A critical review of diet-related surveys in England, 1970-2018

Abstract

Background

Many diet-related surveys have been conducted in England over the past four to five decades. Yet, diet-related ill-health is estimated to cost the NHS £5.8 billion annually. There has been no recent assessment of the diet-related surveys currently available in England. This paper aims to fill this gap in the literature by providing researchers, especially those interested in conducting secondary (quantitative) research on diet, with a detailed overview of the major repeated cross-sectional and longitudinal surveys conducted in England over the last 48 years (1970–2018).

Method

A three-stage review process was used to identify and assess surveys and synthesise the information necessary for achieving the paper’s aim. Surveys were identified using the UK Data Service, Cohort and Longitudinal Studies Enhancement Resources (CLOSER), the Medical Research Council (MRC) Cohort Directory and the Consumer Data Research Centre (CDRC) online data repositories/directories. Surveys were summarised to include a brief background, the survey design and methodology used, variables captured, the target population, level of geography covered, the type of dietary assessment method(s) used, primary data users, data accessibility, availability and costs, as well as key survey features and considerations.

Results

The key considerations identified across the various surveys following the review include: the overall survey design and the different dietary assessment method(s) used in each survey; methodological changes and general inconsistencies in the type and quantity of diet-related questions posed across and within surveys over time; and differences in the level of geography and target groups captured.

Conclusion

It is highly unlikely that any survey dataset will meet all the needs of researchers. Nevertheless, researchers are encouraged to make good use of the secondary data currently available, in order to conduct the research necessary for the creation of more evidence-based diet-related policies and interventions in England. The review process used in this paper is one that can be easily replicated and one which future studies can use to update and expand upon to assist researchers in identifying the survey(s) most aligned to their research questions.

Background

Sub-optimal diet continues to be the most significant contributor to the global burden of disease, accounting for more deaths and disease than physical inactivity, alcohol consumption and smoking combined [4, 8, 10, 14]. Despite a proliferation of interventions which span decades, diet-related ill-health has been estimated to cost the National Health Service (NHS) approximately £5.8 billion annually [21]. In response to this situation, the World Health Organisation (WHO) has urged researchers to make “effective, proper and good use” of the secondary data currently available, in order to conduct the research necessary for the creation of more evidence-based diet-related policies and interventions [27]. Diet-related surveys continue to be the major source of information used by researchers and policymakers to assess dietary patterns, monitor trends over time, evaluate the success/failure of interventions and identify potential inequalities. Although the availability of diet-related survey data is limited in many European countries, England boasts several Government sponsored/endorsed repeated cross-sectional and longitudinal surveys. Surveys such as the National Diet and Nutrition Survey (NDNS), the Health Survey for England (HSE), Understanding Society, and many others, can be easily accessed online from national data repositories such as the UK Data Service, usually at little or no cost. The relative ease with which secondary data can be accessed in England at present means that now, more than ever, researchers are able to explore diet-related topics of interest and forgo what would have been an otherwise time-consuming and costly primary data collection process. Though beneficial, the analysis of secondary data still requires that researchers clearly define their research questions, critically assess diet-related surveys currently available from the outset and identify the survey(s) which best suits their unique research needs, before any data are analysed [3]. Although initially time-consuming, this type of detailed preliminary assessment is essential, as it saves time in the long run and helps to ensure the overall success of diet-related studies undertaken.

Several studies have noted general challenges and practical considerations which researchers often face when analysing diet-related data [1, 12, 13, 16, 26]. Examples of these include: the unavailability of consistent, nationally representative diet-related data, different dietary assessment methods used in surveys and the tendency for surveys to capture data on single food groups/nutrients (such as fruits and vegetables) as opposed to a variety of foods. Rippin et al. [20] previously assessed the current status of nationally representative surveys in Europe. However, the authors of that study only focused on the 53 countries in the WHO European region and not England specifically. Overall, very few studies have outlined and discussed diet-related surveys conducted in England, their characteristics, possible benefits and some of the practical and unique considerations researchers should note when trying to decide the survey dataset(s) most aligned to their research question(s).

