Cohort Profile: The 1946 National Birth Cohort (MRC National Survey of Health and Development)

How did the study come about?

The first of the British National Birth Cohort studies aimed to address two health and social policy questions of importance in the years before the establishment of the National Health Service (NHS) in 1948. The first question, identified by the Royal Commission on Population, asked why the national fertility rate had been falling consistently since the middle of the 19th century. A proposed explanation was that the ‘medical and other costs associated with the birth of a baby may today be a serious deterrent to parenthood.’¹ The second question asked ‘what was the national distribution and use of obstetric medical and midwifery services, and how far do they prevent premature and infant death, and promote the health of mothers and infants?’¹

The study set up to address these questions was promoted by the Royal College of Obstetricians and the Population Investigation Committee, and funded by the Nuffield Foundation and the National Birthday Trust Fund for a data collection from all who delivered births that took place during March 3–9, 1946 throughout England, Wales, and Scotland.

What does it cover?

Policy concerns, as well as scientific interests, have been continued, as described in Table 1.

The first two investigations at ages 2 and 4 years of the sample selected for follow-up were concerned with socioeconomic differences in the infant's growth, development, and morbidity, and with the effect of mothers' and health visitors' care and the family's socioeconomic circumstances on those health outcomes (Table 1). During the school years (ages 5–15 years) the measurement of growth and health continued, and a new strand of work was begun on educational experience and attainment (Table 1). In early adulthood (ages 16–31 years) the collection of health data continued, a study of delinquency was undertaken, and the outcomes of education were studied in terms of income and occupation (Table 1). At the beginning of middle adulthood (ages 32–53 years in Table 1), the study's health data collections were re-oriented in order to make the prime aim the measurement of physical and mental function and the study of pathways to those outcomes, in addition to the study of morbidity and mortality; that aim has continued.

Who is in the sample?

Figure 1 shows that the target sample for the first data collection was all the births in England, Scotland, and Wales that occurred in one week in March 1946. Subsequent data collections have been from a sample of all single births to married women using the selection criteria described in Figure 1. A weighting can be applied in analyses in order to adjust for the sampling procedure. The sample was distributed geographically in proportion to the national population.

How often have they been followed-up?

Data collections were frequent in infancy and the school years because of the pace of development and growth (Table 2).

In adulthood the periods between collections lengthened partly because biological and cognitive change then happens at a slower rate, and also because of cost. Between birth and age 15 years, data collection was undertaken by health care and educational professionals responsible for the child. After a period of postal data collection, and one round of data collection at home visits by professional interviewers at age 26 years, the three subsequent data collections have been at home visits by research nurses.

What has been measured?

Descriptions of measures are given in Table 3 using the same periods as in Table 1.

Attrition

Attrition is summarized in Table 4.²

The greatest overall attrition occurred, not unexpectedly, in the early adult years, when name and address changes were particularly frequent, and five of the seven data collections from the total sample were by postal questionnaire. Sample members are traced annually and a birthday card is sent together with a card to check and return on current address and name.

Unavoidable losses through death were high in infancy, and by age 53 years amounted to 8.7% of the cohort. Similarly, by the same age, unavoidable losses through emigration (8.6%) and living abroad (2.2%) are high in a sample that was affected by the ‘brain drain’ earlier in life, and is beginning to be affected by emigration for retirement. Death and emigration data have been supplied from the NHS Central Register since the cohort was first flagged in 1971, and cancer registrations are now also supplied.

Avoidable and potentially avoidable losses by age 53 years were through permanent refusal (12.4% of the original sample), temporary refusal for this data collection only (5.2%), and failure to trace (6.1%). New permanent refusals (28) were the lowest ever at this most recent data collection.

Of the sample available for contact at age 53 years, that is, all those who were alive and resident in England, Wales, or Scotland, and not already a permanent refusal (N = 3673), 3035 (82.6%) provided information. Among the sample successfully contacted (i.e. providing data) at 53 years, 73% had been successfully contacted at 17 or more of the 20 data collections from the whole cohort since age 2 years. Only 7% of them had taken part in 10 or fewer previous data collections.

Comparisons with Census data of the sample successfully contacted at 53 years and weighted to adjust for the initial sampling procedure, show that that sample represented in most respects the national population of a similar age.² At the most recent data collection 49% of respondents were men and 51% women. The sample also represents the population born in the early post-war period, at the time of the first post-war baby-boom, and will be the first of the generations who lived most of their lives in a welfare state. It represents now the population that will soon become the beginnings of the steeper increase in the national population of those aged ≥65.

Key findings and publications

Findings have been summarized in eight books.^1,3–9 References to all publications are on the study's website (www.nshd.mrc.ac.uk) and only selected references are given here.

Findings and publications during the years before adulthood

Findings on health, survival, growth, development, and morbidity, and their association with family circumstances and health services care, showed the beneficial impact of health visiting (community nursing) and community infant care services,³ and the considerable geographic and SES differences in health, survival, and the cost of child birth.¹

Findings on educational attainment showed the power of parental interest and concern for their cohort child's education independently of school and SES related factors.⁴ It was also clear that adverse SES circumstances reduced the longer-term educational opportunity and attainment, even of children with high measured cognitive ability.⁵

Behaviour problems in adolescence were more frequent in those who had experienced long or repeated hospital admission by age 5 years,¹⁰ and disruption of family life through parental separation was also a risk for that outcome,⁶ as well as for poor educational attainment.¹¹

Findings and publications during the adult years

Our health studies in the adult years, from age 36 years onwards, have concentrated on outcome measures of cardiovascular, musculoskeletal, respiratory and mental health, cognitive function, and women's reproductive health. Our aim is to describe the processes of ageing, and pathways to those processes and to disease risk, as well as to good health and well-being.

