We have demonstrated the preliminary validity of an inductively derived questionnaire measuring Indigenous participation in caring for country activities and described a significant and substantial inverse association with BMI. We found that participation in caring for country activities was significantly associated with greater physical activity, less frequent consumption of takeaway and more frequent consumption of bush foods – health behaviours that contribute to less obesity . These findings are consistent with previous research documenting the health benefits of homelands residence  and reinvigoration of a 'traditional lifestyle' .
Consistent with recent findings in a comparable remote Indigenous community , mean BMI levels in this study indicate a lean population compared to Australia's national prevalence of 51% of overweight and obesity in adults (defined as BMI ≥ 25) . However, this does not imply less risk for development of diabetes and cardiovascular disease, because these diseases occur at lower BMI levels among Indigenous people, with risk increasing incrementally with rising BMI [33, 34, 49, 51].
Male and female participants reported different health behaviours, but similar associations with BMI. For men, health behaviours were associated with BMI as hypothesised. Unexpectedly, given the similar prevalence of smoking for men and non-pregnant women, for women, smoking did not demonstrate an independent relationship with BMI. This may be due to fewer numbers of cigarettes smoked each day (not measured in this study) but this requires examination in further work.
We investigated the reliability and validity of the questionnaire in a challenging setting and it demonstrated satisfactory internal consistency. Reliability was demonstrated through acceptable test-retest and proxy respondent agreement . Content validity was achieved through a two-year collaboration with key Indigenous and non-Indigenous informants from within the study setting. We could not pit our measure against an existing gold standard measure of Indigenous Caring for Country because no such measure is available; indeed, a strength of our research is the development of such a measure. In demonstrating moderate agreement among additional items assessing time on country, it shows concurrent validity. Construct validity was demonstrated by (i) exploratory factor analysis indicating a one-factor solution, consistent with the local Indigenous construct of an-ngurrunga-wana, (ii) a fitted one factor congeneric model, also consistent with our hypothesis and with the Yolngu health tri-partite of land, body and spirit, (iii) higher caring for country scores among expected groups, such as homelands residents, and (iv) the significant and substantial association of the scale score, in the expected directions, with key health behaviours and with our external reference, obesity.
The weighted scale score achieved stronger and more substantial associations with BMI than did the index, indicating that total quantity of participation is more important in achieving a lower BMI than is breadth of participation. This is consistent with the proposition that greater physical activity and a healthier diet, associated with caring for country practices, would deliver a more favourable metabolic state . However, breadth of participation in caring for country activities may be important in other socially mediated outcomes, as it is for mental health . This requires further research.
We observed a strong association in this study between residence in homelands and greater participation in caring for country. Residence also demonstrated significant independent associations with less frequent takeaway consumption, more frequent physical activity and more frequent consumption of bush foods – behaviours that would be expected to contribute to a lower BMI [29, 53]. Unexpectedly, however, residence was not a significant independent predictor of BMI in the final regression model. This may indicate that participation in caring for country activities mediates the relationship between residence and weight, suggesting that homeland residence is associated with lower weight because it engenders a healthier lifestyle. If so, this highlights the value of adequately resourced programs that support Indigenous ranger groups (predominantly based in township locations) and residents in homelands, both of whom maintain caring for country practices [23, 24].
Limitations of this study
We present four main limitations in this study. Firstly, as ours is a cross-sectional study, we are unable to determine the causal direction between caring for country and BMI. However, this was not an aim of our study, which was, instead, to validate a measure of an Indigenous asserted health promotion activity and to relate it to an external reference, obesity. Consistent with a longitudinal study of homelands residents in central Australia that observed significantly lower BMI over time, compared to township residents , our findings indicate that caring for country is associated with health behaviours that are likely to impede weight gain. Given the strength of our findings, a longitudinal study is merited.
Second, there may be a selection bias in this study. Volunteers for a preventive health check may not be representative of the population burden of morbidity as they tend to be more health-conscious . Additionally, those with established disease and receiving treatment may be less likely to participate. However, we purposively sampled just under a quarter of the eligible population, aged 15 – 54 years. This sample did not differ significantly from the census age profile . We also achieved a high questionnaire response rate. Further, if those with established disease or poor health were excluded, the results of this study would constitute (i) a conservative estimate of the health benefits of caring for country and (ii) increased probability that caring for country is linked to better health because those physically unable to care for country were excluded from this study. Finally, while a stratified random sample may have been a desirable alternative sampling strategy, this was impractical in the research setting due to (i) high population mobility, (ii) the absence of an accurate community population list and (iii) the need to obtain a much larger sample and collect data for a wider range of co-morbidities.
Third, several of our measures were crude, reliant on self-report and administered in English. Other measures were Eurocentric; for example, income did not include all forms of subsistence production , and education did not include traditional knowledge, which is equally important in Indigenous communities . While some self-reported health behaviours, such as dietary assessment, are notoriously inaccurate , we could not undertake objective measurement of all behaviours subject to the questionnaire items because participants were widely dispersed and may have found it intrusive. We expect this issue to arise for other research teams. To address it, we have tested the reproducibility of self-reported caring for country activities through test-retest, triangulation with proxy respondent rating and sophisticated statistical modelling. Our methods are consistent with and extend previous research in remote Indigenous communities that have used respondent rating to investigate health behaviours . Unfortunately, translation was not possible due to the lack of qualified interpreters for the eleven language groups. Nevertheless, our questions, piloted and refined with Indigenous health workers in preparation for the study, were considered comprehensible and in a suitable format for this population. Plausible associations between caring for country and health behaviours and obesity support this assessment.
Finally, this scale was developed in a single remote Indigenous community in Arnhem Land fifty years after the founding of the township. Other Indigenous communities with differing linguistic and cultural heritage, or even this community in coming years, may define caring for country differently. However, much of the ethnographic theory underpinning our scale development came from other NT Indigenous communities with longer periods cultural disruption [18, 19]. More broadly, the cultural expression, protection of the environment, healthier lifestyles and participation in society encompassed by the questionnaire items resonate with a Maori model of health promotion  and the health concepts outlined in the Geneva convention on the health and survival of Indigenous peoples .
We argue that the caring for country activities would be relevant to other remote Indigenous populations on their own land in remote areas of Australia. Indigenous Australians possess great diversity in linguistic and cultural traditions and the questionnaire requires further testing in different settings. However, the study cohort was generally representative of remote NT Indigenous peoples in terms of language diversity, residence patterns and a varied participation in customary and contemporary caring for country activities. Further studies are required in similar communities to test the generalisability of this questionnaire and investigate associations between caring for country and other health outcomes.