7.

The Millennium Children

Cohorts Reborn

By 1999 scientists in Britain had a lot to be happy about. They had four birth cohorts rolling along, started in 1946, 1958, 1970 and 1991. This was reason for incredible pride: no other country was following generations of its people in this way, and using them to understand the roots of mental and physical health. And no other country could claim that high-profile reports based on birth cohort studies were finding their way to the heart of government, where they were starting to drive a sweep of policies concerning social mobility, preschool education, adult literacy and much more. Cohort studies were on the up, and all the discoveries that had been made on past generations were slowly seeping into the present day and making life a little better for the next.

And yet, amidst this celebration, the cohort scientists just couldn’t get rid of a nagging regret. They kept on thinking about the missing cohort, the one that should have taken place in 1982 to continue the twelve-year series. That was the one that had died in the Oxford meeting room as a result of Jean Golding’s wildly overambitious plans and its poor timing – pitched when funding was drying up and other scientists were becoming enamoured with randomized controlled trials. So, years later, the joy over the cohorts that Britain did have was still soured by the one that it didn’t. It was as if the scientists were celebrating a family reunion, but were always dwelling ruefully on the one family member whose death had left an empty place.

That Golding had successfully started a fourth cohort later on – the 1991 one – only provided a little salve for the wound, because in many scientists’ minds that study was different and did not properly fill the gap. The fourth cohort only included children in one region, rather than, as the first three cohorts had, spanning Great Britain and representing people across the country. And the 1991 cohort, with its banks of placentas, nails, teeth and DNA, had such a heavy emphasis on modern medical research that it seemed wholly different from its predecessors, which were rooted in previous eras. It was an outsider amongst the cohorts; a study that never really fit.

The growth of science over the last few decades had created a broader split between the four British birth cohorts. Back at the start, James Douglas had been determined that his 1946 cohort should span both social science and medical science, and over the years all the cohorts had vacuumed up reams of information that was relevant to the two disciplines. But, by the millennium, each cohort found itself standing firmly in one discipline or the other – and this was a reflection of the scientists who were running it, using it and paying for it. The 1946 and 1991 cohorts were seen to be biomedical cohorts – focusing on medical research, led by medical researchers and funded by medical research bodies such as the MRC. By contrast, the 1958 and 1970 cohorts were seen to be social science cohorts – led and used by social scientists and mostly funded by the ESRC. The boundaries were certainly not absolute – there was excellent medical research carried out on the social science cohorts and vice versa^aa – but, in general, the cohorts were moving down two parallel roads. It created a divide, because the scientists working in the disciplines tend to talk slightly different languages and follow different customs.

Although the cohort studies are often viewed from a distance as being either ‘biomedical’ or ‘social science’, all of them are involved in both spheres of research. One major example of crossover is the highly successful sweep, starting in 2002, that focused on the biology and health of 1958 cohort members in their forties and that has grown into a major scientific resource.

Mostly this divide was apparent only as an unspoken tension, a slight feeling of ‘my cohort is better than yours’. But sometimes the differences erupted into a spat. One of the biggest points of dispute was over data, and how it was stored. The social scientists had long ago made the decision to deposit all their data in the Essex archive and make it freely available to academics, in the belief that this would lead to wider use of the studies and ensure that scientists could repeat and check each other’s results. But medical researchers had not done this, because the tradition in that field is for scientists leading a project to keep control over the data: they get the first shot at analysing it, publishing the results and then they only share it with others later, if they are asked. This is partly because medical data is viewed as exquisitely sensitive and confidential – no one wants to risk people’s personal health records leaking out – and partly because the field is intensively competitive and scientists are always vying to make discoveries and publish results first. (It is also an increasingly outdated custom because in the last decade or so a movement has sprung up to open up many medical databases for researchers to use, while maintaining the anonymity of people’s data.)

These differing approaches to data meant that the social and medical scientists had a fundamentally different approach to the way in which they ran their cohorts. The social scientists viewed their two cohort studies as a resource that they were running for the benefit of the scientific community: whenever they were planning a sweep, scientists and policy wonks all around the country could suggest hypotheses to be tested by the cohort, help to shape the questionnaires and then use the data. The medical researchers, however, devised their own hypotheses and kept the data close to their chests. This drove a wedge between the studies, because each side tended to think that its approach was best. Some social scientists found it appalling that the medical researchers could run a hugely expensive study, paid for by taxpayers’ money, and not freely share the data. They thought the medical scientists were selfish, guarding the data so that they could get first dibs on it, publish all the exciting findings and further their own academic careers. Some medical researchers, for their part, thought that the social science cohorts lacked a clear scientific vision. They thought that the social scientists were acting like servile data collectors and managers – doing whatever policy-makers told them – rather than steering a course for themselves. Because everyone was reserved and British, none of this was ever said openly. Mostly, cohort scientists are unfailingly polite to each other and then snipe behind each other’s backs.

So, in 1999, the British birth cohorts were a national triumph, but they were also marred by tension and a mournful sense of what could have been. What made this feeling even worse was that it seemed so typical of Britain as a whole. The nation had found a scientific pursuit – following people’s lives en masse – at which it led the world and excelled, but it was now at risk of letting that advantage slip away by not continuing to set up new cohort studies. It was like so many other things that Britain had started – cricket, sandwiches, the computer – that were quickly improved on by others, until we were soundly beaten at our own game. Before too long, some other country would be bound to outdo us at cohort studies too, because we had been too quarrelsome, short-sighted and tight-fisted to start any more.

As it turned out, however, that was not to be the case. And the scientists found this out in the summer of 1999, just a few months before the start of a new millennium. In the wider world, everyone was already making plans for a massive New Year’s Eve party: the supermarkets were stocking up on champagne and planning was under way for a spectacular fireworks display over the Thames. In the birth cohort world, the excitement started when a phone rang on the desolate concrete campus of Essex University, and a social scientist called Jonathan Gershuny picked it up. Gershuny didn’t have too much to do with the British birth cohorts (he was busy running another large longitudinal study at the time^bb) but for a short time in 1999 he had nevertheless been placed in charge of birth cohort strategy and was the point person as far as the ESRC was concerned. Gershuny answered the phone. It was someone from the ESRC headquarters in Swindon, with surprising news: ‘The Cabinet Office has decided there should be a millennium cohort study,’ he was told. It would be a fifth British birth cohort study – and the first in thirty years to span the country.

The British Household Panel Survey began collecting information on all the members of some 5,500 households across the UK in 1991. Like the birth cohort studies, the household study tracks people over time, but the important difference – and the reason that social and economic scientists like such studies – is that, while birth cohort studies follow individuals over time, the household study follows groups of people – all those living in the home – which provides more information on the changing make-up of the household and the relationships between people in it.

Not only that, but the Cabinet Office wanted the cohort to happen right now, and they wanted everyone to know about it. ‘Will you please write the press release by lunchtime?’ Gershuny was told.

It is very hard to work out where the idea of a millennium cohort was born. By now, John Bynner was directing the 1958 and 1970 cohort studies from offices in London’s Institute of Education, which had become known as the Centre for Longitudinal Studies.^cc Certainly the older cohorts were gaining prominence and momentum, and in the higher echelons of funding agencies and the civil service there was a lot of lobbying in favour of cohorts going on behind the scenes, as well as an awareness that the national cohort series needed to be revived. So there was plenty of enthusiasm and support for a new cohort study. All it needed was a spark.

Neville Butler still maintained a presence: his loyal secretary drove him from Bristol to the office once a week. Butler continued to operate his charity to support the birth cohort studies even though it was making very little money and he himself was nudging eighty. His unwavering devotion to the studies continued until his death, in 2007.

According to cohort legend, that spark arrived in the form of Peter Mandelson, who was appointed Minister without Portfolio in Tony Blair’s new government in 1997. Mandelson was one of the masterminds behind the rebranding of the Labour Party as New Labour, helping it to win its landslide election in 1997, and he later became known for his ruthless approach and for being forced to resign twice from the Cabinet due to political scandals but then staging phoenix-like regenerations to secure other high-level posts.

But, before that, it was part of Mandelson’s job to work out how the government should mark the turn of the millennium. The story goes that as he was sorting through proposals for bridges, domes and a rather implausible-sounding big wheel on the bank of the Thames, it seems that someone – but no one knows who – suggested that it might be impressive to mark this once-in-a-thousand-year opportunity with a major study to learn about the children born at the turn of the millennium. Mandelson apparently liked the thought – after all, a study of thousands of adorable British babies born at this momentous moment in time has an undeniable dramatic and political appeal. So the word went out that if scientists could come up with a credible plan for a birth cohort to be started in the year 2000, they would receive a substantial sum of money to get it on the road.

When the news of the cohort reached the scientists, they had two reactions. The first was that they loved the idea. The second was that it was impossible. Anyone who had been in the cohort business for a few weeks, let alone a few decades, knew that you needed at least two or three years to start a new birth cohort study. This was easy to understand if you worked backwards from the births themselves. You have to start contacting mothers about six months before the births to enrol them during pregnancy; you need six months before that to start establishing procedures to recruit mothers at antenatal clinics. Add to that at least a year, but realistically two, for researchers to work out how to recruit people across the country and what information to collect. That adds up to two to three years in all. Back in 1979, in fact, when Golding was proposing her ill-fated cohort, one of the stakes to its heart was the timing: it seemed an unworkable rush to get a cohort together in the three short years before 1982. Now, seventeen years later, the government told the scientists in the summer of 1999 that they wanted a birth cohort to launch the following year – even though a substantial number of the babies who would be born in the year 2000 had already been conceived. That is why, when Gershuny first heard of the plan in that phone call from the ESRC, he found himself holding the phone, saying, ‘Sorry, we can’t do it.’

