In the case of clinical trial evidence, the literature is mixed. There are increasing well-conducted random-assigned, double-blind controlled studies, and even systematic reviews of these. The literature on ginkgo (Ginkgo biloba) (The European Scientific Cooperative on Phytotherapy 2003), which has been among the most widely prescribed of all drugs in continental Europe, and on St John’s wort (Hypericum perforatum) (The European Scientific Cooperative on Phytotherapy 2003), many hundreds of papers in each case, with dozens of these being full clinical trials.

There is little doubt that orthodox evidence requirements are potentially appropriate for most herbal medicines consumed by the public. Over-the-counter (OTC) herbal use in Europe far outstrips practitioner prescription, probably by a factor of more than 20:1 (IMS/PhytoGold 1998). For researching OTC label indications for discrete medicinal products, in which individual responses to remedies are not the critical issue, the double-blind controlled clinical trial is clearly the most applicable method. In addition, as the ‘patient’ is in many cases not being diagnosed professionally and is determining his or her own treatment and prognosis, self-assessment questionnaires are often an appropriate measure of progress. These are not expensive to administer. As this is often ‘outpatient’ medicine research, costs can be saved as close clinical supervision need not be always be necessary throughout the trial.

There are however practical problems in pursuing good clinical research in herbal medicine:

Looking ahead there are new reasons for concern. In Europe, where so much of the herbal clinical literature has been generated, there is a new regulatory regime. The Directive on Traditional Herbal Medicinal Products permits the registration (rather than licensing) as medicines of herbal products on the basis of their traditional use rather than on proven efficacy (Directive 2004). Such products still have to comply with pharmaceutical good manufacturing practice (GMP) and safety monitoring, but there is no longer any incentive to prove their efficacy in clinical trials. In fact, as the Directive does not apply ‘where the competent authorities judge that a traditional herbal medicinal product fulfils the criteria for authorization’, any such data may bar a herbal product from the less expensive registration option and lead to a requirement that it be licensed as a full medicine. There is a concern that the new Directive will lead to a progressive devaluing of herbal medicines and a drying up of clinical research activity.

The herbal practitioner, or phytotherapist, has other reasons to note the limitations of the controlled clinical study. Herbal practitioners emphasize the individuality of their treatments; they most often mix a number of individual herbs in a prescription and in many cases construct a unique mixture. Like other complementary practitioners they might also point out that conventional clinical trials involve the homogenization of the patient population so that only an average effect is confirmed and genuinely important benefits for a minority of the population will be overlooked. Clinical trial data will help with, but still not answer, the fundamental question: ‘is this remedy going to be good for this patient?’

Nevertheless the controlled clinical trial is a notably flexible instrument and clinical trial data at least involve rigorous observations of human use of plants: of the forms of research available they are by far the most valuable in making clinical judgements. Perhaps better hopes lie with the moves towards sustainable economic development in countries around the world. Renewed interest in the potential of indigenous medical traditions, in some cases linked to emerging economies, has increased research activities in a wider range of plants (Blumenthal & Cavaliere 2006). At best standards developed for herbal research in Europe can be exported to improve the quality of investigation elsewhere and new products linked to sustainable ‘freetrade’ arrangements. There is still a hunger for beneficial natural medicines. If there is any substance in the promise, new ways will be found to service that demand.

Investigations of medicinal activity in animals raise particular concerns among those who would otherwise be amenable to using herbal approaches and are therefore often discouraged in the herbal industry. However it is salutary to note that the international scientific literature on plant remedies and their constituents is still dominated by in vivo studies.

As well as challenging moral and ethical sensibilities such studies have other shortcomings. They are always undermined by interspecies differences and often by therapeutically inappropriate mode of application. Observations often relate to unrealistically high doses, and particularly to intraperitoneal or intravenous injection rather than oral administration. In the case of plants, whose constituents are usually transformed by digestive action and hepatic first-pass effect, these latter factors seriously undermine any projection to the effect of the whole plant on oral consumption in humans. Differences in liver processing among animal species add a particular extra uncertainty for complex chemical entities like plants. It has been established, for example, that critical P450 cytochrome metabolic pathways differ significantly in laboratory animals compared to humans (Blumenthal & Cavaliere 2006).

At best oral administration of plants in realistic doses to animals may help to explore some basic pharmacological processes. In most cases these data should be ruthlessly marginalized in projecting clinical effects in humans.

