In November 1981 at a function in London, Neville Butler, a professor of paediatric medicine at Bristol University, contrived to drop a cup of coffee at Margaret Thatcher’s feet. He stooped down to mop it up, then sprang up and asked her for money. ‘I’m Professor Butler,’ he’s reported to have said. ‘We’re doing a national study looking at thousands of children. We need more funding!’ She ordered a member of her entourage to ‘talk to this gentleman’, then pushed on to give her speech. Helen Pearson makes no judgment as to whether splashing Margaret Thatcher with coffee helped Butler to get his funding. She does tell us that he spent his evenings writing begging letters to the worthies listed in Who’s Who, and that he managed to get money from the likes of Robert Maxwell, Cliff Richard and Twiggy.
Since 1958 Butler had been the co-ordinator of a perennially impoverished study which began by examining the medical, social and economic circumstances of 17,000 babies born in the same week in March 1958, then over subsequent decades monitored their health alongside their educational and economic outcomes. This was the second such study; the first, led by James Douglas, followed nearly 14,000 individuals born in the first week of March 1946. By 1981 Butler was running the project with Mia Kellmer Pringle, an educational psychologist born in Vienna who had arrived in Britain as a refugee in the 1930s (Pearson records her telling a colleague that she’d had three hurdles to overcome in Britain: ‘she was a foreigner, she was a woman and she was clever’). Butler and Kellmer Pringle also ran a third study, which followed a cohort of 17,000 born in 1970. Cohort studies are an unusually powerful means of teasing out causes and effects in sociology and medicine. To take just one example, a comparison of the data across the three cohorts shows that a rise in obesity during the 1980s affected all three at the same time, probably due to changes in diet, affluence and an increase in car travel.
Cohort studies are notoriously expensive to run, and vulnerable to fluctuations in the political climate. The early 1980s, like today, were a bad time to ask the government for money. Lord Hailsham had described the social sciences as ‘a happy hunting ground for the bogus and the meretricious’. It was, as Pearson says, a belief widely held in Tory circles: the social sciences, they felt, lacked rigour as a discipline, producing results ‘everybody knew’. In 1982 the Social Science Research Council, the SSRC, had its name changed to ESRC, to emphasise economics over ‘social’ science. Early in Thatcher’s first term, Keith Joseph, then her secretary of state for industry, told researchers: ‘I’ll start funding your research when you start telling me things I want to hear.’ In 1979 Kellmer Pringle had been promised funding for a fourth cohort study, scheduled to begin in 1982, under the care of the mathematician turned social scientist Jean Golding. But an earlier report on the cohort studies had concluded that ‘there is nothing to contradict and everything to support the theory that social class differences are widening rather than diminishing’ – not the kind of thing Keith Joseph wanted to hear. The 1982 study was cancelled, and the chain of data comparing 12-yearly generations was broken.
Another study, smaller in geographical scope (it is focused on the Bristol area) but broader in clinical ambition was launched by Golding in 1991. It has collected 1.5 million tissue and fluid samples from 14,062 children – from their placentas, umbilical cords, milk teeth, urine, hair and plasma – and has been pivotal in the study of genetics because it combines DNA sampling with a record of social and economic outcomes (it is well known that height is correlated with economic power, but the study has been able to make correlations between DNA and foetal growth, eczema, bone density and, startlingly, success at university). The first ALSPAC cohort (Avon Longitudinal Study of Parents and Children) are now having children of their own, making it possible to compare DNA samples across generations. A fifth study began in 2000 – the story is that it was set up at the behest of Peter Mandelson as a more enduring way of marking the millennium than a big wheel or a dome – and has been following around 20,000 children born between late 2000 and early 2002. A sixth study was planned for 2015. The Life Study was to be the biggest yet, following as many as 80,000 children (later scaled down to 16,000) and would have examined air and water pollution, exposure to mobile phone radiation, mothers’ and babies’ intestinal flora, blood, urine, saliva and afterbirth, and video of parents interacting with their babies, as well as all the usual medical and economic demographic data. In 1946 James Douglas managed 98 per cent recruitment for his study; the Millennium Cohort Study got 72 per cent. After six months of recruitment, the Life Study had signed up just 249 people – victim perhaps of its own ambition, or of people’s increasing reluctance to commit their time and their data. In July 2015 the study was abandoned.
One of the earliest cohort studies was conducted in the 1830s by the statistician William Farr, who, working from the Office of the Registrar General, compiled data on mortality figures for various demographic groups. He showed that nearly one in seven babies died within a year of being born, and in 1841 calculated the death rates in England’s lunatic asylums: they were comparable to those in London during the Great Plague. There weren’t many studies of this kind – government had very little data on which to base judgments. In 1901 Seebohm Rowntree showed that in the city of York, 20,000 out of a total population of 46,000 were living in food poverty; before the study it had been assumed this was largely a London problem. In 1946, James Douglas’s study showed that a quarter of families still didn’t have enough food.