This paper is not a systematic review but, rather, a secondary data review which aims to fill a gap in the literature by providing researchers, especially those interested in conducting secondary (quantitative) research on diet and with limited time and resources, with a detailed overview and summary of the strengths and weaknesses of the major repeated cross-sectional and longitudinal surveys conducted in England over the last 48 years (1970–2018). Surveys identified and discussed in this review should not be interpreted as being capable of meeting all the needs of researchers involved/interested in diet-related research. Instead, this review will provide a brief background on some of the major diet-related repeated cross-sectional and longitudinal surveys conducted in England over the past four decades, the survey design and methodology used, variables captured, the target population, level of geography covered, the type of dietary assessment method(s) used, primary users of the data and information related to data accessibility, availability and costs. Additionally, key survey features which could benefit some researchers in answering their particular research question(s) will be highlighted, as well as some practical considerations which should be acknowledged before selecting and analysing data. To the best our knowledge, this is the first paper to provide this type of detailed information on a current snapshot of major repeated cross-sectional and longitudinal diet-related surveys in England. This information could serve as a template or a quick guide which researchers can refer to as a starting point to identify existing diet-related surveys, assess potential survey benefits/issues and the possible impact (positive or negative) this could have on their research. This information will enable researchers to develop separate work-around strategies (where necessary) to suit their unique research needs and will save them time and resources than if it were necessary to compile this information from scratch.

Methods

Preliminary meetings were held with all members of the paper’s Review Team (MC, DS, JB, GM and CV) to discuss the scope, eligibility criteria and analytic strategy of this review. The decision was to include repeated cross-sectional and longitudinal surveys, where quantitative information on diets in England was collected over the 1970–2018 period. A three-stage review process was used to identify, assess and synthesise the information necessary for achieving this paper’s aim (Fig. 1).

Fig. 1

Three-stage research and review process used to identify and assess surveys and synthesise findings

Stage one of the review process (Fig. 1), involved the identification of all major repeated cross-sectional and longitudinal health, diet-related surveys, conducted in England over the period from January 1970 to December 2018. This assessment period (48 years) was thought to be an adequate time span in which a sufficient number of longstanding and current survey datasets (especially longitudinal surveys) could be captured. Surveys were identified using the four major online directories currently available and used by researchers in the UK, namely: the UK Data Service, the Medical Research Council (MRC) Cohort Directory, Cohort and Longitudinal Studies Enhancement Resources (CLOSER) and the Consumer Data Research Centre’s (CDRC) online directory. These four online directories were selected because they provided a comprehensive list of all surveys conducted within the UK over time, a summary of the survey design, variables captured within datasets, links to survey documentation and where relevant, the institutions (academic and research) ultimately responsible for managing and disseminating data.

The search strategy used to identify initial survey results varied, based on how each of the four databases were inherently structured. For the UK Data Service and the CDRC databases, an exact keyword search for “Health behaviour”; “Food consumption;” “Diet consumption”; “Dietary consumption”; “Diet and nutrition”; “Eating habits” and “Diet” was conducted. This was done to ensure that a wide variety of surveys, especially those not directly associated with diet, but which captured aspects of diet-related behaviours, would have been initially identified. The MRC Cohort Directory presented a full list of all major cohort (longitudinal) studies conducted in the UK, from which diet-related surveys relevant to this review were identified using the database’s pre-defined “Dietary Habits” topic filter option. CLOSER was strictly focused on eight longitudinal surveys (the Hertfordshire Cohort Study, 1946 and 1970 British Cohort Study, 1958 Child Development Study, Avon Longitudinal Study of Parents and Children, Southampton Women’s Survey, Understanding Society and Millennium Cohort Study) which captured persons born throughout the 20th and 21st centuries. All eight affiliated surveys were listed in the “Our Studies” section of the CLOSER database, which meant there was no need to filter or conduct any keyword searches. In total, 1836 preliminary results were obtained across the four databases, of which 97% (1785 results) were from the UK Data Service.

