An Application Of Hierarchical Bayes Models To Estimated Prevalence Of Mental Disorder And Mental Health Service Use Among Cook Islanders In New Zealand
The aims of this thesis are twofold. First is the epidemiological goal of reporting patterns and determinants of mental disorder among people of Cook Island ethnic group descent (Cook Islanders) compared to people from other ethnic groups living in New Zealand. In addition to prevalence of disorder, patterns and factors associated with use of treatment services for disorder are identified. The second goal is to develop hierarchical Bayes models based estimates of prevalence that show improved precision for a small sub-population such as Cook Islanders. Two sources of data are used. The New Zealand Mental Health survey (NZMHS), and the Mental Health Information National Collection (MHINC). Hierarchical Bayes models were developed to predict 12-month and lifetime prevalence as well as cumulative lifetime prevalence of mental disorders and service use from both data sets. These models are specified to adapt to complex survey design and question pathways of the NZMHS. The analysis of the NZMHS data, where a subset of people has been excluded from some questions, are analysed as if missing at random. The same principle is further applied to the analysis of MHINC data where the models are adapted to cope with missing ethnic group codes under missing at random assumptions. Bayesian survival models have been used to estimate the cumulative probability of onset of mental disorder and, in addition, that a person will seek treatment or recover without treatment with competing risks. The prevalence of any 12-month mental disorder is 31%, around 50% higher than those from non-Māori, non-Pacific (NMNP) ethnic groups. They have comparatively high prevalence of service use for mental health problems compared with people from other Pacific nations, but less than NMNP. Much of the increased risk of any mental disorder is explained by age and sex differences and further mediated by age at migration. However, the increased risk for some disorders such as mood, alcohol or serious disorders, remain, even after accounting for other factors. Evidence suggests that migrant Cook Islanders were less likely to have disorder while subsequent generation are at greater risk. Rather than reflecting a population with severe risk as a result of their ethnicity, descendants of Cook Islands appear to be a population that is more acculturated to New Zealand society than other Pacific peoples, inheriting not only its benefits but also the levels of mental disorders that accompany a comparative young population with low incomes and higher unemployment. Implications for public health research might involve investigating the potential for this group to return to their original native levels of mental disorder. Further, there is a resistance of those living in areas of high Pacific population density to use mental health services. This may represent an increased burden of care on other parts of the Cook Islands and wider Pacific community with disproportionately high use of mental health services as a result of an acute episode or an enforced intervention. The challenge for New Zealand’s government policy makers is to ensure the needs of those requiring treatment are appropriately met.
Advisor: Wells, Jessie Elisabeth; Graham, Patrick; Poulton, Richie
Degree Name: Doctor of Philosophy
Degree Discipline: Public Health, UOC
Publisher: University of Otago
Keywords: Hierarchical Bayes models; prevalence of mental disorders; Cook Islands and other Pacific; complex survey; competing risks; missing data
Research Type: Thesis