The development and evaluation of a quality assessment framework for reuse of dietary intake data: an FNS-Cloud study

A key aim of the FNS-Cloud project (grant agreement no. 863059) was to overcome fragmentation within food, nutrition and health data through development of tools and services facilitating matching and merging of data to promote increased reuse. However, in an era of increasing data reuse, it is imperative that the scientific quality of data analysis is maintained. Whilst it is true that many datasets <i>can</i> be reused, questions remain regarding whether they <i>should</i> be, thus, there is a need to support researchers making such a decision. This paper describes the development and evaluation of the FNS-Cloud data quality assessment tool for dietary intake datasets. Markers of quality were identified from the literature for dietary intake, lifestyle, demographic, anthropometric, and consumer behavior data at all levels of data generation (data collection, underlying data sources used, dataset management and data analysis). These markers informed the development of a quality assessment framework, which comprised of decision trees and feedback messages relating to each quality parameter. These fed into a report provided to the researcher on completion of the assessment, with considerations to support them in deciding whether the dataset is appropriate for reuse. This quality assessment framework was transformed into an online tool and a user evaluation study undertaken. Participants recruited from three centres (<i>N</i> = 13) were observed and interviewed while using the tool to assess the quality of a dataset they were familiar with. Participants positively rated the assessment format and feedback messages in helping them assess the quality of a dataset. Several participants quoted the tool as being potentially useful in training students and inexperienced researchers in the use of secondary datasets. This quality assessment tool, deployed within FNS-Cloud, is openly accessible to users as one of the first steps in identifying datasets suitable for use in their specific analyses. It is intended to support researchers in their decision-making process of whether previously collected datasets under consideration for reuse are fit their new intended research purposes. While it has been developed and evaluated, further testing and refinement of this resource would improve its applicability to a broader range of users.