The analysis of the herbal products data with a focus on safety submitted to TMDA from 2009 to 2020 showed that the majority of the applicants did not provide evidence to support preclinical and clinical safety. It was observed that most of the herbal medicines submitted for MA were multicomponent herbal products. Most of the applicants did not provide evidence on quality parameters with safety implications.

Features of herbal medicines submitted for marketing authorization

The analysis of the dossiers submitted in this study showed that there was no noticeable difference in the trend of the number of applications received and granted MA from 2009 to 2020 after the introduction of the CTD format in 2017. This shows that introduction of CTD did not affect the rate of herbal dossier submission to TMDA.

In this study, only a few herbal medicines (26%) were granted MA based on complete submissions of evidence for the assessors to determine whether the product is safe to be consumed by the populations. Similar results of low registration of traditional herbal medicine applications were reported by some of the European regulatory authorities in Denmark (11%), France (19%), Romania (24%) and Cyprus (25%) according to the survey report of the European Medicines Agency (EMA) of December 2017 [54]. This low registration rate in Tanzania could be explained by the lack of scientific information to substantiate registration of the products or a lack of knowledge for some manufacturers of herbal products on proper compilation of the dossiers and TMDA requirements, as many queries for submission of additional information were requested.

Assessment of the products submitted showed that 172 plant species were used for the formulation of herbal medicines with the ten most commonly used plants being Z. officinale, G. glabra, A. vasica, E. globulus, P. longum, T. bellirica, C. longa, C. camphora, O. sanctum and P. nigrum. The same plants were found to be commonly registered in Ghana by the Ghana Food and Drugs Authority [55] and in Nigeria by the National Agency for Food and Drug Administration and Control [56].

SafetyCombination of herbal medicines

A higher proportion (68%) of herbal medicines granted MA were formulations that contained several plants or active herbal ingredients. The multicomponent nature of these products is the same as those produced in China and other Asian countries, where manufacturers of herbal medicines are adopting the theory of Asian and Chinese traditional medicine practices whereby complex formulations are used to attain synergism or complementary effects [57]. The situation is different in European Union countries where a study conducted by Wieland Peschel in 2014 showed that the multicomponent herbal products that were granted MA were only 34% for traditional herbal medicinal products and 19% for well-established use herbal medicines [37]. The use of multiple plants in one formulation however must be justified by manufacturers and there are very few studies showing the beneficial effects of complex multiple herbal products in curing diseases [58,59,60,61].

Rigorous scrutiny of multicomponent product dossiers is recommended due to their potential to cause herb–herb interactions that may cause adverse events [62]. This potential for herb–herb interactions has been demonstrated in previous studies, for instance, liquorice (G. glabra) root interacted with either Veratrum nigrum, Sargassum pallidum or Euphorbia pekinensis [19]. In this study, the results showed that G. glabra was combined with multiple different plants in 17 preparations, nonetheless, preclinical safety was submitted in only 35% of the applications and clinical safety in only 41%. Manufacturers have mixed several different plants in their preparations of which most of the combinations have not been studied for their safety as evidence was not available in the submitted dossiers.

The results of this study also showed that piper species (P. longum, P. nigrum and P. cubeca) were combined with Adhatoda vasica in nine preparations together with other herbals. Additionally, Piper species were combined with Curcuma longa (turmeric) in four preparations. Piper species have been documented to interact with herbal and conventional medicines such as R. rosea, C. longa, A. vasica, nevirapine, phenytoin, propranolol, theophylline and verapamil by inhibiting cytochrome P450 or isoenzymes or P-glycoprotein, resulting in an increase or decrease in their bioavailability [63, 64]. However, those studies were limited to in vitro and animal models with very little clinical evidence in humans, which was the same as observed in this study, whereby preclinical safety data were submitted for 90% of the P. longum and 50% of the P. nigrum products. Clinical safety data were submitted in 50% P. longum and 25% P. nigrum products. It was also observed that these documented interactions in the literature were not included in the package inserts to warn consumers.

Safety data

The TMDA guidelines for the submission of documentation for MA of herbal medicinal products clearly state that data to support evidence of safety should be submitted either as reports or from the literature [34]. However, only a few preclinical (38%) and clinical (32%) safety data were submitted in herbal product applications. This lack of evidence from the applicants could be associated with the fact that traditional herbal medicines used for a long time are presumed to be safe and could warrant waivers, as done by the regulatory authorities in some countries. The WHO recommends that there should not be stringent regulatory requirements on safety data if a product has been traditionally used for a long time and is proven to be harmless, however, evidence of safety should be provided through literature sources and references [6].