We are concerned with the relationships between development and ageing, and age-related diseases. For example, we show the effects of the growth trajectory in childhood and body weight change in adulthood on a range of adult outcomes and their change with age, including blood pressure,¹² adult obesity,¹³ cognitive function,¹⁴ mid-life muscle strength¹⁵ and physical capability,¹⁶ pre-menopausal breast cancer,¹⁷ and mammographic density.¹⁸ In other studies of ageing processes we report effects on the timing of the menopause of early growth and breast-feeding,¹⁹ cognition,²⁰ and the early social environment,²¹ in addition to the adult life risks.²² We show the importance of the early social environment, independent of the adult life environment, for many aspects of adult health, such adult obesity,²³ blood pressure,²⁴ cognitive function,²⁵ and survival,²⁶ as well as for socioeconomic outcomes. We show that cognitive function, in childhood and adulthood, is a determinant of physical health,²⁷ and survival,²⁸ and that despite the influences of adult exercise, smoking, and alcohol, adult occupation can protect against age associated cognitive decline.²⁹ Our genetic studies are just beginning, and show, for example in respiratory health, a genetic effect to be of importance independently of early growth, atmospheric pollution, and current lifestyle.³⁰ In our studies of health related behaviour over the life course we show lifetime variation in nutrient intakes,³¹ as well as differences in nutrient intake in childhood in this cohort and more recently born cohorts.³² Comparative collaborative studies are taking place using data from other cohorts, [for example refs (33–34)] and funding begins this year for comparison with the National Birth Cohort born in 1958 of growth trajectory and other effects on adult respiratory and cardiovascular outcomes. New work is beginning on childhood diet in relation to breast cancer and mammographic density. Other aspects of development studied include investigation of trajectories of early bladder control³⁵ and their association with mid-life urinary symptoms, and a study of diet in childhood in relation to mental health outcomes. A range of other life course determinants are being investigated in relation to cognitive function and its age related change in adulthood, including educational attainment and cognitive function at 8 years, which may be sensitive markers for genetic and early environmental exposures (e.g. stress, atmospheric pollution, nutrition) that also have long-term influence on health. New studies of social and behavioural pathways are concerned with musculoskeletal outcomes in terms of physical performance and disability.³⁶ New studies of health care include investigation of the risk factors for and consequences of hysterectomy and the characteristics of women taking HRT or alternative therapies,³⁷ and a study of whether social function is improved in those who received prescribed medication for depression between ages 36 and 53 years, as compared with those who did not.

We continue to develop methods, including the use of traditional epidemiological techniques such as regression to more complex structural equation models, for the analysis of life course data and testing of life course models.³⁸ In particular, we are concerned with the modelling of risk factor trajectories in relation to a later life outcome, distinguishing cumulative models from critical period models and modelling pathways from early risk to later life health. We are also developing approaches to handling missing data.³⁹

Strengths

The study's strengths are as follows:

• Its national and representative sample.

• Its repeated measures, particularly of growth in early life, childhood, and adolescence, of cognitive function in childhood, adolescence, and adulthood, and of physical and mental function three times in adulthood.

• Its source of DNA.

• Its information on diet that includes infant feeding, 24 h recall at age 4 years, and 5–7 day diet diaries at three adult ages; these have been coded into both foods and nutrients. Blood nutrient analytes were also collected at 53 years.

• Its extensive information, throughout life, on occupation, home circumstances, marital and fertility histories, education, and training, and information in adulthood on social support, integration, and networks.

• Its high quality data, which has been ensured through careful training of data collectors, checking with hospital records, and specialist coding, for example of disease events, mammograms, and dietary data.

• Its measures of function and functional change, and of clinically valid measures, in which the treated can be differentiated from the untreated.

The study in retrospect

As Table 1 shows the study is inevitably concerned with the science of its time, and measures that we wish had been taken in childhood and adolescence, for example of blood pressure, respiratory function, mental health, and temperament, were not then usually undertaken on those who were not evidently ill. The adverse health effects of smoking were not clearly demonstrated at the time the study began, and in consequence data on parental smoking was collected in retrospect. We also regret that the initial sampling before the first follow-up, greatly reduced the sample size and excluded births to the unmarried and multiple births, but that was dictated by the available information technology, by lack of access to the Adoption Register at a time when many births to the unmarried were adopted, perceptions of the statistical value of small numbers of multiple births, and by funding.

The data source

The data are the responsibility of the MRC National Survey of Health and Development (www.nshd.mrc.ac.uk), and are available in so far as consent and ethical approval permit, and as it is within the scope of the team's resources to make them available. We hope that, in due course, the Medical Research Council's initiative on data sharing will result in increased resources to make data more widely available.

We are very grateful to the members of this birth cohort for their continuing interest and participation in the study. We are also grateful to past colleagues, particularly to Dr JWB Douglas who initiated the study and directed it for the first 33 years, and to funders, especially the Medical Research Council.

References