Excess mortality in Glasgow and Scotland – a statistical artefact?

Faster life histories (LHs) underlie the excess mortality in Glasgow and Scotland. They are particularly consistent with premature mortality from risky behaviours, and may also contribute to overall mortality from chronic diseases.

Far from being pathological however, faster LHs evolved as an adaptive response to dangerous and unpredictable circumstances, where long-term survival is uncertain. LH theory simply describes the calibration of someone’s life course to the length of time they can expect to stay alive.

Risk-taking behaviour pays off more in unpredictable environments, where the benefits of long-term investment may not be realised. As a result, we have a tendency to “discount the future” and prefer immediate rewards in such situations. The relevance to drug and alcohol abuse is clear – they are used as a short-term escape from ongoing problems.

Future discounting is also reflected in ways of attaining status – education only leads to higher status in the long term, whereas risk-taking behaviours can lead to immediate social status (especially among males). Excess deaths from road traffic accidents are highest in young males, as driving fast is seen as a way of attaining status. Male competition drives the violence that leads to other excess deaths. Fast LHs, which encourage competition, are triggered by factors like low status and resource scarcity. This is because natural selection acts on reproduction as well as survival – we need to do both for our genes to survive. Low status males have nothing to lose, with dire prospects of reproduction and survival.

Lower investment in long-term health contributes to the excess overall mortality from chronic diseases. This is not just the result of conscious decision-making. At the heart of LHs is a trade-off between growth and reproduction. Curtailed growth means faster maturation and so reproduction, but comes at the cost of lower biological investment in the immune system, tissue repair and maintenance. A lifetime of lower investment in health inevitably leads to the earlier onset of chronic diseases, and higher mortality rates because of them. Unfortunately this is mostly irrelevant in evolutionary terms, as after we stop reproducing, it makes little odds to natural selection whether we die at 65 or 85. (An interesting exception may be due to grandmaternal care, which is a theory why the human menopause evolved, and why women live longer).

Faster LHs explain why poverty and inequality cause worse health and social outcomes. But why are LHs in Glasgow and Scotland even faster than would be expected after accounting for these risk factors? The Glasgow Centre for Population Health recently published a new synthesis of evidence on the excess mortality. The hypotheses in it can be compared to established triggers of faster LHs, to see if they are feasible threats to long-term survival and reproduction. Any potential trigger may not actually threaten survival in modern society, but as a species we evolved largely in small bands of hunter-gatherers, and in this context responses to threats can be understood as more adaptive.

The first factor proposed is historical overcrowding. This was a problem in Glasgow and Scotland from around 1939 until after 2000. The extent was significant – in 1971, overcrowding was more than twice as widespread in Glasgow than in Liverpool or Manchester, across every deprivation decile. In the most deprived decile, more than 60% of households were overcrowded in Glasgow, compared to less than 30% in the English cities.

A time lag would be expected between exposure and a rise in mortality, due to life course effects – early exposure can take a lifetime to manifest itself. The gradual increase of the portion of mortality unexplained by deprivation fits with a mini “epidemic” of people affected by overcrowding growing old and appearing on the mortality statistics. In 1981, most of the higher mortality could still be explained by deprivation, but this declined over the next 20 years according to the previous synthesis of evidence by GCPH.

However, the emergence of unexplained mortality since 1981 may also partly be an artefact of the marked decline of overcrowding during this period. Overcrowding is one of four components of the Carstairs Index, along with car ownership, male unemployment and social class. The index is used to account for area deprivation. In 1981, overcrowding was twice as common in Glasgow as in Liverpool and Manchester, but it is unclear why this should immediately have led to higher mortality (indeed it is likely that all components of the index would show a significant lag effect).

There is no reason to expect mortality to follow immediately as overcrowding declines, as its effects can take a lifetime to show up in mortality statistics. The inverse is that higher levels of mortality from 1981 onwards can be seen as the lagged effect of pre-1981 overcrowding. It just happened to be around 1981 when overcrowding started to decline, creating the illusion of a new phenomenon emerging. The reason for the emerging excess isn’t (just) that mortality decreased more slowly in Glasgow over the past few decades; the decline in overcrowding meant that deprivation accounted for less and less of the mortality.

