Professor Snowdon questions the validity of the results on the elderly from the Australian National Survey of Mental Health and Well-Being. He rightly points out that the survey failed adequately to cover the population living in institutional care, which was 9% of Australians over 65 in 1998. This is clearly acknowledged in earlier publications. Indeed, the indigenous people of Australia, people in prison, the homeless, the armed forces and the migrant population were also not included in numbers large enough to give stable prevalence estimates, mainly for the sake of economy in what was already a very large undertaking. We used “an unweighted sample with no group represented in a proportion greater than its frequency in the population” (Reference Henderson, Andrews and HallHenderson et al, 2000).

The lack of information concerning the 22% non-responders is indeed regrettable, but does not detract from the finding, consistent with many of the studies cited in Jorm (Reference Jorm2000), that the community-dwelling elderly displayed significantly lower levels of depressive symptomatology than younger cohorts. This was reflected in the prevalence rates and in all the scales of psychological distress measured in the survey: the 12-item General Health Questionnaire (GHQ-12), the 12-item Short-Form General Health Survey (SF-12), the Kessler-10 scale, the CIDI screen items for depression and finally the neuroticism items from the Eysenck personality questionnaire considered to reflect vulnerability to psychological symptoms. Each of these scales handles symptoms associated with physical disability in a different way. In all cases, the lower levels of symptomatology observed among 65-to 70-year-olds were maintained into the oldest age group (75 years and above), although the pattern is less stable than for younger age groups because of smaller numbers. The interested reader is referred to Jorm (Reference Jorm2000) for a discussion of the possible mechanisms involved.

Information on mental disorders among the oldest old and institutional elderly are of crucial importance for advocacy. But this needs to be addressed in ways other than in large community surveys. This was made explicit from the beginning, where we stated that information on “some of the most significant elements in our society” would need special studies (Reference Henderson, Andrews and HallHenderson et al, 2000). Any concern that our findings might affect decisions about allocation of resources is unlikely to be justified, because it assumes that administrators and policy-makers will make the grave error of extrapolating from community estimates to the special population of the elderly in hostels and nursing homes.