by Wayne Jonas
Like the Great Wall Hospital’s healing environment, the one Norman Cousins produced for himself also had no research to back it up. Subsequent studies of high-dose vitamin C demonstrated very small to negligible effects on the disease. Cousins didn’t know it at the time, but when given at the doses he was taking, vitamin C is a toxin—an oxidant rather than an antioxidant—and so produced repeated stress on his body. As for laughter and the immune system, only a small amount of research had been done. Yet Cousins reported that with these treatments, he improved almost to the point of a complete cure.
HOW COMMON?
I wondered: how common is this pattern—support, stimulus, and belief—in healing? To find out, my team and I conducted a series of field investigations on a variety of practices in several countries. Our goal was to analyze what the practices did and what kind of results they were getting. We visited and performed in-depth evaluations of more than thirty centers around the world. And we saw this same pattern in all of them. Under the right conditions, results like that those experienced by Norma, Bill, Sergeant Martin, Aadi, Xiao, and Mr. Cousins were common. These centers and clinics often produced marked clinical effects. And like the others, they lacked scientific proof. We saw that healing was, in fact, possible and could be induced for many chronic diseases. But when we sought to isolate any one treatment component from the rest and measure its contribution on one outcome—as is required by good science to prove them—the effects diminished, disappeared, or at most contributed only 20% to 30% to a patient’s improvement. The process and ritual of treatment produced the rest. Was this a common pattern when science was applied to healing no matter the treatment—be it herbs, diet, or drugs?
The more I deeply dove into the science, the weaker what I thought I knew from medical training became. I’d seen Norma, Bill, Sergeant Martin, and other patients flounder under my care, while I was using the best evidence—and then get markedly better after using nonscientific methods, despite my skepticism. Seeing patients like Aadi and Xiao, whose diseases are presumed to be incurable, resolve their symptoms using ancient systems of healing, based on no real science, caused my world to fray around the edges.
There was no good evidence that prayer or astrology, massage, cathartics, or herbs cured Parkinson’s disease. Similarly, there was no scientific basis for the belief that needles and tai chi could cure ankylosing spondylitis, or that vitamins cured arthritis, or surgery reversed back pain, or oxygen treated brain injury. How widespread were these phenomena? How often did healing occur? How often did science miss it? And why? Was this simply from lack of research on healing treatments, or was there something in the way we tried to prove healing that interfered with our seeing it? It seemed that the secret to 80% of healing was right under my nose. But how could we test it if we could not even see it? Then I remembered Sarah.
SARAH
Sarah and her baby were not supposed to be in Germany. Her husband, an Army truck mechanic working for an engineering battalion, was on a one-year tour without family. But she came anyway. As a twenty-one-year-old, newly married and a new mother, she did not want to be away from him. So she moved to the town of Dexheim, Germany. I was the physician responsible for the military unit in Dexheim—a small American outpost. Sarah and her husband were both from Kansas, and now Sarah was depressed. This was not surprising, given her environment. She lived off post in a rundown apartment. She hadn’t graduated from high school and didn’t speak any German. She had a new baby. And she was far away from home. When her husband would come home from days in the field, he would often find the house in a mess and his wife sleeping or crying in her bed with the baby. They came to me for help.
I diagnosed Sarah with postpartum depression, began some counseling, and started her on an antidepressant medication, a type of medication called a selective serotonin reuptake inhibitor, or SSRI. About a month later, she came back in and said she had stopped the medication. Since starting it, she had lost all interest in sex or intimacy, which before the baby had been a major part of life for her and her husband. She was sure it was the medication, since the drop-off in interest had occurred right after she started taking it.
“Doctor,” she said, “while my moods are somewhat better, it is making our marriage worse. My husband is understanding, but don’t you have anything better?”
I doubted that her decrease in desire for intimacy was due to the medication. More likely, it was a continuing symptom of her postpartum depression. But to counter her opinion would be counterproductive. I asked them to come in together the following week so we could talk about it and I could do a bit more counseling.
Later that week, I was at a medical meeting and consulted one of my colleagues about what to do. The small clinic I ran was in a remote section of Germany, and German doctors did our emergency transport service when we had car accidents or drug overdoses on the post—something that happened about once a month. I spoke German from living in Germany as a child and knew many of the local doctors. On this occasion, I asked one of my closest German colleagues about Sarah’s case. “Oh yes,” he said, “I’ve seen many women like this—far from home and homesick. Young and without a social support network, they now have the full responsibilities of a baby and the demands of a husband.”
This doctor suggested two things. “First, give her the homeopathic remedy Gelsemium and then add the herb Hypericum,” he said. “Its common name is St. John’s wort. It works just as well as the antidepressants, but doesn’t have the side effects that you describe. Make sure you explain what it is for and what it should do.”
I had never heard of these treatments. I went back to my office and looked for any research on them. Gelsemium was a homeopathic remedy that not been studied but, from what I could tell from its very low dose, had to be a placebo. The homeopathic books said it was good for homesickness and described cases similar to Sarah’s—but with no testing or proof. St. John’s wort, on the other hand, had been studied several times for depression and was slightly better than placebo in randomized controlled trials. It had a good safety profile.
