Learn about the 1946 National Survey for Health and Development (NSHD) and its dietary measurements
Summary of cohort
The Medical Research Council (MRC) National Survey for Health and Development (NSHD), or the 1946 British birth cohort, is the oldest and longest running British birth cohort study [28-31]. NSHD originated from an initial maternity survey of 13,687 births recorded in England, Scotland and Wales during one week in March in 1946, a time when post-war rationing was still underway. Of these births, a socially stratified sample of 5,362 singleton babies born to married parents were selected for follow-up . Participants have been followed up 24 times with the most recent being a postal questionnaire at 68 years and a home visit at 69 years . The initial aim of NSHD was to examine how environmental factors both at home and in school affected physical and mental development and educational attainment. As the cohort has aged interest in how childhood health, development and lifelong social circumstances affect adult health and function grew and the cohort has developed into a life course study of ageing .
Overall participation in this study has remained high . Of the 2,546 (47%) original study members who did not participate in the data collection at 68-69 years, 18% had already died, 12% had withdrawn permanently, 11% lived abroad and 7% remained untraceable for more than 5 years. The majority of participants in NSHD are of white ethnic origin. Participants who previously reported poor general health were less likely to participate in the 68-69 year survey . Lower educational attainment, lower childhood cognition, lifelong smoking, not being married and not owning one’s own home at 53 years were associated with lower response rates at 60-64 years .
Dietary data collection
Information about dietary intake was collected at ages 4 (1950), 36 (1982), 43 (1989), 53 (1999), 60-64 (2006-2011) and 68-69 years (2014-2015). A summary of the different diet-related questions asked at each time point is provided in Table 5.1 and in the paragraphs below.
Diet in childhood
In 1950, diet was assessed during the home visit using a 24-hour recall. The mother or carer of the child was asked “What did this child have for each meal yesterday” with specific reference to breakfast, dinner, tea or high tea, and last thing at night. The quantity of food consumed was not recorded. There was a further question “Do you give this child food between meals” with a yes or no response . It is likely that recorded energy intake may be lower than total energy intake if the child had taken their personal ration of 5oz of sweets per week . The majority of the visits took place in summer (94%) and on a weekday (96%). Therefore, seasonally available fruit and vegetables, e.g. strawberries and lettuces, were probably consumed in greater quantity. Further details about how diet was assessed in 1950, including detailed information about rationing at that time, is outlined by Prynne et al. [33, 34].
Diet in adulthood
Between 1982 and 2006-2011, dietary intake was recorded using 5-day prospective estimated food diaries. In 1982 and 1989 the research nurse gave the participants detailed instructions on how to fill out the food diary and completed an additional 48-hour recall with them. The participant then completed the remaining 5 days of the food diary and returned it by post. If the participant did not send the diary back by post in 1982, the 48-hour recalls were lost. Therefore in 1989, a copy of the 48-hour recall was left with the nurse. In 1999 and 2006-2011, only the 5-day food diary was completed by the participants.
In these food diaries, all food and drink (including alcohol) consumed both at home and away was recorded using household measures to estimate portion sizes. There were detailed guidance notes of how to describe foods and, from 1989 onwards, photographs of portion sizes provided at the beginning of the diary to assist the participant. In the diaries there were seven spaces for each day to record meals and between-meal snacks, and a reminder section about any other snacks and space in which to write recipes. The food diary in 2006-2011 was slightly different in that it had space to record the specific time the food or drink item was consumed.
There was no food diary collected in 2014-2015, but there were a number of diet-related questions asked during the postal questionnaire (see the table below).
|Age 4y||Age 36y||Age 43y||Age 53y||Age 60-64y||Age 68-69y|
|5-day estimated food diary||✔||✔||✔||✔|
|Are you on a special diet?||✔||✔||✔||✔||✔|
|How many days a week do you usually eat breakfast?||✔|
|What kind(s) of milk do you usually have at home either in drinks or on cereal? Do not drink/use milk?**||✔|
|What type(s) of bread do you usually eat?†||✔|
|How often do you eat fruit and on days you eat, how many portions?‡||✔|
|How often do you eat vegetables not including potatoes and on days you eat, how many portions?‡||✔|
* Link provided to CLOSER Discovery (interactive version) or original study PDF as applicable.
