 1 oz of an aqueous or alcoholic extract of Pterocarpus marsupium three times a day. Three patients were followed for less than 7 days, and their data were not included in the study. No significant reduction in blood sugar level was observed. Adverse effects including albuminuria, acetonuria, and an increase in serum glucose levels were noted in Arm I patients. Arm II (n=7) patients were given 1 dram of Eugenia jambolana three times a day for at least 10 days. A number of patients continued for 20 days with three patients continuing for up to 70 days. Five of the patients enrolled in Arm II were patients who had failed to achieve glycemic control during Arm I using the Pterocarpus marsupium treatment. Of the seven patients in Arm II, three patients showed a reduction in blood glucose levels and in urine glucose levels, but no statistical significance was observed. No adverse effects were reported in Arm II patients.

The researchers were encouraged by a nonsignificant reduction in serum glucose levels in 7 percent of the Pterocarpus patients and in 42 percent of the Eugenia patients. However, the absence of statistical significance in such a small sampling of patients makes it impossible to share the authors' enthusiasm for these two study drugs. Given the small sample size, the failure to include the three dropouts in the analysis (i.e., no intention to treat analysis was done) further limits the conclusions that can be drawn from these data. In addition, most of Arm II was composed of treatment failures from Arm I. Presumably, this is a more treatment-resistant group and may not represent a fair assessment of the second herb (Eugenia jambolana) used. A followup study with randomization, blinding, a larger sample size, and a single study drug compared with placebo are essential for any conclusions to be drawn.

Patients were given powdered fenugreek seeds (25 g per day) incorporated into unleavened bread served during lunch and dinner. Both the control and experimental diets were identical except for the protein and fiber content, which was significantly higher in the experimental diet. No adverse effects were observed.

The researchers reported a statistically significant reduction in blood glucose levels at 40, 50, and 60 minutes after a glucose load (p<0.02). The actual values are not reported. Additionally, they reported a significant decrease in the serum glucose area under the curve (AUC) (p<0.05) and in the glucose half-life (p<0.02), as well as significant increases in the serum glucose metabolic clearance rate (p<0.02) and in glucose erythrocyte insulin receptors (p<0.02).

The researchers concluded that the addition of 25 g of fenugreek seed to the daily diet of NIDDM (type 2) patients can be an effective supportive therapy in the management of diabetes. Additionally, they propose that fenugreek may modulate plasma glucose levels by delaying gastric emptying, by direct interference with glucose absorption, and by systemic effects as indicated by the statistically significant reduction in AUC, half-life, metabolic clearance rate, and erythrocyte insulin receptors. The study's design flaws make it difficult to support such a statement. However, this study suggests that the use of fenugreek seed in the management of NIDDM (type 2) patients merits further investigation. The Jadad score for this trial is 1.

Sharma and Raghuram (1990), in an RCT, studied the effects of a diet enriched with fenugreek seeds on NIDDM (type 2) patients at the National Institute of Nutrition, Hyderabad, India, using a randomized, crossover design in two trials. The first arm consisted of 15 patients randomized to receive a fenugreek-enriched diet either during the first 10 days of the trial or during the second 10 days. These patients (33 percent female, ages 32 to 60) had diabetes from 2 months to 16 years. The second arm consisted of five NIDDM (type 2) patients (ages 35 to 58) who were also randomized to receive a fenugreek-enriched diet for either the first or second 20 days of the trial. Gender and other demographic data were not reported for the second arm. Patients not on the intervention diet were instructed to eat a control diet. The two diets were nutritionally identical, except the protein and fiber content was significantly higher in the experimental diet. During both of these trials, patients received their entire diet from the institute. There was no blinding of either the patients or the researchers. It was not noted whether or not the study patients were taking any concurrent allopathic medications. The experimental diet contained defatted fenugreek seed powder (100 g per day) incorporated into unleavened bread served during lunch and dinner. A protein isolate from groundnut was added to the control diet to supplement the protein content.

Patients given the intervention diet in the first arm had a statistically significant reduction in mean fasting blood sugar of 42 mg/dl (p<0.05). Their mean serum insulin level dropped 17.6 μU/l (p<0.05), a significant reduction, and their mean 2-hour post-prandial blood glucose dropped by 110 mg/dl (p<0.01). The mean urinary glucose excretion was reduced by 64 percent (p<0.05). There was also a significant reduction of 29 mg/dl (p<0.001) in total serum cholesterol and of 23 mg/dl (p<0.001) in low-density lipoprotein (LDL) and very low-density lipoprotein (VLDL) cholesterol.

