Chinese Herbs for Memory Disorders: A Review and Systematic Analysis of Classical Herbal Literature

Text mining and other literature-based investigations can assist in identifying natural products for experimental and clinical research. This article details a method for systematically analyzing data derived from the classical Chinese medical literature. We present the results of electronic searches of Zhong Hua Yi Dian (“Encyclopaedia of Traditional Chinese Medicine”), a CD of 1000 premodern (before 1950) medical books, for single herbs, and other natural products used for dementia, memory disorders, and memory improvement. This review explores how the terminology for these disorders has changed over time and which herbs have been used more or less frequently, and compares the results from the premodern literature with the herbs indexed for memory disorders in a modern pharmacopoeia. The searches located 731 citations deriving from 127 different books written between ca. 188 AD and ca. 1920. Of the 110 different natural products identified, those most frequently cited for forgetfulness were yuan zhi (Polygala tenuifolia), fu shen (Poria cocos), and chang pu (Acorus spp.), all of which have been cited repeatedly over the past 1800 years and appear among the 31 herbs indexed in a modern pharmacopoeia. By providing a complete, hierarchically organized list of herbs for a specific disorder, this approach can assist researchers in selecting herbs for research.

Medical traditions are an important source of information in the application of natural products in health and disease. A considerable number of bioactive compounds used in modern medicine and in research have a rich history of use in traditional medicine [1–3]. Ethnobotanical knowledge is also of assistance in drug discovery [4–6], and a number of studies have reported the results of text mining and other literature-based investigations [7–12]. In traditional Chinese medicine (TCM), as collections of the classical literature become more accessible, opportunities arise for examining the content of multiple works in the search for effective remedies or inspiration for drug development [13].

Of the genres of TCM literature, those most relevant to natural products research are books on materia medica (ben cao lei 本草类), formulae (fang shu lei 方书类), and clinical medicine (lin zheng ge ke lei 临证各科类). In general, books in the materia medica genre are primarily concerned with the identity and uses of individual natural products, hereafter referred to “herbs”, although they may include items of animal or mineral origin. Books on formulae are mostly concerned with combinations of herbs, how they are prepared, and their therapeutic applications, whereas books on clinical medicine focus on the diagnosis and differentiation of particular disorders and their management using herbal formulae and other methods. In this study, we focus on individual items of materia medica (i.e., “herbs”) and on the materia medica genre of TCM literature, for which we use the term “pharmacopoeia”.

1.1. Outline of the development of traditional Chinese pharmacopoeia

Over the past 2000 or more years, hundreds of pharmacopoeias have been compiled in China. This genre continues to be a cornerstone of clinical practice in TCM, with modern pharmacopoeias referencing much of their therapeutic information to their traditional predecessors as well as to the results of scientific experimentation and clinical trials [14–20]. Traditional pharmacopoeia list natural products of plant, mineral, and animal origin together with their therapeutic applications, properties, functions, and a diversity of other information.

Although the term ben cao appeared in ca. 4 _AD in Qian Han Shu 前汉书 (“History of the Former Han Dynasty”) [21], the earliest received pharmacopoeia, Shen Nong Ben Cao Jing 神农本草经 (“Divine Farmer’s Pharmacopoeia”), was probably compiled during the Latter Han Dynasty (ca. 25–220 _AD) [22]. It contains 365 items classified according to therapeutic application [23,24].

In traditional pharmacopoeias, referencing to the source appears to have been an innovation by Tao Hong-Jing (ca. 456–536), who combined Shen Nong Ben Cao Jing with other sources into Ben Cao Jing Ji Zhu 本草经集注 (“Supplemented Classic of Pharmacopoeia”), which contains 730 items. He wrote the original text in red and the additions in black to distinguish these from the ancient text [22,24,25]. Tang Shen-Wei refined this approach in his Jing Shi Zheng Lei Bei Ji Ben Cao 经史证类备急本草 (“Pharmacopoeia Based upon the Classics and Historical Works Arranged by Types for Speedy Use”; ca. 1082) [22,24]. He consulted approximately 250 previous books, compiled the information under 1744 items and referenced it to the sources [25]. This approach was further refined by Li Shi-Zhen in Ben Cao Gang Mu 本草纲目 (“Pharmacopoeia Classified by Section and Sub-section”; ca. 1593) [22,24]. It contains monographs on 1892 items, about 1160 illustrations, and more than 11,096 formulae, references 952 previous authors, and includes Li’s own views and observations [22,24–27].

