Evaluation of the school-based ‘PhunkyFoods’ intervention: a cluster randomised controlled trial in the UK

26 schools (13 intervention, 13 control), 631 children (307 intervention, 324 control)

Children aged 6-9 in primary schools across North Yorkshire, UK; September 2022 to March 2023.

The closest thing this archive has to a randomised test of home economics — a whole-school food-literacy and cooking-skills programme, cluster-randomised, with pre-registered primary outcomes measured on validated instruments (TFLAC-UK for food literacy, CooC11 for cooking skills) and 24-hour dietary assessment for intake. That combination — cluster randomisation, an independently validated outcome instrument, and a behavioural outcome measured by diet recall rather than by asking children whether they liked the lessons — is exactly what the cooking-skills literature otherwise lacks, and it is why one small trial carries so much weight here. Limitations: 12 months is short for a habit outcome, the trial is single-region, and dietary recall in 6-9 year-olds is noisy (the vegetable and fruit confidence intervals are very wide). Read at abstract depth.

Null on everything. Food literacy total score differed by -1.13 points (95% CI -2.87 to 0.62, P = 0.2) on a 0-40 scale; cooking skills by -0.86 points (95% CI -5.17 to 3.45, P = 0.69) on a 0-55 scale; vegetable intake by -52.36 g (95% CI -122.75 to 18.03, P = 0.14) and fruit intake by -79.21 g (95% CI -250.34 to 91.92, P = 0.34). Every point estimate is negative and none is significant. The programme did not even move the knowledge outcome, which is the outcome life-skills instruction normally does move.

None material. School-level randomisation.

Unreplicated as a trial. It sits inside a wider systematic-review literature on school cooking classes reporting small positive effects on cooking self-efficacy and vegetable intake — but that literature is dominated by uncontrolled and self-report designs (roughly a quarter of the included studies are randomised), which is the standard pattern where a randomised test then returns nothing.