Preliminary search results obtained were manually assessed by the Review Team (MC, DS, JB, GM and CV) to filter out duplicates (205 of the 1836 total preliminary results) and surveys which did not meet the paper’s eligibility criteria (1614 out of the 1836 total preliminary results). Ineligible surveys included: discontinued surveys, non-repeated cross-sectional surveys conducted only once, surveys which although diet-related, had no data for England (e.g. international studies or studies focused on a particular UK constituent country such as Scotland only), surveys which fell outside the 1970–2018 search period, surveys which had little or no quantitative diet-related data (e.g. qualitative/perception studies, gene/twin studies, general health studies with no diet-related data) and surveys which could not have been properly assessed due to limited documentation at the time of assessment. The removal of duplicate and ineligible surveys (1819 results omitted), reduced the results from 1836 to 17 surveys eligible for inclusion in the current review (Fig. 1).

In Stage two of the review process, questionnaires, documents and technical reports for the 17 eligible surveys were retrieved online from the UK Data Service and the official website of the responsible academic and research institutions. Academic and research institutions included: the CDRC; UK Biobank; University of Oxford; University of Cambridge; University College London (UCL); University of Bristol; the MRC Lifecourse Epidemiology Unit, University of Southampton; Understanding Society; the Bradford Institute for Health Research; University of Leeds and the Centre for Longitudinal Studies (CLS). Where necessary, follow-up emails were sent directly to the UK Data Service and institutions to collect additional information not available on official websites. Documents (inclusive of questionnaires used across survey waves/periods) received either from websites or via email were thoroughly reviewed in order to identify: the survey design and methodology used, diet-related questions/variables captured, the target population, level of geography covered, the type of dietary assessment method(s) used, primary users of the data, accessibility, availability and data costs, as well as the key survey features/potential benefits and key considerations for each survey.

Finally, Stage three involved the compilation of findings, which were cross-validated with all members of the Review Team (MC, DS, JB, GM and CV) and tabulated (see Table 1) in order to capture the detailed information on all 17 surveys in an easy to understand and user-friendly manner.

Table 1 Summary/review of 17 major repeated cross-sectional and longitudinal surveys conducted in England over the January, 1970 to December, 2018 period

Results

Overall, 17 surveys (5 repeated cross-sectional and 12 longitudinal) were identified and deemed relevant for inclusion within this paper (Table 1). The five repeated cross-sectional surveys were the Living Cost and Food Survey (LCFS), Active Lives Survey (ALS), National Diet and Nutrition Survey (NDNS), Health Survey for England (HSE) and Food and You, all of which were accessible via the UK Data Service (Table 1). The Centre for Longitudinal Studies (CLS) and Understanding Society were the primary institutions responsible for collecting, managing and disseminating data related to the British Cohort Study 1970 (BCS70), Millennium Cohort Study (MCS) and Understanding Society, respectively. However, these were also the only longitudinal surveys which were accessible through the UK Data Service. The nine remaining longitudinal surveys assessed (the Southampton Women’s Survey (SWS), Born in Bradford (BiB), Avon Longitudinal Survey of Parents and Children (ALSPAC), UK Women’s Cohort Study (UKWCS), European Prospective Investigation into Cancer and Nutrition (EPIC Norfolk/Oxford), UK Biobank, Whitehall II, British Regional Heart Study (BRHS) and British Women’s Heart and Health Study (BWHHS)) were primarily accessible through the respective academic and research institutions listed in Table 1. Table 1 provides a detailed summary of each of the 17 surveys reviewed, inclusive of their key features/potential benefits and some key considerations which researchers should note, if or when using any of the following surveys to conduct secondary data analysis.

The HSE remains the primary source of information used by the English Government to monitor and assess changes in the overall health and lifestyle of children (0–15 years) and adults (16 years and over) living in England. Although a sports and recreation survey, the ALS captured annual fruit and vegetable consumption for over 198,000 persons (aged 14 years and over) living in England. The NDNS, on the other hand, is currently the only annual, nationally representative survey which provides detailed information on all foods and beverages consumed by persons 18 months of age and older. Food and You was the only repeated cross-sectional survey which was not conducted annually, but every 2 y (bi-annually).