And the ESRC said, ‘Look, if it’s not the millennium cohort, it’s no cohort at all.’

From the government’s point of view, the idea of a millennium cohort had flair, and there was money to blow on millennium events. By contrast, a 2001 or 2002 cohort didn’t have the same appeal and the money would be spent elsewhere. So the scientists were stuck: either they raced to launch a study in the year 2000 and risked compromising the science, or there would be no study at all.

The scientists took the first route. They didn’t want to let the precious chance of a new cohort slip away so easily; nor did they want to lose the opportunity to restart the cohort series. Peter Elias, an economist and adviser to the ESRC, remembers the situation being discussed at a meeting at the research council. ‘We said, “This is crazy, but we can do it. We know we can do it. Let’s do it.”’ So they did. Gershuny wrote the press release in a hurry and the ESRC commissioned him to lead a quick scoping study to show that the cohort was a feasible proposition. (It concluded that it was.) Then the ESRC ran a competition in which it invited any group of scientists to put in a bid to run the new cohort, which would be judged so that the best plan would win. By this time it was already February 2000. The New Year’s party was well over, the millennium was already two months old, and no one had even been appointed to take charge. It was going to be very tight.

Bynner and the cohort scientists scrambled to get a plausible-sounding plan together. Starting in February, they had just a few weeks to design the gargantuan study, which would be drastically different from the last national cohort study, some three decades before. This time the births would be collected over a year or more, so that scientists could examine the sometimes substantial effects of being born in different seasons. The data would be entered directly into a computer for the first time, which meant that the researchers needed a crew of expensive, professional interviewers. (There was no way, in the year 2000, that the nation’s midwives and health visitors could be convinced to collect information on top of already crippling workloads, nor did they have the time or the expertise.) This study would make it a priority to include the growing populations of ethnic minorities in Britain. And, hard though it is to believe, this would be the first national cohort to survey fathers, as well as mothers, from the start.

There were however a few things to be examined in the survey that were very familiar. In 1999 poverty and inequality were just as hot a topic as they ever had been, and social scientists were keen to learn more about the disadvantaged children. They decided to do this by deliberately over-representing them in the count. Although roughly one quarter of children in the UK lived in poverty at that time, around 30% of those included in the millennium cohort were poor, which increased the power of the study to describe the impact of a difficult start.^dd

The scientists achieved this by over-sampling in regions of the country where a high proportion of children live in families receiving means-tested benefits. They also over-represented ethnic minorities and children born in Northern Ireland, Scotland and Wales.

All this was written into the plan, and in May 2000 the team received word that they had won the competition to run the study. A social scientist called Heather Joshi would lead it: she was one of the very few scientists to have worked with all three national birth cohorts, which she had used in pioneering work on women’s pay and employment. And beside impeccable scientific credentials, she had integrity and determination to her name. But by now, Joshi and her team of scientists were facing a hopeless task. Nearly five months’ worth of millennium babies had already been born and the last of the millennium children – the ones who would arrive in December – had been conceived and were making their presence known on pregnancy tests. It was politically unpalatable to rename the study, and therefore it had to include babies born in the millennium year – yet there was no practical way the scientists could start collecting data before the year was up. The team was under immense pressure, the atmosphere was fraught. It was a ‘mission impossible’ for the cohort scientists: would they be able to save the day?

Joshi rolled up her sleeves and got the job done. She and her team devised a workaround – a massive compromise that would allow them to pull off the cohort and keep the name. They decided to recruit babies who were born from September 2000 onwards – so that they could enrol at least some millennium-born babies and justify the cohort’s name – but they wouldn’t actually interview the parents until the year 2001, when the children were around nine months old, giving them time to get the study together. (Joshi identified the eligible children when families registered for child benefit, which most parents in the UK did.) In fact, Joshi realized about halfway through the recruitment phase that she was going to miss the target of 20,000 children because not enough babies were being born. So she stretched out the study to include ‘millennium’ births in some regions all the way up to January 2002.

This decision to recruit babies so late in their lives marked a major change from the previous cohort studies, which had all recruited families through doctors, midwives and health visitors, and had collected a ream of medical information about pregnancy and birth. Not this time. The impossible deadline for the study meant that the mothers were not monitored during pregnancy, far less information was collected about the medical side of birth, and there was no collection of tissues and DNA. The social scientists didn’t worry too much about that – but the medical scientists certainly did.

The earlier cohorts had been ground-breaking in their detailed study of birth; they had established the link between smoking and reduced birth weight, they had been influential in changing obstetric practice, and they had become landmarks in the field of perinatal epidemiology. The birth cohorts had shown that collecting detailed information on pregnancy and birth was scientifically valuable, and it was now well established that foetal and infant development shaped people’s health for the rest of their lives. At the same time, genetic research was exploding and it seemed incomprehensible to most biomedical researchers that you would start a huge, expensive, national birth cohort study without collecting a wealth of medical information during pregnancy and birth as well as the mothers’ and children’s DNA. The chance was missed – jettisoned because of politicians’ fixation on having a splashy-sounding millennium cohort. This still causes some medical scientists to press their lips into a disappointed line whenever the millennium cohort comes up in conversation; to them, the start of the study was the squandering of a fantastic opportunity. But it was all very well to criticize from the side lines: Joshi was working her heart out to just get the show on the road before the money and momentum disappeared. ‘It was a miracle that we collected as much as we did,’ she says now.^ee

The social scientists did work with medical researchers when they were designing and launching the cohort – but the tight timetable severely limited what they could do. What’s more, the cohort was funded at the start by the ESRC, whose focus is on social science, not medical research.

Meanwhile, the progression of time and science was creating other problems for this cohort. In 1946 giving ‘consent’ simply involved Gertrude Palmer inviting the health visitor into her house. In March 2001, just as the researchers were about to visit the first millennium children, the Medical Research Ethics Committee nearly put a kibosh on the entire study over consent, which had become a hot-button issue in research. They found out that prospective parents were being sent a letter informing them that they could opt out of the study (‘please write and let us know if you don’t want to do it’), rather than opt in (‘please let us know if you do want to do it’). The committee considered this too coercive. It was an ‘absolutely hair-raising experience’, says Joshi, who remembers making frantic telephone calls to prevent the cohort from going down. The scientists and ethicists eventually reached a compromise: parents were asked to confirm that they opted in to the study when the interviewers turned up on their doorstep.

So when the interviewers set out, they were armed with consent forms, computers and questionnaires that testify to their moment in time. (Are you married to baby’s father/separated/divorced/closely involved/just friends/not in any relationship? How many hours of childcare do you pay for each week? Did you have any medical fertility treatment for this pregnancy? When it comes to changing baby’s nappies, do you do most of it/does your partner do most of it/do you share it more or less equally/does someone else do it?)

Despite the impossible deadline, Joshi’s army of interviewers managed to recruit 18,818 children – 72% of those eligible. It was a far cry from the 98% that the 1958 cohort signed up, but that’s because lives and attitudes had changed. When the first birth cohort was recruiting, all the mothers took part because they felt an obligation to do their bit for the country. By the time of the fifth, that sentiment seemed as laughably out of date as rationing and hand-knitted vests. Today’s parents feel no obligation to take part in national science projects: they are busy, and they tend to wonder what’s in it for them. They also move house more often, which makes them that much harder to track. And even as the cohort scientists were working to keep parents involved, they had to maintain the interest of the government too – if it had started the cohort on a whim, there was nothing to stop it from turning around just as suddenly and cancelling the entire thing. The scientists knew they would have to fight for funds every time they wanted to do another sweep of the children. But they successfully won money to survey the children at 3, 5, 7 and 11, and as I write they are surveying them at 14.

When the millennium cohort started, then, Britain had over fifty years of experience in birth cohort studies – and yet this newest one was rushed, a searing disappointment to biologists and it had an uncertain future. But it was a triumph at the same time. Scientists had finally got another cohort study on the road, and had gone a long way towards cementing Britain’s position as a world leader in cohort research. Now there were five cohort studies, and another 18,818 children to put under the microscope. The scientists couldn’t wait to plunge into the data to see how the country had changed.

Rebecca Wood remembers when the interviewers turned up on her doorstep asking her, her partner and her son Thomas to take part in a study of children born at the turn of the millennium (or thereabouts). She agreed – in part because she thought she might get a diary of his achievements at the end of it. It was only some years later that she realized this wasn’t the case: that all the information was going into some kind of statistical research and that she wasn’t going to get anything back. But she wanted to stick with the study. Once she’s made a commitment, she says, ‘It’s not in my nature to pull out.’

When Thomas was born in 2001, Wood was working full time in a bank. The pregnancy went pretty smoothly, but the birth did not. Her baby was facing the wrong way, which meant that the doctors anticipated a difficult birth – and they were right. When she was five days overdue, Wood had her labour induced, endured thirteen hours of pain and was eventually given an epidural anaesthetic to quell her urge to push him out too soon. When he finally emerged at 9.27 a.m., she remembers being utterly exhausted and shocked that there were nine medical staff all crammed into the room to help. The whole experience was unforgettable, she says, although she still keeps all her hospital discharge notes upstairs in a box.

Wood would have liked to take more than six months of maternity leave, but the family needed the money and so she went back to work. There was no way she could afford the extortionate cost of childcare, but she was lucky that her parents lived close enough to be able to help with that. The family has not always been able to afford a holiday, but they do have a mortgage and manage to get the children (Thomas now has a younger brother) what they need. And she knows that many people around her are worse off; the area where she lives is one of the more deprived in the UK.

