An investigation of dietary patterns in UK adults as a method for developing a brief diet quality.

Background
Dietary patterns analyses reduce detailed dietary intake data to underlying patterns, derived empirically or based on a priori knowledge. Brief tools to assess diet quality are needed for dietary surveillance and research in the UK. This study explores dietary patterns in UK adults as a method for developing such a tool.

Methods
Data analyses were conducted in National Diet and Nutrition Survey (NDNS) data (n=2083). Dietary patterns were derived through Principal Component Analysis. Associations with socio-demographic characteristics, lifestyle factors, nutrient intake and biomarkers were explored.
A nutrient-based diet quality score (NDQS) was developed based on a priori knowledge and validated against nutrient biomarkers. NDNS respondents were scored against the NDQS. Backwards elimination regression identified variables independently predictive of the NDQS.
Confirmatory analysis compared the indicator variables generated from empirically and theoretically driven methods as predictors of the NDQS. Regressions identified the most parsimonious model predictive of the NDQS.

Results
Four dietary patterns, explaining13.4% of the sample variance, were labelled according to the foods characterising them: ‘fruit, vegetables and oily fish’ (FVOF), ‘snacks, fast food and fizzy drinks’ (SFFFD), ‘sugary foods and dairy (SFD)’ and ‘meat, potatoes and beer (MPB)’. FVOF was positively associated with being female, non-white, older, a non-smoker and a higher NDQS score. SFFFD was the inverse.
The NDQS was positively associated with biomarkers of vitamins C, D, B6, total carotenoids and age and negatively with urinary sodium, being white and smoking. Both methods generated models predictive of diet quality (adjusted R2 = 0.29-0.31, 0.33 respectively). A tool with 5 items, fruit, vegetables, wholemeal bread, sugary drinks and coated chicken/turkey, was predictive of diet quality (adjusted R2 = 0.26).

Conclusion
Diet quality is an important measure for informing public health policy. This study found dietary patterns analyses to be useful methods for developing a brief, diet quality assessment tool.