Improving mental health inequality? Some initial evidence from Australia

Abstract

This study statistically analyses the location and distribution of some aspects of mental health in Queensland. The health status measured here is not the conventional ‘count’ approach, i.e. counting people diagnosed with/without illness, but rather an approach that was developed in the context of mortality, by Silber and subsequently Le Grand. The present context is morbidity. We measure the years lived free of serious mental illness, for people who subsequently contract these illnesses. A complete enumeration dataset (i.e. not a sample), namely, the Queensland hospital admissions dataset commencing in 1964, is available. Specific illness codings are for conditions so consistently serious that hospital admission invariably occurred throughout the 40-year period. The present study is partial, but has two major advantages. First, the measure incorporates the notion that the later in life is the onset of serious mental illness, the better. Second, although age distribution measurement is not possible with ‘count’ data, it is possible with this approach. Inequality measures, such as the Gini coefficient, are applied to measure the inequality in the distribution. Time series on mental health and mental health inequality, for males and females, are thus generated and statistically analysed.

Acknowledgement

The authors thank the anonymous referees.

Notes

¹ Studies that are concerned with mortality focus on one point of the total product curve for ‘healthy time’, which may be conceived as Health Status_min, while a focus upon morbidity, as in the case here, involves other points on the total product curve. For an exposition in the present context, see Doessel and Williams (2005).

² The number of age groups in the data set occasionally varied throughout the 40-year period. Account was taken of this in the statistical modelling.

³ Such data sets do not, however, identify persons uniquely, and it is not possible to track individuals' lifetime experiences of mental morbidity subsequent to onset. Furthermore, the measure does not include all the people of Queensland who have no mental illness. In addition, no account is taken of re-admission. To take that factor into account requires data on cohorts of people (born in different years) that traces their experience through time. Such data are rare; only one such data set exists (Deaton, 2001), to the best of the authors' knowledge.

⁴ Lorenz Curves were also constructed and are available in Doessel and Williams (2005).

⁵ Having a near-zero value for ∊ reflects a view of social welfare that aversion to the inequality is low, whereas a relatively high value of ∊ reflects a society that has a high level of aversion to the inequality under study. The value of the parameter for aversion to mental illness inequality is not known empirically, either for Australia or internationally.

⁶ Details about the diagnostic testing are available from the authors on request.

⁷ Although, it is not our purpose to engage in causal, or explanatory, modelling of the phenomenon here, we generated various additional graphs, which are available from the authors on request, which provide some visual depictions of some of the factors that may be underlying the phenomenon under study here. These plots are: the ratio of numbers of admissions in each age group and the population in each age group per year, provided at 5 yearly intervals (a broad ?SMI incidence indicator'); and time series on the coefficients of variation and the Gini coefficients on the age distribution of the population.

⁸ The words here derive from those in the Gospel of Mark, chapter 14, verse 7.