Before I embarked upon medical training, I aspired to be a clinical psychologist. One of the several reasons I changed track was when, after approximately 5 years of studying a self-consciously empirical and academic discipline, my course recommended a popular self-help book with a daisy (or maybe it was a sunflower) on the front as a key reference.

Perhaps it’s little surprise then, that several years later once I was a medical doctor and designing my own lifestyle intervention (with the initial goal of treating heart disease), that I dismissed the idea that stress was a relevant consideration. Both Dr Dean Ornish and Dr Caldwell Esselstyn Jr have published intervention studies demonstrating reversal of atherosclerosis^1,2. A key difference is the intervention used: Dr Ornish’s program involved a low-fat vegetarian diet, moderate exercise, and stress management, whereas Dr Esselstyn’s involved just a low-fat vegan diet. Given that Dr Esselstyn’s results were arguably more impressive than Dr Ornish’s, and perhaps also my pre-existing bias for biological causes of physical disease, I was very happy to apply Occam’s razor and dismiss at least the stress management aspect of Ornish’s program as superfluous.

Had I at the time delved further into Dr Ornish’s writings and reasoning, I’m certain I would have been surprised to learn of Nerem et al.’s now classic 1980 interventional study examining atherosclerosis in rabbits fed a 2% cholesterol diet: those which were individually petted, held, played with, and talked to on a regular basis developed 60% less atherosclerosis despite similar serum cholesterol, blood pressure, and heart rate³. The INTERHEART study’s finding that almost as much of the population attributable risk for myocardial infarction was due to psychosocial factors as to smoking (32.5% vs 35.7%) ought also to have provided pause for thought⁴. Of course neither demonstrate such factors would be of importance over and above Esselstyn’s ‘plant-perfect’ cardio-protective diet, but the idea that thoughts or feelings might in fact lead to physical disease is nonetheless intriguing. Additionally, ‘stress management’ and ‘positive social connection’ have been proposed as two of six factors of focus for lifestyle medicine⁵, so let’s take a further look.

‘Psychosocial factors’, as the term would imply, could include any psychological or social influences on health. For example, the INTERHEART measure included depression, stress at work, financial stress, life events, and locus of control. Despite the potentially broad definition, factors that have been the major focus of research include stress (especially), psychological distress (e.g. depression, anxiety, and hopelessness), job control, hostility, and connectedness.

In a recent TED talk, Canadian psychologist and author Susan Pinker discusses the Sardinia Blue Zone, which is apparently the only place where men live as long as women⁶. Pinker believes social connection is the cause. She also cites the work of another psychologist, Julianne Holt-Lunstad, who has recently argued that social connection must be seen as a public health priority, given the compelling evidence for its influence on health outcomes. In fact, not only is loneliness more prevalent in the United States than obesity, effect sizes estimated from meta-analyses suggest that social connection is significantly more effective in reducing mortality than not being obese (or being physically active, or abstaining from alcohol), and comparable to quitting smoking⁷.

Although the effect of stress in adulthood is less marked than the ‘classic’ biological cardiovascular risk factors, those with work or private life stress nonetheless have a 1.1 to 1.6-fold increased risk of incident coronary heart disease and stroke (see Kivimäki and Steptoe for an excellent review, including summary of stress’s potential pathophysiological effects⁸). Furthermore, the effect of multiple severe stressful experiences in childhood is more significant, and adulthood stress is an important disease trigger in people with high atherosclerotic plaque burden and a determinant of prognosis and outcomes in those with pre-existing cardiovascular or cerebrovascular disease⁸. Similarly, biological responses to stress observed in animal studies that would appear to be diabetogenic are supported by findings in humans that raised evening levels of cortisol predict new onset Type 2 diabetes, and epidemiological studies that implicate depression, chronic work stress, and early life adversity as risk factors (see Hackett and Steptoe for another excellent review⁹).

There is even a substantial body of research implicating psychosocial risk factors in outcomes of HIV/AIDS and cancer¹⁰. Positive coping strategies such as finding meaning have been linked to slower HIV disease progression, and social support has increased survival, CD4+ count, and viral suppression¹⁰. Psychosocial risk factors that signal poor psychological adaptation to cancer have been associated with poorer clinical outcomes in breast, skin, ovarian, renal, hepatocellular, colorectal, and haematologic cancers, and there is also evidence that positive psychosocial characteristics are related to better long-term health outcomes¹⁰. It is suggested that effects on these diseases are mediated by the ability of psychosocial factors to affect the immune system.

For obvious reasons, randomised interventional studies involving humans are rare. However, some enterprising researchers used a crossover trial to demonstrate that married couples’ blister wounds healed more slowly, and that local cytokine production was lower at wound sites following marital conflicts rather than positive social support interactions¹¹. In results that may be of even more interest given the current coronavirus pandemic, the Pittsburgh Common Cold Studies have famously demonstrated that psychological stress is associated with increased risk for developing respiratory illness in persons intentionally exposed to a common cold virus¹². They also discovered social integration was associated with reduced risk irrespective of stress level.