Of the 12 longitudinal surveys assessed, five (SWS, BiB, BCS70, ALSPAC and MCS) were birth cohort surveys which followed the same group of individuals from birth through to adulthood (Table 1). With the exception of Understanding Society, the remainder of the longitudinal surveys reviewed (UKWCS, EPIC, UK Biobank, BRHS, BWHHS and Whitehall II) were primarily focused on exploring the relationship between diet and health outcomes such as cancer and heart disease, amongst middle- aged persons (aged 35 years and over at the time of recruitment). Understanding Society was the only large-scale, multi-topic longitudinal study, which followed individuals across all age groups (children and adults), living in over 40,000 households in the UK, on an annual basis. As such, one of its key features was its large annual sample size and its national representativeness.

In terms of dietary assessment methods used, the Food and You, BiB, BRHS, BWHHS, MCS and Whitehall II used Food Frequency Questionnaires (FFQs) solely, whereas the SWS, ALSPAC, UKWCS and EPIC used a combination of methods (inclusive of FFQs, across different survey waves). A key feature of the LCFS was the availability of two-week expenditure diaries which captured purchased quantities of food and drink. However, it should be noted that the survey does not capture foods actually consumed by individuals, but rather household food purchasing and expenditure. Understanding Society primarily captured the frequency of fruit and vegetable consumption using a brief dietary instrument. Besides their large annual sample sizes, the HSE and ALS captured the consumption of fruits and vegetables using a single 24-h shortened dietary instrument/screener. The NDNS’ consistent use of the food diary assessment method across the survey waves was a feature which set it apart from the remainder of the surveys which used FFQ, shortened dietary screener instruments, 24-h diet recalls or a combination of these methods across the different survey waves. The use of this method meant that the survey provided detailed information, including nutrient content and portion size, on all foods and beverages actually consumed by individuals, over a four-day period.

Besides methodological changes to the NDNS, and LCFS noteworthy changes to the type and number of diet-related questions asked across the survey waves were observed for 13 of the 17 surveys reviewed (HSE, Food and You, BIB, Understanding Society, BCS70, ALSPAC, UKWCS, MCS, EPIC, UK Biobank, Whitehall II, BRHS and BWHHS).

Of all the surveys reviewed, nine (SWS, BiB, ALSPAC, UKWCS, Whitehall II, EPIC, UK Biobank, BRHS and BWHHS) were not representative of the general English population, all of which were longitudinal surveys. BiB focused on Bradford in the North of England, whereas SWS and ALSPAC were limited to Southampton and Bristol in South East and South West England, respectively. Besides the study’s focus on middle-aged persons, EPIC Norfolk/Oxford was also limited in terms of its focus on the geographical areas of Norwich, Greater Manchester, Oxfordshire and Buckinghamshire. Data captured in BRHS and BWHHS were not representative of the English population and were limited to middle-aged males and females from only 24 and 23 towns (respectively) across Scotland, Wales and England. Although the UK Biobank followed 500,000 persons across the UK, the survey was focused on middle-aged persons. Overall, BCS70, MCS and Understanding Society were the only longitudinal surveys reviewed which were nationally representative.

Discussion

The primary aim of this paper was to provide researchers, especially those interested in conducting secondary data analysis, with a detailed overview of 17 major diet-related repeated cross-sectional and longitudinal surveys conducted in England over the past 48 years (1970–2018). Following this review, three broad thematic areas were identified. These included: the overall survey design and the different dietary assessment method(s) used in each survey; methodological changes and general inconsistencies in the type and quantity of diet-related questions posed across and within surveys over time; and differences in the level of geography and target groups captured across the surveys.