Traditional pharmacopoeia further expanded in the modern era with the publication of Zhong Yao Da Ci Dian 中药大辞典 (ZYDCD; “Great Compendium of Chinese Medicines”) in 1977, which contains monographs on 5767 individual herbs together with illustrations and Latin binomials [15]. Information is referenced to premodern and modern books, and it includes chemical constituents and the results of research. Even larger is Zhong Hua Ben Cao 中华本草(ZHBC; “Chinese Pharmacopoeia”). It was published in 1999 in 34 volumes, the last four of which cover materia medica from Tibet, Mongolia, and other regions, and includes monographs on 8980 items based on information referenced to premodern and modern sources [20]. An abridged version of the Chinese section also is available [16]. Besides these large comprehensive pharmacopoeias, there are numerous smaller works that may specialize in regional materia medica, only list herbs in common use, or deal with foodstuffs with a few focusing on famine foods [22,25].

Throughout their history, the comprehensive pharmacopoeias tend to accumulate new items and more information on existing items. As such, they provide a wealth of possibilities for the researcher interested in the therapeutic applications of natural products. However, pharmacopoeias are organized by item of materia medica, not by the disorder for which the items have been used, which can make locating information on specific disorders problematic. Ben Cao Gang Mu includes a list of 114 categories of disorders and the applicable herbs [28], but most premodern pharmacopoeias only provide lists of their included herbs. ZYDCD contains an index of disorders and the applicable herbs, but there is no corresponding index in ZHBC [15,20].

1.2. Objectives of this study

In this study, our aim was to develop an approach to using multiple full-text versions of classical pharmacopoeia to generate hierarchical lists of herbs for specific disorders in order to inform ongoing research. The searches focused on age-related memory disorders and the specific objectives were to:

1. 1.

   Develop a systematic approach to searching multiple Chinese pharmacopoeia and classifying the retrieved information.

2. 2.

   Identify which individual herbs were most frequently used for dementia and/or memory disorders in these pharmacopoeias.

3. 3.

   Determine the time spans over which these herbs have been used for these disorders.

4. 4.

   Compare the results of searches of multiple classical pharmacopoeias and premodern books from other genres, with the results from a modern indexed pharmacopoeia.

2.1. Selection of search terms

The standard nomenclature for disease names and syndromes in modern TCM includes the terms chi dai 痴呆(“dementia”) and jian wang 健忘 (“forgetfulness”) [29]. However, conducting a search of the premodern literature required a broader range of terms, so dictionaries and medical nomenclatures [29–33] were consulted as well as the historical discussions in contemporary clinical manuals [34–44]. Based on these sources, a list of search terms relevant to dementia and memory disorders and applicable to the traditional literature was compiled. These were used in the initial searches and supplemented with additional terms identified during the search process. The search terms fell into three groups: (1) wang group (“forgetfulness”) jian wang 健忘, shan wang 善忘, duo wang 多忘, xi wang 喜忘, hao wang 好忘, yi wang 易忘, bu wang 不忘, etc.; (2) chi dai group (“dementia”) chi dai 癡/痴呆, dai chi 呆痴, bai chi 白痴, chi bing 痴病, dai bing 呆病, dai zheng 呆证, etc.; and (3) ji group (“memory improvement”) qiang ji 强记 and bu shan ji 不善记.

2.2. Selection of resources

Zhong Guo Ben Cao Quan Shu 中国本草全书 (ZGBCQS; “The Complete Collection of Traditional Texts on Chinese Materia Medica”) includes 2026 titles and is the largest published collection of books available on Chinese pharmacopoeia [13,45]. However, it is difficult to search ZGBCQS for information on specific disorders, because most books only list the names of the included herbs and have no indices to diseases or symptoms. Zhong Hua Yi Dian 中华医典 (ZHYD; “Encyclopaedia of Traditional Chinese Medicine,” 4th edition) includes about 1000 titles with 67 different books in Section 3, which is specifically devoted to pharmacopoeias [46]. It was selected as the source collection of classical literature because it is a CD with a search function, constitutes the largest single published collection of complete TCM books, and includes 48 of the pharmacopoeias found in ZGBCQS [13,47]. ZYDCD was selected as the modern indexed compendium for comparison because it is one of the two largest modern pharmacopoeias, includes an index of herbs indicated for specific disorders including jian wang, and is well indexed to herb name in Chinese and Latin binomials [15].

