Aboriginal and Torres Strait Islander (hereafter, respectfully Aboriginal) people, the First Nations peoples of Australia, comprise 3.8% of the total Australian population.¹ Aboriginal Australians live in varying geographical contexts and are culturally, linguistically and socio-economically diverse, contributing richly to Australian life. The health of Aboriginal children is a priority for Aboriginal communities, families, peak bodies and governments,² with the early years being particularly important. However, aggregate national statistics consistently reflect an excess disease burden for Aboriginal children compared with other Australian children.³ The focus in the literature has been primarily on describing deficits rather than on trying to understand them, such that the complex pathways of Aboriginal paediatric health disadvantage are often poorly understood.

For many Aboriginal parents and carers, hospitals are alienating environments, reminiscent of a traumatic history of colonisation in which children were forcibly removed and Aboriginal families treated as second-class citizens.^{4, 5} To maximise the health and well-being of Aboriginal children, it is important to recognise that the way they and their families interact with the health system today is often influenced by these past (and sometimes ongoing) experiences.

Discharge against medical advice (DAMA) occurs when patients (parent or carer when the patient is a child) leave an in-patient facility or emergency department before being formally discharged by the clinical team; this can put patients at an increased risk of readmission or poor health outcome.⁶ Complex person, hospital and contextual factors contribute to DAMA. In Australia, DAMA occurs more frequently among Aboriginal patients, with DAMA rates considered a measure of the responsiveness of hospitals to the needs of Aboriginal patients. This includes the level of cultural security felt by patients in the way they are treated, and how their needs are met or not met in the system.⁶ Australian hospitals are regularly audited regarding DAMA rates. To date, most existing research focuses on adults.

In the current issue of Paediatric and Perinatal Epidemiology, Christensen and colleagues⁷ leverage the linked Western Australian (WA) administrative data from their ‘Defying the Odds’ study. They investigated associations between DAMA in inpatient and emergency department (ED) settings and child, family, and episode-of-service characteristics in a population-based birth cohort of Aboriginal children <5 years. State-wide ED DAMA (predominantly ‘did-not-waits’) and hospital DAMA were analysed separately, a prudent decision given the mostly contrasting results.

Unsurprisingly, DAMA was more common in ED (4.7%) than in inpatient settings (1.6%). After adjustment for pertinent covariates, perinatal and maternal factors generally were not associated with DAMA during hospital episodes—although DAMA risk was borderline increased among those with birthweight <2500 g (adjusted odds ratio [aOR] 1.31, 95% confidence interval [CI] 1.00, 1.72). In contrast, ED DAMA was inversely associated with multiple births (aOR 0.80) and three or more previous births (aOR 0.92), while maternal age at birth <20 years was associated with increased DAMA risk (aOR 1.14). Thus, ED DAMA related more to mothers and their circumstances.

Children under 1 year at admission were more likely (aOR 1.24) than those 1–4 years to DAMA from the hospital but less likely to DAMA from ED (aOR 0.74). As expected, those with emergency admissions were 3.4 times more likely to hospital DAMA than elective admissions from waitlists, and a gradient was seen regarding the urgency of the ED triage category, with aORs of about 11 for semi- and non-urgent versus urgent presentations.

Geographical context was shown to be an important determinant of DAMA. This was no surprise: WA covers a vast and varying geographical area, with respect to accessibility to services, population density, climate, and socio-economic status. For hospitalisations, all levels of non-metropolitan areas of residence at birth had an increased risk of DAMA. There was evidence of a gradient effect with increasing remoteness such that children residing in very remote areas had 4.7 times the risk of DAMA compared with metropolitan Perth. This was also true of hospital location (rural hospitals aOR 6.99). However, for ED presentations, a very remote area of residence was associated with 40% lower odds of DAMA – with no associations with other levels of remoteness. Rural EDs had 12% lower odds (aOR 0.88), mainly reflecting limited primary healthcare options for very remote-residing children.