Patients given the experimental diet in the second arm showed a significant decrease in mean fasting blood glucose of 41 mg/dl (p<0.05) as well as significant reductions in 24-hour urinary glucose excretion, serum cholesterol, and triglyceride levels (p<0.05). No adverse effects were reported in either arm.

The experimental diet contained defatted fenugreek seed powder (100 g per day) incorporated into unleavened bread served during lunch and dinner. A protein isolate from groundnut was added to the control diet to adjust the protein content. Both the control diet and the experimental diet were identical except for the fiber content, which was significantly higher in the experimental diet. The patients were not told when they were on the experimental diet or the control diet.

Four patients complained of minor gastrointestinal irritation (diarrhea, excess flatus) while on the fenugreek diet, but no other adverse effects were reported.

On the experimental diet, patients had a statistically significant reduction of 84 mg/dl (p<0.01) in mean FBS and 94 mg/dl (p<0.055) in the 1½-hour PPBS. There was no significant reduction in the serum insulin levels. The mean urinary glucose excretion was reduced by 54 percent (p<0.01). There was also a significant reduction in total serum cholesterol (p<0.001) and in LDL and VLDL cholesterol (p<0.01).

The researchers concluded that the addition of 100 g of fenugreek seeds to the daily diet of IDDM (type 1) patients can be an effective supportive therapy by improving glucose tolerance and the serum lipid profile. In addition, the crossover design did not incorporate a wash-out period for both arms, which may have affected the results observed in the second trial period. However, this study does suggest that the use of fenugreek seed in the management of IDDM (type 1) patients merits further investigation. This trial had a Jadad score of 1.

Sharma, Sarkar, Hazra, et al. (1996b) conducted a case series evaluating fenugreek on patients with NIDDM. Patients were recruited from the diabetes clinics of S. N. Medical College, Agra, India. There were 60 patients, 25 percent female, with an age range from 30 to 70 years. Inclusion/exclusion criteria were not described. Diagnostic criteria included fasting blood sugar and glucose tolerance test. Twenty-two patients had mild diabetes, 31 had moderate diabetes, and 7 had severe diabetes. Of the 60 patients, 40 took oral hypoglycemic agents. Each subject underwent two test periods. There was one control period during which patients were given isocaloric diets without fenugreek for 7 days. This was followed by a study period of 24 weeks when patients were given the same diets along with fenugreek seeds powdered and made into a soup with water. The dose was 12.5 g of fenugreek twice a day.

Sharma (1986) performed a series of pre/post studies looking at the effect of fenugreek on serum glucose and insulin levels. Demographics (age and gender), recruitment, inclusion and exclusion criteria, and diagnostic criteria of the study group were not given.

Six protocols to test the acute effects of fenugreek were performed. These protocols involved administering various preparations of fenugreek seeds to healthy subjects. Whole fenugreek seeds, extracted fenugreek seed powder, gum isolate from fenugreek seed, degummed fenugreek seed, cooked fenugreek seed, and cooked fenugreek leaves were all tested. Outcomes studied were blood glucose and insulin levels. The gum isolate of fenugreek significantly reduced plasma glucose at 30 and 60 minutes (p<0.05), and it significantly altered the serum insulin levels at 30 minutes (p<0.01) and at 60 minutes (p<0.05). The other preparations did show a consistent reduction in plasma glucose with serum insulin.

The author believed that fenugreek exerted both acute and chronic effects on blood glucose. The herb produced lower insulin and glucose responses to the standard glucose tolerance test load following a single dose of the herb. The author judged this effect to be the result of the fiber content of the fenugreek. Chronic administration of fenugreek was useful in lowering both blood sugar and cholesterol. The anti-diabetic effects of fenugreek were not destroyed by cooking; and consequently the author felt that this herb would be a useful addition to a diabetic diet.

Smith and Holm (1982) enrolled a series of healthy and diabetic patients in a study to test the effect of gum guar on serum glucose and lipid levels. Seventeen diabetic patients (6 insulin dependent and 11 on oral hypoglycemic medication, ages 22–73 years) were recruited along with 6 healthy volunteers (ages 20–40 years, 67 percent female) from a Western mainstream practice in Sweden. Diagnosis of diabetes was confirmed by fasting blood sugar. The demographics of the study group were not given nor were inclusion/exclusion criteria described.

A standard meal to be eaten at home was provided to all patients. They were instructed to take 10 g of granulated guar gum stirred in 50 ml of water prior to meals twice a day for 1 or 3 weeks, and in some cases up to 13 weeks. The average duration of administration was 3 weeks. The authors noted that the granulated layer of the guar gum solution is fairly rapidly dissolved in liquid, thus producing a highly viscous product in the stomach, which seems necessary for its effect.