2.3. Search procedure

Boolean operators cannot be used when searching ZHYD, so each search term was separately entered into the search boxes for “headings” (目錄搜尋全部) and “body text” (內容搜尋全部), and the hit numbers were recorded. Each hit was screened to determine if it referred to an item of materia medica of plant, mineral, or animal origin (i.e., a “herb”) that was used for treating the particular search term. This was defined as a citation. Citations of herbal formulae and combinations of herbs were excluded. When a citation referred to a list of herbs that were all used for the same condition, each herb was considered a separate citation. When the citation indicated that two or more herbs were to be used in combination to treat the disorder, this was considered to be a formula, even when it was not named and no amounts were ascribed to the individual herbs. This article reports on citations of individual herbs.

For each citation, the following information was extracted: herb name, search term(s), any reference to aging, actions/indications, book name, ZHYD section. It was expected that most entries would derive from ZHYD Section 3—Pharmacopoeia. However, we also wished to determine whether searching Section 3 alone would produce different results to searches of the full CD. Duplicate citations were defined as multiple citations of the same herb for the same or similar search term within the same book. When the same passage was located using different search terms, these hits were considered as one citation. In cases where there were multiple citations within the same book, which varied in the scope of the information provided, these were combined into one citation. An identical passage in a different book was considered a separate citation.

For ZYDCD, all herbs listed under the index term jian wang were identified. Chi dai is not listed as an index term, but because it was possible that chi dai had been included under dian kuang 癲狂 (“psychosis”), as it is in Jing Yue Quan Shu 景岳全书 (“Complete Works of Jing-Yue”), all herbs indexed under dian kuang were also searched.

2.4. Data synthesis

Each citation was extracted from ZHYD into an Excel spreadsheet along with the search term(s) that had located it, the book name, ZHYD section, author(s), the year it was written, and dynasty. When this information was not available in ZHYD, other sources were consulted [24,48,49] and the most recent plausible year was used. If the year was not available, it was inputed as the year of the author’s death, and if this was unavailable, as the last year of the dynasty. For the purposes of analysis, the start and end years of dynasties have been adjusted to eliminate overlap and intervening periods (see Table 1). When the citation indicated that the herb was also used to promote longevity, restore youthfulness, or other statements that suggested it was useful in disorders of aging, this information was recorded and scored separately. Data were checked by another researcher, and duplicates were removed. The citations retrieved from ZHYD were checked against the passages in the same books in ZGBCQS, when available, to verify their accuracy. Each citation was coded and entered into Predictive Analytics SoftWare (PASW), Version 18 (also known as SPSS), for analysis.

The 40 search terms yielded 114 hits under “headings” and 3281 hits under “body text”. Hits that were simply a mention of the search term or were in relation to the use of the herb in a formula were excluded. In total, 792 citations that related to the use of a single herb for one or more of the search terms were extracted. Of these, 57 were multiple references to the same herb in the same book. When the information was complementary, the multiple references were combined into one citation for that particular book. Otherwise, duplicates were removed. In four cases, the same book appears in more than one section of ZHYD, for example, Yao Xing Si Bai Wei Ge Kuo 药性四百味歌括 is included in Shou Shi Bao Yuan 寿世保元, so these were regarded as duplicates [47]. After combining references and removing duplicates, 731 citations deriving from 127 different books written between ca. 188 AD and ca. 1920 formed the primary data set.