Wood has always read to Thomas, taught him right from wrong, and made sure he’s ready for school and polite. (She didn’t need a cohort study to tell her that, she says; ‘I think that comes from within.’) When Thomas was in infant school she noticed that he was falling behind, but the school told her that it was probably because he was a daydreamer and a summer-born child, so one of the youngest in his school year. It was only last year that he was diagnosed with dyslexia – words appear blurred to him – and he now has assistance at school to read and write. It’s something that worries Wood, who wishes the condition had been spotted earlier and wonders if it could affect his future progress and his ability to find a job.

On top of that, Thomas is about to turn fourteen and has entered a stage where he either doesn’t talk at all or communicates by grunts. ‘I don’t know what’s worse, giving birth or having a teenager,’ she says now.

The millennium children were born into an era of relative prosperity and economic growth, and they were benefiting from a host of changes that the earlier cohorts had helped to introduce. The first cohort had made the case for good maternity benefits and healthcare; by 2000 pregnant women enjoyed free medical care on the NHS as well as a minimum of four months’ maternity leave supported by an employer or government stipend. The cohorts had exposed the unacceptably high numbers of children dying at birth; now, thanks in part to the NHS and radical improvements in healthcare, infant mortality had plummeted.

All the earlier cohorts had also shown the lasting impact on children of being born into poverty and disadvantage. Now living standards had risen enormously, and few children suffered the same absolute levels of disadvantage that some children once faced. But in real terms, poverty was still a huge problem. In 1999 New Labour made a prominent pledge to eradicate child poverty in a generation and started pumping money into benefits for families with children. And work on the birth cohorts had shown that educational gaps emerge between upper- and lower-class children even before they reach school, suggesting that intervention to close these gaps had to start very early in life. That was part of the driver for the Labour government when it introduced free nursery school places for three-to-four-year-olds, and the Sure Start programme to support families with young children.

So the country had changed: the Britain that the millennium children were born into was almost unrecognizably different from the one in which the first cohort had arrived, just after the war. These children were taller and healthier; they had refrigerators, computers, TVs, mobile phones and food from all over the world. The new millennium really looked like a wonderful time in which to be born. But, as scientists started combing through the data on the millennium cohort, they found an all too familiar picture: there were signs of inequality everywhere they looked.

By this point, the London economists were infatuated with the cohort data and they leapt at the chance to use the new figures on the millennium children. When they examined the vocabulary of millennium cohort children at the ages of three and five, they found that those from the poorest families were already an entire year behind those in the richest ones – a cognitive chasm just as large as any measured in the generations before. They also plugged the new data from the millennium cohort into their controversial work on social mobility, looking to see whether the decline in social mobility that they had observed between 1958 and 1970 was continuing. They examined the association between children’s cognitive test scores and their parents’ income and found no sign that social mobility was getting worse – which suggested that the raft of initiatives to support children in the preschool years was helping rich and poor children progress at a more equal rate than they once had. But the economists knew that it was early days and they would have to watch the children pass through the education system, and get jobs, before they could say for sure whether these millennium children were truly less tied to the wealth of their parents than previous generations.

There were other ways in which the data conjured a feeling of déjà vu – particularly when the cohort scientists looked at the disadvantaged children, the ones with the toughest start. One particularly striking study emerged from two social scientists, Ricardo Sabates and Shirley Dex, when they compiled a list of ‘risks’ that a millennium child might face at the start of life that could put him or her on a difficult trajectory. This included overcrowding, financial difficulties, teenage motherhood, domestic violence, disability, parents’ lack of basic reading or numerical skills, alcoholism, unemployment, depression and smoking during pregnancy. Then the scientists added up how many children in the cohort experienced two or more of these risks. The study was, in essence, a rerun of the work in the 1970s bestseller Born to Fail?, which had focused on children in the second cohort who had everything stacked against them – the ones in the worst-off families, who lived in houses with no bathrooms or hot water, and received less than £15 net income per week. Since then, living standards had shot up, but many millennium families were still poor, living in overcrowded homes or struggling in other ways. The question that the scientists took on, using their new criteria, was how many millennium children might be born to fail too?

The answer was, quite a lot. The two researchers showed that 31% of babies in the millennium cohort were facing one of the risks in their list, and 8% had three. These figures were comparable to those in Born to Fail?, when 36% of children had one risk on the list and 6% had all three. Extrapolated across the country, this meant that at least 83,000 children born into this glorious new millennium – and hundreds of thousands more if you factored in their siblings – were growing up in an environment that, if history were anything to go by, was likely to give them a difficult time ahead. Over sixty years after James Douglas had started to show the difficulties faced by disadvantaged children in the first birth cohort, scientists were discovering that a massive swathe of the most recent generation was facing disadvantages at the very start.

Sabates and Dex then examined the data to see if these children were already showing signs of difficulty as they grew up – and found that they were. Compared with children with one or no risks, those with two or more had a poorer vocabulary at the ages of three and five, and scored worse on measures of behaviour, emotional development, social development and hyperactivity. The scientists only had to glance at the mountain of evidence from the earlier birth cohorts to know that these early-warning signs were pointing them towards trouble. They predicted that these millennium children were more likely than their peers to leave school at the minimum age, win few if any qualifications, end up unemployed or earning a low wage, commit crime, drink dangerously or otherwise rack up measures of adversity as adults. They were headed straight on to the trajectory of disadvantage that Bynner had described based on the earlier cohorts, but this time they would follow that trajectory into the distant future of 2070, 2080 or later.

Things looked worse when, in their analysis, Sabates and Dex also revealed that there were no particular types of risks that tended to cluster together and so could be tackled by policies en masse. The most common combination – smoking, financial stress and teenage motherhood – applied to just 6% of children, but most families had an assortment of risks quite different from those to be found next door. This was a dismal conclusion as far as policy-makers were concerned, because it meant there was no quick fix that could pull these children out of disadvantage. Every troubled family was troubled in its own way – and therefore every family would need a different type of support to escape.

Or, as one cohort scientist summed up the study, ‘Crikey, we’ve got a big problem.’

It was true: Britain did indeed have a problem, and it was hardly new. That there are lots of disadvantaged children, and that they tend to follow a difficult trajectory, hadn’t changed since the early days of the birth cohorts when James Douglas exposed the wide inequalities between children in the first cohort. And there was another thing that was being rediscovered – parenting could help.

Back at the start, Douglas had carried out a pioneering analysis on his cohort showing that good parenting was strongly linked to a child’s success in school; crucially, the effect of having interested parents appeared to go some way towards overcoming the disadvantages of a working-class start. Over the years, Douglas had collected information from teachers in an effort to assess how interested parents were in their children’s education, including how often they visited school to discuss their child’s progress. He found that children whose parents were very interested in their children’s education secured many more grammar school places than would be expected based on ability alone. In the 1960s, however, when Douglas reported the results, they were widely overlooked – ‘unforgivably’, as one scientist says now.

But things were different in millennial Britain. A wealth of evidence by this point – from the British cohorts and elsewhere – suggested that having interested and aspirational parents helped children to overcome the setbacks presented by poverty, low class or other measures of disadvantage; it was associated with better outcomes in school and a generally smoother path through life. But, on the ground, out there in the arduous world of parenting, what everyone wanted to know is what being an engaged and interested parent actually meant. If parents – and society – wanted to do the best for their kids, what were they supposed to do?

Scientists had made some progress on this front, often using cohort studies, since Douglas’s day. They had taken the association between ‘interested’ parents and good outcomes, and broken it down into more specific behaviours that interested parents tend to display. It is now widely accepted, for example, that reading to a child every day, from a very young age, is strongly correlated with good performance at school. And so is a host of other things. One cohort study in the UK has followed 3,000 children since the age of three and focused intensively on best practices in bringing up children during the early years. It concluded that whatever parents did to create a good ‘learning environment’ at home was more important for children’s intellectual and social development than their parents’ job, education or income. Reading with a child, teaching them songs and nursery rhymes, painting and drawing, showing them the alphabet and numbers, visiting the library, taking children on trips and visits – all of these were associated with higher intellectual, social and behavioural scores as the children grew up. Other studies have padded this out, and highlighted the importance of parental involvement as children get older too. Talking to children about what they are learning at school, supporting homework, expressing ambitions for their future and talking up the idea of further education are all associated with improved academic achievement. Pretty much any attention paid to a child’s education seems to be highly predictive of success.

The millennium cohort has looked at a broader spectrum of ‘good’ parenting behaviours than those associated with learning. It found that talking and listening to a child, responding to them warmly, instituting regular mealtimes and bedtimes, and authoritative discipline are all strongly correlated with a brighter future. (Harsh discipline, such as smacking, was associated with poor outcomes.) To some extent, the motivation to parent children well seems to be as important as the exact methods by which it is carried out. Inspiring them, reading to them, taking them out and spending time with them may ultimately be more productive than sinking energy into moving to the catchment area for a particular school.

The difficulty in all of these studies lies in filtering out the confounders. Parents who read to their children, take them to the library, put them to bed on time and do all the other ‘good’ parenting behaviours are also more likely to be of higher socioeconomic status – with more money, more education and better jobs. So does ‘good’ parenting really lead to better outcomes in children, or are they actually a result of having more money and better-educated parents at home? This is something that social scientists are still working to understand, and work from the millennium cohort has helped.