The natural response is to look for any conditions which worsened around that time, which could have caused the increasing excess mortality. If though it is just an artefact of lagged deprivation, the opposite in fact applies: the rapid improvement in overcrowding created the statistical excess. A more ecologically valid way of accounting for deprivation may be to introduce a lag period in the analysis itself. This could be done by adjusting mortality statistics with the deprivation measure from a certain number of years previous.

The decline in the percentage of mortality in Glasgow which the Carstairs Index can explain between 1981 and 2001 correlates almost perfectly with the decline in overcrowding during the same period. There is no equivalent unexplained excess of poor health, as there is no (or a much shorter) time lag from exposure to outcome. Conversely, people don’t die young solely because of what happens to them in the last year or even decade of their life.

The authors don’t seem to see the excess as the result of the decrease in the deprivation profile: However, the causal pathways are complex, and the relationship [of overcrowding] with excess mortality less clear, given that the excess has increased over a period in which Scotland has become relatively less deprived compared with the rest of Britain.” The two measures are partially confounded however, as the measure of deprivation is used to establish the level of excess mortality. In addition, two of the four components of deprivation (overcrowding and car ownership) have seen Scotland close the gap to England and Wales between 1981-2011.

People born in 1991 would be amongst the first to experience lower levels of overcrowding from birth. In 2011, they would have been 20 – barely old enough to register on premature mortality. As their generation ages though, there would be an expected plateau or decrease in premature mortality, if early years exposure to overcrowding and deprivation more generally is driving the excess mortality. This process of improvement may be slowed by intergenerational effects, including epigenetic effects, which perpetuate the influence of a stressor even after it has been removed. Rats born to depressed or anxious mothers showed increased DNA methylation at the glucocorticoid receptor gene, and increased stress reactivity. Overcrowding is a risk factor for poor mental health, suggesting a potential mechanism for overcrowding to harm health both directly and through intergenerational epigenetic effects.

The lagged effect of overcrowding might predict that premature mortality would show an excess before all age mortality did. If this model is correct, exposure to overcrowding would move through the population as it ages, and premature mortality from overcrowding would be expected to decline before a similar reduction in overall mortality.

Another effect would be that it is the same population exposed to overcrowding that shows up in both premature mortality and all age mortality. This could explain the excess being bigger for premature mortality (around 30% compared to 15%), as exposure shifts mortality towards younger ages. This has a disproportionate effect on premature mortality simply because many fewer people die before the age of 65.

The same process could also contribute to the strong socioeconomic gradient that is shown for premature mortality. Within the exposed cohort, people of low SES face a double disadvantage, and so are much more likely to die before 65. Over-65s in the exposed cohort are therefore disproportionately of higher SES, which cancels out the social gradient in mortality. In overall mortality, this translates into the flat pattern of excess mortality across deprivation deciles. There is still an excess as exposed, high-SES people die younger than non-exposed people of equivalent SES in the comparator areas.

A life course approach may also shed light on low SES groups suffering more early deaths. Early life SES is a strong predictor of SES in early adulthood, and young people experiencing low status both during childhood and currently are at much higher risk. As people age, the correlation between childhood and current SES weakens (although social mobility has declined since the 1980s). People who have moved up or down the social ladder over their life cloud the picture, and flatten the social gradient in all-age mortality. This is consistent with the “health constraint” hypothesis, which posits that socially mobile individuals have health characteristics of both SES groups they move between, minimising health differences between groups.

The other side of the lagged deprivation argument is that it might predict there would be a time period when overcrowding had begun to rise higher in Glasgow, but before it had time to have an impact on mortality. Mortality would then actually be lower than expected given contemporary deprivation as measured by overcrowding.

Early exposure is the core tenet of a cohort effect like this. People who grow up in Scotland and move away still show “Scottish” levels of excess mortality. Conversely, people who grow up elsewhere in the UK and move to Scotland retain lower levels of adult mortality. Both lines of evidence support the theory that early life exposure of some kind is driving the excess mortality in Scotland. Early exposure is also consistent with a “critical period”, during which LHs are calibrated.

The causes of death seem to reflect a cohort process too. The earliest causes, like drug and alcohol abuse and suicide, show a much bigger excess. The lower excess for later-life causes, like cancer and heart disease, may reflect the smaller cohort surviving long enough to contract and die from these diseases.

