Landscape of clinical trial activity focusing on Indigenous health in Australia: an overview using clinical trial registry data from 2008-2018

We extracted all interventional studies registered on ANZCTR or CTgov from 1 November 2008 to 31 October 2018 that listed Australia as the only country of recruitment. Publicly available data on ANZCTR.​org.​au and Clinicaltrials.​gov were used to perform this study, where the combined and cleaned dataset could be supplied upon request. Trials were included based on their date of registration as commencement or completion date may not be reported. Multi-national trials were excluded due to concerns that they would have less proportional representation of Indigenous participants from Australia. The final set of sampled trials was termed ‘All-Australian trials’. Within this sample, we searched electronically for a subset of trials with an Indigenous focus, which was distinguished by a specific emphasis on Indigenous health, involving either a high percentage of Indigenous participants or Indigenous service providers, or with dedicated subgroup analyses for Indigenous Australians. We included trials that had terms such as ‘Indigenous’, ‘Aboriginal’ or ‘Torres Strait’ in the trial registration record’s ‘Public Title’, ‘Inclusion Criteria’, ‘Brief Summary’, ‘Description of intervention(s) / procedure’ and ‘Ethics committee name’ fields. To validate our scope, clan names from the Australian Standard Classification of Cultural and Ethnic Groups were searched in the sampled All-Australian trials [10]. Eligibility of the included Indigenous-focused trials were then assessed independently by two reviewers with high agreement (kappa = 0.88, 96% agreement). The extracted subset of trials was termed ‘Indigenous-Australian trials’ and the remaining subset was termed ‘General Australian trials’. To test the reliability of our Indigenous trial search strategy, 200 trials were randomly selected from all included trials and manually screened for eligibility in the Indigenous trial subset. No additional Indigenous-Australian trials were identified via manual screening that had not already been captured via our electronic search.

Trials in the Indigenous-Australian and General-Australian groups were compared in terms of numbers and characteristics, see Table 1. For our secondary objective, we compared the conditions studied in Indigenous-Australian trials against the top ten burden of disease conditions for Indigenous Australians using Australian Institute of Health and Welfare (AIHW) data [11]. Burden of disease for Indigenous Australians was measured by totalling Disability Adjusted Life Years (DALYs; the number of years lived with or lost due to a certain disease or injury) for each disease.

Data were analysed using the open-source software R 3.5.1 [13]. Key trial characteristics were compared between Indigenous trials and General trials by calculating proportions per category for binary and categorical measures, and medians with interquartile range for continuous measures. Within each category of comparison, percentage calculations were adjusted by the total number of trials, not the number of entries for conditions as each trial could list up to fourteen condition codes and up to three intervention codes. We derived odds ratios (OR) with 95% confidence intervals (CI) in a logistic regression analysis, p-values for categorical comparisons using χ² test and for nonparametric binary comparisons via Mann-Whitney U test. Co-variates were not adjusted since our aim was to provide a descriptive overview of Indigenous-Australian trials. The study methodology was designed in consultation with our Indigenous researcher and co-author. Ethical approval was not required as all trial data were publicly available and no human participants were recruited.

Our study provides only a descriptive analysis of Indigenous-Australian trials in the decade from 2008 to 2018 and should thus not be used to draw causal inferences between Indigenous research and Indigenous health. Second, our study only captured registered trials and whilst registration rates on ANZCTR have been reported as being as high as 95% of all trials conducted [6], we may have potentially missed locally-conducted studies that were not registered. Additionally, we included trials based on their year of registration, which may vary from the year of commencement or completion. This would also affect the actual participant sample size which may be missing on initial entry, or different to the target participant sample size on registration. It is important to note that Indigenous Australians were/are likely eligible for most trials classified as General, with an unclear participation rate due to no explicit data. It was also beyond the scope of our study to critically appraise the research outcomes identified from each trial, and their overall impact on general health and health service usage. Last, our study had a limited capacity to reflect the social, environmental and cultural complexities of Indigenous health research using the traditional quantitative research framework we employed. For the Indigenous population, primary healthcare centres are at the forefront of disease prevention and management [25]. Our analysis compared research categories to health priority areas that were determined by DALYs which may be incongruent with the health priorities determined by local Indigenous communities or primary care physicians caring for Indigenous people.