Method

There are difficulties in presenting information about overall prevalence. The first concerns the disorders to be covered. There is little dispute about the inclusion of common mental conditions such as anxiety and depressive disorder, and rarer but severe disorders like psychosis and bipolar disorder. Other disorders recognised in international classifications are also candidates for inclusion, for instance post-traumatic stress disorder (PTSD) and attention-deficit hyperactivity disorder (ADHD). Likewise, including dependence on illicit drugs and alcohol is unlikely to be contentious. However, in the analyses presented here, we excluded the hazardous use of alcohol and nicotine from analysis because, although clearly a behavioural problem and a social concern, it may not conform to lay ideas of mental disorder. The inclusion of hazardous use would of course result in higher values for the prevalence of mental disorder, to a degree that might be treated with suspicion by potential consumers of this information.

The new analyses presented here are based on the three most recent APM surveys, carried out in 2000, 2007 and 2014.^{Reference Singleton, Bumpstead, O'Brien, Lee and Meltzer2–Reference McManus, Bebbington, Jenkins and Brugha4} The two later surveys cover only England, and we therefore restrict the analyses of the 2000 survey to respondents from England. Structured interviews were used to provide a logical, valid and clinically meaningful classification, thereby formalising the process of clinical description and case identification. These measures were not adjusted to accommodate intervening changes in the dominant classificatory schemes^5,6 and are thus comparable across surveys. They are referenced and described in detail in the reports.^{Reference Singleton, Bumpstead, O'Brien, Lee and Meltzer2–Reference McManus, Bebbington, Jenkins and Brugha4} The values presented here are based on point prevalence for common mental disorders (CMDs) and a prevalence period of up to 1 year for the other disorders. Again, this follows the procedures adopted in the reports.

Comorbidity between psychiatric disorders is common, with the consequence that adding in a new category increases the overall percentage of people affected by less than the individual prevalence of that category. For this reason, we quantified people suffering from mental illness by incorporating disorders sequentially. Such a sequence must inevitably have an arbitrary element. We chose to start with psychotic disorder and CMDs, then moving on to include dependence disorders, as these were quantified identically in all three surveys. Disorders assessed only in the later surveys were included later in the sequence of analysis.

We first identified those suffering from probable psychosis. From the remaining group, we then added people with an identified CMD, in sequence according to their severity. Severity was estimated from their score on the instrument used to identify them, the revised Clinical Interview Schedule.^{Reference Lewis, Pelosi, Araya and Dunn7} Scores of 12–17 indicate a significant mental health problem requiring assessment, whereas scores of ≥18 imply treatment is almost certainly needed. From the remaining sample, we then incorporated people with signs of alcohol or drug dependence.^{Reference Saunders, Aasland, Babor, Dela Fuente and Grant8,Reference Drummond, McBride, Fear, Fuller, McManus, Bebbington, Jenkins and Brugha9} Next, we added those who screened positive for ADHD,^{Reference Kessler, Adler, Ames, Demler, Faraone and Hiripi10} and finally people screening positive for personality disorder,^{Reference Moran, Leese, Lee, Walters, Thornicroft and Mann11} PTSD^{Reference Blanchard, Jones-Alexander, Buckley and Forneris12} or bipolar disorder.^{Reference Hirschfeld, Williams, Spitzer, Calabrese, Flynn and Keck13} The ADHD screen was not used until the 2007 survey, and the personality disorder, PTSD and bipolar disorder screens were only introduced in 2014.