Interpretation of DSM-5 alcohol use disorder criteria in self-report surveys may change with age. A longitudinal analysis of young Swiss men

Background: General population surveys using self-reported measures show alcohol use disorder (AUD) to be most prevalent in young adulthood, but this may be due to misinterpretations of AUD criteria among inexperienced drinkers. The present study investigated changes in prevalence of criteria during emerging adulthood.Methods: 4924 young Swiss men were followed across three waves (age at wave 1 (w1): 20; w2: 21; w3: 25 years). We measured AUD according to the 11DSM-5 criteria and estimated Item Response Theory models for each wave and differential item functioning across waves, related to the cohort growing older.Results: Endorsement of several AUD criteria varied considerably as a function of age in a period of only five years: Five criteria showed differential item functioning between waves 1 and 3 (i.e. between the age of 20 and 25), including the three most frequently endorsed criteria. Prevalence of tolerance (w1, 57.8%; w3, 29.6%) decreased in relation to the AUD score (Mantel-Haenszel OR=0.26), whereas the use of alcohol larger/longer than intended (w1, 73.7%; w3, 84.8%; OR=1.93) and hazardous use (w1, 62.7%; w3, 68.4%; OR=1.31) increased, resulting in an unchanged mean AUD score and prevalence, but changing combinations of criteria to qualify AUD.Conclusions: Considerable differential item functioning over five years among several of the most endorsed AUD criteria suggests shifts in the interpretation of DSM-5 criteria during emerging adulthood. Self-reported measures of DSM-5 AUD criteria may require reformulation to account for young people's different perceptions and to yield comparable diagnoses over time and across age groups.