Australian Institute of Criminology

Skip to content


The mental health measures detailed above were trialled during the first quarterly data collection of 2010. They were incorporated into the DUMA survey as a set of addendum questions. Data were collected at the Queensland (Southport and Brisbane), New South Wales (Bankstown and Kings Cross), Western Australia (East Perth) and Northern Territory (Darwin) DUMA sites. Details of the methodology and data have been published elsewhere (Forsythe & Gaffney 2012) and as such, this report will focus on conceptual and pragmatic issues.

It became apparent during data collection that the Indigenous people who constituted the majority of Darwin detainees either did not understand the mental health questions or found them inappropriate. The Northern Territory DUMA Site Manager reported that the Indigenous detainees at the Darwin site tended to have English as their second or subsequent language and limited English literacy. In addition to the actual language limitations, the terms mental health and diagnosed seemed conceptually unfamiliar to the majority of Indigenous detainees. These issues resulted in a large amount of missing data and questions about the validity of the data recorded. Given these quality issues, the data collected at the Darwin site were excluded from analysis (Forsythe & Gaffney 2012). This experience highlights the importance of developing culturally valid mental health measures for Indigenous people and in hindsight, it is clear that Indigenous expertise is vital to such a process (Vicary & Westerman 2004).

The CMHS (see Appendix 1) was completed by 690 detainees. Interviewers reported that the questions were quite long and some detainees required the questions to be repeated. Overall however, the questions appeared to be understood and were readily answered by detainees. The CMHS was easily scored—‘yes’ responses being given a score of one (1). A cut-off of five or more ‘yes’ responses for men and four or more ‘yes’ responses for women was used to determine detainees who screened in, with 46 percent of male detainees and 64 percent of female detainees subsequently qualifying as ‘screened in’ (see Forsythe & Gaffney 2012 for more details).

The authors of the CMHS provide supplemental tables detailing the predictive utility of a variety of cut-off points to allow those using the tool to set the cut-off points at the most appropriate level for their purpose (Ford et al. 2009). For example, if using the CMHS as a clinical tool to identify detainees to be referred for detailed psychiatric assessment, the cut-off may be set at a lower level. This may mean some false positives screen in; however, if the aim is to maximise the chance of mentally ill people being referred and minimise the chance of mentally ill people not being referred then this would be appropriate. Alternatively, if the CMHS is to be utilised by researchers to estimate the prevalence of mental disorders, then higher cut-off may be used in order to avoid inflated estimates. Overall, the CMHS appeared to perform well and it would be useful to conduct a validation study in order to test whether it performs as accurately in the Australian context as it has in the United States.

The purpose-built question asking whether a detainee had ever been diagnosed with a mental health problem by a doctor, psychiatrist, psychologist or nurse appeared at face value to be understood by detainees. Six hundred and eighty-seven detainees answered the free recall question—281 said they had been diagnosed with a mental health problem while 406 detainees indicated no previous diagnosis had been given.

When cued with the list of mental disorders, 29 detainees who originally indicated that they had not been diagnosed recognised at least one disorder they had been diagnosed with. Therefore, cueing increased the number of detainees who indicated that they had been diagnosed with a mental health problem from 281 to 310.

Of the 29 detainees who were reminded by cueing that they had been diagnosed with a mental health problem, seven reported diagnosis of a learning disorder, eight reported diagnosis of attention deficit or other behavioural disorder, six reported substance-related disorder, four reported mood disorder, three reported eating disorder, two reported sleeping disorder, two reported anxiety, and one each adjustment disorder and psychotic disorder (some respondents reported more than 1 diagnosis).

Of the 281 detainees who reported they had been diagnosed with a mental health problem in response to the free recall question, nine were not able to recall what disorder they had been diagnosed with, even with the aid of cueing.

Table 1 shows a comparison of diagnoses reported in response to the free and cued recall response versions of the question. Up to three responses could be recorded.

Table 1: Mental disorders reported by detainees—free recall versus free plus cued recall
Diagnostic category Free recall only Free plus cued recall
Learning disorders 2 34
ADHD and behavioural disorders 37 71
Substance-related disorders 3 39
Schizophrenia and other psychotic disorders 21 36
Mood disorders 185 226
Anxiety disorders 40 103
Sleep disorders 3 33
Personality disorders 9 17
Other disorders 0 22
Total diagnoses 300 581

Source: AIC DUMA Q1 2010 [computer file]

It is hypothesised that learning, behavioural and substance-related disorders may not have been recognised by some detainees as mental health problems and therefore not mentioned in response to the free recall version of the question.

This trial suggests that cueing responses served two functions; first, it defined what was meant by the term mental health problems and therefore increased the number of detainees who reported having been diagnosed and second, it appeared to remind detainees of additional diagnoses. The data generated by the free recall version of the question would appear to suggest that most detainees who experienced a mental illness were given a single diagnosis. However, the cued recall data is more suggestive of a high prevalence of comorbidity. This suggests that if the information sought relates to the types and number of diagnoses, then the cued response version of the question is more likely to yield this information. If the aim is simply to identify respondents who have been diagnosed, then the free recall question is adequate.

The final question on this trial asked detainees who had reported being diagnosed to report how old they had been the first time they were diagnosed with a mental health problem. This information was found to be of very limited use as there are some disorders that by definition can only be diagnosed in childhood (eg some developmental and behavioural disorders). In hindsight, if age at diagnosis is of research interest, better data would be generated if age diagnosed was recorded for each diagnostic category.

Related links

Measuring mental health in criminology research: Lessons from the Drug Use Monitoring in Australia program: