Juniper: IELTS Academic Reading practice test

Open a hospital pharmacy and a surprising proportion of what is inside can be traced to a plant. Aspirin descends from salicin, a compound found in willow bark and in meadowsweet, both of which were used against fever long before anybody could say why. Morphine comes from the opium poppy and quinine from the bark of the cinchona tree, which for three centuries was the only treatment for malaria that worked at all. Digoxin, still prescribed for certain disorders of heart rhythm, comes from the foxglove. The list is not a historical curiosity: something between a quarter and a half of the drugs approved in recent decades are either natural products or close chemical relatives of them.

There is a reason plants are such productive chemists, and it has nothing to do with our welfare. A plant cannot run away from whatever is eating it, and cannot fight. What it can do is make the eater ill. Over several hundred million years plants have evolved compounds designed to interfere with the nervous systems, the digestion and the reproduction of animals, and the precision of those compounds is the result of exactly the same selection pressure that produced the animals' countermeasures. A molecule shaped over that long a period to act on animal physiology is, from a pharmacologist's point of view, a lead that has already passed its first test.

Finding such a molecule can be done in two ways. The first is to collect plants more or less at random, extract them and test the extracts against a panel of diseases. The second is to ask. Communities that have used a plant for generations have, in effect, been running an uncontrolled clinical trial for centuries, and the rate at which promising compounds turn up among plants named by traditional healers is several times the rate among plants collected at random. Artemisinin, now the basis of malaria treatment worldwide, was isolated in the 1970s by a Chinese research team that worked through classical medical texts and found the decisive hint, including the instruction to steep the plant rather than boil it, in a manual written in the fourth century.

For most of the 1990s the pharmaceutical industry lost interest in all of this. New techniques allowed chemists to generate enormous libraries of synthetic compounds and to test them automatically at great speed, and a library sitting in a freezer has obvious advantages over a rainforest: no expeditions, no permits, no negotiations, no seasons. The results disappointed. The compounds produced this way were structurally simpler than natural products and worked less often, and the output of genuinely new medicines fell rather than rose. Interest in natural products has since recovered, helped by analytical methods that can now identify a compound from a quantity far too small to have been useful thirty years ago.

The difficulties of the older approach were real, though, as one famous case shows. Paclitaxel, one of the most effective drugs against several cancers, was isolated from the bark of the Pacific yew, a slow-growing tree of the north-western American forests. Stripping the bark kills the tree, and a single course of treatment for one patient required the bark of roughly three mature yews. Scaling that to clinical demand was impossible, and for some years the supply was the limiting factor in the treatment. The solution came in two stages: chemists learned to build the molecule from a precursor extracted from the needles of a common European yew, which regrow, and the drug is now produced in tanks of cultured plant cells without any tree being involved.

The legal position changed as well. Until 1992 the genetic resources of a country were widely treated as the common heritage of humanity, which in practice meant that anybody could collect them and anybody could patent what they found. An international convention of that year established instead that states have sovereign rights over the biological resources within their borders, and a later protocol set out what a collector must now do: obtain the consent of the country in advance, agree terms for sharing whatever benefits arise, and respect the rights of communities whose knowledge is being used. The principle is hard to argue with. The practice has produced a great deal of paperwork, a marked reduction in the number of collecting expeditions, and remarkably few cases in which substantial money has actually flowed back to a source country.

Meanwhile the material itself is disappearing, and not only through extinction. Knowledge of what a plant does is held in the language of the people who use it, and languages are being lost faster than species. A study published a few years ago examined medicinal plant knowledge across several regions and found that about three quarters of the uses recorded were known in one language only, so that the loss of that language would take the knowledge with it whether or not the plant survived. The point is easy to miss because the plants remain standing. What vanishes is the information about which of them is worth examining.

It is tempting to conclude that every unexamined plant is a potential medicine, and that is not what the evidence supports; most extracts do nothing useful, and the failure rate in this field is as brutal as anywhere in drug discovery. The correct statement is narrower and, I think, more serious. A plant that has not been screened has not been shown to be useless. It has not been asked. Every species lost and every language lost closes a question before it was put, and the number of questions available to us is not obviously large enough to be spending them at this rate.