One recent study on 10,000 of the millennium cohort children found an association between regular bedtimes and good behaviour – one of the biggest studies to examine the impact of sleep schedules on young children. What made this study stand out was the evidence that the bedtimes were a cause of the behavioural problems, rather than just being associated with them. The scientists showed that children who experienced irregular bedtimes throughout their early years – when they were 3, 5 and 7 – were more likely to have problem behaviours than those who had irregular bedtimes at only one of those ages, even when they stripped away the possible confounders. And they showed that children who had once had irregular bedtimes, but who later in life switched to a regular sleep schedule, showed a significant improvement in behaviour – all supporting the idea that instituting consistent bedtimes could actually bring about an improvement in children’s lives. Scientists even had a good idea why: disrupted sleep is thought to upset the brain’s innate 24-hour clock, leaving children in a state much like perpetual jet lag, which interferes with the working of their brains and messes with their behaviour.

This is hardly rocket science. The sleep study made headlines not because it was surprising but because it threw the weight of science behind what is already parenting common sense. The same is true of many of the cohort findings, which offer evidence to support what people already know they should be doing – reading to their children, talking to them and so on – rather than rewriting the rules. Still, knowing this can strengthen parents’ resolve. One cohort scientist has found the birth cohorts to be invaluable when she is battling to get her headstrong teenager to eat breakfast or get to bed. In these types of arguments, it really helps to have the scientific literature on your side.

That said, all these findings about parenting come with the usual caveat from cohort studies, which is that there are no guarantees of success. Reading regularly to a child is associated with good educational outcomes when examined across a group; but this says little about what will happen if you read regularly to an individual child. It might do them some good, and it’s unlikely to do them any harm. What it does not do is guarantee them success in life.

From the parents’ point of view, all this risks fanning the everyday, smouldering worries about the quality of their parenting into a guilty blaze. As so often in science, there are no simple take-home messages and no guarantees. But the likelihood is that if you’re reading this, or thinking about how to do the best for your child, you’re already a long way towards being the engaged, interested parent that you want to be.

Right from the start, birth cohorts had shown that both poverty and parenting mattered. In the new millennium, inequality was still a huge topic of discussion, and politicians were still making efforts to reduce it, although the conversation had a lot less energy than it once had: there was no appetite for tackling inequality with the ambition present after the war, when the welfare state had been introduced. The actual policies to reduce inequality – such as offering free hours at nursery school – looked a little lacklustre by comparison. But, while poverty was still important, the attention being paid to the parenting part of the equation had ramped up.

All the work on the subject, from the cohorts and elsewhere, was thrust firmly into the limelight. There was some sense that the pendulum had swung too far in one direction, and that now parenting should be the priority. After all, it can be convenient to put the emphasis on parenting; that way, you can say disadvantaged children could do just as well as those of the well-to-do if their parents simply did a better job at home. According to this line of thinking, a lot of educational inequality would go away if only poor parents started to show more of an interest in their children, reading to them every day and getting them to bed on time. This idea received an airing in a speech by David Cameron in 2010, when he pointed to some of the recent research. The differences in outcome between a child born in poverty and a child born in wealth were no longer statistically significant when both had been raised by confident and able parents, he said. The discovery that parenting could compensate for poverty was the new law of social mobility. Parenting, Cameron asserted, mattered more than poverty.

When they heard this, some cohort scientists were aghast. As usual, the scientific picture was far more nuanced than it might seem. In fact, some scientists had come to question a number of old studies suggesting that having interested parents mattered quite so much. When they sat down and thought a bit more carefully about the way in which ‘parental interest’ had been measured in the past cohort studies – when teachers were asked to rate how interested parents were in their child’s education – they saw a problem: teachers weren’t always in a good position to judge. In many cases, parents weren’t encouraged to go past the school gate and they typically met the teachers during the single annual parents’ evening at school. This meant that the teachers barely knew the parents, and they were very likely to be judging their interest based on what they thought of the child, including his or her behaviour, appearance, social class and performance at school. So some of these early measurements of parental interest were far more subjective and unreliable than most scientists had realized. This didn’t negate the idea that parents were important in their children’s performance at school, but it did say that the early studies should be viewed with caution, and that interested parents might not be as important as some studies had made out.

And anyway it could hardly be said that today’s parents were not interested in, and ambitious for, their children’s future. When researchers surveyed parents of the millennium cohort when their children were seven, they found that almost all of them had high aspirations – 97% wanted their child to attend university, for example. This meant that parental ambition on its own was a useless indicator of future success, because clearly not all of those children would do well enough at school to go on to attend university. This statistic alone defied the idea that difficulties faced by disadvantaged children could be overcome if only their parents were more engaged and ambitious for them; according to this data, their parents already were.

One study on the millennium cohort children has shown quite clearly that both poverty and parenting matter a lot. It examined the effect of poverty and the quality of parenting on the children’s achievement in their first year of school, when they were aged between four and five. Children living in persistent poverty were falling behind those who were better off from an early age, but good parenting appeared to offset the disadvantage to some extent. (The scientists rated parenting based on a series of measures that included how often parents read to a child, did home learning activities, took them to the library, attended parents’ evenings, as well as whether they interacted warmly with them, breastfed them, instituted regular bed times and meal times, and fed them a healthy diet.) But good parenting reduced the gap by only about 50%; it didn’t close it. The scientists argued that focusing on good parenting without efforts to tackle poverty would therefore never close the gap between rich and poor.

Parents on their own cannot provide all the support that children require: they need the right landscape around them to help. Many experts in education want to see major investment in initiatives to support children in their early years, including access to affordable, full-time childcare for disadvantaged children; good-quality, widely available preschool education; parenting courses; and a concurrent effort to establish better evidence on what type of interventions to support children really work. The one thing that is clear is that you can’t just tell parents to be better parents, because most of them are already trying as hard as they can.

Poverty and how to escape it is a very old problem indeed for Britain. Before too long, however, the millennium cohort was pointing up a very new problem besetting the country: children were growing obese.

In the early 2000s, when the scientists were holding a powwow to plan what to ask of the millennium children at age three, one of those around the table was a paediatric doctor and researcher called Carol Dezateux, who was about to be sucked into the cohort world as if by quicksand. At that meeting she put up her hand and said, ‘My goodness, you’ve got to measure these children’s height and weight.’

The issue at the forefront of Dezateux’s mind was, of course, obesity. The earlier cohorts had shown how quickly British adults were gaining weight and by now it was clear that children were too. An epidemic of childhood obesity was well under way, and the millennium cohort offered an ideal opportunity to investigate the causes by tracking which children gained weight. Doing so turned out to be Dezateux’s first lesson in the challenges of collecting human data by the thousand. Even something as simple as measuring height and weight is very difficult when it has to be done very precisely for several thousand children scattered around Britain. It involved buying and calibrating hundreds of sets of scales and height gauges, briefing the field workers who would take the measurements and then working out whether the workers would break any health and safety regulations – which are famously overzealous in the UK – if they had to lug the heavy kit up the stairs to a council flat.

The effort was worth it, however, for it produced alarming results: 23% of children were either overweight or obese by the age of just three, and a similar percentage by the age of five. The work opened many people’s eyes to a looming medical crisis, because overweight children are more likely to grow up into overweight adults, with all the associated high risks of chronic diseases such as heart disease and diabetes. So this generation, aside from being just as divided by inequality as the ones before, was also a whole lot fatter. Researchers are still desperately trying to unpick the tangle of factors associated with a higher risk of being overweight as a child, which include lower levels of parental education, introducing solid food to babies very early in life, and children watching more than three hours of television a day.

The cohort scientists also found an association between childhood obesity and having a mother who works. They are still trying to figure out why this should be the case. It could be that children of families with working mothers rely more on unhealthy, pre-prepared foods, or have less time to get out and be active, or it may be that the association is spurious and can be explained by some unknown confounder. What is clear is that the study hit a raw nerve, because it implied that working mothers were neglecting their children and failing to feed them properly. The results of the study were so deeply unpopular that when the BBC website ran a news story on it, the page received around 1,000 comments within three days – and many of those were vitriolic rants.

A few years later Dezateux and her collaborators filled out the picture on childhood obesity when they posted accelerometers – little gadgets that measure how much a person walks and runs – to seven-year-olds in the millennium cohort and asked them to wear the devices around their waists for a week so they could record how active they are. The children seemed to love it, and nearly 10,000 of them wore the accelerometers and posted them back. The scientists then spent about eighteen months trying to make sense of the data, which was far more taxing than it might seem. For one thing, they had to work out whether some children had really been sitting very still or whether they had taken the device off. One child seemed to have a very peculiar pattern of activity, and when a researcher went out to visit the family the child confessed that he had strapped the accelerometer to the family dog. The results were published in August 2013, and the study made headlines by showing that only 51% of children overall – 38% of girls and 63% of boys – took part in an hour of physical activity each day, as the current guidelines recommend. For once, the social-class gradient favours those at the bottom of the socioeconomic scale, because children of mothers who were not employed were slightly more active than those of mothers who worked. This may be because better-off parents are more likely to drive their kids around and put computers in their rooms, which means they are sitting around for more of the day.

As the years have passed, the rushed start to the millennium cohort has been forgotten and forgiven in the torrent of invaluable research that the study has produced. It may be the youngest of the cohort studies, but it is already prized by scientists around the world. And some of those hasty decisions about its design have paid off handsomely – such as the one to collect the births over the course of a year, rather than over one week. Several studies have examined how the month in which children are born has an impact on their progression through school. This is a big problem: a body of work has shown that English children born in the summer months are most likely to fare poorly on academic tests and to suffer other problems at school, in comparison with those born at other times of year. This is because schools in England use the end of August as the cut-off date for dividing one school year from another, so children born in June, July and August become the youngest in the school year while those born in September, October and November become the oldest. (A similar ‘month of birth effect’ can be seen for the youngest children in a school year in other countries, even when the cut-off dates are different.) The advantages of being September-born are so well known that sometimes, when a child is born in the dying hours of 31 August, midwives will do parents a favour and record a baby’s birth as being in the early hours of 1 September.