Survey design and dietary assessment methods used

Repeated cross-sectional surveys such as the NDNS, HSE, ALS, LCFS and Food and You, are inherently designed to provide researchers with a snapshot of diet and related behaviours for a particular group of individuals (sample), at a particular point in time. With the exception of Food and You (conducted bi-annually), the remaining repeated cross-sectional surveys were conducted annually. Longitudinal surveys (such as SWS, BiB, Understanding Society, BCS70, ALSPAC, UKWCS, Whitehall II, MCS, EPIC, UK Biobank, BRHS and BWHHS) however, are primarily designed to follow the same group of individuals over an extended period of time or across the lifecourse (in the case of birth cohort studies such the SWS, BiB, BCS70, ALSPAC and MCS). It is possible to pool data from individual survey waves/years for repeated-cross-sectional surveys. This could help to increase the overall sample size (where deemed necessary) and could be a means of exploring possible differences in diet and related behaviours across survey waves. However, because repeated cross-sectional surveys capture a different group of individuals at each survey wave, they may be more appropriate for researchers interested in assessing current diet-related behaviours, than those interested in tracking possible changes amongst the same group of individuals over time.

Aside from survey design, it was known that the dietary assessment method(s) used in all 17 surveys would have inherent strengths and weaknesses, depending on the context in which they are used. Unlike previous studies [18, 19, 22], providing a detailed description of the pros and cons of the different dietary assessment methods used in surveys was not within the scope of this review. Nevertheless, similar to those studies, this review found that the type of dietary assessment method(s) used in surveys is another area researchers should closely consider, especially when trying to decide the secondary data sources(s) most aligned to their research questions. For instance, the LCFS captures data on the amount (quantity) of food and drink purchased by households, via 2 week/14-day expenditure diaries (found in the survey’s Family Food Module). This type of information is particularly useful for persons interested in exploring household-level shopping and eating habits, household-level socio-economic variations in diet [5] or evaluating population level food purchasing-focused interventions [2]. Researchers in the Department for Environment, Food and Rural Affairs (DEFRA) rely on LCFS data to calculate cost of living indices and to produce the Government’s annual Family Food Report, which provides estimates of nutrient content and statistics on household food purchases by food type. Although beneficial in these circumstances, because the LCFS is an expenditure survey, its design and focus are not the diet of individuals. Whilst it is possible to use expenditure data (as captured in the LCFS) to estimate the quantity of food consumed and the nutrient intake of individuals within households (proxy measure), this is mostly done in low resource settings, specifically in countries which have limited diet-related data other than that captured in household expenditure surveys [7]. Given that the LCFS is not the only source of diet-related data in England, researchers interested in exploring the actual consumption of individuals and potential demographic and socio-economic differences (e.g. age, sex, educational attainment) in diet in England (using repeated cross-sectional survey data), should consider more appropriate surveys such as the NDNS or others, which have data on the actual diet of individuals.

The NDNS’ consistent use of the food diary assessment method across the nine survey waves (nine waves were completed at the time of this review/assessment) meant that the survey captured detailed information on all foods and beverages actually consumed by individuals, over a four-day period. A key feature of the food diary method is that recording of data is done at the time of consumption, which helps to reduce recall bias or the reliance on memory and improves the quality and accuracy of data collected [22]. Respondents are trained to estimate and record amounts consumed using household measures (e.g. one tablespoon of baked beans) and photographs included in the survey. This type of data could be useful to researchers interested in fully exploring the overall diet, nutrients or portion sizes (not only single food groups such as fruits and vegetables) of individuals living in England and possible socio-demographic differences. However, the food diary method, although beneficial, requires significant financial, physical and human resources to implement, especially on an annual basis, and requires that survey participants be literate and committed to completing the entire process [7, 22]. As a result, individuals with low levels of literacy and those from lower socio-economic groups may be under-represented.

Another key consideration is that the NDNS currently targets 1000 persons (500 adults and 500 children) annually, across the entire UK (England, Scotland, Northern Ireland and Wales). Although customary for surveys which use the food diary method, the survey’s relatively low annual sample size could be seen as a limitation. Nevertheless, the pooling of data across the survey waves is one means of increasing the overall sample size and a possible workaround for researchers desirous of investigating diet across the survey waves. Similar to the NDNS, a key feature of the BCS70 was the availability of food diary data for the 1986 and 2016 survey wave. The use of this method meant that diet-related information captured was detailed and as a longitudinal survey, interested researchers could possibly assess differences or changes in the diet of cohort members over time. However, researchers keen on accessing BCS70 food diary data should note that data for both the 1986 and 2016 waves were being processed at the time of assessment and the expected date of release is yet to be determined.