The authors reported that there were statistically significant decreases in the fasting blood sugar and post-prandial blood sugar for all diabetic patients (p<0.05). However, the actual glucose levels were not reported. The serum cholesterol decreased from 5.46 mM to 4.68 mM (p<0.005). The serum triglycerides were not significantly lowered.

The most frequent side effects were increased flatulence and occasional abdominal cramps. Two of the insulin-dependent patients had more frequent hypoglycemic episodes, and their insulin was lowered by 4 to 8 units/day. The authors concluded that guar gum leads to lowering of fasting and post-prandial blood sugars as well as serum cholesterol levels, with minimal side effects. Thus, they believed that guar gum represents an interesting and possibly beneficial fiber supplement for diabetic patients.

Balasubramaniam, Arasaratnam, Seevaratnam, et al. (1988) studied the effects of Gymnema sylvestre on eight NIDDM (type 2) patients. No age, gender, or other information was available for these patients. There was no mention of concurrent hypoglycemic agents, nor was there any blinding or controls for this study.

Patients were given 10 g of Gymnema sylvestre leaf powder taken orally once a day for 21 days.

At the end of treatment, the patients showed a statistically significant 50.5 mg/dl reduction in mean fasting blood glucose and a 74.2 mg/dl reduction in mean 2-hour blood glucose levels (p<0.05) compared to baseline. The study also found a nonsignificant increase in mean body weight (0.9 kg) during this same 21-day period. No adverse effects were reported, and serum glutamic pyruvic transaminase (SGPT) levels were described as being within the normal range for all patients after treatment.

The researchers concluded that Gymnema sylvestre has a hypoglycemic effect by reducing the fasting glucose levels as well as the 2-hour blood glucose levels. The authors added that further studies should be undertaken to confirm these results and to determine the herb's mechanism of action. Such studies will benefit from the addition of blinding, randomization, and a larger patient pool, as well as a comparison arm.

Kothe and Uppal (1997) studied the effects of a homeopathic preparation of Gymnema sylvestre on 21 diabetic patients. The patients (ages 36–74 years, 33 percent female) included 7 patients with diabetic complications and 14 patients without complications. The study's inclusion/exclusion criteria, demographics, and diagnostic criteria were not described. The test drug, Gymnema sylvestre Q (mother tincture), was given in the following doses: 6 to 7 drops twice a day for patients with FBS of 80–100 mg/dl and PPBS of 160–180 mg/dl; 10 drops three times a day for patients with FBS of 100–120 mg/dl and PPBS of 180–200 mg/dl; and 15 drops four times a day for patients with FBS of 120 mg/dl and greater and PPBS of 200 and greater. Suitable homeopathic remedies were also prescribed for patients with complications.

The study was carried out for 6 months. No quantitative results were reported. The authors state that out of the seven patients with complications, three showed excellent results in controlling blood sugar, two showed moderate control, and two remained uncontrolled. Of the patients without complications (n=14), nine responded well, two showed moderate control, and three remained uncontrolled. The nonreporting of any quantitative data, lack of control, small number of cases, and the unspecified variations in dosage are serious design flaws of this study.

Balasubramaniam, Arasaratnam, Nageswaran, et al. (1992), in a CCT, studied the effects of Gymnema sylvestre on the fasting and post-prandial blood sugar of 16 normal subjects and 43 type 2 diabetic patients. The subjects' ages ranged from 43 to 68 years; gender and socioeconomic status of the patients were not mentioned, nor were the inclusion and exclusion criteria. All the patients received powdered leaf extract of Gymnemra sylvestre at a dose of 10 g/day for 7 days. Arm I consisted of the normal patients (n=16), who did not receive any treatment other than the initial one. The 43 diabetic patients, divided between Arm II and Arm III, did receive further treatment. Arm II (n=7) received Gymnema sylvestre powder for an additional 14 days, and Arm III (n=36) received an oral hypoglycemic agent (Tolbutamide) for an additional 7 days but no more Gymnema sylvestre.