The millennium cohort study came up with explanations as to why summer-born children developed such disadvantages. It showed that because children born in the summer months often perform worse at school, they are more likely to be placed in low-ability groups by the age of just seven, which may then entrench the differences between summer- and autumn-borns. The older children gain confidence and ability in the high-ability group, while the younger children may lose confidence and have less opportunity in the lower one, which deepens and perpetuates the differences between them. Studies on the millennium cohort have also shown that summer-born children show social and emotional differences, as well as educational ones, to autumn-borns. They are more likely to doubt their own ability, to dislike school and to be bullied. Education experts are still debating what to do about the month-of-birth effect: shifting the cut-off date would simply shift the problem on to another group of children. The recommendation from cohort researchers is to adjust test scores based on the child’s age, something that policy-makers have shown no sign of doing as of yet. In general, many educational differences narrow a lot by the time children have reached their late teens, although they don’t seem to disappear altogether.

Collecting data on births over the course of a year also proved very useful in demonstrating the benefits of breastfeeding. The roots of this study can be traced all the way back to the first maternity survey in 1946, when James Douglas found that women were more likely to start and continue breastfeeding when they had good antenatal support from doctors, midwives and health visitors. This is widely accepted today: the entire NHS is committed to encouraging women to breastfeed, and many hospitals employ a midwife to patrol the postnatal wards in order to help women establish breastfeeding in the first day or two after the birth – the critical window for the practice. But when an economist called Emla Fitzsimons spent a Saturday night having her baby at a London hospital, she noticed that the service was patchy: there was very little support for breastfeeding at the weekend, whereas on Monday there was plenty. The NHS, under permanent pressure from the government to cut costs, pares down to only the most essential staff at the weekends and Fitzsimons wondered if breastfeeding support was one of those things that are cut.

Breastfeeding is associated with many measures of improved health, development and cognition in children – but mothers who breastfeed are also more likely to be of higher socioeconomic class, so the challenge lies in working out whether breastfeeding really causes better outcomes when class is discounted. Fitzsimons and her husband, an economist called Marco Vera-Hernandez, realized that the staffing patterns of British hospitals provided a natural experiment that would allow them to study the causative associations with breastfeeding. Women who had their babies during the week were more likely to have good postnatal support and would be more likely to continue breastfeeding than those who happened to have their babies at the weekend, regardless of class. All they had to do was to compare the two groups. The previous national cohorts were no good for this, because they spanned one week and so some chance event in that week could have muddied the waters. They needed to examine births over many weeks in the year, which is exactly what the millennium cohort had done.

The two economists divided up the millennium cohort children into those who had been born on a weekday (good support, more likely to have been breastfed) and those who had been born on a weekend (less support, less likely to have been breastfed) and then looked at various measures of health and development as the children grew up. They found that breastfeeding was causally associated with better cognition – but the associations with other measures of good health did not hold up. (This mirrors the findings of George Davey Smith, in Chapter 5, who identified causal associations with breastfeeding by comparing cohorts from different countries, with different class structures.) Fitzsimons and Vera-Hernandez’s clever study also exposed the massive, unexpected repercussions of money-saving efforts in healthcare: the simple decision to cut back support staff at the weekends leads to less breastfeeding of weekend-born babies, which potentially affects brain development for the rest of these children’s lives.

While the fifth cohort was growing up, something quiet but dramatic had been happening at the helm of the British birth cohort studies, something that is best illustrated if we think back to the very start of the story, when the Population Investigation Committee came together to launch the 1946 maternity survey. At that time almost all the doctors and academics around the table were men, and the two key scientists who got the cohort going – the doctor James Douglas and the social scientist David Glass – were men. Since then, the cohorts had passed through the hands of a string of men: Neville Butler, John Bynner and Mike Wadsworth. There had been two notable exceptions in Mia Kellmer Pringle – who complained that being a woman held her back – and Jean Golding, who hardly let anything hold her back. But, nevertheless, the cohorts and academia had largely been a man’s game. Through the 2000s, however, there was a sea-change: the cohorts had moved almost exclusively into the hands of women.^ff

The exception was the 1991 cohort, which moved from the hands of a woman – Jean Golding – into the hands of a man, George Davey Smith.

The cohort studies had produced a ream of influential research showing the steady rise of women in the workplace since the 1970s, and now this came to be mirrored in the leadership of the studies themselves. In 2004 Jane Elliott took charge of the 1958 and 1970 cohorts. Elliott was an unwitting beneficiary of John Bynner’s decision in the 1980s to draw new scientists into the cohorts by creating sample data sets for them to play with. Bynner had no idea that this decision would actually hook his own successor, but it did. In 1986 Elliott was a student at Cambridge, where she had switched from studying mathematics to social sciences, and she was given one of those data sets as part of her course. At that time, the only way to use computers was to walk to the university’s central computer room, which was filled with men and the odour of old socks. Elliott didn’t mind. She was captivated by this generation of people who had been born only eight years before her and yet seemed so grown up. The data left such an impression on her that, nearly thirty years later, she can still recite some of the codes by heart.

As she entered academia, Elliott went on to investigate divorce; she was one of the researchers who produced the surprise finding that the impacts of divorce could be seen on children’s cognition and behaviour before their parents actually divorced (as mentioned in Chapter 2). After Elliott took the reins of the 1958 and 1970 cohorts, she soon sensibly realized that two cohorts were too much for one person to handle. She kept the 1958 cohort, and brought in Alice Sullivan, a researcher in sociology and education, to lead the 1970 one. Sullivan also mirrors some features of the cohort that she directs: she is roughly the same age – born four years after them – and she has no desire to have children, just like many women in her generation.

The newest additions to the team are Alissa Goodman, an economist who in 2012 assumed leadership of the 1958 cohort, and Emla Fitzsimons, who took over the millennium cohort the following year. Goodman was involved in cohort work that demonstrated how a higher education could pay off richly in terms of securing more lucrative employment: women who had a university degree earned nearly 40% more per hour than women who had A-levels alone by the time they were thirty-three.^gg As for Fitzsimons, her studies on breastfeeding convinced her that the millennium study was strong. The women work in a maze of ageing offices on London’s Gordon Square, where the 1958, 1970 and millennium cohorts are based. The birth cohorts – these unique studies built up over decades of work – are run out of humdrum offices with worn carpets that look and smell like so much of British academia, chronically deprived of recognition and cash.

Cohort work showed so convincingly that people with a university degree earn more over their lifetime that it helped to convince the government to introduce tuition fees for university in 1998.

Sullivan, Elliott, Goodman and Fitzsimons are all extraordinarily clever, articulate, talented, efficient scientists who love their work – and yet they couldn’t be more different from Douglas and Butler, the men who dominated the early life of the cohorts and kept them going with wartime resolve, charm, self-assurance and a big dose of luck. But those characteristics won’t keep a cohort study going these days. What the studies need now, if they are to stay alive, is people who can fight for them in the modern world: people who can not only lead the science, but also write convincing reports for the government that list all the ways in which they have influenced policy; and who can send in dense grant applications, on time, to make a financial case to funders that they should support the next sweep. That is one of the many things that these women do exceptionally well, and that is why they are absolutely essential for the cohorts – without them, the entire scientific legacy would be lost. Just as vital is the team that works behind the scenes: the data experts, survey managers, communications staff and huge fieldwork agencies that ensure the studies run with the tight professionalism necessary in the research business today.

There are other ways in which the cohort leaders now are different from those of the past. They don’t necessarily see the job of running a cohort as a job for life. This has produced some turnover at the top. When Joshi relinquished leadership of the millennium cohort in 2011,^hh her successor Lucinda Platt stayed in the job for less than three years before leaving to become a professor elsewhere. In 2014 Jane Elliott quit the cohorts to become head of the ESRC, so that she is now directing the organization that funds most of the country’s social science research, including the cohorts. This turnover is good and bad: good, because it means that scientists steeped in the cohort studies are rising upwards, so the cohorts have support at the highest levels of British science; bad, because some cohort leaders have a shorter, more distant relationship to their studies than the early leaders did, and because some of the rich history of the studies is inevitably forgotten and lost when new scientists come in.

Like many other cohort scientists passionate about their subject, Joshi didn’t actually stop work when she retired: she continued her research and went on spreading the word about cohort studies around the world.

But the struggle to secure the next round of funding for the cohort studies goes on and on. No sooner have the scientists secured one short-term tranche of money than they have to start pleading for the next. This is time-consuming, exhausting and emotionally draining for them, and it means that they can never make plans with the surety that the studies have a future. It has to be said that running a birth cohort today sounds like a lot less fun than it did back at the start, when the scientists were tallying every baby born in a single week, watching the girls play netball and desperately trying to raise money by spilling coffee on to Mrs Thatcher. Today a cadre of scientists sit in offices with computers, devising smart hypotheses and striving to get grant applications in on time.

On the other hand, the cohort scientists no longer have to work quite so hard to sell the idea of birth cohorts. The entire world seems to have noticed that they are a splendid thing to do.

