Abstract
Australian and New Zealand researchers have engaged with the challenge of estimating population rates of mental disorders over the last 20 years, aligned with the World Health Organization’s World Mental Health (WMH) Survey Initiative (see www.hcp.med.harvard.edu/wmh). One outcome of this work has been the provision of estimates of how these rates may differ between age-group cohorts. The Australian National Mental Health and Well-Being surveys of 1997 and 2007 have both generated publications in respect of the older people surveyed (Sunderland et al., 2015; Troller et al., 2007, respectively). Te Rau Hinengaro, the New Zealand Mental Health Survey of 2003/2004, has also yielded prevalence estimates for different age-bands (Oakley Browne et al., 2006). All three studies have yielded the same finding: older people surveyed with these methods are detected as having very significantly lower lifetime, 12-month and 1-month prevalence figures than younger adults for all Axis 1 Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) disorders, excluding neurocognitive disorders (see Table 1). This is a surprising result for clinicians.
Prevalence estimates (% of cohort) for selected DSM-IV Axis I classes across national mental health surveys for 65- to 84-year-olds and younger NZ comparator group.
DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; n/r: not reported; NZ: New Zealand.
Practically all major depressive episodes (MDEs), especially for older people.
Practically all alcohol use disorders, especially for older people.
While broadly similar, Te Rau Hinengaro data in comparison with the Australian surveys paint a picture of a slightly more anxious Kiwi cohort (14.2% lifetime prevalence vs 10.2% in Australia in 2007, and 6.0% 12-month prevalence vs 4.1% in Australia in 2007, with most of the difference coming from increased rates of generalized anxiety disorder [GAD] and posttraumatic stress disorder [PTSD], and only obsessive compulsive disorder [OCD] bucking this trend with much higher Australian rates), very much less alcohol use disorder detected in New Zealand (4.7% lifetime prevalence vs 11.7% in Australia 2007, and 0.1% 12-month prevalence vs 0.7% in Australia 2007) and less trans-Tasman difference overall for mood disorders and for any mental disorder (see Table 1). New Zealand data for 25- to 45-year-olds demonstrate the order of difference in mental disorder rates compared with their elders, with lifetime prevalence for the younger group being around 1.5–2 times higher for non-alcohol-related problems and four times higher for alcohol, and 12-month prevalence being around 3–4.5 times higher for non-alcohol-related problems and 50 times higher for alcohol.
The question of whether or not these estimates can possibly be accurate is just as salient in New Zealand as it is across the Tasman, yet there are as yet no good New Zealand data to bring to bear to address it. Sunderland et al. (2015) and the key critics (O’Connor and Parslow, 2009; Snowdon et al., 2015) agree on the potential problems that might deliver falsely low estimates of psychopathology for our older citizens, and in respect of the lifetime data at least, Sunderland and colleagues are appropriately wary of their findings. There are three main potential sources of bias: older people who chose not to participate may have significantly higher rates of psychopathology than the surveyed group, older people excluded from the sample due to the method used may have significantly higher rates than their peers who were included, and the instruments used to detect mental health problems in the surveys may be more likely to fail to detect cases in the older participants than in younger people. These arguments have been well-rehearsed by these authors and appear to have face validity, although whether or not the very large age-band rate differences the surveys report can be explained in their entirety by these factors remains unknown. It seems reasonable to propose that simply repeating this kind of WMH-compatible methodology in either country is unlikely to shed light on these questions, with the exception of looking at recall bias by cohort.
The New Zealand methodology is broadly similar to the Australian surveys in that it rests upon a face-to-face interview using the Composite International Diagnostic Interview with a household-based sampling frame. Between a quarter and one-third of eligible participants declined to participate in each of the surveys (26.7% refused in the 2003/2004 New Zealand process), and no data have been provided in any of the survey publications showing how age was associated with this. People living in any kind of institution, people unable to be interviewed in English, the oldest old (people over 85 years of age) and people with significant cognitive or physical disabilities were excluded, most notably people with neurodegenerative disorders. The proportion of the eligible elderly population excluded is therefore very likely to be high. Of concern, these excluded people are most likely to have higher rates of mental health problems than the general population of elders, not least because the primary influence of age on risk of neurodegenerative disorder is beyond question.
As an example of how a single factor from this list can exert a relatively large effect, consider the exclusion of people in aged residential care (ARC). In Canterbury, a health region of New Zealand with one of the largest numbers of people aged 65 years and over, the proportion of this population living in ARC in 2013 was estimated to be 7.6% out of 72,000 (Keightley and Brailsford, 2015). A calculation using these figures and the 7.1% 12-month estimate for mental disorder shows that each additional 13% excess in the true ARC psychopathology rate over this non-institutional estimate could increase the total community rate by 1 percentage point. Moreover, it can be seen that the contention that at least 10% of the eligible population may have been excluded from these studies is not far-fetched (Snowdon et al., 2015), especially considering the somewhat overlapping cohort of people aged over 85 years, which in the same year in Canterbury was measured at 12.5% of the 65+ population. Maori, Pasifika, South East Asian and other ethnic group under-representation is likely to give rise to another important missing group as are all people living with neurodegenerative disorders. These are our most psychiatrically vulnerable older citizens, and health services have a duty to engage with them, arguably a special duty.
Therefore, let our Ministries of Health and academics get serious about measuring the burden of mental disorder for all older people in our communities using age-appropriate methods; let our Service Planners routinely account for the oldest old, people of all ethnicities and people with psychiatric complications of neurocognitive disorders when we plan older persons mental health services; and let clinicians and the community itself continue to advocate in the meantime for services for older citizens based on the measures we do have in respect of local community need rather than relying on population-based estimates from WMH-related surveys.
See Research by Sunderland et al., 49(2): 145–155.
Footnotes
Declaration of interest
The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.
Funding
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.