It seemed that there was little downside to these treatments, so I made note of these two additional tools for when I met with Sarah and her husband.
Because her husband had described how their home was quite cluttered and dirty because of her depression, I suggested that we have someone come in and provide household help. I also asked if they would be willing to have someone come watch the baby a couple times a week so Sarah could go to events on the post; she agreed. I then told them about the two remedies that my German colleague had recommended.
I read to her the indications from the book about Gelsemium (good for homesickness) and translated the description of St. John’s wort from a German flyer. It was “an ancient herb with a beautiful yellow flower—like a ray of sunshine in a plant.” Not only did studies show that it lifted mood, but it would be less likely to produce a decline in sexual interest. After hearing that these didn’t have the side effects Sarah thought were produced by the antidepressant, they both said, “Absolutely, Doc, we want to try that.”
The two medications were not in the military pharmacy, so I told them to go to their local German apothecary to pick them up. I asked them to come back in three weeks.
When she returned, Sarah reported feeling much better. Her husband was out on another a field exercise, but she had not been crying as much and had gone to several of the women’s support sessions on base. “I met another woman from Kansas,” she said. “She grew up a couple hours from where I did. We meet for coffee now between the women’s meetings. And get this,” she said, with more animation than I had ever heard from her before. “She is three months pregnant!”
Sarah made a plan to clean up the house with a friend while her husband was gone. I asked her to continue the medicines and come back again in another three weeks, this time with her husband. Three weeks later—now six weeks after the change in environment and treatment—they came in, both smilin
g.
“Doc,” said her husband, “that St. Germ’s wort really works! She is feeling much better. Thank you so much.”
I never found out how their love life was going, but I assumed a bit better. I followed them for another six months before they returned to the United States, and Sarah continued to do well and the baby began to thrive. I found out on one of the last visits before they left to go home that she had stopped taking both Gelsemium and St. John’s wort but had continued to function and cope.
Certainly, something had helped Sarah. Perhaps it was the change in the physical environment in her house after someone came in to clean it. Perhaps it was the friend she had made. Perhaps it was because their sex life had improved. Perhaps it was the homeopathic Gelsemium (most likely a placebo) or the herb St. John’s wort (a mild antidepressant). I wondered, though: was it really the St. John’s wort? Her improvement was quite dramatic.
When I looked at the research literature, the evidence indicated that the herb did help with mild to moderate depression, but the effect was small—not much larger than what you’d see with placebo. Was it the herb, the friend, the clean house, or the presumed intimacy that healed her? Why had it worked when the drug did not?
ST. JOHN’S WORT
A few years later, I had the great fortune to test this question directly by helping to design and fund a large clinical study of the same herb and drug for depression that I had given Sarah. It was an unusual type of study, more rigorous than most drug tests. Normally, new drugs are tested in two-armed studies in which patients are randomly assigned to the active treatment (drug or herb, for example) or an identical-looking placebo. These types of clinical studies are expensive, so before a drug gets to such a test, a series of laboratory studies are usually done to show the drug is absorbed into the brain and affects the chemicals thought to be involved in depression. If these first studies are not done, a clinical study comparing a drug to placebo is usually not done. The scientific community was skeptical about St. John’s wort’s effectiveness because these types of preliminary studies either had not been done or did not show direct effects in the brain. While antidepressants could be explained because of the known effect they had on certain chemicals in the brain—SSRIs, for example, impacted a neurotransmitter called serotonin—there was no specific chemical in a sufficient amount in St. John’s wort to affect any known brain chemical related to depression. The herb contained small amounts of several chemicals, and one in particular, called hypericin, seemed to impact the brain in a variety of ways, but the amounts of hypericin in the herb—including the amount I gave to Sarah—were too low to reasonably explain these effects. Most scientists in the United States thought the studies done in Germany, which were funded by the herbal companies who made them, were probably biased, and that the data showing benefit was wrong. Even though I offered to fund the study from my budget at the NIH, the other NIH institutes were hesitant to undertake the test.
Finally, Dr. Bob Temple, one of the most respected researchers at the FDA, had a solution. We should do a three-armed study in which patients were randomly assigned to the herb, a placebo, or a proven FDA-approved antidepressant drug: sertraline (brand name Zoloft), one of the SSRIs with a known mechanism of action and clinical effects, which I had prescribed for Sarah. With this design, the director of the National Institute of Mental Health at NIH agreed to conduct the study. He had always wanted to have his institute run an independent drug study on depression (most were done by pharmaceutical companies), and a direct comparison with the herb and placebo had never been done. We found one of the most respected mental health researchers in the country to do it—Dr. Jonathan Davidson from Duke University. I had known Dr. Davidson for years. A wonderful psychiatrist, originally from England, he was not only a great researcher but also a compassionate and careful physician, the type of psychiatrist who listens carefully to you and spends the time needed to understand you. Like Dr. Manu, he possessed an unusual presence—a healing presence—and his British accent lent an air of sophistication and authority to any encounter.
