**5. Limitations**

Some limitations should be underscored. The retrospective design and data gathering through clinical charts entail some limitations. Some information could not be retrieved, for example, detailed descriptions of type of self-harm or aggressive behavior. Psychiatric diagnoses were grouped into broad categories (affective disorders, schizophrenia and other psychosis, personality disorders, anxiety disorders, and others), and we did not discriminate between bipolar and unipolar affective disorders. We did not include data about laboratory tests objectively detecting drugs; nonetheless, it has been suggested that a urine drug screening can only identify a small additional rate (5%) of substance users [52]. Although it would have been interesting to assess details about reason for "acute" inpatient psychiatric admission, in this study, we focused specifically on a "snapshot" of comorbidity in a psychiatric ward over a 10 year period. Last, we cannot exclude that our results might have been influenced by broader systemic differences in the treatment of the DD population across time. Anyway, in our country, in the study period, there have neither been relevant changes in treatment options available for DD patients, not in the legal policies about drugs.