In this study, it was found that applications for Echinacea species (E. angustifolia and E. purpurea) products lacked preclinical and clinical safety data despite the availability of safety data in the literature conducted over the years since the 1950s, as well as case studies and spontaneous reports from the regulatory authorities [41]. In this case, the inclusion of literature data to support safety would have prevented unnecessary delays and resubmissions to the TMDA.

A similar observation was made for products containing eucalyptus oil, where most of the dossiers lacked preclinical and clinical safety data despite literature documentation on its toxicity, AEs with prolonged use and in children [65, 66]. Inclusion of the safety data from the literature and a precautionary warning to “avoid prolonged use and not to give to children under 2 years of age” might have been sufficient evidence to support the product’s safety for MA.

Other herbal medicines with documented adverse effects and/or toxicity in the literature, such as P. ginseng, G. biloba and S. nigrum [42, 52], were used in the formulations in the submissions and should have been accompanied by documentation to support the safety of the products. However, only half of the products for P. ginseng and G. biloba had submitted preclinical and clinical safety data, and S. nigrum had no clinical safety data even though there were some cases of toxicity in the literature [52].

In this study, products containing V. officinalis, A. belladonna and M. fragrans were granted MA based on the submission of clinical safety and efficacy data, despite the lack of submission of preclinical data according to TMDA and WHO guidelines [6, 34]. This shows the importance of systematic clinical trials to make evidence-based decisions on the benefit–risk profile of a herbal product [7]. Although manufacturers are expected to understand this importance, the results showed that the majority (70%) of them did not conduct clinical studies of their products and relied only on the literature data. The same was observed in a study conducted in the Netherlands in 2001, whereby most of the evidence from the applications was sourced from the literature [36].

Toxicity data were only submitted for 38% of products as part of preclinical safety studies, which were mainly on acute toxicity, followed by chronic, repeat dose/ sub-chronic, carcinogenicity and developmental toxicity. This low submission rate of toxicity data was also observed in a previous study in the Netherlands [36]. Submission of toxicity studies is important since carcinogenicity, developmental and reproductive toxicities cannot be determined by long-term use alone without conducting studies [67]. Gingko was used for many years however, it was found to have carcinogenicity potential in preclinical studies in vitro and in vivo studies done in rodents [68, 69]. Ephedra products have also been used for a long time and it was discovered that they can cause cardiovascular toxicity [70, 71].

Despite the WHO recommendation on granting waivers for submission of toxicity studies for long traditional used herbal products, they also recommend that for herbal products that have toxicity, documented risk assessment information should be provided together with documentation on long-term use safety otherwise toxicity studies should be performed [6]. The applicants did not follow the recommendation since they did not provide any documentation from the literature to support long-term use safety or the absence of toxicity risks.

As with conventional medicines, documented evidence of developmental and reproductive toxicity may be required if the herbal product is intended to be used by women with childbearing potential, pregnant, or lactating. [34]. In this study, developmental and reproductive toxicity studies were not submitted in the majority (96%) of the applications. This can be explained by the fact that these studies are usually not necessary for products with documentation in the literature on long-term traditional use unless there is some safety concern or the product changes fertility, hormones or has an effect on the endocrine system [72].

Herbal medicines such as C. camphora essential oils, when used in high doses have been observed to cause maternal toxicity in pregnant animal studies and in clinical studies which emphasizes the necessity of conducting reproductive toxicity studies [73]. In contrast, in this study, there were no preclinical data submitted for C. camphora with few (25%) clinical safety data. Furthermore, package inserts did not include any warnings or precautions for consumers in the regarding the potential toxicity.

Quality data with safety implications

The results showed that tests on contaminants such as heavy metals, pesticides, fumigants, mycotoxins, microbial counts, fungal and mould were submitted in only half (52%) of the active herbal substances and in 77% of the finished herbal product specifications. This is contrary to the TMDA guidelines which require specifications for both active herbal substances and finished herbal products to include tests to verify purity [34]. The absence of such important tests explains the low rate of approvals in this study since contaminants have a direct influence on the outcomes of the safety assessment of a herbal product [74].

Heavy metal contamination in herbal medicines may originate from the environment, manufacturing processes, or may be intentionally added by manufacturers in belief of their medicinal potential [75]. Tests for the control of heavy metals are critical for the authorization of a herbal product since even very small amounts can be very toxic [76]. In this study, tests for heavy metals (mercury, lead, cadmium and arsenic) were included in only 35% and 14% of the active substances and finished herbal product specifications, respectively. A study conducted in Malaysia by the Drug Regulatory Authority found that 22% of Eugenia dyeriana preparations on the market were contaminated with lead [77]. It is therefore important to monitor heavy metals both during MA assessments and post-market to ensure continuous compliance.

