While a growing number of epidemiological studies have shown a positive association of unhealthy food retail access, with poor diet, greater body weight and odds of obesity, and poor health outcomes, the evidence base remains equivocal. The majority of published research evidence has been cross-sectional and observational, which limits scope for causal inference. Studies have also focussed on broad classes of food outlets (e.g. takeaways, supermarkets, restaurants), with little discrimination between outlet cuisine types, exposure to which may be differentially associated with body weight, and specific dietary and health outcomes. Moreover, there is currently no evidence regarding the effectiveness of local authority actions to restrict proliferation of hot food takeaway outlets, for example, where these have been implemented in England. These knowledge gaps exist in part due to the lack of accessible, complete, usable, detailed, longitudinal secondary food outlet location databases in the UK.
Since 2014, the Food Standards Agency (FSA) have made their Food Hygiene Rating System (FHRS) food outlet database available freely online. The database contains FHRS data submitted by local authorities, who collect this information to enable routine food hygiene inspections and ratings. The database contains information, including geographic coordinates, on all establishments that serve food, including those where food retail is not the primary purpose, e.g. high street clothing retailers. The number of local authorities contributing their FHRS to this platform has grown over time, such that all local authorities in the UK are now contributing their data on a fortnightly basis (according to unpublished data held by the research team). FSA FHRS data, which now have national, historic (beginning in 2014) and regularly-updated coverage going forwards, promise to help researchers overcome many of the critical limitations associated with existing food outlet datasets for research, which include lack of national coverage, resource intensive data collation, infrequent updates, restrictive terms and conditions of use, and financial cost. These data have the potential to mark a step change in food environments research, forming a valuable resource for longitudinal and pre- and post-intervention quasi-experimental research into neighbourhood ‘effects’, from which causal inference can be more strongly concluded. However, as a new dataset, the validity of FSA FHRS data have not yet been evaluated.
Not having longitudinal data to permit these scientific studies is more than just a scientific concern. The National Planning Policy Framework, for example, requires local risk factors be taken into account alongside scientific research evidence when developing local authority planning policies, however there is no national surveillance data to permit this local quantification of risk. This lack of local data may serve as a barrier to policy uptake, which may in turn compromise the public’s health. Real-time FSA FHRS could meet this data need, and serve to inform local decision making in support of the public’s health. Existing secondary data sources of food outlet locations commonly used by researchers are often prohibitively expensive for surveillance purposes, subject to terms and conditions of use that restrict sharing with those in local government, and integration with other datasets. With some data-processing to ensure usability and utility, FSA FHRS data could be made available online, free of charge to all users, either standalone or through a platform such as the Food environment assessment tool (Feat), which is currently part-funded by SPHR. However, it is unclear whether it is feasible to cache and categorise (process) large volumes of FSA FHRS ‘big’ data repeatedly over time, for surveillance purposes.
Therefore, formative work is required to scrutinise the validity of FSA FHRS data, and the feasibility of its application in both research and surveillance roles. Without funding to research these important aspects of the FSA FHRS data, it may have limited use in underpinning evidence based public health policy.