The traditional 24-h diet recall method captures all foods and beverages consumed the preceding day, ideally, over multiple or repeated assessment periods. Dietary screeners or shortened instruments, however, only assess one or two nutrients/food groups, such as fruits and vegetables or calcium/dairy products [7, 19]. The UK Biobank was the only survey in which 24-h diet recalls were conducted on four separate occasions over a 10-day and 16-month period, respectively. Conversely, respondents in the HSE and ALS were asked to recall their consumption of fruits and vegetables, over a single 24-h period. This meant that a brief dietary assessment instrument (screener) was used in both surveys, and not the traditional 24-h diet recall method as initially assumed. The traditional 24-h diet recall method is beneficial in that it provides more precise estimates of nutrients/food and estimates which are more representative of usual dietary consumption. Given that this method captures all foods and beverages consumed over repeated assessment periods, it may be useful to researchers interested in exploring total diet, rather than just key food groups such as fruit and vegetables. The fruit and vegetable screener used in the HSE and ALS may be more beneficial to researchers interested in assessing current adherence to the national “5-A-Day” (fruit and vegetable) dietary target or those interested in exploring the association between fruit and vegetable consumption, physical activity and related health outcomes/chronic diseases. However, because the HSE and ALS only captured consumption over a single 24-h period, researchers should also bear in mind that day-to-day variations in consumption cannot be accounted for.

FFQs often require that respondents indicate how much and/or how often (e.g. daily, weekly) they consume a set of listed foods over a specific period (e.g. over a week, the last 12 months). Unlike the food diary and 24-h diet recall method, surveys which used FFQs are beneficial as they are usually less burdensome and are able to assess the usual diet of individuals over a long-term period, with the added benefit of larger sample sizes [23]. However, because some FFQs are comprised of a short, pre-selected list of foods, (sometimes referred to as dietary screeners) many aspects of diet are not measured, which may make them prone to systematic errors and not be entirely reflective of diet consumption at the population level [9]. For instance, in Understanding Society, respondents were primarily asked about the number of days in a week they eat fruits and vegetables and the number of portions consumed on those days. Although this captures some elements of diet, the survey’s emphasis on fruits and vegetables may make it inappropriate for researchers more interested in exploring diet in its entirety.

Methodological changes and changes to survey questions over time

As expected, more than a half of the surveys reviewed either had changes made to the type and number of questions asked and the level of detailed captured over time or the survey design/methodology used. For instance, the NDNS, established in 1992, initially consisted of four separate cross-sectional surveys which captured data for individuals from specific age groups (e.g. persons aged 19–64 years in 2000–2001), 18 months and older, across the 1992–2001 period. However, with the introduction of the rolling programme (the NDNS RP) in 2008, the survey changed from a series of ad-hoc age-group specific surveys, to an annual repeated cross-sectional survey for all age groups. As a result, data captured prior to 2008 may not be easily compared with NDNS RP data, which could affect researchers interested in assessing food consumption in England, especially by age. Besides methodological changes, as expected there were notable changes to the type and number of diet-related questions posed across the survey waves. However, the most noteworthy were those made to the HSE across the survey waves. Prior to 2009, the HSE had a “Fruit and Vegetable Consumption” module in addition to an “Eating Habits” module, which captured the frequency of consumption for at least 12 food items via a FFQ. Food categories included: cheese, red and white meat, fried food, sweets, fizzy drinks, among others. However, since 2009, the HSE only captures data on fruit and vegetable consumption, as it is currently the primary survey used by Public Health England to monitor the Government’s national “5-A-Day” target [17]. Although the survey is currently focused on fruit and vegetable consumption, it should be noted that the “Fruit and Vegetable Consumption” module was completely omitted from the survey in 2012, for all age groups and was omitted in 2014 for persons 16 years and older. These changes could possibly affect researchers interested in monitoring fruit and vegetable consumption specifically for the 2012 and 2014 survey period, as well as persons interested in merging and analysing data across several survey waves, inclusive of the 2012 and 2014 waves.