Mycotoxins such as aflatoxins are by-products of fungal contamination in plants and are a human health hazard with fatal outcomes in some cases [78]. Contamination can occur during post-harvesting processes, storage or transportation [79]. The results of this study showed that the majority of the submissions (91%) did not include mycotoxin tests in the raw material specifications despite being among the regulatory requirements for MA of herbal products [34, 80]. Furthermore, evidence of possible contamination has been reported in previous studies of medicinal herbs in Spain where over 96% were found to be contaminated with aflatoxins and other mycotoxins [81] and aflatoxin contamination in 43% of crude herbs and 64% of finished herbal products in India [82].

Agrochemical contaminants such as insecticides, herbicides and fungicides should be controlled in herbal medicine raw materials due to possible contamination from soil, farming/cultivation, water sources and post-harvest processing [79]. The absence of these contaminants therefore indicates the manufacturer’s compliance with GACP and GMP practices during all stages, from cultivation to production and storage. However, in this study, few (12%) manufacturers included tests for pesticides or fumigant residues in the herbal substance specifications. This could imply that manufacturers did not follow GACP and GMP standards as only a few applicants (4%) provided evidence of GACP for the raw materials and 42% for GMP compliance. A study conducted in the USA on herbal medicines sourced from China showed that 36% of the samples were contaminated with pesticide residues [83] and a study on five ginseng plants in China found high levels of four types of organochlorinated pesticide residues [84]. The importance of verification tests on acceptable levels of these contaminants by the manufacturers cannot be emphasized enough.

Few submissions in this study included test results for total microbial, fungal, mould, and yeast counts of active herbal substances and finished herbal products. These tests are critical in ensuring consumer safety [79]. A study in Brazil showed that more than 50% of the herbal medicines that circulate on the market were contaminated with microorganisms [85], while another study identified 42 fungal species contaminating raw materials of T. cordifolia and M. fragrans [86]. Based on the observations made, it is evident that these critical tests could not be waived by the regulatory assessors.

Safety data in patient information leaflets or summary of product characteristics

Most of the key safety data were not included in the patient information leaflets, with little information on warnings, precautions, side effects, overdose, pregnancy and lactation, or interactions. For example, G. glabra (liquorice) has been documented to cause ARs in the cardiovascular system including cardiac arrhythmias, hypertension, hypokalemia myopathy and hypermineralcorticoidism and is therefore not recommended for use for a long period of time [87]. However, the applicants did not include these reactions in the package insert for consumers and healthcare workers to be cautious except for one product which mentioned “the adverse events were uncommon, and the consumers should avoid prolonged use” without providing reasons. This pattern was also observed for C. camphora which has been documented to cause hepatotoxicity but was not mentioned in the package inserts [88]. Asian ginseng has been documented to interact with digoxin [89], however no documentation of any interaction was provided in the package inserts or in the summary of product characteristics. This lack of information could be explained by the fact that most of the toxicity or adverse reactions were not supported by evidence from sufficiently large systematic RCTs.

A similar lack of safety information was observed in a United Kingdom study, whereby 75% of the five commonly used herbal products on the market did not contain key information on safety [90]. Patient information is important to ensure that consumers understand important safety information such as possible ARs, side effects, contraindications, warnings and precautions to avoid unexpected outcomes from the medicines [91]. A survey conducted among consumers of herbal medicines in the UK showed that most of the herbal products’ consumers had little knowledge about possible safety issues of the products they consumed, and 40% presumed herbal medicines to be safe [92]. This lack of awareness of possible ARs might lead to few reports of adverse reactions due to herbal medicines being reported to the regulatory authorities.

Strengths and limitations

A major strength of this study is the access to the dossiers and the TMDA database for analysis of the information submitted over a wide span of 12 years. A major limitation of this study is the possibility of bias, as with any document analysis, one must work with the data that are available.

Recommendations

Manufacturers should build capacity on knowledge of herbal medicine regulations and guidelines for MA in various countries to address the serious problem of data deficiencies. Regulatory authorities should conduct a series of sessions with manufacturers and marketing authorization holders that include support, training, and informing them about the requirements in different sections of the guidelines to reduce future resubmissions and delays in the application process. This will also help to avoid the circulation of the same products in the market as food supplements since once the products are queried or refused due to a lack of adequate data, there is a high probability that they end up in the market with new claims as food supplements. There is also a need for manufacturers and stakeholders to invest more in research and development for herbal medicines and generate safety and efficacy data to support market authorization.

A detailed future study could be carried out on the comments issued to the applicants and their responses to identify frequently occurring deficiencies issued in each section and the reasons for not providing the data to inform all stakeholders.