Inter-observer reliability of DSM-5 substance use disorders.

Area of Science:

• Psychiatry
• Psychology
• Neuroscience

Background:

• The Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) introduced changes to substance use disorder (SUD) diagnoses.
• Limited evidence exists on the inter-observer reliability of DSM-5 SUD diagnoses.
• Previous research evaluated DSM-IV SUD reliability in the same sample.

Purpose of the Study:

• To evaluate the inter-observer reliability of four DSM-5 SUD diagnoses: alcohol, opioid, cocaine, and cannabis use disorders.
• To compare the reliability of DSM-5 SUD diagnoses with DSM-IV SUD diagnoses.
• To assess the impact of comorbid SUDs and mood/anxiety disorders on DSM-5 SUD reliability.

Main Methods:

• Two independent interviewers assessed 173 subjects using the Semi-Structured Assessment for Drug Dependence and Alcoholism (SSADDA).
• Reliability was measured using percent agreement and kappa (κ) coefficients for DSM-5 SUDs.
• DSM-5 SUD reliability was compared to SSADDA-derived DSM-IV SUD diagnoses.

Main Results:

• Good to excellent inter-observer reliability (κ = 0.65–0.94) was found for the four DSM-5 SUDs.
• Reliability was lower for mild severity SUDs compared to moderate or severe disorders.
• DSM-5 SUD diagnoses exhibited higher reliability than DSM-IV abuse/dependence diagnoses.

Conclusions:

• DSM-5 criteria for alcohol, opioid, cocaine, and cannabis use disorders are at least as reliable as DSM-IV criteria.
• Comorbid SUDs and mood/anxiety disorders had a modest impact on DSM-5 SUD reliability.