In historical terms, the largest number of citations (273) derived from 37 books written during the Ming dynasty, followed by the Qing dynasty with 52 books and 248 citations (Table 1). Not surprisingly, the Ming Dynasty work Ben Cao Gang Mu (ca. 1578), which remains the largest and best known of the premodern pharmacopoeias, provided the largest number of citations (73). This was followed by Zheng Lei Ben Cao 证类本草 from the Song Dynasty (ca. 1082) with 40, and another Ming work Ben Cao Pin Hui Jing Yao 本草品汇精要 (ca. 1505) with 37. The earliest work was Shen Nong Ben Cao Jing (ca. 188) with seven citations, and the most recent work was Ben Cao Zheng Yi 本草正义 (ca. 1920) with four citations. The dominance of Ming dynasty works as sources for citations reflects the history of Chinese pharmacopoeias, which reached a peak during the Ming dynasty in terms of number of included herbs and level of detail. Although the pharmacopoeias written during the Qing dynasty tended to be shorter, there was an overall increase in the number of books available [22,24,25].

The most productive search term was jian wang with 306 citations from 992 AD onward (Table 2). The next most productive terms were bu wang (203 citations) and shan wang (78 citations), both of which were used in Shen Nong Ben Cao Jing (ca. 188), followed by xi wang, which first appeared in Ming Yi Bie Lu 名医别录 (ca. 220). Although jian wang has been the dominant term since the Song dynasty, most of the earlier terms persisted in the literature until at least the 1890s. This was mainly attributable to the tendency of Chinese medical works to quote verbatim from earlier resources, thereby retaining outdated terminology. An exception was the term hao wang, which was not found after 1505.

Over the period 1047–1901, there were 32 citations of herbs being used for memory improvement, 16 of which were also for “forgetfulness”. No herb was listed as used for the main terms in the “dementia” group, chi dai and dai bing, but the less common term chi bing found two entries, both to fei zao jia 肥皂荚 (Gymnocladus chinensis). However, these were in relation to its inclusion in a formula, so these were excluded. In addition, four citations were located for hun chi 昏痴, all of which originally derived from a case reported in Ru Men Shi Qin 儒门事亲(ca. 1228). This described a woman who had suffered from seizures (feng xian 风痫) since she was 6 or 7 years old, which had become increasingly frequent until they were occurring every day, and as often as 10 times a day, when she was aged between 30 and 40. Her condition was described as hun chi jian wang 昏痴健忘, which suggests that her ability to think and remember things was severely impaired. The text mentioned that she wished to die. She picked, steamed, and ate a plant resembling Chinese leek (cong 葱) and afterward vomited thick fluids for several days and sweated copiously. After this, her condition improved and she asked people what the plant was. She was told it was called han cong miao 憨葱苗, which the Ru Men Shi Qin said was the shoots of the toxic plant li lu 藜芦(Veratrum spp.). Although this was not a typical citation and her condition was clearly not dementia or Age Associated Memory Impairment (AAMI), it illustrates how poisoning due to plant misidentification could occur and enter the literature.

As expected, the most productive section of ZHYD was Section 3 (Pharmacopoeia) with 441 citations derived from 52 books (Table 3). However, more than 21% of citations derived from Section 8 (Comprehensive medical works), because some books in this section also covered aspects of materia medica. Citations from the clinical medicine genre were unexpectedly infrequent (2.5%). Even though formulae were excluded, almost 8% of citations were from Section 4, because some formula books also discuss the uses of individual herbs. Therefore, 89.3% of the citations could be located by searching Sections 3, 4, and 8.

In ZYDCD, 30 different herbs were indexed under the term jian wang. Under the index term dian kuang 癫狂(“psychosis”) (63 entries), one citation contained the term chi but it was unrelated to dementia. However, an additional herb for duo wang was located (i.e., zhu sha cinnabar), so 31 herbs were found using these two index terms.

For ZHYD, after removing the herbs that cause forget-fulness or dementia, 661 citations remained. Of these, 627 were for “forgetfulness”, 32 were for “improving memory”, and 142 citations were also for aging. The citations specified 110 different herbs, of which 43 were only cited once, 14 were cited twice, five were cited thrice, and 14 were cited four times. A total of 408 citations of 83 different herbs were from Section 3. In general, the herbs found frequently in Section 3 were also frequent in the full CD. Exceptions were zhu li 竹沥 (14 times in total but none in Section 3) and tao zhi 桃枝 (12 times in total, 3 times in Section 3). Neither herb is indexed in ZYDCD. Despite these exceptions, in the case of the higher frequency herbs, a similar list of herbs would have been returned if only Section 3 had been searched.