Dr. Davidson carefully constructed a study designed to separate the placebo and his own healing effects from those of the herb and drug. However, when we approached the companies to supply the herbs and drug—standard procedure even for government-run studies—the drug company balked. They did not want to participate in the study. They made more than a billion dollars a year on the sale of sertraline. When I offered to make public their noncooperation, they relented and agreed to supply the drug. However, before the NIH study began they launched their own, two-armed study, comparing the herb to placebo—exactly the type of study not recommended by the FDA, because it did not have a positive (proven drug) control group. In that study, they selected patients who had worse depression than we planned to test—worse than Sarah had and worse than the subjects of the German studies. Those patients were less likely to respond to the herb. By spending a great deal of money, they sped the trial to completion in an attempt to beat the NIH and show that St. John’s wort did not work. And they did, publishing a negative study—St. John’s wort and placebo had equivalent effects—a full year before the NIH study was done. The conclusions of this study—that St. John’s wort did not work—spread widely. Sales of the herb dropped.
Unlike the public, my colleagues held out for the results of the more rigorous, three-armed study by Dr. Davidson. His study picked the right types of patients; gave them the right doses; did proper blinding with placebo, so no one knew who was getting the herb, drug, or placebo; and included a large enough number of patients to rule out chance. What would it show? Would the herb beat the drug? Likely not, I thought. Would it beat placebo? It should. Would it have fewer side effects than the drug? If Sarah’s experience was any indication, it would.
When the data was analyzed and they had broken the blinding code to see the effect in each group, I waited with anticipation. It had been nearly ten years since I had treated Sarah with both the drug and the herb, and it took three years to complete the study. My bet was that the drug and herb would work better than placebo but the herb would not work as well as the drug—but with fewer side effects.
Turns out I was wrong. All three groups—whether they were taking the herb, drug, or placebo—got better at the same rate. There was no difference in the rate or degree of improvement in depression. The herb and placebo had fewer side effects than the drug, however, confirming what I had seen with Sarah.
I was flummoxed. When the study was published in the prestigious Journal of the American Medical Association (JAMA), the news headlines around the world reported that a major NIH study had shown that St. John’s wort did not work. Sales of the herb dropped further. What few people picked up on, however, was that the proven drug had also not worked any better than placebo. This was the most important finding of the study, and it was totally missed. The healing effect of the ritual—of just getting treated and seen in the placebo group—was as powerful as both the drug and the herb.
Knowing Dr. Davidson and his bedside manner, I was not surprised that many of the patients got better, including the ones on placebo. I had seen it in my own studies and with patients such as Norma. But the scientists and the public were so focused on whether the herb or drug added more than placebo—by even a small amount—that the actual reasons for healing were completely overlooked. Something about the way we were going about our science—always attempting to reduce it to the smallest, most specific focus—was causing us to miss out on healing.
THE DECLINE EFFECT
The publicity around the JAMA-reported study implied that drug treatment was more effective than St. John’s wort, and it heightened the likelihood that the drug and not the herb would be prescribed by other physicians. I interpreted it quite differently—as did many of my FDA colleagues. To me, it was further evidence that something other than the agent produced healing. Previous research had demonstrated that between 70% and 80% of the improvement seen in clinical studies wi
th both St. John’s wort and antidepressants was also seen in those taking placebo. The single study by Dr. Davidson had simply confirmed this. And this, as it turns out, is more often the rule than the exception. The more rigorously one looks at almost any treatment, the smaller and smaller the effect size becomes compared to placebo; as the science gets better, the difference between placebo and the real treatment diminishes.
This is known as the “decline effect.” We see it over and over again in clinical research. The more rigorous and larger a study, the smaller the actual contribution of the active treatment. Early studies, especially smaller pilot studies, frequently show large effects, which encourage physicians and patients to use the treatment. Usually those smaller studies are not enough for FDA approval or to be accepted by the mainstream community as standard of care; therefore, additional and larger studies follow. As those larger and more rigorous studies are done, the effects diminish. When you combine the results of those studies in a method called meta-analysis, frequently the effects became so small as to become irrelevant for use in practice.
What’s more, the effects of even proven treatments frequently cannot be replicated by others when taken out of the hands of the original investigators. This “replicability problem” has been extensively reported by Dr. John Ioannidis, chair of the Department of Medicine at Stanford University, and others. In a startling analysis of clinical research published in the Journal of the American Medical Association in 2012, Ioannidis showed that only about one-third of proven results can be replicated. These were not just pilot studies, whose test treatments often show a decline effect in subsequent studies; these were failures at replications of established and proven therapies—like the antidepressant I had prescribed to Sarah. Soon, others began to look at research outside of clinical medicine and found that replicability was a general problem in science. Even laboratory and basic research—where we control many more factors than in clinical research—can be replicated only about 30% to 40% of the time. While the decline effect shows that initial findings often shrink or even disappear in the end, the replicability problem shows that even what are thought to be proven effects are usually not replicable. If someone else tries to repeat a study, often the effects of the active agent disappear. What is left are other nonspecific or unknown factors variously attributed to the “placebo effect.” As I explain later, I prefer to call these effects—the component that hold the majority of healing—“the meaning effect.”