Douglas was the first to examine multiple outcomes across a broadly distributed, representative sample of the population, and he used his findings to make the case that the state should support the health of pregnant women. He investigated the costs of pregnancy and childbirth, and calculated that before the introduction of the NHS, an average pregnancy and delivery cost the equivalent of six weeks’ wages. He demonstrated how much more likely babies of unskilled workers were to die than those of the well-to-do, and tabulated the risk of death for mothers in childbirth: highest during the first delivery, as anticipated, but then rising again after four or five pregnancies. He was able to show that continuing to work during early pregnancy carried no risk to the foetus, but that working late in the final trimester increased the likelihood that your baby would be born underweight. He showed that having a mother who worked carried no social disadvantage (more recent studies have shown that having two parents at work is associated with childhood obesity – a correlation that hasn’t yet been properly explored). He also found that bedwetting as a child was linked to low mood and emotional problems as an adult, and that bright working-class children had half the chance of getting a grammar school place as their equally bright middle-class peers. Analysis of the 1958 cohort indicated that, in the primary years, it was doubtful whether private schooling conferred ‘any advantage in measured attainment’ that ‘could justify the additional costs involved for parents’. Data from that cohort informed the 1967 Plowden Report, which recommended the introduction of free nursery places for all children over three years old, and the banning of corporal punishment (recommendations that took, respectively, twenty and thirty years to implement).
When he was putting together the questionnaire for the 1958 cohort, Butler decided at the last minute to ask whether mothers had smoked during pregnancy. In 1950 Richard Doll and Austin Bradford Hill’s study of British doctors had shown that smoking increased the risk of lung cancer, and in 1957 an American, W.J. Simpson, had shown that smoking when pregnant might be associated with premature birth. Butler’s study showed that smoking was associated with poorer outcomes for the baby; Harvey Goldstein, the study’s statistician, quit smoking while crunching the data, as did his wife. The effect was limited to later pregnancy; mothers who gave up by the end of the first trimester had the same outcomes as lifelong non-smokers. The knowledge took decades to take hold. As late as September 1973 an editorial in Nature claimed that Goldstein and Butler’s study poured unjustifiable guilt on any mother who was a smoker and had lost a baby: ‘Mothers-to-be have always been under pressure to avoid excessive weight gains and this pressure, it is well known, frequently causes distress … Cigarettes often keep both weight and nerves under control.’ At that time 39 per cent of pregnant mothers smoked; the figure now is around 12 per cent. When I delivered babies as a junior doctor it was usually obvious which women were in that 12 per cent: the placenta would be more fragile and friable, and mottled with pale clots.
In the 1940s there was growing awareness that high blood pressure could be a risk to health. It wasn’t enough to save Franklin D. Roosevelt. In 1941 his blood pressure was measured at 188/105 (the norm is around 150/90 or lower); it wasn’t checked again until 1944, when it hadn’t improved. By the time of the Yalta Conference in February 1945 it was 260/150, and shortly before he died (of a brain haemorrhage) in April that year it was a catastrophic 300/190. In 1948 the National Heart Institute in Boston set up the Framingham Heart Study, which followed a cohort of around five thousand men and women between the ages of 30 and 62 in the town of Framingham, Massachusetts, calling them in every two years for blood pressure tests as well as gathering data on smoking, obesity, diabetes, exercise and cholesterol levels. A glance through the Framingham findings over subsequent decades shows the influence they have had – consolidated by the British cohorts – on modern preventative medicine. In 1960, cigarette smoking was correlated with heart disease, and in 1961 with cholesterol and high blood pressure; in 1970 the study found that high blood pressure and irregular heart rhythms were associated with stroke, and in 1988 that high levels of HDL (‘good’ cholesterol) protected against stroke.
By the mid-1980s Douglas had retired, his cohort taken over by Michael Wadsworth. The Douglas ‘babies’ were by then in their late thirties, and Wadsworth showed that high blood pressure and obesity were more common among those who’d been underweight as babies. There were, it seemed, patterns and cycles to disadvantage. Around the same time David Barker began to show the ways in which nutrition during pregnancy had long-lasting effects not just on babies’ growth, but also on their risk as adults of developing heart disease, diabetes and obesity. Barker argued that to improve the health of the next generation it was vital to improve the health of young women and girls before they became pregnant. As early as 1986, in the Lancet, Barker suggested that those with poor nutrition as children were the least able to cope with an affluent diet in adulthood, and so were doubly disadvantaged. When Jean Golding, having been refused funding for a 1982 cohort, began planning the ALSPAC study, Barker urged her to ‘get the placentas’. In 2013 it was shown that placental structure and blood pressure in childhood are correlated.