The appeal of tracking human lives has led scientists to start other cohorts of different shapes and sizes in all corners of the world. The British scientists helped to set up some of them. In the 1970s Butler helped to start a study of births and perinatal deaths in Cuba that went on to become a cohort study. (This led to one memorable linguistic misunderstanding when Butler drove into a village to find a big banner strung across the road reading WELCOME, AND LONG LIVE PERINATAL MORTALITY.) In the 1980s Golding started a perinatal survey in Jamaica encompassing over 10,000 births, which eventually helped bring down the number of deaths from pre-eclampsia. South Africa has been running a study since 1990 called Mandela’s Children, because the 3,273 babies enrolled in it arrived in the year that Nelson Mandela was released. There are many more like these.

What happened as the millennium approached, however, was that everyone’s cohort ambitions seemed to grow. Suddenly massive birth cohort studies were the thing to have. It was as if everyone wanted one.

There were two major reasons for this. The first was that scientists had developed a wider appreciation that tracking people’s lives from cradle to grave was a genuinely valuable pursuit, rather than just an odd thing that the British did. A lot of the impetus came from medical research and epidemiology – and particularly from the rapid spread of David Barker’s ideas showing that chronic diseases have their origins in pregnancy and the first few years of life. If pregnancy and childhood are crucial to our lifelong health, this meant that it was important to study people from their conception and on through their lives.

Second, technology had made it possible to carry out cohort studies more easily than ever before. James Douglas had only followed 5,362 children in his 1946 study, because that was all that one man and a tabulating machine could handle. Those constraints are now laughable. A laptop computer can crunch statistics on tens of thousands of children within minutes; a robot can process thousands of biological samples; and advanced statistical programmes have made it possible to eliminate confounding factors from an analysis while the scientists pop out for a bite to eat. Dealing with data from cohort studies is no longer the laborious, cumbersome task it once was. What’s more, scientists had developed a taste for big, splashy projects in science. The Human Genome Project paved the way by showing that, with enough money and chutzpah, it was possible to pull off an ambitious, mega-biology project. This heralded a wave of projects – sometimes called Big Science – involving lots of people, money and big ideas. Birth cohort studies were certainly big, bold and expensive, and now they seemed like a great thing to have. So, as Britain was gearing up to launch the millennium study, scientists around the world began to establish birth cohorts that were far bigger than any that had gone before.

In 1999 scientists in Norway started recruiting women into what has become the largest birth cohort study so far, with over 110,000 children. Denmark started to track about 100,000 of its children a few years before that. Giant cohorts like this are easier to pull off in countries with pre-existing, national databases of personal, health and educational data, because researchers can simply tap into the databases and sweep up swathes of information from them on the cheap, as long as they get their permission from the participants first. The Norwegian cohort, for example, used national identification numbers to link up its study members with the country’s extensive databases including birth and death registries, a cancer registry, a prescription database and vaccination registry.

The cohort craze spread wider than this. France recruited over 18,000 children from maternity hospitals in 2011, and scientists in the Netherlands have enrolled what is now nearly 10,000 children in a birth cohort called Generation R in the city of Rotterdam. When scientists in China recorded the births of over 13,000 children in Anhui province, starting in 2008, they emerged with the staggering statistics that 67% of children were born by caesarean section and that 115 males were born for every 100 females. This lopsided ratio is a product of China’s one-child policy and its preference for sons; once ultrasound scans could be used to identify the sex of children before birth, abortion of female foetuses became widespread. Closer to home, in 2007 researchers started a project to follow a few hundred children born in the city of Bradford, and some years later had recruited 13,500 of them instead.ⁱⁱ There are other birth cohorts under way in Canada, Japan, Australia, New Zealand, Germany, Finland, India, Guatemala, the Philippines and Brazil.^jj One inventory of birth cohorts listed eighty-seven at the last count; if you wrap in cohort studies that recruited children after birth, the number shoots up.

This effort, called Born in Bradford, has been run so cheaply that its director calls it the ‘Poundland of birth cohorts’. It is designed to understand the high rates of childhood death and disability in a multi-ethnic population that includes some of the most deprived regions in the UK.

The city of Pelotas, in Brazil, has a series of birth cohorts separated by eleven years, starting in 1982, 1993 and 2004. (A fourth one was getting underway in 2015.) This is by far the most impressive series of birth cohorts outside Britain. George Davey Smith used one of these cohorts to find the causal associations with breastfeeding.

Scientists found themselves in the middle of a birth cohort boom. They began gathering at an international meeting on longitudinal studies – and having a dedicated conference is a sure sign that a scientific field has come of age. There, they swapped lessons and results. They also sighed over shared enemies: the drop-out of participants, which can kill a cohort if it’s not kept in check; the dread that a breach of confidentiality could expose the identities of the cohort members and jeopardize their trust; and the eternal struggle to keep politicians interested enough to continue stumping up the cash.

This was all excellent news for the British birth cohorts, which, after years of being side-lined or ignored, had become the subject of great interest abroad. The 1946 cohort was, by this stage, the longest continuously running birth cohort in the world – and the whole British series was considered a pioneer by scientists in the field. It wasn’t unusual to hear scientists elsewhere refer to them in awestruck terms. ‘They’re kind of one of the wonders of the world, you know?’ said one. ‘I don’t think you can beat the British cohorts.’

This flurry of birth cohorts might create the impression that starting one has become a completely trivial exercise, when nothing could be further from the truth. If you want an example of how difficult it is to start a major birth cohort, the United States is one place to look. There, researchers and policy-makers spent over a decade and more than $1 billion mired in arguments about how to track 100,000 children from birth to twenty-one, and eventually gave up. This effort, called the National Children’s Study, had the potential to be one of the biggest and best. But it all went horribly wrong.

The concept of a US birth cohort emerged around the same time as the British millennium cohort, and when many other major studies were taking shape. In 2000 Congress asked the National Institutes of Health to launch a national longitudinal study that would examine the health and development of children. On paper this seemed do-able. The Americans could draw on the fifty-plus years of experience that Britain and other countries had amassed in starting birth cohort studies, and they could see the enormous difficulties that come if you try to start a cohort with sprawling, unfocused hypotheses and overly ambitious plans for recruiting mothers. But few people in the US were taking much notice of what a bunch of quirky British scientists had been doing with its studies of children since the end of the war. And anyway, cohort studies, just like babies, tend to encounter their own unique troubles at the time of birth. So, whatever mistakes the British cohort scientists had made over the decades, the US scientists proceeded to make in a more spectacular fashion and with a few extra twists of their own.

The Americans wanted a cohort that was as big and bold as the best of them, and so they embarked on a colossal planning and consultation exercise spanning several years and involving, by one estimate, some 2,500 experts. They wanted a sample of 100,000 children chosen at random from across the country so that any conclusions could be generalized to all American children. This made sense – but the process they came up with for finding potential mothers was extraordinarily labour intensive. In the UK, the NHS provides care for almost all pregnant women, and so offers a centralized means for recruiting them, following them and collecting their medical information. But there is no such system in the United States, where healthcare is provided by a patchwork of privately financed providers. So the scientists had to devise a different plan. They decided to select 105 counties scattered across the US, choose random blocks of houses there, and then send out interviewers to walk along and knock on doors until they found women who were pregnant or planning to get pregnant. This was complicated, and was made more so by the scientists’ aim to recruit some women before they even conceived – which meant they would have to follow around subjects for many months, as they had no idea who would conceive or when. And it seems highly unlikely that a federal government official standing on the doorstep and asking if a woman was thinking about getting pregnant would be warmly invited inside.

The cohort study also became completely bogged down by the huge number of scientists involved, who all had their own interests and wanted to see their questions crammed into the questionnaires. It became overloaded with hypotheses and plans to collect an immense amount of data. The aim was to carry out detailed interviews with all 100,000 sets of prospective parents and to collect environmental samples – such as drinking water, air, dust and soil – as well as a full set of biological samples from the very start of pregnancy. Many of the scientists involved were openly voicing doubts that it could work.

Still, the study kicked off, and started recruiting pregnant women into pilot studies in 2009, by which point the estimated costs had doubled to some $6.9 billion – and the effort was already consuming over $190 million each year. (These types of sums make the British cohort scientists weep; for comparison, the sum assigned for the last sweep of the millennium cohort was £3.5 million, or about $5.5 million.) But soon the plans started to unravel, and then they came undone. The pilot studies confirmed that the doorstep strategy wouldn’t work, the study had yet to find a clear scientific direction, and politicians were baulking at the spiralling costs.

In December 2014, with several thousand children already enrolled and over $1.3 billion spent, word came that the study was to be scrapped. Mothers and their children who had been drafted into the pilot study were left in limbo, and felt betrayed by the scientists who had talked them into taking part. The scientists involved, some of whom had been working on the study for nearly fifteen years, were left angry and shocked. Many researchers feel that the effort has been a costly train wreck and a national embarrassment. Over the time it took for the US to plan, pilot and then cancel its birth cohort, the Norwegians and Danes both reached their target of recruiting 100,000 children into their studies, and the British scientists visited their smaller millennium cohort five times. Back in Britain, no one in the birth cohort community took any pleasure from the fiasco across the water, it just made them realize how fortunate and thrifty they had been. Britain is more used to playing David to the United States’ Goliath when it comes to scientific research – but in this case it was leading the way, running a string of birth cohorts for just a fraction of the price of the projected American one. Epidemiologist Michael Marmot compares the British economical style of doing things to a sharp shooter with a pistol, and the US approach to a fleet of B52 carpet bombers. Theirs was supposed to be ‘the biggest, bestest thing ever’, he says. ‘Well, the British did it differently and carried it off.’

The failure of the National Children’s Study gave some scientists pause for thought. They started to wonder where the birth cohort boom would end. Not every country needs a giant 100,000-child cohort. Even the most ardent cohort fans questioned how much of the cohort boom was now being driven by science, and how much by a sense of national pride. There was a sense that some countries wanted a birth cohort just because that was the done thing.