Results were reported as follows. In Arm II there was a reduction in FBS level from 151.7 mg/dl to 101.2 mg/dl at 21 days and a reduction in PPBS level from 215.7 mg/dl to 141.5 mg/dl at 21 days. Both these changes were statistically significant (p values were not reported). In Arm III, there was a decrease in FBS from 157.8 mg/dl to 136.3 mg/dl at 7 days, and the FBS continued to be low (133.1 mg/dl) on day 14. The PPBS decreased from 244.8 mg/dl to 144.7 mg/dl at day 7 and was 135.5 mg/dl at day 14. The difference in FBS and PPBS from day 1 to day 7 was statistically significant, as was the difference from day 1 to day 14. There was no significant change from day 7 to day 14. In the arm of normal patients (Arm I), the FBS changed from 80.8 mg/dl to 71.6 mg/dl at day 7. This change was statistically significant.

The authors concluded that Gymnema sylvestre leaf powder has a definite hypoglycemic effect in both normal and diabetic patients. They further concluded—based on the differences from day 7 to day 14 in FBS and PPBS values between Arm II and Arm III—that Gymnema sylvestre was at least as effective as oral hypoglycemic agents. The short duration arm is a weakness of this study. The Jadad score was 0.

Khare, Tondon, and Tewari (1983) published a report of the hypoglycemic effects of Gymnema sylvestre leaves on diabetic and normal healthy adults. Arm I (n=10, 40 percent female, ages 19–25 years) consisted of healthy student volunteers with no diabetes who were recruited from G. S. V. M. Medical College, Kanpur, India. Arm II (n=6, 33 percent female, ages 35–50 years) consisted of type 2 diabetic patients who were recruited from the diabetic clinics of L. L. R. and associated hospitals in Kanpur, India. The diabetic patients were identified as having mild-to-moderate hyperglycemia without any diabetic complications. No mention of socioeconomic characteristics or demographics or clinical equivalence between the arms was made. The patients were given an aqueous decoction of the shade-dried leaves of Gymnema sylvestre in a dose of 2 g three times a day. Subjects in Arm I received the preparation for 10 days, and subjects in Arm II for 15 days.

Akhtar (1982) studied the effects of karela (the fruit of Momordica charantia) on eight patients with uncomplicated NIDDM (type 2). Patients (50 percent female, ages 38 to 50 years) were randomly recruited from the outpatient clinic at Agriculture University Hospital, Punjab, India. All anti-diabetic medications were stopped 48 hours prior to the initiation of the study, and all patients were placed on a low-carbohydrate diabetic diet.

Patients were given 50 mg/kg of body weight of powdered karela obtained from a local vegetable market, twice a day for 7 days. There was no blinding of either the patients or the researchers.

The study showed a statistically significant 100 percent (p<0.001) reduction in the mean urinary excretion of glucose. There was also a significant decrease of 93 mg/dl (p<0.001) in mean fasting glucose levels and 130 mg/dl (p<0.001) in mean 2-hour post-prandial glucose levels. No adverse effects were reported.

The researchers concluded that karela produced significant reductions in hyperglycemia and could be recommended for the oral treatment of patients with uncomplicated NIDDM (type 2). They also concluded that clinical studies on a larger scale are needed to further establish karela's antidiabetic efficacy.

This novel insulin product demonstrated a consistent hypoglycemic action in a small number of diabetic patients. It had a pharmacokinetic pattern between regular and NPH types of insulin. No adverse events were recorded. The authors suggested that this product had possible value when animal insulin cannot be used. The advent of recombinant insulin, however, places the utility of this observation in some doubt, except possibly in countries where recombinant insulin is prohibitively expensive. The Jadad score for this trial was 1.

Leatherdale, Panesar, Singh, et al. (1981) studied the effects of karela (the fruit of Momordica charantia, which is indigenous to South America and Asia) on glucose and insulin concentrations in nine NIDDM (type 2) patients (33 percent female, age not reported). The patients were taking various allopathic hypoglycemic agents, including chlorpropamide, tolbutamide, glibencalmide, and glymidine. They were given three oral glucose tolerance tests: one at the beginning of the study; one immediately after taking 50 ml of karela juice; and a final test 8 to 11 weeks after eating 0.23 kg of fried karela daily. Patients were instructed to stop their hypoglycemic agents for 48 hours before each test.

For the juice test, patients were given 250 ml of fresh squeezed juice concentrated by rotary evaporation. For the fried karela test, patients were given slices of karela (0.23 g/day) and instructed to fry them in vegetable oil. No other dietary restrictions were imposed.