In laboratory experiments it is extremely difficult to reproduce the balance of plant constituents that will actually reach internal tissues after digestion, absorption and the hepatic first-pass effects. In cell and tissue cultures concentrations of active principles at cell membranes and receptors usually far exceed those ever reached in vivo.

Most in vitro studies are on isolated fractions of the plant. It is very doubtful that the action of the whole plant can be determined by knowing the effects of individual constituents. Interaction between dozens or even hundreds of these limits the inferences that can be drawn. There have been some calculations demonstrating the synergistic effects of plant constituents in the whole plant which reinforce this point (Bogaards et al. 2000).

One particular hurdle arises from the historical choice of glass (‘vitro’) to conduct these experiments. Some key plant constituents are adherent to glass surfaces and quickly coat glassware, so compromising observations. Plants therefore have to be ‘cleaned up’ before being tested, leaving out important principles like resins, tannins, gums and mucilages.

The best that can be claimed for this work is that it may suggest pharmacological mechanisms. It can have almost no impact on judgements in clinical practice.

Accounts of other people’s use of medicines, particularly by observers outside the culture concerned, are vulnerable to observer error and usually relate to isolated cases or reputations in which non-specific influences on efficacy may be overwhelming. Without some form of confirmation any such report is at best an anecdote and when the additional cultural contexts are added it is difficult to generalize to the wider human community. Organized traditional records, as in the classic texts from earlier cultures, may be more systematized but suffer from ‘shibboleth’ status: they are very rarely subject to rigorous review and have usually been transmitted didactically in a way that may enhance the powerful impact of suggestion on efficacy. Nevertheless, even with these caveats, clinical experience is the prime source of information on which most practitioners in any tradition base their judgements. Although personal perceptions of treatment benefit are obscured by non-specific effects, including placebo, collective impressions have great potency in determining the therapeutic profile of any remedy. Indeed it has been proposed that herbal texts may provide resources for modern data-mining techniques to identify novel pharmacotherapeutic leads for bioactive compounds (Williamson, 2001 and Buenz et al., 2006). Substantial accounts, for example as seen in systematic compilations of ethnobotanic records (Moerman, 1998, Allen and Hatfield, 2004 and Buenz et al., 2004), can provide valuable sources for processing and distillation. Organized pre-modern physician texts, such as those of Eclectic physicians of 19th-century North America (Perry 1980), have relevance in modern clinical experience and should not be dismissed. The vast classical literature from China, India, Islamic and other regions is still the basis of medical practice and education in these countries today. There is potential – largely unexplored – scope for a distillation of such experiences. One form of correlation between traditional use and pharmacological research is outlined in Table 8.2 and referred to on page 161.

There is an application of the controlled clinical trial that could be particularly appropriate for assessing the impact of traditional herbal treatments: observing physiological responses to treatments in human subjects rather than direct effects on morbidity.

Traditional views of herbal remedies emphasize their primary influence on transient body functions, e.g. they are classed as diaphoretics, expectorants, circulatory stimulants, diuretics, digestive stimulants, laxatives and so on. In other words, contrary to common belief, many herbs may have almost immediate results on the body. It is possible to devise methods by which such effects can be detected. Activity on biological markers, physiological functions and tissue or fluid constitution can be monitored directly in healthy or morbid populations and could provide much useful information on the effects on the body of herbal products. With advances in non-invasive monitoring technologies it is possible to conceive of important trials in human subjects, both in observational and controlled studies.

Against such ideas it has been pointed out that measures of efficacy tend to be based on a medicine’s effects on morbidity or mortality. ‘Surrogate parameters’ may restrict possible indications as labelled claims for herbal products. It would need to be argued in each individual case, with the support of other medical evidence, that a verified effect in changing some physiological parameter would be likely to have an effect on morbidity or mortality.

The observational study design has emerged as a research tool to complement controlled clinical trials (Dobre et al. 2007). It is more suited to establish benefit in the real world (‘effectiveness’) (McKee et al. 1999) and may help to assess such effectiveness in subgroups not studied in controlled trials, or over the long term. In some cases, careful observations of non-controlled clinical events may be the best or even only practicable source of evidence. These include circumstances where controlled studies are difficult, for example: (1) rare diseases; (2) where there are many clinical complexities; and (3) for cases where blinding is impossible.

Observational studies may be a particularly appropriate method for the study of herbal medicines. A persistent tradition is that these medicines may support self-corrective functions in the body, perhaps more readily than synthetic pharmaceuticals. New insights into complex dynamic systems suggest that there may be benefits in observing humans as ecosystems. It is apparent in this case that different research methodologies are required. These should: (1) have regard to the widest range of symptoms and signs together rather than particular variables in isolation; (2) aim to measure quantifiable components of health, rather than of morbidity; and (3) involve minimal intervention.

