A further advantage of the 1995 NHS data is the capacity they provide to extend the analysis of the relationship between income status and expenditure, and to explore the links between these factors and health status for both the Indigenous and non-Indigenous populations. While such analysis is desirable, legitimate concerns surround the extent to which a suitable measure of health status is available from the NHS to enable meaningful comparison between Indigenous and non-Indigenous populations. In effect, the exclusion of sparsely settled areas from the NHS sample partly resolves this issue, as argued below.

This same dilemma regarding the potential usefulness of self-assessed health status was raised and extensively investigated subsequent to the release of results from the 1994 NATSIS (Cunningham, Sibthorpe & Anderson 1997). This survey (as well as the 1995 NHS) asked a global question on self-assessed health status as follows:

Given objective evidence of higher Indigenous morbidity and mortality, an apparent similarity in self-reported rates of poor to fair health among the Indigenous and non-Indigenous respondents in the NATSIS (around 17 per cent each) suggests that differential thresholds for reporting poor or fair health were being applied. If this were so, it would potentially undermine the utility of data on self-assessed health status as a proxy for comparison of health needs between groups. However, as Cunningham, Sibthorpe & Anderson (1997: 26) have pointed out, expected differences in self-assessed health status were evident between the two populations after accounting for age structure. This variation was sufficient to suggest that the limitations of the data did not override their utility for comparative purposes. Further evidence from the NATSIS also supports the utility of responses to the global question on self-assessed health status. This is drawn from the fact that individuals who reported that they had a long-term health condition were significantly more likely to report poor or fair health than those who indicated that they had no long-term condition (Cunningham, Sibthorpe & Anderson 1997: 18).

While, in principle, the utility of the global question may be accepted, one concern remains to be overcome. This is based on the observations that the level of reported poor or fair health in the NATSIS was markedly lower for people who indicated that they did not speak English as their main language, and that other estimates for this group also displayed a large degree of response error (Cunningham, Sibthorpe & Anderson 1997: 19–21). Similar error among Indigenous respondents whose main language was not English was found in the ABS evaluation of Indigenous data quality issues in the 1995 NHS (Gray 1997). Indeed, it was concern over data quality for this group that led to their exclusion from the calculation of final published estimates. Thus, by focusing the sample on respondents from non-sparsely settled areas, residual doubt about the utility of the global question on self-assessed health status is largely overcome.