Patients showed a significant reduction in plasma glucose levels during the glucose tolerance test at 30 minutes (p<0.05), at 60 minutes (p<0.01), and at 90 minutes (p<0.001) compared with controls. When the results of this first test were reported, the glucose values themselves were not given; only the area under the curve of the glucose tolerance test was reported. In the second study, patients who ate fried karela also showed a statistically significant 1.7 percent (p<0.01) reduction in glycosylated hemoglobin compared to pre-study levels. Additionally, insulin concentrations were statistically higher by 6.7 μU/l (p<0.05) when the patients were given karela juice compared to pre-study levels. Interestingly, insulin levels were significantly lower by 32.9 μU/l (p<0.05) after patients ate fried karela for 8 to 11 weeks. No adverse effects were reported.

The researchers concluded that karela, taken as both a juice extract and as a dietary supplement in a fried form, has a significant hypoglycemic effect that does not appear to be a result of hyper-insulinemia. The study was weakened by the fact that hypoglycemic medications were removed for only 48 hours prior to each test; however, it does demonstrate that karela causes a significant reduction in hyperglycemic response following a glucose load. This plant merits further study.

Welihinda, Karunanayake, Sheriff (1986) studied the effects of Momordica charantia on 18 new-onset NIDDM (type 2) patients (percent female not reported, average age 38.2 years) who were free from any diabetic complications and were not taking any concurrent hypoglycemic agents.

A standard oral glucose tolerance test (50 g glucose load in 100 ml distilled water) was performed on all patients after an overnight fast. The test was repeated the next day with all patients receiving 100 ml of Momordica charantia juice 30 minutes before the oral glucose load. The juice was made from fresh unripe fruits of M. charantia obtained from the local market. All of the juice was extracted at the same time, and this fresh juice was frozen. As needed, the appropriate amounts were thawed, the supernatant was decanted, and 100 ml aliquots of clear juice were given orally to the patients.

Glucose tolerance curves were obtained after all 18 patients ingested the M. charantia juice. These were then compared to the corresponding curves after the patients drank water. This analysis revealed that 13 patients showed a significant improvement in glucose tolerance, and 5 patients showed no improvement. The mean total area (187.0 cm²) under the juice-test glucose tolerance curves was significantly smaller than the area (243.6 cm²) under the water-test curves, which served as controls (p<0.001).

The researchers concluded that M. charantia has hypoglycemic activity. The failure to measure fasting blood glucose levels make this study inconclusive.

Srivastava, Venkatakrishna-Bhatt, Verma, et al. (1993) reported a case series evaluating the properties of Momordica charantia extract (powder vs. aqueous extract) on glucose levels of diabetic patients. Glucose tolerance tests confirmed the diagnosis of diabetes, and all participants maintained a diabetic diet. The first set of diabetic patients (n=5) was given a powder of dried Momordica charantia fruit (the plant is also known as bitter melon) at a dose of 5 g three times per day for 3 weeks. The patients treated with the powdered form of Momordica charantia showed a trend toward lowered blood glucose levels, but this did not reach statistical significance. An aqueous extract from Momordica charantia was obtained by boiling 100 g of fruit in 200 ml of water. It was given to a second set of diabetic patients (n=7, all male, ages 42–70) at a dose of 100 ml once per day for 3 weeks. The subjects treated with the aqueous extract showed a significant decrease in the hemoglobin A_1c from 8.37 percent to 6.95 percent (p<0.01). The absence of female participants limit the ability to generalize from this data.

Srividya and Periwal (1995) reported a study of the effects of Phyllanthus amarus ⁶ on blood glucose in 14 patients. The patients were recruited from the clinics of local physicians in Anantpur, India. Their ages ranged from 40 to 60 years, and the arm was 56 percent female. Socioeconomic characteristics of the patients and inclusion/exclusion criteria were not reported. The patients were classified into three arms: Arm I consisted of five normal control subjects; Arm II consisted of five hypertensive patients; and Arm III consisted of four patients with diabetes and hypertension. The patients in Arm II and III received Phyllanthus amarus (powdered whole plant) at a dose of 1.6 g three times a day with meals for 10 days.

Results were reported as follows. Mean fasting blood sugar in Arm II and Arm III patients decreased from 126.4 to 109.4 mg/dl (p<0.001). The authors concluded that Phyllanthus amarus has definite potential as a hypoglycemic drug for humans, with no side effects.

Mathew and Augusti (1975) studied the effects of onion (Allium cepa) on three NIDDM (type 2) patients (33 percent female, ages 45 to 62). No other socioeconomic or demographic data were reported. Patients were instructed to stop taking their allopathic hypoglycemic agents 5 days prior to initiating onion therapy.