And even if the US scientists did get a cohort going,^kk they would still be seventy years behind. Because in the birth cohort business, you can’t play catch-up, no matter how much money you have to throw around. It’s true that the British birth cohorts have had a troubled history, routinely starved of cash and frequently on the verge of collapse. But what the British scientists have done is to push on, get the job done and keep collecting data whenever they could.

This could yet happen, because although the National Children’s Study was cancelled, the idea of a major longitudinal study of children’s health is still alive. The most recent plan is to do this by knitting together existing cohort studies rather than start a brand new one.

The reason that the British cohorts are still here is because the scientists did what the British have always done: kept calm and carried on.

At the start of the twenty-first century, birth cohorts turned full circle. As new ones were springing up around the world, the children in the first and original study – the ones born in March 1946 – were reaching old age. In 2011 they would turn sixty-five, the age at which many people in the UK retire and, as such, a milestone in British life. This raised a question: should the scientists throw a birthday party to celebrate the study and its members having got this far? As the birthday approached, this was being turned over in the mind of Diana Kuh, the economist who had developed the idea of life course epidemiology while working with the cohort. In 2007, when Mike Wadsworth retired, Kuh took charge of the 1946 study. It was a heavy mantle of responsibility. This cohort was the grandfather of them all, and scientists around the world were in awe that it had survived this long.

However, none of this stopped the MRC, which was still paying for the study, from threatening to close it down when Wadsworth retired. The research council is under permanent pressure to control its costs, and so it took the opportunity of a change of leadership at the cohort to consider whether the study should go on at all. But, once again, chance, politics and grim determination helped to keep it on the road. At just the time when the MRC was considering the cohort’s future, a House of Lords report called attention to the country’s ageing population, and asked the research councils why on earth they weren’t doing more to investigate the situation. Suddenly, a cohort study of Brits in their sixties sounded like a terribly worthy pursuit.

To Kuh, it was inconceivable that the cohort would not go on. One of Kuh’s colleagues describes her as a mother tiger for her protective yet aggressive care of the cohort, and if you can superimpose sensible bobbed hair and spectacles on that idea, you have a rather good description of her. Kuh cares deeply for the cohort, and she’ll claw out the eyes of anyone who threatens it.

Kuh set about taking the cohort on to its next stage, the one that would study the cohort members as they grew old. Until that time, nurses had still been travelling regularly to the homes of the cohort members to carry out medical examinations, but by this time they were staggering under the weight of all the equipment. If the study was to keep up with the rapid developments in medical research, Kuh argued, she needed to get people to a clinic where scientists could carry out more thorough examinations. By 2008 she had convinced the MRC to pay for every willing cohort member to spend half a day in one of a number of clinics around the country and complete their most intensive interrogation to date, including ultrasound scans to examine their hearts and body scans to look for weakening bones and thickening fat. (They were also put through a battery of low-tech tests to measure how well their bodies and brains were holding up. Please can you balance on one leg with your eyes closed? How many words can you remember from this list?) The study members loved the attention and like to call it their MOT. The MRC also reinstated the cohort as a dedicated research unit, just as in Douglas’s time, and it is now housed in a Georgian terrace on Bedford Place, sandwiched in between upmarket London hotels. Here, Kuh goes up and down carpeted steps all day to and from her office on the second floor.

As the sixty-fifth birthday approached, however, the prospect of a party was causing Kuh great concern. None of the national birth cohorts had held a birthday party since 1989 – when Neville Butler threw his wild bash for the third cohort at Alton Towers – because scientists now tend to think that they should be passive observers of their cohort. If they were to bring them together at a party, they risked influencing the participants’ life course in some way, which is a bit like tampering with your own experiment. (The main concern was that someone might get drunk and get off with someone else in the study, so that the party inadvertently led to a marriage breaking up.) And even if Kuh did throw a party, how could she accommodate all the thousands of cohort members across the country if everybody wanted to come? Or, worse, what if nobody came?

In the end, Kuh decided that the benefits outweighed the risks, because the party would recognize and reward the cohort members for sticking with the study for so many years. She solved the inclusivity problem by planning four parties across the UK. (She also wrote a letter to Buckingham Palace to request an invitation for the study members to one of the annual royal garden parties. A few months later the palace wrote back: would fifteen cohort members like to come? The chosen ones were over the moon.) Meanwhile, the other parties were getting oversubscribed. The cohort members were curious to meet others like them, even if they had a few reservations about signing up. They worried that a guest list made up of scientists and pensioners would make for a really dull evening, and that the MRC would be stingy when it came to paying for the drinks. ‘I thought one glass and that would be it,’ Philip Cheetham said. The first party was set for 3 March 2011, in a function room of London’s British Library, sixty-five years to the day after the first Douglas Babies were born. All the cohort members would turn sixty-five that week.

In the end, all the worries were for nothing: the evening was a triumph and the cohort members were abuzz. They wished each other a happy birthday and talked about rationing and the terrible diets of their grandchildren. Kuh, who was too busy to be nervous, made a heartfelt speech. There was plenty of wine, and it didn’t look like any cohort members were getting off with each other either, although you never really know. There were glowing articles in the newspapers about this extraordinary treasure of British science, and the remarkable scientists who had kept it alive. It was a high point in the life of the study. And, because it was a party full of 65-year-olds and scientists, everyone arrived on time – and then left early, to make sure they caught the last train.

About a year later, the buzz of the party had faded and Kuh was close to getting ill trying to keep the cohort going for another few years. The MRC subjects all its research units to a rigorous performance assessment, called the quinquennial review, every five years. It is a deadly serious business: if scientists are not up to par, they can be kicked out and their funding can be slashed. Douglas and Wadsworth, her predecessors, had sweated blood every time they had to get through one. But the 2012 review would be the first for which Kuh was solely responsible, and she did not want to go down in history as the one cohort director who hadn’t managed to get the study funded. The pressure was enormous; she had the 65-year legacy of the study resting on her shoulders.

In order to pass the review, Kuh had to write a telephone directory of a document summarizing the cohort’s achievements – which by that point added up to eight books and 700 publications – as well as her future plans. She rented a flat near the research unit in London partly so that she could focus on the review. She would get up at 5 or 6 a.m. thinking about the cohort, walk the short distance to the office, work all day on the review, and be the one locking up the building at night. She would go back to the flat, collapse in bed and get up to start all over again. She sent in the document in June, responded to forty-five pages of comments on it, and then waited for a crucial two days in November, when a panel of reviewers would visit the unit and decide its future. They were scheduled to meet in the boardroom, downstairs at the unit in Bedford Place.

On the morning, Kuh was exhausted and yet wired on adrenalin at the same time. In the boardroom, the scientists each gave a short talk and then the review panellists talked privately among themselves, before calling her back in and giving her the verdict – a short report with scores for the unit out of 10. She got a string of 9s, which is as close to perfection as anyone in science is going to get. The reviewers recommended that the unit should receive nearly £9 million to take it through until 2018, when the cohort would be turning seventy-two. The study had won the latest reprieve in its long life. The scientists celebrated – they popped champagne and went to the pub. Then, the next day, they were in early to work again. They have so much to do in such a short time. Because, now that she has money, Kuh has the sad scientific task of carefully watching the cohort members grow old, fall ill and die.

The truth is that if funders don’t kill this cohort, time, inevitably, will. Some 13% of the subjects had died by that point – and the study has plenty to say about the fate of the rest. Now that the scientists have traced the cohort for so long, it is possible to draw associations between some of the earliest events in people’s lives and their state of health today, as if they were sketching lines through decades of time. In one of their recent papers, Kuh’s team looked for correlations between the cohort members’ social class as children and how well they performed on those medical MOTs in their sixties. It seems amazing to think that the social class of your parents at birth could, over six decades later, have any bearing on how long you can balance on one foot with your eyes closed, or how well you can remember a list of words. But it does. The cohort members who had been born into lower socioeconomic positions tended to perform the worst on every one of the tests.

This fulfilled a prediction made by James Douglas, when the cohort members were still in their twenties. He had said that the differences between those in the upper and lower classes would become more prominent with the passage of time. The idea was that those who were already scarred by disadvantage as children would more rapidly succumb to disease and the deterioration that arrived with old age, while those who’d had a more comfortable life would enjoy a healthy trajectory for more years. No one would argue with that now, when study after study from the cohort has shown that those born into the lower classes have been more likely to develop almost every problem in the medical book.

Still, the beauty of scientific inquiry is that it often kicks up a curious stone, and when it comes to death, the correlations haven’t emerged as neatly as you might expect. A few years ago Kuh put together graphs showing the proportion of the cohort members surviving up to age sixty, categorized according to their social class at birth. Disadvantage never stops showing its hand: those study members from the poorest backgrounds were 60% more likely to have died than those from the better-off ones. But the picture was not so simple when the survival rates were divided not just by class but by sex. When the figures were diced this way, it became clear that middle- and upper-class men were dying at roughly the same rate as lower-class men and women – whereas women from better-off backgrounds had a death rate about half that of everyone else. The scientists have not yet explained why this is – it doesn’t seem to be because of anything obvious such as these women smoking less than the other groups. Kuh suspects that these middle- and upper-class women were better able to take advantage of the opportunities afforded by post-war Britain to improve themselves: attending grammar school and securing good healthcare on the NHS. The opportunities given to them by Beveridge’s reforms allowed them to get educated, get a job and stay healthy, which have all helped them stay alive longer than others in the study. Kuh has a personal interest in the graph, because she is only six years younger than the cohort and she knows that she is a middle-class woman who benefited from grammar school and the welfare state – so the graph tells her something about her own chances of survival too.