The ALSPAC study has generated some unlikely associations as well as reassuring ones. A diet rich in fish can have a beneficial effect on IQ; low iodine levels in urine are associated with poorer IQ. Analysis of trace contaminants present in the umbilical cords of infants showed that mercury-based vaccine preservatives are unlikely to be harmful. We already know that the presence of particular genes is predictive of weight gain, but the ALSPAC study shows that having two copies of one particular ‘fat’ gene, FTO, leads to just a 3kg weight difference – the suggestion here is that environmental factors may be more significant than genetics in determining obesity.
Douglas’s study of 1946 and Golding’s of 1991 have been primarily interested in medical data; they have been funded largely, if patchily, by the Medical Research Council. The 1958 and 1970 cohort studies have been more heavily influenced by the interests of social science; they have been funded by the SSRC and ESRC and have generated vast quantities of data about social mobility and education. (The loss of the 1982 cohort has significantly limited the scope of these studies.) Absolute social mobility improved over the course of the 20th century, largely because of an expansion in the number of higher-status professional roles. But relative social mobility – the likelihood of moving up or down the socioeconomic ladder relative to your peers – hasn’t changed at all. Between 1958 and 1970 mobility actually decreased: adult income became more tightly correlated to parental income, and home ownership fell. Members of the 1970 cohort who were born high up the social ladder have sprinted even further ahead: in Pearson’s words, poverty has got ‘stickier’ while the beneficial effects of wealth have grown. The expansion of university places in the 1990s and 2000s for the most part benefited those least in need of it, and a study in 2007 showed that more than 50 per cent of the leading figures in law, politics, medicine, journalism and business had attended private schools, though only 7 per cent of children are privately educated.
Parental interest can compensate for a lack of financial power to some degree: the children of the most interested and involved parents on a low income may do ‘better’ than wealthy children whose parents are less interested (outsourcing involvement in your children to a boarding school skews the data). Reading for pleasure as a child has a stronger influence on school test results than having a parent with a university degree, and school catchment effects are less important than parental attention. These findings were picked up by David Cameron in a speech in 2010, when he suggested that poor parenting mattered more than poverty when it came to cramping childrens’ potential – a grossly slanted interpretation of the data. ‘You can’t just tell parents to be better parents,’ Pearson concludes, ‘because most of them are already trying as hard as they can.’
The early data from the studies were held on punch cards. By the 1980s this was causing difficulties for social scientists and in 1984 the ESRC asked John Bynner to investigate; he recommended that the cohort data be more widely advertised, and transcribed into electronic formats. In the 1990s the data were amalgamated and digitised at the University of Essex. Economists and social researchers from all over the world could now get access to, and play with, these databases; some 2500 papers have been published drawing on the 1958 cohort’s data, and 770 examining the 1970 cohort. In 2003 Leon Feinstein of the LSE published a contentious graph dividing the 1970s children broadly into four groups: rich and bright, rich and dim, poor and bright, poor and dim. He showed that by the age of six or seven, the rich dim kids had caught up with the poor bright ones in terms of test results, and that by age ten they had consolidated their advantage. The analysis remains hugely controversial because of the raggedness of the data and the varying ways in which wealth and intelligence can be measured, but it was cited by Labour in support of its Sure Start pre-school funding programme, and in 2011 found its way into the coalition government’s Strategy for Social Mobility.
‘The continuance of social evils is not due to the fact that we do not know what is right,’ the socialist and economic historian R.H. Tawney said in 1913, ‘but that we prefer to continue doing what is wrong. Those who have the power to remove them do not have the will, and those who have the will have not, as yet, the power.’ A hundred years later John Major made a speech to the South Norfolk Conservative Association in which he registered how little had changed. ‘In every single sphere of British influence,’ he said, ‘the upper echelons of power in 2013 are held overwhelmingly by the privately educated or the affluent middle class … To me from my background, I find that truly shocking.’ Without cohort studies, the extent of our social problems – and potential ways of solving them – will remain opaque. As a general practitioner my work is informed daily by the medical conclusions of these studies. Their findings in the field of economics and social science have had a significant influence on policies intended to help the less privileged, though clearly not to the degree that R.H. Tawney (or John Major) would have hoped.
In February, Pearson took part in a discussion on Radio Four’s Start the Week exploring the distinction between nature and nurture. She hardly got a word in, which is a pity because in The Life Project she supplies an elegant explanation of the light cast by cohort studies on that ancient crux:
Health is a consequence of everything that has happened to us through life: on the biological side, that includes the genes we inherit from our parents, our development in the womb, growth in childhood, maturation during adolescence and our behaviour as adults; on the social side, it encompasses our social class at birth, our parents, homes, schools, jobs and socioeconomic status as adults. One thing influences the next, which influences the next, and so on in a chain reaction until all the benefits and risks have accumulated to produce the state of health that we have today.
In time the medical and social aspects of the ALSPAC and Millennium Studies will probably be brought together. Thousands of individual subjects have already had their DNA sequenced, and as the data is correlated with what have traditionally been understood as social outcomes, we can expect to gain a much deeper understanding of the interplay between genes and environment.