This review aims to summarise the evidence of diagnostic accuracy for screening and diagnostic tests to be used in populations with Substance Use Disorder (SUD) and Severe Mental Illness (SMI). The review includes studies which use a reference standard and research methods enabling the calculation of diagnostic accuracy. The evidence base is mainly single studies. This approach delivers a high level of accuracy and considerable precision in results, however it should be noted that some measurement tools which may be clinically relevant, may not be included in the review due to the research methodology employed.

Patients receiving treatment for SUD often have SMI, and patients recieving treatment for SMI often have SUD. Comorbidity between the two diagnoses presents a challenge to both diagnostic procedures and treatment.

We appraised 121 studies in full text and 19 with high or moderate methodological quality were included.

Screening tests for SUD for patients with SMI: According to the evidence, CAGE (Chemical Use, Abuse and Dependence Scale) is able to identify alcohol disorder (LR+ 13), both current and lifetime. The AUDIT (Alcohol Use Disorders Identification Test) reports an area under the curve ROC-kurven of 0.95. This means that the tests’ ability to identify patients with alcohol disorder and not identify those without, is high. The rest of the studies did not report likelihood ratio, or the likelihood was reported as low.

Screening tests for SMI for patients with SUD: The evidence base suggests that PDSQ (Psychiatric Diagnostic Screening Questionnaire) could identify mania well (SaR+ 21). For depression, psychotic disorders and personality disorders the evidence base did not report figures which could decide diagnostic accuracy.

Diagnostic tests The two tests C-DIS (Computerized Diagnostic Interview Schedule) and MMPI (Minnesota Multiphasic Personality Inventory) identified both showed low diagnostic accuracy (LR+ below 10) when compared to reference standard (CIDI, LEAD, MINI, SCID or thorough clinical evaluation). PRISM (Psychiatric Research Interview for Substance and Mental Disorders) shows good concordance with reference standard, reporting a kappa between 0.63 and 0.90 (usually considered good) for borderline personality disorders, alcohol dependence, heroin dependence and severe depression.