Introduction: The primary aim was to investigate feasibility of a web-based cross-over Paleolithic diet intervention in the general population. The secondary aim was to calculate the sample size needed to reach a statistically significant difference in effect of a Paleolithic-like diet on psychological and somatic symptoms compared with the Dutch consensus diet.

Methods: 35 participants from the general population were recruited using social media. Participants filled-out an intake questionnaire to assess representativeness of the sample. Participants were instructed to consume both a Paleolithic-like diet and the Dutch consensus diet during four weeks, the order was randomly assigned. After each period, participants filled-out a questionnaire to assess compliance to the dietary instructions, the Depression Anxiety Stress Scale to assess psychological symptoms and the Rosmalen Somatization Index to assess somatic symptoms.

Results: It took 42 days to recruit 35 participants, the drop-out rate was 20% and compliance to the diets was at least 68%. Participants were representative of the general Dutch population regarding age, BMI, marital and work status and symptoms of depression and anxiety. Participants showed greater reductions in psychological and somatic symptoms when consuming a Paleolithic-like diet compared to the Dutch consensus diet. Sample size calculations showed that at least 30 participants need to be recruited for statistical significance.

Conclusion: Conducting a web-based cross-over dietary intervention in the general population appeared feasible. Given the ease of increasing sample size in web-based studies, we recommend recruiting more participants than estimated when comparing a Paleolithic-like diet to the Dutch consensus diet.