Non-invasive monitoring of physiological functions may be applied (as above), perhaps coupled with patient self-rated questionnaires, clinical observations of overall behaviour and epidemiological methods, to establish as far as possible what actually happens to individual patients when they seek treatments.

Routine collection of patient and clinical data may be practicable at a teaching clinic, for example. These could include self-rating questionnaires, for general health and for target conditions, perhaps combined with a number of other non-invasive observations, compared with general remission rates. With computer technology it is possible to accrue considerable quantities of useful observational data by involving patients and practitioners in simultaneous recording of treatments and questionnaire-derived outcomes (using simple touch-screen check-box entry forms, for example).

Although it is difficult to establish cause and effect in observational or field studies, or specifically to separate specific from non-specific treatment effects, there are a number of ways that observational studies could productively be used in herbal research. For example:

Such information however suffers from one clear problem: it is rarely subject to independent validation or review, so includes partisan judgements. Its use without supporting controlled data in determining efficacy of treatment is therefore limited. Nevertheless there is some evidence that good observational studies generate similar results to controlled clinical trials (Radford & Foody 2001), and in combination with other studies could provide very useful information.

The main charge against single case studies is that they cannot credibly select out real effects from confusing variables, and specific from non-specific treatment effects. Further investigation of such research however shows both that it can have more credibility than might be supposed, and that it is not a soft option.

The criteria for validity of such trials have been well reviewed by Aldridge (1991): a good single case study design can be very rigorous. It includes providing as many points of view on the event as possible, clarifying operational definitions and recycling observed data around the researchers (including the patient as co-researcher) for checking and possible refutation. It is possible to conduct double-blind, placebo-controlled studies in a series of such individual case studies with each patient being his or her control. These could allow for a useful database of reliable case histories to be assembled over the years, as both an educational and research exercise.

Perhaps most importantly, the single case study is the raw caseload of each practitioner. A routine of rigorously collating studies is a commendable continuing education exercise that may encourage practitioners to become more active in the generation of efficacy data.

An effective methodology for clinical studies could usefully combine elements of all three research designs above.

The persistent theme running through any discussion on the efficacy of herbal medicine is that it has been used over centuries and millennia. It is claimed with some justification that at least some value will have been distilled out by this vast store of human bioassay data, especially as there would have been little room for sentiment and idealism in the life-and-death situations that prevailed through most of herbal use.

However without a rigorous screening the record of traditional use can appear as little more that a motley and primitive hotchpotch of folk fancies. The review of historical methodologies highlights some limitations, but the power of cultural placebo alone renders any individual observation almost worthless. What is needed is the identification of themes, a structure for assessing traditional claims. Fortunately this may be possible.

Clinical insight into the traditional record leads to the realization that from the very earliest and most primitive accounts of herb use, humans classified the plant material into relatively consistent pharmaceutical categories. The classification was based on subjective properties of taste or appearance and the immediate impact they made on consumption. These categories, encountered repeatedly in the ethnobotanical records, became the basis of core therapeutic principles in the classic texts of China, India, Greco-Roman Europe, Islam and almost all other written traditions. Many of these categories survive today as recognizable phytochemical groups. They constitute a primal pharmaceutics of surprising potency. It had to be: there was usually no other choice. Going to a traditional herbal practitioner was often a very robust experience!

The pharmaceutical principles of traditional medicine provide a potentially robust correlation with modern herbal research, provided the latter can be linked to these phytochemical subgroups; in other words, provided both the scientific and the traditional data relate to the same common elements in comparable contexts (it is important for example that dosage and pharmacokinetic criteria in the scientific studies are consistent with traditional application).

Some of the most consistently encountered pharmaceutical categories in history are listed in Table 8.2, with the traditional experience of the category as a whole followed by the established modern pharmacological assessment. There are as many variations on these themes as there are plants included within them, but the properties listed predominate across each category. Modern ethnopharmacological studies show countless examples where pharmacological activity can be demonstrated in traditional remedies and practices. Such correlations are clearly interesting; in effect the early reputation provides the ‘human bioassay data’, the clinical evidence before the preclinical studies, rather than the other way around, as is usual in modern pharmaceutical research. These correspondences lift the findings of laboratory research into a real clinical context.

In addition to providing a second guess on the veracity of traditional reputations, laboratory studies can cast light on some of the persistent puzzles in herb use. Some of the more relevant questions follow.