Long-term follow-up study of low-weight avoidant restrictive food intake disorder and childhood-onset anorexia nervosa: Comparison of autistic eating behaviours

Background There is growing evidence of a link between anorexia nervosa (AN) and autism, but little research exists on the long term outcome of autism in AN. In avoidant/restrictive food intake disorders (ARFID), assumptions regarding links with autism seem warranted but there is as yet a scarcity of empirical support and long-term follow-up. This study compares long term outcome between childhood-onset anorexia nervosa (AN) and low-weight avoidant/restrictive food intake disorder (ARFID) in regard to autistic eating behaviours. Would an ARFID-group display more autistic eating behaviours than an AN-group? Methods 19 patients with ARFID and 37 patients with AN were followed-up after a mean of 15.9 years. Autistic eating behaviours were assessed using the SWedish Eating Assessment for Autism disorders questionnaire (SWEAA). Morgan-Russell Outcome Assessment Schedule and EDE-Q were also completed. SCID-interview was performed. Weight and height were recorderd. ARFID-diagnoses were assigned retrospectively using all available information. Mann-Whitney U-test, Kruskal-Wallis analysis and linear regression were used to analyse outcome, alongside descriptive statistics. Results The AN and ARFID-groups both had elevated SWEAA-scores, near the clinical cut-off for SWEAA. On-going Eating Disorder (ED) had the highest SWEAA-scores, above cut off, having Another Psychiatric Diagnosis (but no ED) had intermediate scores (below cut-off) and No Diagnosis had the lowest scores, for both AN and ARFID-groups. SWEAA-outcome by BMI-category, showed lowest SWEAA-scores for BMI < 20 (below cut-off) for both AN and ARFID, the BMI 20–23 category had high SWEAA scores for AN, whereas the BMI > 23 category had high SWEAA-scores for ARFID. ARFID showed a positive linear relationship between BMI and SWEAA-score. Morgan-Russell Outcome Assessment Schedule results differentiated between the diagnostic categories at follow-up (eating disorder, other diagnosis, no diagnosis), lending validity to the follow-up procedure. Conclusions Autistic eating behaviours were elevated in both AN and ARFID-groups. Overall, autistic eating behaviours were not more pronounced in the ARFID-group. Low BMI had fewer autistic eating behaviours in both AN and ARFID. High SWEAA-score indicated poor prognosis. The study provides further support for a link between autistic eating behaviours in both AN and ARFID.