Assessment of quality of information available over the internet about vegan diet

This paper aims to assess the quality of health information available to patients seeking online advice about the vegan diet.

A cross-sectional sample of patient-oriented websites was selected by searching for “Vegan diet” in the three most popular search engines. The first 50 websites from each search were examined. Quality of information was assessed using the DISCERN instrument, a questionnaire tool designed to judge the quality of written information on treatment choices. Readability was determined with the Flesch Reading Ease score (FRES) and Flesch–Kincaid Grade Level (FKGL). Relevance to health and disease was assessed by counting the appearances of ten related keywords, generated by searching the query term “Vegan diet” into PubMed and recording the top ten health-related words.

Of 150 websites retrieved, 67 (44.7 per cent) met inclusion criteria. Of these, 42 (62.7 per cent) were non-pharmaceutical commercial, 7 (10.4 per cent) institutional, 6 (9.0 per cent) magazines or newspapers, 4 (6.0 per cent) support websites, 4 (6.0 per cent) charitable websites, 2 (3.0 per cent) encyclopedias and 2 (3.0 per cent) personal blogs. The overall DISCERN rating of the websites was fair (mean 41.6 ± 15.4 on an 80-point scale), but nearly half (31/67) of the websites were assessed as having “poor” or “very poor” quality of information. FRES and FKGL readability indices met the recommended standards on average (means 63.3 ± 9.6 and 6.6 ± 1.7, respectively), but did not correlate with high DISCERN ratings. Analysis of variance on DISCERN scores (F(6,60) = 6.536, p < 0.001) and FRES (F(6,60) = 2.733, p = 0.021) yielded significant variation according to website source type.

Quality standards of health information available on the internet about the vegan diet vary greatly. Patients are at risk of exposure to low quality and potentially misleading information over the internet and should be consulting dietitians or physicians to avoid being misled.