Although the DSM-5 was adopted in 2013, the validity of the new substance use disorder (SUD) diagnosis and craving criterion has not been investigated systematically across substances.

Adults (N = 588) who engaged in binge drinking or illicit drug use and endorsed at least one DSM-5 SUD criterion were included. DSM-5 SUD criteria were assessed for alcohol, tobacco, cannabis, cocaine, heroin, and opioids. Craving was considered positive if “wanted to use so badly that could not think of anything else” (severe craving) or “felt a very strong desire or urge to use” (moderate craving) was endorsed. Baseline information on substance-related variables and psychopathology was collected, and electronic daily assessment queried substance use for the following 90 days. For each substance, logistic regression estimated the association between craving and validators, i.e. variables expected to be related to craving/SUD, and whether association with the validators differed for DSM-5 SUD diagnosed with craving as a criterion v. without.

Across substances, craving was associated with most baseline validators (p values<0.05); neither moderate nor severe craving consistently showed greater associations. Baseline craving predicted subsequent use [odds ratios (OR): 4.2 (alcohol) – 234.3 (heroin); p's ⩽ 0.0001], with stronger associations for moderate than severe craving (p's < 0.05). Baseline DSM-5 SUD showed stronger associations with subsequent use when diagnosed with craving than without (p's < 0.05).

The DSM-5 craving criterion as operationalized in this study is valid. Including craving improves the validity of DSM-5 SUD diagnoses, and clinical relevance, since craving may cause impaired control over use and development and maintenance of SUD.