** Do no drink/use milk, whole milk, semi-skimmed, skimmed, other (specify).
† White, brown, granary, wholemeal, don’t often eat bread, other (specify).
‡ Rarely or never, sometimes or not every day, every day or most days.
Estimation of nutrient intake
Nutrients from the food diaries were estimated following the approach outlined in Estimating nutrient intakes from DATs. The coding of the NSHD dietary diary data has advanced over the years through the development by MRC Human Nutrition Research in Cambridge of two in-house programmes: Diet In Data Out (DIDO) and Diet in Nutrients Out (DINO) [35, 36]. In 1982, the dietary data were originally manually coded in Bristol . DIDO was developed to code the 1989 diary data and was also used to code the 1950 24-hour recall and to convert the previously coded 1982 dietary dairy data and the 1999 diet diary data.
DIDO is a specially developed data entry system written in the C programming language . It is designed around a hierarchical food menu consisting of nearly 2000 food and drink items arranged by major food groups and sub-groups as listed in food composition tables. It generates a food code and associated weight in grams for each item recorded. The food codes are taken from the British food composition tables and the portion weights can be chosen from a list of standard weights attached to descriptions appropriate for each food e.g. teaspoon, tablespoon, medium slice etc, These portion weights vary according to the type of food and can be informed by manufacturers information. Once each diary is coded using DIDO, it is linked to British food composition tables [38-41] using a separate in-house suite of programmes to estimate nutrient intakes. Since the nutrient composition of food items were likely to have changed over time, time-appropriate food composition tables were used to estimate the nutrient intake for each diary as outlined previously . For the 2006-2011 dietary data coding, DIDO was updated to a Microsoft Access based system, DINO . DINO includes >6000 food items with their estimated portion sizes and is directly linked with food composition tables to estimate nutrient intakes . The previous diet diaries have now been transferred to DINO; however, there is some minor discrepancy in how the foods were categorised between the years. Vitamin and mineral supplements were coded separately. There were some specific elements of the 1950 24-hour recall that required special attention when coding and these are outlined in detail by Prynne et al. [33, 34].
It is not possible to estimate nutrient composition from the 2014-2015 diet-related questions due to insufficient detail captured.
The majority of participants who responded to the diet diaries at each age (see the table below) completed the full 5 days (83-99%). Up to 1999, the majority of diet diaries were completed in summer and autumn (68-94%) and excluding winter (<5%) when shortages of fresh fruit and vegetables would have been most apparent. In 2006-2011, diet diaries were completed equally throughout all four seasons (20-28%).
|Year||Age (y)||N interviewed||Response to diet question (n(%))||≥3 days†‡|
* N restricted to 5-day diet diaries only.
† In 1950 and 2014-2015, no food diaries were collected.
‡ N from postal questionnaire.
The participants of NSHD lived through a time of post-war austerity in their early years which directly affected their dietary intake. By 1950, the number of foods available was still in short supply. More information about rationing during this period in relation to NSHD has been discussed by Prynne et al. [33, 34]. Three studies examined the diet of NSHD participants in 1950 [33, 34, 43]. In comparison to children in the 1990s, children in 1950 ate a more homogenous diet which contained more bread and vegetables but less sugar and soft drinks . Their diets were higher in fibre and vitamin K but also higher in fat compared to children in the 1990s [33, 43]. Food sources of major nutrients were also different. For example, in 1950, iron came from red meat, but in 1992 it mainly came from fortified breakfast cereals . Rationing was designed to reduce inequalities in food intake, but despite this, there was some evidence of small social and regional differences remaining . Consumption of fruit and vegetables, which were not rationed, was higher among children whose father was classified as being in a non-manual occupational social class compared with manual social class. Some rationed foods (bacon, orange juice and tea) were also associated with social class, but meat and spreading fats were not. Children in Scotland tended to have a lower energy intake and vitamin K than other regions and retained traditional Scottish diets such as porridge and soups [34, 43].