After the patients had followed a standardized diet for 7 days, they were given 50 g of juice-expressed onion residue. The onion was prepared by drying 100 g of fresh onion in the sun and then powdering the resulting dried material. That powder was then extracted with ether, and the extracted fraction was again dried to a powder. A further extraction was done with petroleum, and the petroleum-insoluble fractions, once dried, were administered to the patients along with a standardized diet for 1 week. At the end of the study, glucose tolerance tests were performed.

The authors reported that the study patients had a rise in blood sugar of 75 to 90 mg/dl before the onion intervention but only 25 to 50 mg/dl after onion was used. Moreover, the blood sugar values came down to normal after 4 hours when onion was used. This was not the case when onion was not used. No adverse effects were reported from this study.

The researchers concluded that the common onion may contain hypoglycemic properties.

Sharma, Gupta, Gupta, et al. (1977) studied the effects of an aqueous extract of onion on 20 healthy volunteers (10 percent female, ages 20 to 30 years) at the Medical College of Jhansi, India. The volunteers were subjected to five interventions that compared different concentrations of the onion extract. These patients were also compared to patients receiving tolbutamide. In a separate study, the researchers also studied the effects of onion on glucose tolerance tests in patients with and without adrenaline-induced hyperglycemia.

The water-based extract used in the studies was prepared from locally bought onions that were blended into a pulp and then filtered twice. The volume of the final filtrate was corrected by the addition of distilled water so that 1 ml of extract represented 1 g of onion by weight. Patients were given between 25 and 200 g of this aqueous onion extract.

Aqueous extract of onion did not produce a change in fasting blood sugar in normal volunteers, even with doses that increased from 25 g to 200 g. However, the authors also reported that when onion extract was administered in various doses along with 50 g of glucose during an oral glucose tolerance test, it reduced the normal rise in glucose in a dose-dependent manner. The study showed no effect of onion on fasting blood glucose levels. The blood sugar lowering effect of various doses of onion extract was reported to be statistically significant (p<0.05 to <0.01), and at the highest doses of onion, it was comparable to the effect of tolbutamide.

The researchers concluded that aqueous onion extract has an anti-hyperglycemic, as opposed to a hypoglycemic, effect on healthy volunteers. They based this conclusion on the observation that the main effect of the onion extract was not on fasting blood glucose but on the rise in glucose following a glucose load. They also reported that the active principle in the onion is not degraded by heat because boiling the onion extract did not change its anti-hyperglycemic effect. As a preliminary investigation of onion as a possible anti-hyperglycemic agent, this study presents dose-response curves and elucidates some potentially useful areas for additional research.

Rajasekharan and Tuli (1976) studied the effects of vijaysar (Pterocarpus marsupium) on 35 patients with NIDDM (type 2) (40 percent female, ages 20 to 60+ years). It is unknown if these patients were taking any concurrent allopathic medications.

Vijaysar is made from the heartwood of Pterocarpus marsupium, which is decocted in water and then dried into a powder. Patients were given varying doses of 250 mg capsules of vijaysar depending on their age, stamina, and what the authors describe as “intensity of disease” for 1 month.

The majority of patients reported good to fair improvement in symptoms of polyphagia, polydipsia, and polyuria. Out of 23 patients for whom blood sugar levels were reported, only 8 showed a good to fair response in reducing blood glucose. The mean values and statistical tests were not reported. The study reported reductions in blood glucose levels, urinary glucose excretion, and symptoms of polyuria, polydipsia, and polyphagia. No adverse effects were reported.

The researchers concluded that vijaysar had an anti-diabetic effect on the treated patients. The lack of statistical analysis makes it impossible to draw conclusions from this study. Further investigation is needed to assess the anti-diabetic effects of vijaysar.

Combinations and Formulas

Shankar and Aggarwal (1988) studied the effects of Abraga Chendooram on 60 patients with NIDDM (type 2). Patients were selected from the diabetes clinic at Safdarjang Hospital in New Delhi, India. Potential participants were given a glucose tolerance test to confirm diagnosis and were then treated with dietary control for 15 days. Only those patients (n=60, ages 21–70 years, 45 percent female) who failed dietary control and who did not show evidence of serious complications of diabetes (ketoacidosis, nephropathy, neuropathy, or retinopathy) were eligible for the study.

Patients were given a formula called Abraga Chendooram that consisted of Abragam (purified black mica); Vengaram (dehydrated borax); Saranaiver Charu (juice of the root of Trianthema decandra); Adathondaielai Charu (juice of the leaves of Adhatoda zeylanica); and Alam vizhuthu kudineer (root of Ficus benghalensis). Patients were given 200 mg of the drug in gelatin capsules twice daily for 45 days. During this period, caloric intake was restricted to 25 calories per kg of ideal body weight. No mention was made of whether or not any patients were taking any concurrent allopathic medications for diabetes.