In 32.8% of citations, a corresponding book could not be located in ZGBCQS and in a further 1.6% the book in ZGBCQS was incomplete and missing the relevant section. The remaining 478 citations (65.5%) could be checked against one or more copies of the same book in ZGBCQS. Of these, 35 were found to differ in some way, but 18 were just variations in the text that did not affect the meaning. In three cases, the item could not be located in the ZGBCQS editions; four made no mention of the search term; in one case the search term appeared in one version of the book but not in another; in one case a different character, also pronounced “wang”, was used but this appears to be an error; however, in another case the characters jian and wang appear to belong to different sentences; thus, nine of these citations may be due to errors. Seven of the citations from Ming Yi Bie Lu in ZHYD appeared in the Ben Cao Jing Ji Zhu (ca. 536) in ZGBCQS but not as references to Ming Yi Bie Lu. Therefore, 16 citations may be erroneous and seven may be referenced incorrectly. Overall, 3.6% of included citations may have been included based on textual errors. In cases where discrepancies were evident, the ZHYD citation was retained unaltered, because it would have required a lengthy process of textual research to determine which version was more true to the original.

3.1. Herbs frequently cited for treating memory disorders

The 34 herbs with five or more citations in total are listed in Table 4 along with their total citation frequencies for forgetfulness, memory improvement, and aging, and in Section 3. Of the 31 herbs indexed in ZYDCD, 28 were included in the total ZHYD list. Of these, the 17 included in Table 4 are marked with an asterisk. Of the herbs listed in Table 4, 18 were also used for the elderly.

Yuan zhi 遠志 (Polygala tenuifolia) was the most frequently cited herb for forgetfulness in total, in Section 3, and for aging. It first appeared in Shen Nong Ben Cao Jing (ca. 188) and is included in ZYDCD. The second highest frequency item was the fungus fu shen (Poria cocos). It first appeared in the Ming Yi Bie Lu, which is of uncertain origin; however, its content can be dated to before the end of the Han dynasty (ca. 220 AD). Fu shen was indicated for shan wang 善忘. Closely related to fu shen is fu ling. This herb first appeared for jian wang in Zheng Lei Ben Cao (ca. 1082), but its frequency was considerably lower than that of fu shen in subsequent pharmacopoeia. Both fu shen and fu ling are from the same plant, Poria cocos, the only difference being that fu shen includes parts of the root of the pine trees upon which the fungus grows. The pine root itself was first included as a separate item called huang song jie 黃松節 in Zeng Guang He Ji Ju Fang Yao Xing Zong Lun 增广和剂局方药性总论 (1368). It is also called fu shen mu 茯神木 and appears in subsequent pharmacopoeia for jian wang until 1862. It seems that the combination of the fungus and the pine root in the form of fu shen or fu shen mu was considered more effective than the fu ling fungus alone. Fu ling typically grows on Pinus massoniana (ma wei song 馬尾松) and Pinus densiflora (chi song 赤松) as well as on other species of pine. All three herbs are indexed in ZYDCD [15,50]. When researching fu shen, it may be productive to ensure that there are significant root inclusions in the test samples and to identify the tree species.

Also appearing in Shen Nong Ben Cao Jing (ca. 188) and continuing to the modern era was chang pu. In modern pharmacopoeia it is divided into shi chang pu 石菖蒲, which derives from Acorus graminus or Acorus tatarinowii, and shui chang pu 水菖蒲 or bai chang pu 白菖蒲 from Acorus calamus. There is also an herb called jiu jie chang pu 九节菖蒲, which is now generally derived from Anemone altaica; in earlier times, however, this name may just have referred to large Calamus rhizomes. In ZHYD, it is not clear which species is referred to when the terms chang pu and jiu jie chang pu are used. However, for shi chang pu, Li Shi-Zhen’s description in Ben Cao Gang Mu (ca. 1578) indicates that the species referred to was A. graminus or A. tatarinowii. All three herbs are indexed in ZYDCD [15].

Two common foods appear in Table 4. Shan yao 山药(Dioscorea spp.) appeared from the Song dynasty onward, whereas long yan rou 龙眼肉 (longan fruit) appeared during the Ming dynasty, and both continue to be listed for forget-fulness. Also frequently used in both food and medicine is gan cao 甘草 (Glycyrrhiza spp.), with citations for forgetfulness in seven books from ca. 1082 onward as well as in ZYDCD.