The rapid and ever-evolving field of nutrition science could possibly explain some of the changes observed in the surveys reviewed over the paper’s 1970–2018 review period. However, it should also be acknowledged that survey content, questions asked over time and the methodology used is ultimately based on the overall purpose and intended use of the survey, and the priorities, interests and needs of survey administrators/Governmental Departments/primary stakeholders, rather than the research interests of researchers/users of secondary data. For instance, although surveys such as the HSE capture some aspects of diet, researchers should recall that the survey’s main purpose or focus is not on diet, but on capturing the overall health status of the population and associated risk factors. Also, changes to the type of survey questions asked and the level of detail captured over time, is heavily dependent on the financial, physical and human resources available. Whilst funders and data collectors are cognisant of some of the general interests and data needs of secondary data users, they are also faced with the tremendous challenge of balancing the needs of primary stakeholders and reducing survey costs and participant burden [15]. Researchers therefore need to be aware and constantly keep abreast of survey changes (such as those highlighted in this paper) and their potential impact (positive or negative) on research and devise workaround strategies needed to meet their unique research needs, as far as possible.

Geographical areas and groups targeted across the surveys

Another major consideration which researchers should acknowledge is the different geographical areas/regions and target groups captured across the surveys. All repeated cross-sectional surveys reviewed were nationally representative and the BCS70, MCS and Understanding Society were the only nationally representative longitudinal studies reviewed. SWS, BiB and ALSPAC could be beneficial for researchers interested in tracking changes in the diet-related behaviours of cohort members from birth through to adulthood. However, it should be noted that these surveys were only focused on certain regions of England, (specifically Southampton, Bradford and Bristol/Avon Health Authority and surrounding areas, respectively), of interest to the respective survey administrators/academic institutions. Similarly, the UKWCS, Whitehall II, EPIC, BRHS, BWHHS and UK Biobank were not representative of the English population, as they targeted certain groups within the population, such as women, middle-aged persons, middle-class persons, vegetarians or members of the Civil Service. Groups which although of possible interest to some researchers, were specifically aligned to the interests and needs of the administrators/academic institutions responsible for these surveys.

In terms of the repeated cross-sectional surveys reviewed, a key feature of the ALS was its annual sample size of over 198,000 individuals and data at the local authority level. The ALS was the only repeated cross-sectional survey reviewed in which data below the Government Office Region (GOR) level was readily available in survey datasets. The ALS could be especially beneficial to researchers (e.g. public health geographers) interested in exploring diet (fruit and vegetable consumption) and possible variations at the national, regional and sub-regional/local authority level. However, persons interested in accessing data below the regional (GOR) level should note that this information is not included in the general End User License for the HSE, NDNS, LCFS or Food and You survey datasets. This type of information needs to be specially requested and approved, and in some instances (in case of the HSE), at an additional cost, to cover data processing and administration fees. Based on the General Data Protection Regulation (GDPR) and other disclosure guidelines, the UK Data Service has instituted strict measures regarding access to sensitive data (e.g. lower-level/sub-regional geographical data), which could be used to reveal the identity of participants [17]. These are other considerations researchers need to acknowledge when trying to decide the survey(s) best aligned to their unique research questions/interests.

