The Sickening Mind: Brain, Behaviour, Immunity and Disease. Paul Martin

of life events include the death of a partner or family member, divorce, marriage, starting a new job, moving house or financial problems. At the other end of the scale, minor upheavals such as family holidays and Christmas are also classified as life events. The basic hypothesis underlying this work is that any disruptive changes, whether desirable or undesirable, are potentially stressful and can increase our chances of falling ill.

      Thousands of research projects have investigated the relationships between life events and health. The majority of these studies have used a standardized method for assessing life-event stress called the Social Readjustment Rating Scale. In its simplest form this involves asking each individual to record which of forty-three types of life event they have experienced over a specified period, usually between six months and two years.

      Each type of life event is assigned a standard score according to its supposed severity, rated on a scale from o (least severe) to 100 (most severe). The maximum rating of 100 is awarded to the death of a spouse; divorce is rated 73; marriage, 50; changing to a different line of work, 36; moving house, 20; Christmas, 12; and so on. (Personally, I would rate Christmas at around 60, and anyone who has recently experienced the horrors of moving house may be excused for wondering at its modest rating.) A composite score is then calculated for each individual, taking account of both the total number of life events they have experienced and the relative awfulness of those life events. A high score can denote a few serious life events or a multitude of minor ones.²

      If it is true that life events act as risk factors for illness then people who register high life-event scores should, on average, have more illnesses than those whose lives have been undisturbed by change. Simple. By and large, this is what the research has found.

      A seminal early investigation looked at the effects of life-event stress on US Navy personnel during the Vietnam War. The results showed that individuals with the highest life-event scores suffered almost twice the number of illnesses over the following months as those with low scores. In another study scientists asked young men in a navy submarine training establishment to record the life events they had experienced over the previous twelve months; again, the incidence of life events correlated with subsequent illness.

      The general conclusion from several thousand such studies is that people who have been exposed to lots of life-event stress have a slightly greater risk of illness. This increased risk applies across the board and seems to encompass virtually every form of ailment and disease under the sun, ranging from headaches, common colds, allergies and inflammation of the gums to mental illness, coronary heart disease, leukaemia, diabetes, tuberculosis and multiple sclerosis. Life-event stress also has an impact on childbirth; women who register high stress ratings during the year or so before pregnancy tend to give birth to babies with slightly lower birth weights and a slightly poorer overall state. Life events are even associated with an increased risk of minor accidents and sports injuries.

      As well as suffering more episodes of illness, people with high life-event scores also tend to be ill for longer, have more severe symptoms and take longer to recover.

      Not surprisingly, the adverse effects of life events are generally worse when the life events are severe, undesirable and clustered together in time. In the early days of life event research it was widely assumed that ‘good’ life events, such as getting married or starting a new job, were potentially just as damaging to health as ‘bad’ life events of comparable disruptiveness. However, more recent research has tended to support the common-sense assumption that, other things being equal, undesirable life events are inherently more damaging than desirable ones.

      It has to be said that the link between life events and later illness is not as neat and simple as it sometimes appears. Some of the research on life events has been justifiably criticized for a variety of reasons. This is not the right place to debate the abstruse technicalities of research methodology. Nonetheless, the difficulties inherent in life event research are of broader relevance and therefore merit our attention.

      First of all, the statistical correlation between life events and illness is highly consistent but it is also fairly weak. Life events do have a bearing on health, but not a very major bearing. Typically, life events account for only about 10–15 per cent of the total variation in the incidence of illness. A number of those who are exposed to stressful life events become ill, but most do not. Conversely, it is possible to fall ill despite living a life of unruffled stability. A phenomenon that is highly significant in a strictly statistical sense – meaning that the patterns in the data are more than just chance variations – may not necessarily be highly significant in a clinical or scientific sense.

      A second fundamental point is that correlation is not the same as causation. The existence of a statistical association between two things is not proof that one of them causes the other. The population of the world and the age of the current pope are correlated, but there is no causal connection between the two. They both happen to be independently related to a third variable – time. So the correlation between life-event scores and illness does not by itself prove that life events are a direct cause of illness. The causation might even work the opposite way round; that is, chronic illness might conceivably precipitate life events. For instance, someone’s marriage or career might run into problems because they are ill. And it may be the case that things which are classified as life events, such as sexual problems or changes in sleep patterns, could in fact be symptoms of an existing but undiagnosed illness.

      In order to disentangle cause and effect in this type of research it is vital to establish which came first, the life events or the illness. There is plenty of evidence that life events do indeed tend to precede illness, which suggests that they may genuinely contribute to ill health.

      A third pitfall with life event research, especially in its early days, has been its retrospective nature. When investigators ask subjects to recall their life events during, say, the previous year, great reliance is placed on frail and faulty memories. And therein lies a weakness. It is an awkward fact of life that most of us grossly over-estimate our ability to recall the past accurately and objectively. Ask any policeman, lawyer or judge about the reliability of witnesses to crimes. Psychologists have found that after a period of ten months people are typically able to recall life events with an accuracy of only 25 per cent. Conclusions that depend on people’s memories of what happened to them one or two years ago are therefore bound to be suspect.

      As well as the inherent difficulty of recalling past events accurately there is also a danger of systematic bias. People who are unwell may focus on a particular trauma in their past and assume it must have been responsible for their illness. We all have a basic need to find explanations for our illnesses and some people understandably attribute their poor health to traumatic experiences. But in doing so they inadvertently undermine the objectivity of the research data.

      Fortunately, not all research on life events has had to rely on faulty memories. Instead, scientists have monitored groups of initially healthy subjects over a period of time, recording their life events and illnesses as and when they happen. This style of research is referred to as prospective, in contrast to the backward-looking retrospective method. And plenty of these prospective studies have borne out the link between life events and subsequent illness.

      Another potential pitfall lies in failing to distinguish between an interviewee’s actual health, as measured according to objective, clinical criteria, and what they say or think about their health. The problem here is not that people deliberately lie; the majority of those who volunteer to take part in scientific research try hard to be truthful. The real problem is that few of us are capable of being entirely objective about our own health. We all perceive and interpret our physical symptoms in different ways; something that would constitute a distressing malady for one person might not even be noticed by another.

      Problems also arise if we attempt to measure health in terms of what is called sickness-related behaviour. This means behaviour like going to the doctor or taking sick leave from work. Sickness-related behaviour is obviously not the same thing as actual sickness.

      The way humans respond when they think they are ill depends on other factors besides their state of health,