Defining "caseness," that is, which individuals have the condition of interest, is extremely important in obtaining accurate estimates of prevalence and incidence in psychiatric comorbidity. The gold standard is the standardized psychiatric interview, which yields DSM IV diagnoses (Manchanda et al., 1996 ; Davies et al.,
2003 ; Tellez-Zenteno et al., 2005) . This is difficult to apply to large populations, and it is more often used in studies of selected populations. A commonly used method, attractive for its simplicity, relies on self-reported symptoms (Kobau et al.) 2006) obtained through validated screening questionnaires, for example the Beck Depression Inventory (Grabowska-Grzyb et al.) 2006) , the Hospital Anxiety and Depression Scale (HADS) (Mensah et al., 2006) and others (Strine et al., 2005) . These methods are less accurate, often overestimate the prevalence of psychiatric conditions, and should be explicitly considered as measures of symptom endorsement only. Not all psychiatric symptoms necessarily represent psychiatric disorders. The concept of a disorder requires symptoms to be present, but also imposes additional requirements such as persistence, and associated distress and dysfunction. Some studies have used non-standardized interviews to identify psychiatric disorders. This lack of standardization may under or overestimate prevalence depending on the type of interview, but it often yields higher estimates as compared with studies using standardized interviews (Gudmundsson, 1966) . Another method of ascertainment used in recent years relies on coded administrative data using the World Health Organization International Classification of Diseases (ICD) (Gaitatzis et al., 2004b). This method is very attractive because large databases already in existence can be readily accessed and analyzed. However, the sensitivity and specificity of ICD coding is highly variable among psychiatric conditions and also among different studies, due to variations in ICD coding practices.
Was this article helpful?