Dubious findings concerning the prevalence of late-life depression

Recently published estimates concerning the prevalence of late-life mood disorders among Australians living in the community (Sunderland et al., 2015) are questionable. The authors came to conclusions based on data obtained from persons aged 65–85 years (n = 1905) in the 2007 Australian National Survey of Mental Health and Wellbeing (MHWB; Slade et al., 2009). The 1-month prevalence of ‘any affective disorder’ was reported as 0.9% (0.9% major depression, 0.4% dysthymia). The 12-month prevalence was 2.4% (3% of the women, 1.7% of the men); bipolar disorder was reported in 0.1% of the men (none of the women). The 12-months affective disorders rate was higher among the 65–69 years age group (3.8%) than at age 70–85 years (around 1.7%), but anxiety disorders were decreasingly prevalent (5.6–1.6%) across the four age groups from 65–69 to 80–85 years. At age 75–85 years, the 12-month prevalence of any mental disorder (excluding cognitive impairment) was reported as 5%.

Sunderland et al. (2015) did not provide comparisons of the late-life figures with those recorded for younger people in the same survey. Slade et al. (2009) had reported that the 12-month prevalence of any mental disorder in the MHWB community sample of people aged 16–85 years (n = 8841) was 20%, and that of ‘any affective disorder’ was 6.2%.

The finding that mood disorder is less common in old age than earlier in life accords with evidence from several cross-age studies but not with all (Snowdon, 2003). Methodologies varied. The MHWB results concerning the prevalence of Diagnostic and Statistical Manual of Mental Disorders (DSM) major depression in old age contrasted with those from most community studies that confined their investigations to elderly populations and used age-appropriate tools (Snowdon, 2003). We acknowledge that DSM-defined major depressive episodes may be less prevalent in old age, and this may be partly attributable to the higher mortality rate of older people with depressive disorder (see Almeida et al., 2014). However, a chief reason why the MHWB studies found lower rates of depression in late life was that they used more restrictive definitions of affective disorders. Thus, depressive conditions associated with stroke, Parkinson’s or other age-associated medical illnesses, which have been shown to be as clinically serious and prevalent in late life as major depression, were not recorded by MHWB data-gatherers as affective disorders. When they noted pointers to organic pathology at interview, cases were reviewed with a psychiatrist to determine whether organic exclusion criteria were met (Slade et al., 2009).

Attention has been drawn to numerous flaws in the methodology of the 1997 MHWB study (Snowdon et al., 1998), and some of these extended to the 2007 survey (O’Connor and Parslow, 2009). Respondent selection processes tend to exclude older people who are frail, disabled, physically unwell or with difficulties in comprehension (McCaul et al., 2015). In addition, the prevalence of depressive disorders among those who decline invitations to complete health surveys is nearly twice as high as for those who agree to participate (Almeida et al., 2014). The World Health Organization’s structured Composite International Diagnostic Interview (CIDI) has been criticised for its under-identification of depressive disorders among elderly people. Because of difficulty in registering and comprehending complex or lengthy questions, they may deny having experienced symptoms. O’Connor and Parslow (2009) found that positive responses to CIDI depression screening items ranged from 11% in young adults to 3% in the oldest group, even though scores on two simpler distress scales differed little between different age groups. The CIDI needs validation among older people. Snowdon et al. (1998) also raised the possibility of ageism in the design of surveys.

Community residents with dementia were excluded from the MHWB studies. The 5.7% scoring <24 on the Mini-Mental State Examination were labelled as cognitively impaired, even if some fulfilled criteria for depression. Those living outside domestic environments (but not in hospital) were excluded, as were those who could not speak English. Sunderland et al. (2015) drew attention to various limitations of the 2007 MHWB study, including that the survey did not attempt to detect dementia. They referred to the response rate (60%) as sub-optimal. In 1997, the response rate was 78%. In neither study did investigators report response rate differences between age groups.

Because the MHWB study was of a non-representative community sample of older Australians (having excluded at least 10% because they were non-English-speaking, in residential care, dementing or too physically disabled to take part), because the response rate was low and because of methodological limitations, there is good reason to question its findings concerning the community prevalence of depression in Australia. We believe that 0.9% is a considerable underestimate of the point prevalence of late-life affective disorder, even if Australians in nursing homes are excluded. Sadly, bad and unfair policies may arise if the MHWB results are taken at face value.

See Research by Sunderland et al., 49(2): 145–155.