Dietary trends and dietary patterns in adulthood
At 36 years, higher educational attainment was associated with better dietary habits, but in women, this was also associated with higher intakes of fat and alcohol . Disadvantaged social class and low educational attainment were associated with the worst dietary habits .
In the 17 year period between 36 and 53 years, there were changes in key nutrient intakes . For example, fat, sodium and iron intakes have fallen while calcium, carotene, folic acid, vitamin C and fibre intakes have increased [44, 45]. Total haem and non-haem iron rose from 36 until 43 years then decreased at 53 years . There was a decline in haem iron from beef with an increase in that from poultry, possibly reflecting the bovine spongiform encephalopathy (BSE) outbreak from 1990 [46, 47]. Similarly, over a 30 year period between 36 and 60-64 years, white bread was replaced by granary and wholemeal bread, while there was a reduction in the consumption of red and processed meats and an increase in the consumption of vegetables . These changes could be due to ageing, cohort effects or a response to government dietary recommendations and greater availability of foods such as wholegrain bread in the UK.
Five distinct dietary patterns (DPs) were identified using factor analysis on the 48-hour recall at 43 years: “health aware”, “dinner party”, “traditional”, “refined”, “sandwich” . Social class in childhood was associated with the DPs at 43 years, however, social mobility also had an impact. For example, participants who made the transition from manual to non-manual social class partly adopted the “health aware” and “dinner party” DPs of the non-manual SEP . Change in other DPs between 43 and 53 years were also observed . Three DPs among women (“fruit, vegetable and dairy”; “ethnic foods and alcohol”; “meat, potatoes and sweet foods”) and two in men (‘”ethnic foods and alcohol”; “mixed”) were identified from factor analysis of 126 food groups at 43 and 53 years . There was an increase in adherence to the DPs over time with only the “meat, potatoes and sweet foods” DP in women showing a decline.
The dietary diary data at 43 years was used to develop an index to discriminate healthy and unhealthy foods; the Eating Choice Index (ECI) . The index consists of information about breakfast and fruit consumption and type of bread and milk. Higher scores on the ECI are positively associated with protein, carbohydrate, fibre, vitamin C, iron, calcium and folate and negatively associated with fat. Participants with a lower ECI were also more likely to be in a lower social class, obese and less active .
Since diet diaries were used to collect the dietary data, NSHD provides a unique opportunity to examine a new discipline termed “chrononutrition” which investigates the timing and regularity of food intake [51-55]. Results from these studies have found that the proportion of energy and macronutrients consumed at lunch declined between 36 and 53 years, with greater intakes occurring in mid-afternoon . There was an association between increasing carbohydrate intake in the morning while simultaneously reducing fat intake at 43 years with metabolic syndrome  and diabetes  at 53 years. Irregular energy intakes at and between meals decreased with increasing age between 36 and 53 years . There was a cross-sectional and longitudinal association between meal irregularity and cardio-metabolic risk [54, 55].
Learn about the other studies covered by this guide and their dietary measurements:
- Overview of dietary information in selected CLOSER studies
- Hertfordshire Cohort Study (HCS)
- 1958 National Child Development Study (NCDS)
- 1970 British Cohort Study (BCS70)
- Understanding Society: The UK Household Longitudinal Study (UKHLS)
- The Avon Longitudinal Study of Parents and Children (ALSPAC)
- Southampton Women’s Survey (SWS)
- Millennium Cohort Study (MCS)
Get background detail on this guide:
- Objective and outline of this guide
- Dietary research in context
- Dietary assessment tools (DATs)
- Estimating nutrient intakes from DATs
Learn about harmonisation in the context of dietary data:
- Acknowledgements and copyright information for this guide
- References for this guide
- Download the full guide as a PDF
This page is part of the CLOSER resource: ‘A guide to the dietary data in eight CLOSER studies’.