In terms of the mechanism by which overcrowding harms later health, established disease mechanisms have already been identified. Risk factors of overcrowding like early infections, damp, mould and disturbed sleep, all documented by a report by Shelter into poor housing, could all lead to worse later-life cardiovascular health.

High exposure to infectious diseases is however known to be a vital LH variable. Populations around the world with exceptionally high pathogen loads have independently come to the solution of radically curtailing growth, and become what are commonly known as pygmies. They die much younger on average than other populations, so have to prioritise early maturation and reproduction, at the expense of adult height.

Clearly, Glaswegians aren’t exposed to dangerous levels of tropical diseases. But socioeconomic status (SES) is a strong predictor of both adult height, and of exposure to overcrowding. Height itself correlates with health status. Again, faster LHs can be seen as the overarching factor, reducing height and worsening long-term health.

The effect of infections on LHs may be twofold: firstly, as an indicator of higher general pathogen load in the environment, which predicts more frequent future illnesses. Secondly, early infections can compromise ongoing health status, increasing susceptibility to future insults. Both are important predictors of expected lifespan, and lead to the perhaps counterintuitive outcome that in conditions where health is at risk, the adaptive solution is to sacrifice investment in long-term health.

Another theory put forward in the new synthesis of evidence is that housing policy, especially the building of New Towns and peripheral council estates, increased Glasgow’s vulnerability to poverty and deprivation. The socially selective process of populating the New Towns must have broken up established communities. Slums were cleared, and housing estates and high-rise flats were built on a much larger, and arguably inhuman scale in Glasgow than in Liverpool or Manchester. This suggests the scattering of traditional communities must have been greater. The resulting loss of community described by concepts like social capital and connectedness is thought to be an important social determinant of health.

From a LH perspective, social support is crucial in a social species to long-term survival and reproduction. Any perceived lack of affiliative relationships is likely to affect LHs by speeding them up, as throughout evolutionary history, threats to personal health and safety like victimisation or famine were better resisted with support from kith and kin. Normally, levels of support may be captured by measures of SES, but the fracturing of communities may have decimated support networks, without increasing deprivation per se.

Perhaps significantly, the resettlement was carried out along class lines, with the New Towns “skimming the cream” off Glasgow. The resulting social stratification may have undermined feelings of social solidarity, encouraging a more individualistic, hierarchical outlook, leading to increased status anxiety.

The democratic deficit argument, that Scots and Glaswegians in particular felt a lack of control over their lives from the 1980s onwards, chimes with the psychosocial risk of feeling out of control of one’s life, as the authors point out. LHs may be the mediating factor here, accelerating people’s lives in response to the inherent unpredictability of circumstances at the time.

The huge increase in premature mortality from around 1980 would appear to need a cause that emerged at the same time. Thatcherism is the obvious candidate, but any explanation needs to show why Glasgow and Scotland were particularly disadvantaged. The response of local government is cited as exacerbating UK economic policy in Glasgow, whereas it was mitigated in the other cities. This is linked to the lack of social capital, as lower levels of politicisation meant that there was less opposition to commercial development ahead of social investment in Glasgow.

A final specific risk factor identified was negative physical environment, in particular the amount of vacant and derelict land. This has already been explicitly linked with LHs, with the density of dilapidated structures found to be independently correlated with premature births and low birth weight in the area, two fast LH traits. These are bad health outcomes for infants, but they are also associated with worse lifelong health outcomes, and so increased mortality. The rationale is that dereliction signifies the unpredictability of future outcomes, in turn encouraging riskier reproduction.

Faster LHs are clearly implicated in the biological, cognitive and behavioural strategies which result in the excess mortality in Glasgow and Scotland as a whole. Several of the new hypotheses put forward to account for the excess are feasible triggers of faster strategies, independent of SES. However, just as the overwhelming majority of worse health and social outcomes and inequalities are caused by already understood socioeconomic factors, LHs are also overwhelmingly determined by SES. As the authors note, the definition of deprivation may need to be updated. Factors may need to be added that haven’t been considered part of the concept before, but which predictably lead to faster LHs, and so earlier deaths.

By Breck MacGregor

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s