This study did not provide information on the actual fasting blood sugar results. Instead, patients were categorized into “complete relief” (normal blood sugar levels), “partial relief” (blood sugar levels slightly above normal; no actual range given), and “no relief” (no change in blood sugar levels). At 45 days, 35 patients (59 percent) demonstrated complete relief; 20 (33 percent) demonstrated partial relief; and 5 (8 percent) showed no relief. No adverse effects were reported from this study.

The researchers concluded that Abraga Chendooram demonstrates hypoglycemic activity and that further clinical studies are required. The concurrent implementation of a strict dietary regimen reduces the ability to draw any meaningful conclusions from this report.

Kumar and Kumar (1995) studied the effects of the herbal preparation M-93 on 30 NIDDM (type 2) patients. Seventy-three percent of the patient population was between the ages of 41 and 60, and 34 percent were female. No information was given on selection criteria other than the fact that patients with FBS > 300, chronic renal failure, diabetic neuropathy, or diabetic ketoacidosis were excluded from this study.

M-93 consists of Bilva (Aegle marmelos); Nimba, also known as neem (Azadirachta indica); Tulsi (Ocimum sanctum); and Kalimircha (Piper nigrum). M-93 was given orally at a dose of 1 g three times a day for 1 month.

After 1 month of treatment with M-93, the mean FBS dropped 31.84 mg/dl, from 160.2 mg/dl to 128.36 mg/dl (p values were not reported), while the mean PPBS dropped 56.9 mg/dl, from 229.5 mg/dl to 172.6 mg/dl (p values not reported). No adverse effects were reported from this study.

The authors concluded that all patients treated with M-93 showed a positive response after 1 month of therapy.

Nanda, Chopra, Sahu, et al. (1998) studied the effects of an Ayurvedic herbal combination, Nishamalaki, on blood sugar and symptoms of an arm of type 2 diabetic patients. The patients (n=100, 20 percent female, ages 31–70 years) were recruited from the outpatient department of the Central Research Institute (Ayurveda) in Bhubaneshwar, India. The diagnosis of diabetes was made on the basis of an elevated fasting blood sugar and presence of symptoms. The socioeconomic status of the patients was not reported. Patients who were younger than 30 years and older than 70 years, and patients who had hypertension or diabetic renal and cardiac complications were excluded.

Details of the preparation of Nishamalaki or its ingredients were not reported in the study. The herbal combination was given at a dose of 1 g of powder with water twice daily for 6 weeks. All patients also received diet and lifestyle modification advice.

Based on these results, the authors concluded that Nishamalaki is effective in lowering fasting blood sugar levels as well as reducing symptoms in diabetic patients.

Jain and Chakraborty (1987) studied the effects of an Ayurvedic herbal product called Nosulin on 20 NIDDM (type 2) patients (35 percent female, ages 35 to 70 years). The patients were classified into two arms: Arm I had 15 patients who were not taking any other medicine; Arm II had 5 patients who were concurrently taking 10 mg/day of the oral hypoglycemic agent glibenclamide. All patients were given 5 g of Nosulin twice daily, after lunch and dinner, and were also placed on an 1,800-calorie ADA diet. There was no blinding of either the patients or the researchers.

Each 5 g of Nosulin contains 4.5 g of Cyamopsis tetragonoloba (guar gum), 0.2 g of Cephalandra indica (tundika), 0.2 mg of Gymnema sylvestre (meshasringi), and a sufficient amount of excipients.

The authors reported a reduction of 25 to 30 percent in the mean blood sugar level for Arm I patients and a 50-percent reduction for Arm II patients. The actual numerical values and statistical significance of these results were not reported. The authors also reported a marked reduction in symptoms in all cases, as well as a decrease in the frequency and quantity of urine and a general sense of well being. No adverse effects were observed.

The researchers concluded that Nosulin appears to be a suitable therapeutic adjunct to oral hypoglycemic agents and dietary restrictions for diabetes. They also noted that further studies of Nosulin are currently underway. This preliminary study does not report any statistically significant results, and thus no conclusions can be inferred regarding Nosulin in the management of NIDDM (type 2) patients.

Sadhukhan, Roychowdhury, Banerjee, et al. (1994) reported on the hypoglycemic effects of an herbal anti-diabetic compound in a series of three studies. The patients (n=67) were recruited from the outpatient departments of the SSKM Hospital, Calcutta, India.