We could perhaps be forgiven at this point for concluding that psychosocial factors would be an important consideration for public health intervention, not only ostensibly to improve psychological health in the population, but also to decrease the incidence and/or improve the outcomes of ‘physical’ diseases of everything from the common cold to cancer. Not so fast.  McLeod and Davey Smith have argued that the observed associations may not be causal¹³. Albeit writing in 2003, their interpretation of the evidence at that point was that it provided ‘… little support for a direct causal relation and no basis to propose psychosocial interventions as a public health strategy to improve health in general or to reduce health inequalities.’ They suggested the apparent relationship might reflect issues of reverse causation, reporting bias, and especially ‘…confounding by aspects of the material environment…’ that was typically associated with adverse psychosocial exposure. McLeod and Davey Smith are particularly critical of the fact that psychosocial interventions have shown little effect on physical disease outcomes, a point that is worth exploring as it has implications for how we address psychosocial factors, even if we accept they have a role in chronic disease causation or outcomes.

I would hazard a guess that most people reading this article, along with myself, would be fairly satisfied that the aforementioned research (much of which is published well after McLeod and Davey Smith’s critique) is compatible with psychosocial factors playing at least a minor role in chronic disease, separate from confounding. How then can we explain the nonetheless mixed at best results from intervention studies? I feel McLeod and Davey Smith’s concern over the influence of the material environment as a confounder points us in the right direction: psychosocial factors such as stress are exquisitely complex in their aetiology and interventions aimed at ‘reducing stress’ may be particularly ineffective if these are largely a predictable reaction to a particular set of environmental circumstances. Practitioners and people have little if any control over important factors such as finances, work circumstances, and life events for example, and most interventions cannot hope to address these generally, and much less so the complex combination of these each individual may experience. For instance, with a classic intervention such as cognitive behavioural therapy we are intervening at the proximal level only to try to affect factors that are likely to be caused and maintained primarily by distal determinants – stress isn’t caused by a lack of meditation or ‘mindfulness’in the same way that hypercholesterolaemia isn’t caused by a lack of statins.

Furthermore we can predict that chronic stress, which is of course more so the kind of stress that is likely to lead to chronic disease is nearly entirely due to environmental factors, and particularly the material factors that McLeod and Davey Smith are concerned about being labelled as ‘psychosocial’ factors the individual is seen by society to have responsibility for:

“Psychosocial solutions may seem attractive to some policy makers—because they permit location of responsibility for health at the level of the individual and their unhealthy feelings. Unlike material solutions, psychosocial solutions do not necessitate fundamental social change…”

This is congruent with my experience as a General Practitioner, and I imagine rings true for the rest of us in the healthcare sector. In the end we have precious little influence on factors that are clearly causing and maintaining the ‘stress’, ‘work control’, ‘depression’, or ‘connectedness’ of our patients. There isn’t so much utility in asking a homeless patient who is sleeping in their car to reframe their thoughts and core beliefs, or telling a solo parent they need to take more time for themselves.

So then, given that we have now established that psychosocial factors are likely to be associated with chronic disease, what can we do with this information as practitioners? My suspicion is not a lot. Perhaps the first step is to realise that these factors tend to be substantially less important than other factors that patients have more control over, particularly diet and physical activity, which of course both can improve psychosocial factors such as stress, anxiety, and depression in any case. One of the reasons Dr Esselstyn decided not to have his patients add physical activity or stress management is that he believes people have a fixed capacity for what they can be changing at once – I tend to agree with this, and because we know both that diet is so important in the aetiology of chronic disease and that all of our patients must eat regardless, I feel this is the best place to start. In some cases it may seem appropriate to suggest patients could try a meditation app like Headspace or Smiling Mind, or increase their social connection by joining an interest group or club for example (the Citizens Advice Bureau maintains a community directory, or the Meetup app also provides a platform for becoming involved with others with shared interests).

Clearly sustained stress and disconnection are not natural states for humans, but these are the new normal and are part and parcel of our capitalist societies, driven by the desire for relentless GDP growth regardless of the social and environmental costs. The inequality that is a natural and necessary consequence of our so-called meritocratic distribution of wealth and resources contributes inestimably to both physical and psychological ill health and myriad other negative social outcomes (see Wilkinson and Pickett’s books The Spirit Level and The Inner Level for a persuasive analysis). Over and above our work as health practitioners, advocating for a transition to a new model such as that proposed by Kate Raworth in Doughnut Economics is ultimately what is required in order to address the psychosocial factors that have been associated with chronic disease. The economic disruption caused by the current coronavirus pandemic and ever worsening climate crisis provide the perfect context for our policy makers to implement such a transition.

Although I feel McLeod and Davey Smith are ultimately incorrect in their denial of a causal role for psychosocial factors in chronic disease, their assertion that these must not become a target themselves of public health intervention in a way that trivialises and ignores the actual determinants and places the blame for this on individuals is correct. In other words, the way to deal with the psychosocial factors that impact health in a meaningful fashion in the longterm is not to hand out advice from self-help books with daisies on the front. We must address the true causes of these stressors, namely material deprivation, social inequality, and an outdated system that glorifies paid work and places value on revenue generation above the leisure time, friendship and community needed to live a healthy, happy and meaningful existence.