A few years after the death graphs, the scientists published another simple but powerful analysis of the cohort’s decline. They took all the detailed MOTs that Kuh had collected and counted up how many medical disorders each person had out of a list of fifteen, including cardiovascular disease, hypertension, raised cholesterol, diabetes, obesity, osteoporosis, psychiatric problems, cancers and respiratory disease. They found that a whopping 85% of the cohort had at least one of these conditions and that, on average, they had two disorders apiece – even though most of those people, when asked, said that they were in good health and a large number of the conditions had never been previously diagnosed. It was a sobering picture of ageing, tout ensemble, and one with important implications for Britain and for the rest of the world. The 1946 birth cohort contains the very first members of the baby boom, the surge in births that started after the end of the Second World War. Now the boomers are starting to enter retirement and so the population as a whole contains more and more elderly people. (Some 17% of people in the UK were aged sixty-five and over in 2010, a figure that is expected to rise to 23% by 2035.) A similar phenomenon is happening elsewhere, where growing life expectancy and falling fertility rates mean that the over-sixties are the fastest-growing proportion of the population in countries around the world. The cohort study shows that many of them will have some type of chronic disease. It is acting as a bellwether, an early indicator of the tsunami of illness that this rapidly ageing population is going to bring in its wake.

It could also be a best-case scenario, because the people in the cohort study were cushioned by the brand-new welfare state and brought up on healthy, if meagre, rations. Those coming along later, when obesity was arriving on the scene, could find themselves suffering more disease. Kuh also sees a way for the cohort study to help, and at the same time for it to return to its roots. At the start of its life, the study shaped maternity services in the fledgling NHS; now its observations of pensioners could guide the NHS as it prepares for the onslaught of age-associated disorders that is heading our way. Quite how the NHS is going to cope with the enormous expense of a sick and ageing population is the topic of regular talks and hand-wringing at very high levels of the health service, but no one seems to have come up with any clear solutions to the problem.

At present the cohort scientists are also busy working out how to continue tracking the cohort members as they age. They are particularly interested in carrying out a further MOT that focuses on their brains, because a big slice of the cohort is expected to develop dementia over time. The team has just started to bring about 500 of the cohort members into London to carry out state-of-the-art scans of their brains, looking for early signs of Alzheimer’s disease and other deterioration of the mind.

All that is more than enough to fill the next five years, but Kuh won’t take the cohort through its next big review, because she’ll be close to retirement age herself – and, much as she loves the cohort, she doesn’t think she has the energy to see it through another round. Like her cohort, Kuh has her two clinical conditions and she doesn’t want any more. She is thinking a lot about who will take over from her. Her goal is to hand on the baton, passing the cohort study safely into another scientist’s hands.

It’s hard to resist the morbid temptation to look far into the cohort’s future when, one day, someone will have a very hard decision to make. When, finally, should the cohort study be allowed to die? There are currently about 3,000 study members left – because of those who have died or dropped out – and with a little extrapolation of her survival curves, Kuh predicts that there will be about 1,400 left by the time they are eighty-four and just 300 around to celebrate their hundredth birthday in 2046. When the scientists strip away whatever emotional attachment they have to the cohort, they know that at some point it won’t make scientific sense to keep collecting the data any more. There will be too few surviving members for the study to have any statistical power and therefore to produce meaningful results.

Most cohort members don’t know the detailed plans that the scientists are making in order to watch them die. Even if they do think about it, they are completely unruffled about science accompanying them all the way to the grave. Most like the attention and feel that if their mothers signed them up for the study, it’s their duty to see it through to the last. ‘It’s something you said you would do, and you do it,’ says one cohort member stoically. And very occasionally, someone who had dropped out of the cohort gets back in touch. One long-lost member e-mailed the scientists in 2015. He’d been in Canada for the previous forty years, but now he was back in Britain and had seen the study in the news. Could he please join up again? Kuh, who treasures all her remaining cohort members, couldn’t have been more pleased.

Dropping out hasn’t even occurred to Patricia Palmer, who was born into the first birth cohort, in March 1946. Palmer is one of those who should, statistically speaking, have had a difficult life because of her tough start: born into a poor, working-class household, with a father who drank and left home when she was five. She failed her 11-plus and, although she wanted to train to be a teacher, knew her mother wouldn’t be able to afford it and left school at sixteen. It’s something that she still bitterly regrets.

Sixty-five years later, Palmer lives in a well-to-do suburb on the fringes of Cheltenham, just a mile or two from where she was born. Somewhere in her house is a collection of birthday cards that the cohort scientists have sent her almost every year since she was born, but she can’t quite remember which drawer it’s in. In the decades that her cohort has been going, four more British birth cohorts have been started and she has become just one of the more than 70,000 people that the scientists have now enrolled. But Palmer has been busy living her own life, which is the stuff from which the study is built.

In her case, it’s hard to find a neat explanation for the way life turned out apart from chance and a strong work ethic – something that she inherited from her own tireless mother. She flourished in secondary school and, after leaving at sixteen, learnt secretarial skills at GCHQ and went on to sell cosmetics in department stores. But she gave up work at twenty-one when she married a divorced man and found herself stepmother to his two young children.^ll (Her mother was absolutely horrified, she recalls.) Within a few years, Palmer bore a daughter of her own. Her husband earned a good wage and eventually Palmer went back to work, and ended up managing the finances of a large school. She moved to Luxembourg with him, and it was there that – like many of her cohort – she started to gain weight. She thinks it was the pâté, baguettes and all the hormonal changes that accompany the menopause that pushed her slowly up from 11.5 stone to 15.

Palmer changed her surname to Malvern when she married.

Now that she’s back in the UK, the weight has come off. (She knows that she has to keep going to Slimming World and make morning visits to use the cross-trainer at the gym.) It helps that she spends her afternoons picking up three of her grandchildren from school, which she does to support her daughter and son-in-law, who both have busy jobs. It’s exhausting, she says, but she loves seeing so much of the children too.

Although Palmer was born in difficult circumstances, she thinks that she could have chosen almost whatever she wanted to do because of the time in which she was born. Everyone found it tough to some extent after the war, but people were mostly happy with their lot. ‘I feel very lucky, actually, the way that my life has gone,’ she says. She thinks that it’s harder for children who are born into disadvantage today, because they are surrounded by toys and computers and they expect so much more. She also worries that the opportunities for her grandchildren aren’t as great. The cost of living is spiralling upwards, and the price of a university education could put it almost out of reach for them. It’s sad, she reflects, to see anyone denied an education because of money, just as she was over fifty years ago.

For most of Palmer’s life, the cohort study was just like a shadow that followed her around – always present, but something that was attached to her rather than the other way around. But that changed when she received the invitation to attend one of the cohort’s sixty-fifth birthday parties. She decided to go to the one in Birmingham, where she met Diana Kuh and the other scientists, and she even agreed to make a short speech. She told a little joke about the scientists following her until the end. ‘I wonder,’ she said, ‘when we reach the pearly gates, whether there’ll be somebody from the study standing there, you know, with their clipboard, waiting for our last comments.’

^a Although the cohort studies are often viewed from a distance as being either ‘biomedical’ or ‘social science’, all of them are involved in both spheres of research. One major example of crossover is the highly successful sweep, starting in 2002, that focused on the biology and health of 1958 cohort members in their forties and that has grown into a major scientific resource.

^b The British Household Panel Survey began collecting information on all the members of some 5,500 households across the UK in 1991. Like the birth cohort studies, the household study tracks people over time, but the important difference – and the reason that social and economic scientists like such studies – is that, while birth cohort studies follow individuals over time, the household study follows groups of people – all those living in the home – which provides more information on the changing make-up of the household and the relationships between people in it.

^c Neville Butler still maintained a presence: his loyal secretary drove him from Bristol to the office once a week. Butler continued to operate his charity to support the birth cohort studies even though it was making very little money and he himself was nudging eighty. His unwavering devotion to the studies continued until his death, in 2007.

^d The scientists achieved this by over-sampling in regions of the country where a high proportion of children live in families receiving means-tested benefits. They also over-represented ethnic minorities and children born in Northern Ireland, Scotland and Wales.

^e The social scientists did work with medical researchers when they were designing and launching the cohort – but the tight timetable severely limited what they could do. What’s more, the cohort was funded at the start by the ESRC, whose focus is on social science, not medical research.

^f The exception was the 1991 cohort, which moved from the hands of a woman – Jean Golding – into the hands of a man, George Davey Smith.

^g Cohort work showed so convincingly that people with a university degree earn more over their lifetime that it helped to convince the government to introduce tuition fees for university in 1998.

^h Like many other cohort scientists passionate about their subject, Joshi didn’t actually stop work when she retired: she continued her research and went on spreading the word about cohort studies around the world.

ⁱ This effort, called Born in Bradford, has been run so cheaply that its director calls it the ‘Poundland of birth cohorts’. It is designed to understand the high rates of childhood death and disability in a multi-ethnic population that includes some of the most deprived regions in the UK.

^j The city of Pelotas, in Brazil, has a series of birth cohorts separated by eleven years, starting in 1982, 1993 and 2004. (A fourth one was getting underway in 2015.) This is by far the most impressive series of birth cohorts outside Britain. George Davey Smith used one of these cohorts to find the causal associations with breastfeeding.

^k This could yet happen, because although the National Children’s Study was cancelled, the idea of a major longitudinal study of children’s health is still alive. The most recent plan is to do this by knitting together existing cohort studies rather than start a brand new one.

^l Palmer changed her surname to Malvern when she married.