Health care service use by people diagnosed with invasive melanoma in Queensland: a benefit incidence analysis

In reply: We thank Bailie and colleagues1 for their commentary on our article.2 Our aim was not to undermine the well documented inequities in access to health care experienced by people with lower socio‐economic status and those living rurally. We reported significant disparities in health care use in our article, with socio‐economically advantaged individuals using more specialist services, and those living more remotely using more hospital‐based health care, mirroring Australian trends.3

We minimised using the term “equity” in our article (and did not mention “inequality” at all), as we focused on variations in health care use, which may arise from inequities, although one can never be sure without access to more detailed data. It is worth noting our use of the term “equity” in our original submission was downplayed at the request of reviewers. Given our results, instead of concluding that “… access to health care for people with melanoma is fairly equitable in Queensland”, we may have more accurately concluded that health care use is fairly equitable across these factors.

The aim of our article differed from the re‐analyses presented, as we focused on variation in health care use based on socio‐economic status and rurality, and did not explore the cancer stage data. An assumption is made in the re‐analyses that differences in cancer stage prevalence across the socio‐economic groups indicate poorer access to care, which cannot be definitively concluded from the data available.

Opinions will differ on whether, and when, to use corrections to reduce the risk of type I error. On reviewer advice, we corrected for multiple comparisons in our analyses, as, a priori, we were agnostic as to how the use of the 12 health care services would vary based on socio‐economic status or rurality.4 We agree that removing the correction post hoc will inevitably increase the likelihood that any particular association is deemed significant. However, applying the Bonferroni correction reduced the likelihood of ascribing significance to associations reflecting the play of chance.

We readily acknowledge the limitations of using postcode to classify socio‐economic status and rurality measures; indeed, we stated in our article that “levels of remoteness and socio‐economic disadvantage can vary within a postcode”. We removed people who died within three years in our sample because we wanted to avoid quantifying elevated health care use at the end of life.5 We used the Australian Statistical Geography Standard – Remoteness Area (ASGS‐RA) to classify remoteness based on a recent systematic review that found this was the most commonly used spatial measure to classify remoteness in Australia; the Modified Monash Model was not used by any articles in that review.6

Melanoma remains a major health problem in Australia. Equitable access to health services is essential to ensure all Australians benefit from high quality care. We share our correspondents’ desire to contribute to such care through the careful analysis of data.