Of the uncommon substances, an early inclusion was the spider zhi zhu 蜘蛛. Ming Yi Bie Lu instructs the gathering of its web on the 7th day of the 7th month, to which Ben Cao Gang Mu added that you should hide it in the collar of your clothes and not let anyone know. Despite such talis-manic usage, it persists into the modern ZYDCD, in which it is identified as Aranea ventricosa. Used similarly are bie zhua 鳖爪 (tortoise claw/foot), which was collected on the 5^th day of the 5th month, and tao zhi 桃枝 (Prunus persica twig), which was carried, although references to these practices do not appear after 1625. Consuming the hearts of horses, oxen, and other animals was also considered beneficial for forgetfulness, as was eating bat brain (fu yi nao 伏翼脑). Whereas bat brain fell out of favor after the Ming dynasty, horse heart continues to be listed in ZYDCD.

Also first appearing in Ming Yi Bie Lu (ca. 220) was lü ru. It was subsequently cited 12 times, lastly in Shen Nong Ben Cao Jing Zan 神农本草经赞 (1850). The name appears synonymous with lan ru 蘭茹, but no herb under either of these names could be located in the modern pharmacopoeia consulted [15,20]. Wright [51] tentatively identified lü ru as Euphorbia adenochlora, but the identity of this plant remains unclear.

Two high frequency herbs, huang lian 黄连 (Coptis chinensis) and ren shen 人参 (Panax ginseng), which have a long history of use for forgetfulness, were not listed in the ZYDCD index. When their individual entries in ZYDCD were consulted, however, they were both indicated for bu wang, as was the mineral kong qing 空青 (azurite). Also not indexed in ZYDCD was xian mao 仙茅 (Curculigo orchoides), which was cited 13 times from 1082 to 1832 for memory enhancement (qiang ji 強記) and also twice for forgetfulness. Therefore, it appears that herbs for memory improvement or preventing forgetfulness (i.e., bu wang) were not indexed under jian wang in ZYDCD. Aside from these herbs, yin yang huo 淫羊藿, suan zao ren 酸枣仁, and xuan ming fen 玄明粉 are all indicated for jian wangor related terms in their respective ZYDCD entries but do not appear in the index. For the remaining herbs in Table 4, there is no mention of any of the search terms under their respective ZYDCD entries, and in the case of niu xin and bie zhua the herbs are not listed in ZYDCD.

The herbs that appeared in both the ZYDCD index and in the ZHYD search results but are not listed in Table 4 owing to low frequency are as follows: xuan shen 玄参 Scrophularia ningpoensis root (n = 4), chai hu 柴胡 Bupleurum chinense root (n = 4); cinnabar (n = 4); long gu 龙骨 fossil bone (n = 4); bai shi ying 白石英 quartz (n = 4); sang piao xiao 桑螵蛸 Paratenodera sinensis egg case (n = 2); ma hua 麻花 Cannabis sativa flower (n = 1); hu rou 狐肉 fox meat (n = 1); and di gu pi 地骨皮 Lycium chinense or L. bar-barum root bark (n = 1). In addition, the following three herbs were indexed for jian wang in ZYDCD but not included in the final ZHYD list: hu po 琥珀 amber, gua zi jin 瓜子金Polygala japonica, and da zao 大枣 Ziziphus jujuba. The main reason for these omissions was that the source books that listed these three herbs for forgetfulness were not included in ZHYD. In the case of hu po, citations were found in ZHYD, but these were in reference to formulae and were therefore excluded from the analysis.

There was a considerable overlap in the herbs indexed in ZYDCD and the high-frequency herbs found in the searches of ZHYD; however, the ZYDCD index is limited to herbs for jian wang and close synonyms, excluding bu wang and other terms relating to memory improvement. Furthermore, it is not a complete listing of all herbs used for jian wang in the ZYDCD entries. In comparison, the ZHYD search resulted in more than three times as many different herbs and also provided a hierarchical listing in terms of citation frequency.