Strengths and weaknesses of this review

The research presented involved a detailed process to provide researchers, especially those interested in conducting secondary data analysis, with an overview (inclusive of key features and practical considerations) of 17 major diet-related repeated cross-sectional and longitudinal surveys conducted in England over the past 48 years (1970–2018). A major strength is that the findings presented in this paper should save researchers interested in diet-related research, time and well-needed resources in compiling this type of information from scratch. This structure is one that may be easily replicated as a follow-up as resources change, providing a clear template for the evaluation of available sources for secondary data analysis of population diet in England. This review did not discuss new and emerging technology-based dietary assessment methods (e.g. web-based and mobile device applications or the use of “big data”), which is a limitation. However, such methods are still not clearly defined and not comprehensively captured in repositories or widely available for re-use [24, 25]. Also, the surveys reviewed may not be exhaustive of all diet-related surveys conducted in England over the 1970–2018 period. The paper’s focus on longitudinal and repeated cross-sectional surveys meant that surveys conducted only once were not included within this review. Therefore, cross-sectional surveys such as the Low-income Diet and Nutrition Survey (LIDNS) and What about Youth (WAY), conducted in 2003–2005 and 2014–2015 (respectively) were not assessed. The detailed description of the pros and cons of the different dietary assessment methods used in surveys was not within the scope of this review. As a result, the review’s failure to discuss the availability of biomarker data in surveys such as the NDNS and the usefulness of this kind of information for validating self-reported dietary data, was another limitation.

The review process used in this paper was time consuming but was a task which assisted the paper’s Review Team (MC, DS, JB, GM and CV) in identifying the surveys most appropriate for their individual research projects. During this process, the need for a review of the current status of diet-related surveys conducted in England over time was identified, particularly if benefits and the practical considerations to using surveys datasets could be incorporated as part of a review. Although the survey documentation required to conduct the review was readily available online from the UK Data Service, CLOSER, CDRC and the MRC Cohort Directory, to the best of our knowledge, no resource exists which provides a comprehensive list and background on the major repeated cross-sectional and longitudinal surveys in England. Although Rippin et al. [20] previously assessed the current status of nationally representative surveys in Europe, it focused on the 53 countries in the WHO European region and not England specifically. Griffith, O’Connell & Smith [11] noted some benefits and possible limitations of diet-related surveys in England. However, unlike this review, their assessment was limited to only three data sources: the NDNS, LCFS and Kantar Worldpanel. Coleman [6] comprehensively summarised 16 longitudinal surveys conducted in England, over the 2005–2015 period. However, Coleman’s report was not focused on diet-related behaviours because it was intended to provide the Department of Education with the information necessary to plan interventions and meet the educational needs of children and young persons under age 19. This review has helped to fill this gap in the literature. Overall, the findings presented indicate that although several diet-related surveys have been conducted over the years, each with their own unique benefits/features, there are still several practical considerations which researchers should note when considering the survey(s) best suited to their research interests.

Conclusion

Diet-related surveys continue to be the major source of information used by researchers and policymakers to assess dietary patterns, monitor trends over time, evaluate the success/failure of interventions and identify potential inequalities. It is highly unlikely that any survey conducted will meet all the needs of researchers. Additionally, data-related challenges faced by researchers will inevitably vary based on the nature of the research question(s). Regardless, it is still vital that researchers clearly define their research question(s), critically analyse the secondary survey data available (as done in this paper), gain a full understanding of the unique survey characteristics and note key considerations, before delving into data sets. In some instances this may mean that initial research questions may have to be modified or refined, where data of interest may be limited, unavailable, inconsistently captured across survey waves, captured/defined in a manner not befitting to research questions or perhaps too costly to access based on financial constraints. Although not ideal, this is one possible strategy which may help to save time and money and could help researchers to make the best use of the data currently available.

Enhanced communication and engagement between data collectors, data users (existing and new/emerging), data repositories, funding agencies and policy makers could help to ensure that the data being collected is appropriate and cost-effective to inform policy and intervention development. However, researchers using secondary data must acknowledge that change is inevitable and that the type of dietary assessment used, the type of questions included and the level of detail captured in surveys over time, ultimately depends on the priorities and interests of primary stakeholders, the overall purpose and intended use of the survey, and the financial, physical and human resources available. With the increasing prevalence of sub-optimal diet and as research budgets continue to tighten, funding agencies, governments and research institutions are constantly having to consider new, cost-effective and creative methods (e.g. big data and digital technology) of maintaining existing repeated cross-sectional and cohort studies, retaining survey participants and overcoming geographical constraints. In light of these challenges, researchers therefore need be cognisant of these practical considerations, and as far as possible, make every effort to make “effective, proper and good use” of the secondary data currently available, in order to conduct the research necessary for the creation of more evidence-based diet-related policies and interventions in England.