Correction to ‘Prevalence of food addiction determined by the Yale Food Addiction Scale and associated factors: A systematic review with meta-analysis’

Praxedes DRS, Silva-Júnior AE, Macena ML, Oliveira AD, Cardoso KS, Nunes LO, Monteiro MB, Melo ISV, Gearhardt AN, Bueno NB. Prevalence of food addiction determined by the Yale Food Addiction Scale and associated factors: A systematic review with meta-analysis. Eur Eat Disord Rev. 2022; 30(2): 85–95. https://doi.org/10.1002/erv.2878

A letter by Dr. Meule (2023) indicated that some results in our paper regarding the prevalence of food addiction (FA) in samples with eating disorders are potentially wrong. Indeed, after a revision of our spreadsheet we were able to identify the mistake in the analysis involving studies that had subsamples with eating and weight disorders. Our mistake was that we had included in the analysis studies with subsamples with different conditions (e.g. individuals with binge eating, bulimia, anorexia) but the prevalence used to run the analysis was the prevalence of the whole sample from each study (and not from each subsample). Hence, some pooled prevalence arising from the analysis of these subsamples is wrong, and usually underestimated.

We undertook an extensive review in every single included study and conducted a new analysis of the data. Prevalence of FA in anorexia, binge eating and bulimia were wrong in our original paper and underestimated. Prevalence of FA in anorexia increased from 44% to 55%, in binge eating from 55% to 63%, and in bulimia from 48% to 84%. Regarding the prevalence of FA in other subsamples, especially in those with weight disorders, there were only minor changes: the prevalence of FA in obesity increased from 28% to 30% and in bariatric surgery from 29% to 31%. As a result of such increases, the prevalence of FA in ‘clinical samples’ as a whole also increased from 31% to 40%. The prevalence of FA in non-clinical samples was maintained (14%). All these findings may be seen in the new table below:

After the commentary by Dr. Meule and the new analysis, we decided to present the pooled prevalence of eating disorders without stratifying by clinical and non-clinical samples, since there were almost no studies with ‘non-clinical’ diagnosis of the eating disorders. Hence, the last paragraph of the paper in the 2.3 section ‘Data extraction’, readers should disregard the stratification of eating disorders studies in samples with clinical and non-clinical diagnosis.

Hence, some findings of our original publication are invalid: (a) the overall pooled prevalence in all studies is 24% and not 20%; (b) the prevalence in clinical samples is 40% and not 31%; (c) bulimia is the eating disorder with higher prevalence of FA (84%) followed by binge eating (63%) and anorexia (53%). On the other hand, some other findings are maintained, especially the pooled prevalence in non-clinical samples which was 14%.

We also have uploaded a new spreadsheet with all data used to generate the new analysis and to replace the supplementary file that was uploaded with the original publication.

We would like to acknowledge Dr. Adrian Meule for his important inputs and to apologise for our mistake.