In the first study, 25 patients with type 1 or type 2 diabetes (ages 10–60 years, 32 percent female) were recruited. The socioeconomic characteristics and the inclusion/exclusion criteria were not reported. Patients who were taking oral hypoglycemic agents were asked to discontinue their medications 2 days prior to the study. Standard diet was prescribed to all the patients. The intervention consisted of an herbal preparation that contained guar gum (Cyamopsis tetragonoloba), 4 g; Trigonella foenum-graecum, 0.6 g; Cephalandra indica, 0.1 g; and Gymnema sylvestre, 0.2 g. The test drug was given for 2 days. An oral glucose tolerance test was done prior to starting therapy and after completion of therapy.

Results were reported for 11 out of the 25 patients as follows. The FBS improved from 221.18 mg/dl to 211.8 mg/dl. The 1-hour PPBS improved from 401.63 mg/dl to 316.0 mg/dl, and the 2-hour PPBS improved from 427.81 mg/dl to 341.45 mg/dl. The results for the remaining 14 patients were not reported and were said to be insignificant.

The authors concluded that the blood-sugar lowering effects of the combination test drug, as well as the two separate herbs, were significant. The possibility of overlapping hypoglycemic effect from the previous drugs is a major limitation of this study.

The authors concluded that the combination test drug did not produce a significant decrease in blood sugar but that the individual herbs—Cyamopsis tetragonoloba and Trigonella foenum graecum—did produce a significant reduction in the fasting and post-prandial blood sugars. The possibility of overlapping hypoglycemic effects across arms is a limitation of these studies.

Diet Studies

Acharya, Upadhyay, and Dwivedi (1996) reported the results of a study of a particular diet for diabetic patients (n=30, percent female and ages not given). Two type 1 diabetic patients and 28 type 2 diabetic patients were included. Twenty-five of the patients were newly diagnosed, and five patients were known to be diabetic. Inclusion/exclusion criteria and socioeconomic status were not reported. The diagnosis of diabetes was made on the basis of fasting and post-prandial blood sugar levels. The patients were classified into three categories, depending on their dosha body type: kaphaja (n=15), pittaja (n=10), and vataja (n=5).

The most significant decrease in FBS and PPBS was reported in the kaphaja arm followed by the pittaja arm and, last, the vataja arm. The researchers also noted that the patients with newly diagnosed diabetes had a significant decrease in their blood sugars, but the patients with previously known diabetes, including type 1 diabetic patients, did not. A statistically significant improvement in the symptoms of polyuria, polydipsia, polyphagia, and weakness was reported. No significant improvement in cramping, decreased libido, and joint pains was noted.

The authors concluded that dietary management is highly effective in newly diagnosed diabetic patients and in kapha-predominant patients. They added that diet has a supportive role in the treatment of vata-predominant patients and in chronic cases of diabetes.

Pandya, Nirmal, and Mistry (1990) conducted a case series on the role of Pathya Ahara, a particular Ayurvedic diet, in the treatment of diabetes. The diet is characterized as being low in madhura rasa (sweet taste) items and high in complex carbohydrates such as kidney beans, barley, and wheat. The caloric value was approximately 1,500 to 2,000 calories per day. Sixteen patients were recruited from an Ayurvedic practice; but their demographics, including age and gender, and inclusion/exclusion criteria were not given. The diagnosis of diabetes was based on blood glucose level. The patients were classified into two arms. Of the 16 patients initially enrolled, 6 dropped out because of noncompliance with the diet.

During the 15-day study, Arm I received Pathyhara as the sole treatment. Arm II received the diet plus an Ayurvedic preparation consisting of Banga bhasma and Tulsirasa (holy basil or Ocinum sanctum). Based on a glucose tolerance test, the fasting blood sugar in Arm I was 169.3 mg/dl before treatment and 138.52 mg/dl after treatment (SD = 39.42, p <0.05). In Arm II, fasting blood sugar decreased from 238.25 mg/dl to 222.40 mg/dl, and the 2-hour post-prandial blood sugar decreased from 368.93 mg/dl to 317.43 mg/dl. All other urinary and blood glucose measures for both arms did not reach statistical significance.

The high percentage of dropouts was not included in the analysis (i.e., an intention to treat analysis was not done), and this could distort the positive results noted.

The three treatments described in this study appear to have some promise in the reduction of post-prandial blood glucose and lipid levels. The small number of patients and the number of different substances studied within a single study as well as the lack of a placebo comparison arm make it difficult to draw meaningful conclusions. However, this study does indicate that the gurmar preparation, beeja wood water, and tamarind seed merit further investigation. The Jadad score for this study is 1.