The cost-effectiveness of primary care for Indigenous Australians with diabetes living in remote Northern Territory communities

To the Editor: Thomas and colleagues1 attribute differences in hospital use among patients with diabetes to differences in use of primary care at remote state-run primary care clinics. We believe this conclusion to be unsupported by the evidence presented.

First, the study fails to accurately measure primary care use. Although study participants were restricted to those with remote residences, this ignores the mobility of such a population, and their subsequent access to primary care services not captured by the remote clinics' primary care information system. Additionally, there are 26 Aboriginal community controlled health services, many of which have primary or satellite locations in remote areas.

Second, there are other known associations, unadjusted for in this study, that may explain differing hospitalisation rates in low and high users of the remote clinics. These include social acceptability,2 socioeconomics,3 behaviour of health care providers and patients4 and access to health services. It is known that access to hospital drives hospital use.5 Differences in social norms between the two groups may drive different choices in site of health care delivery. Stigmatised health problems of a social or spiritual nature may affect a patient's willingness to see health workers from within their community.

Without adjusting for these variables, differences in hospital use cannot be simply attributed to differences in the use of (some) primary care services. Without such attribution, no realistic cost-effectiveness analysis can be undertaken. This study should not be used to guide policy or planning.