A search of Chinese herbal formulae for memory disorders found yuan zhi to be the most commonly used herb [10]. However, the second most commonly used herb was di huang (Rehmannia glutinosa), which only appears once in this search. Mai dong (Ophiopogon japonicas, Ophiopogon bodinieri or Liriope spp.) was the sixth most frequently used herb in the formulae but only appeared four times in this search. The herbs chang pu, shi chang pu, fu ling, fu shen, ren shen, suan zao ren, bai zi ren, long yan rou, and gan cao all appeared as higher frequency items in both the formulae search and this search, but the following 24 items did not appear on the list of herbs commonly used in formulae for memory disorders: zhi zhu, huang lian, shan yao, niu huang, zhu li, niu xin, xian mao, long dan cao, tong cao, lü ru, fu shen mu, yin yang huo, ling zhi, shang lu, ma xin, tao zhi, tie hua fen, mu tong, du ruo, kong qing, dan shen, fu yi nao, xuan ming fen, or bie zha. One likely reason for this is that the included formulae were all for both memory impairment (excluding memory improvement) and aging, which significantly restricted the range of formulae included in the analysis. Nevertheless, the following seven herbs satisfied both criteria but did not appear frequently in the formulae: shan yao, niu huang, long dan cao, yin yang huo, shang lu, tie hua fen, and xuan ming fen. Therefore, the herbs frequently cited in the pharmacopoeia as useful for memory disorders and aging did not necessarily appear frequently in premodern herbal formulae.

3.2. Herbs cited as causing memory impairment

Unexpectedly, 68 citations were for herbs that caused “memory impairment” (i.e., ling ren jian wang 令人健忘and related terms) and two citations noted that frequent consumption of dan sha 丹砂 (cinnabar) causes chi dai 多服令人痴呆 (“dementia”). The most commonly listed herb for causing forgetfulness was hu sui 胡荽 (n = 29, ca. 682–1886), which is the usual name for coriander (Coriandrum sativum). Hu sui is often used in both medicine and food, but many pharmacopoeias warn that long-term consumption causes forgetfulness [15,51]. The next most frequently cited herb was shu jiao 蜀椒 or chuan jiao 川椒(Zanthoxylum species) (n = 10, ca. 1082–1862) [15], followed by hu cong 胡蔥 (n = 9, ca.739–1578), which appears to be Allium ascalonicum, Allium fistulosum, or Allium ledebourianum [15,52]—about which Shi Liao Ben Cao 食疗本草 (ca. 739) said “long term consumption causes duo wang” (久服令人多忘). The herb fan lu 繁縷 (Stellaria media or Stellaria chinensis) [15,51,52] received four citations (ca. 682–1578), as did (zao) qing gua (早)青瓜 (ca. 682–1556), which is now a name for cucumber [52]. The remaining 10 herbs received only one or two citations.

It is not surprising that the frequent consumption of cinnabar was linked with chi dai, because mercuric compounds have long been known to adversely affect the brain [53,54]. However, the plants mentioned above all have a long history of use in food, and the basis for their causing forgetfulness was not explained. Common chick-weed (Stellaria media) is used as food in China, and a number of Zanthoxylum species are sources of “Sichuan pepper” [52]. These plants do not appear to be associated with toxicity, but in earlier times they would have been gathered from the wild, so the mistaken use of other species may have led to these warnings.

It is also possible that different plants were once known under these names. Qing gua 青瓜 is now a name for cucumber (Cucumis sativus), but cucumbers probably originated in India and were introduced to China [55]. Cucumber was called hu gua 胡瓜 (i.e., Tartar melon) and is now usually called huang gua 黃瓜 (yellow melon) [52], so the zao qing gua referred to in Qian Jin Shi Zhi 千金食治and cited in subsequent works may have been a different kind of melon to that which is now called qing gua. Similarly, the hu cong (i.e., Tartar leek) in Shi Liao Ben Cao may have been a foreign plant resembling the Chinese cong 葱but not any of the Allium spp. now known under this name. It is notable that in the Veratrum poisoning case mentioned above, the herb that was mistakenly eaten was evidently considered a kind of cong by the local people.