A placebo is a treatment with no active ingredient, given so that the effect of a real treatment can be measured against something. The phrase placebo effect, however, is used loosely to mean whatever happens to the patients who receive it, and that is not at all the same thing. People enrol in trials when their symptoms are at their worst, and symptoms at their worst tend to improve whatever is done. Many conditions get better by themselves. Patients who know they are being watched change what they do and what they report. Every one of these produces improvement in a placebo group without the placebo having caused anything, and separating them requires a comparison that most trials do not contain.

The confusion has a specific origin. A paper of 1955 entitled The Powerful Placebo reviewed a set of trials and concluded that about a third of patients respond to an inert treatment, a figure that entered the textbooks and has never entirely left them. Reanalysis half a century later showed that none of the trials reviewed had included an untreated group, so the celebrated third was the sum of the placebo response and everything described in the previous paragraph. When investigators later gathered the much smaller number of trials that did compare placebo with no treatment at all, the placebo arm showed little advantage on objectively measured outcomes such as blood pressure or cholesterol, and a real but moderate advantage on what patients reported about how they felt.

That distinction turns out to be the useful one. Where the outcome is a person's experience, particularly pain, nausea, fatigue and itching, placebo responses are consistent and reproducible, and they behave in ways that indicate a mechanism rather than an artefact. They are larger from an injection than from a tablet, and larger from two tablets than from one. They vary with the manner of the clinician. Most tellingly, placebo pain relief can be blocked by a drug that blocks the body's own opioid system, which is difficult to explain unless the expectation of relief is causing the release of those opioids. In Parkinson's disease, an inert treatment given in the expectation of benefit produces measurable dopamine release in the relevant part of the brain.

The same machinery runs in reverse, and the reverse is a larger clinical problem than the forward version. Negative expectation produces real symptoms. The clearest demonstration involved patients who had stopped taking cholesterol-lowering drugs because of muscle pain; in a trial in which each patient took the drug, an identical placebo and nothing at all in randomised months, the muscle symptoms reported during placebo months were almost as severe as during drug months, and both were considerably worse than during the empty months. Reading a list of possible side effects reliably increases the rate at which those side effects are reported, which places the warning leaflet in an awkward position: it is legally required, ethically defensible, and a cause of the harm it describes.

One line of work appears to defy the whole framework. In several trials, patients were given inert pills and told in so many words that the pills contained no active medicine, that they were made of an inert substance, and that placebos have nonetheless been shown to help some people. They improved anyway, relative to patients given nothing, in irritable bowel syndrome, chronic back pain and cancer-related fatigue. The effect sizes are moderate, the trials have generally been small, and the comparison is almost always with no treatment rather than with an effective drug. But the results have been repeated by independent groups, and they undermine the assumption that deception is what makes a placebo work.

Ingrid Vasquez, a specialist in pain medicine, thinks the field's enthusiasm has outrun its evidence in a way that carries a cost. The effects, she notes, are on reported symptoms rather than on disease, they are usually short-lived, and almost nothing in this literature shows a placebo altering the course of an illness. Her worry is about what clinicians take from it. An instruction to harness the placebo response can be heard as permission to offer attention in place of treatment, particularly to patients whose complaints are already dismissed as psychological, and it invites the suggestion that a person who has not improved has failed to believe properly. That is an old and destructive idea returning in a laboratory coat.

Her warning is well aimed and it does not follow that the research should be ignored. What it supports is narrower than the popular version and more practical. The context in which a treatment is given is not neutral packaging around the active ingredient; it is itself a variable that affects the outcome, and unlike most variables in medicine it is almost free to change. A consultation long enough for the patient to be sure they have been understood, an explanation of what the medicine will do and when, and a warning about side effects phrased so that it informs without suggesting, all measurably alter results. None of that involves giving anybody a sugar pill, and all of it is a use of the same evidence.

My own conclusion is that the placebo response is neither an illusion to be subtracted nor a power of mind over matter. It is what a measurement contains when the thing being measured is a person's report of their own condition, and a person's condition genuinely includes their expectations about it. That is a fact about the nature of symptoms rather than a loophole in the laws of physiology, and it has two consequences. Trials must be designed so that the effect can be separated out, which they often are not. And clinicians should understand that they are administering the context whether they attend to it or not.