Coriander appears to have been introduced into China during the Han dynasty and has been known under many names including xiang sui 香荽, hu cai 胡菜, and yuan sui 芫荽[15,51]. The cultivar Coriandrum sativum is now the main species, but Coriandrum tordylium and Biflora radicans are also used [56]. Therefore, the botanical identity of the hu sui of ancient times may have been diverse. The part of the plant responsible for causing forgetfulness is not specified in the earlier citations, but Zheng Lei Ben Cao (ca. 1082) lists this warning for the seeds, whereas Ben Cao Gang Mu (ca. 1578) mentions the leaf and/or root. In the European literature, there are also warnings against the excessive consumption of coriander seed, which was said to cause dizziness [51,56]. The citations of qing gua and fan lu as causing forgetfulness ceased in the Ming dynasty, and this may reflect a change in the plants referred to under these names. Only hu cong and hu sui retain their warnings in ZYDCD.

3.3. Methodological considerations

The most straightforward method of obtaining a list of herbs used for a particular disorder is to look up an indexed pharmacopoeia. In the case of memory disorders, this yielded 30 herbs under jian wang in ZYDCD but none under chi dai, because the term was not indexed, and one extra herb when dian kuang was used instead. Of the 30 herbs identified under jian wang, 28 were found by the electronic search, although one of these was in the context of a formula. In comparison, the electronic search located 110 different herbs and provided citation frequencies, which enabled a hierarchical listing, so there was considerable value added by this approach. Searches of ZHYD Section 3 yielded a similar result to that obtained when searching the entire CD. It could be argued that besides requiring less time, this approach was superior, because it eliminates the uncommon items bie zhua 鳖爪, tao zhi 桃枝, and zhu li 竹沥 from the high-frequency list.

Checking citations against a printed source proved very time-consuming and for 252 citations (34.5%) was fruitless, because a corresponding printed version of the passage could not be located in ZGBCQS. Nevertheless, all but a few of the checked citations could be verified using the ZGBCQS editions, and the few discrepancies had little effect on the higher frequency items. A number of issues with the lists of books included in ZHYD have been identified [47], but the content of the included editions appears to correspond well with the editions in ZGBCQS. Therefore, provided that the characteristics of the ZHYD collection are known and taken into account when calculating frequencies, researchers can be confident of the results of systematic searches, at least in the case of the higher frequency items. The main drawback with this approach is the lack of a Boolean search facility in ZHYD. An improved search interface would substantially reduce the time required for searching and enhance the utility of this CD as a research tool.

The herbs most frequently cited in classical Chinese pharmacopoeia for memory disorders are yuan zhi 遠志 (Polygala tenuifolia root), fu shen 茯神 (Poria cocos sclerotium with pine root inclusions), and chang pu 菖蒲 (Acorus graminus or A. calamus rhizome), all of which were cited over long periods and were also used for disorders associated with aging. These herbs are listed for memory disorders in the contemporary Chinese pharmacopoeia ZYDCD. Overall, there was considerable similarity in the results obtained from the two approaches, but the electronic search of ZHYD identified more than three times as many herbs as listed in the ZYDCD index and also provided a method of organizing these herbs according to frequency and time spanofcitation. This hierarchical organization may assist researchers in selecting herbs for further research.

We thank Ms Chang Su-Yueh, Mr David Lu, Dr Angela Yang, and Dr Iris Zhou for their work in data extraction, compilation and scoring. We also wish to thank the library staff at Hong Kong Baptist University and at the Guangzhou University of Chinese Medicine for their kind assistance with the literature search process. Funding support: We thank the Preventive Health National Flagship Program, Commonwealth Scientific and Industrial Research Organization (CSIRO) for providing partial funding support. The project is also partially supported by an International Grant from the Guangdong Provincial Academy of Chinese Medical Sciences, Guangdong Province, China.

Authors and Affiliations

1. Traditional & Complementary Medicine Research Program, Health Innovations Research Institute; WHO Collaborating Centre for Traditional Medicine, Discipline of Chinese Medicine, School of Health Sciences, RMIT University, PO Box 71, 3083, Bundoora, Victoria, Australia

   Brian H. May, Anthony L. Zhang & Charlie C. L. Xue

2. Guangdong Provincial Academy of Chinese Medical Sciences, Guangzhou, China

   Chuanjian Lu, Yubo Lu & Charlie C. L. Xue

3. Guangdong Provincial Hospital of Chinese Medicine, Guangzhou, China

   Chuanjian Lu, Yubo Lu & Charlie C. L. Xue

Corresponding author

Correspondence to Charlie C. L. Xue.

Reprints and permissions