An important finding was the decrease in hospitalisation DAMA during the study period, but an increase in ED DAMA. This suggests that in-patient systems for Aboriginal children have substantially improved. This may be due to the introduction of mandatory online cultural security training for staff across the WA Health system⁸ accompanied by focused strategies and monitoring to reduce DAMA rates. These same strategies do not seem to have been sufficient to impact ED rates. Indeed, the increase in ED DAMA over time suggests broader contextual factors outside of the hospital system, in particular, reduced access to primary health care with consequent increases in ED presentations by 3.2% per annum before the COVID-19 pandemic in 2019–2020.⁹ Many changes over the study period, including reduced after-hour GP services, absence of after-hours services in Aboriginal Medical Services, increased service costs resulting in increased out-of-pocket expenses (due to reduced ‘bulk-billing’ by general practices where fees are completely government subsidised) and increased living costs.¹⁰

Does it matter if patients DAMA? With the focus on the adjusted results only, the paper does not adequately emphasise that the unadjusted odds of readmission among hospital DAMA is 40% higher than other patients—even the adjusted estimates suggest increased odds. This means that all things being equal, hospital DAMA is worthy of concern—even if a range of factors can explain the DAMA. The reduced unadjusted odds of readmission in ED DAMA supports the assertion that those who DAMA in ED could often be adequately treated in primary care.

This study was technically exemplary. A comprehensive and unbiased cohort was achieved using a population-based birth cohort covering an entire State, with multiple data sources to provide rich contextual information. Indigenous identification was rigorous using family relational data in addition to best-practice algorithms. Such methods of identification are rarely possible in most other countries with large minority Indigenous populations. The focus on explanatory models allowed within-Aboriginal comparisons to identify strengths and risks. The complex statistical analysis also considered the nested data structure—multiple levels of service episodes, children, and mothers. A Directed Acyclic Graph was explicitly used to assess potentially confounding relations. This allowed the generation of a minimal sufficient adjustment set to reduce confounding of each exposure, although the rationale for particular sets was not clearly provided and may have over- or under-adjusted in the analysis.

A few issues require consideration. First, the recording of DAMA involves a judgment based on clinical experience, the effort put into understanding the parents' context/needs, and the interpretation of coders. Second, local-level qualitative information about parents' experiences, reasons, attitudes, and suggestions for system improvement can add value to understanding the forces that influence the decision to DAMA. Practical, co-designed changes to systems, enhanced communication, and support can make a difference in reducing DAMA. Moreover, they also serve to improve the hospital experiences of Aboriginal children and their families, and the morale of staff working in the health system.

This policy-relevant study has been undertaken by a strong Aboriginal and non-Aboriginal epidemiological team committed to making a difference in Aboriginal child health. The findings from this study shed a positive light on DAMA among paediatric Aboriginal inpatients: the practice is relatively rare and has significantly and substantially improved since 2002–2005. Improvements coincide with emphasis on staff training and the introduction of the Closing the Gap policy, which aims to substantially increase the life expectancy of Australia's First Nations peoples.² While the lack of cultural security can exacerbate situations in which hospital and ED DAMA occur, society-level and broader system-level challenges are likely to be major contributors to ED DAMA rather than hospital systems and staff alone. Improved access to culturally secure primary care, better resourcing of rural hospitals and the recruitment and cultural safety training of suitable staff together have the potential to reduce DAMA—particularly in ED contexts where service demands and DAMA are increasing.

原住民和托雷斯海峡岛民（以下简称原住民），即澳大利亚原住民，占澳大利亚总人口的 3.8%。¹澳大利亚原住民生活在不同的地理环境中，在文化、语言和社会经济上具有多样性，为澳大利亚的生活做出了丰富的贡献。原住民儿童的健康是原住民社区、家庭、最高机构和政府的首要任务² ，其中儿童早期尤为重要。然而，国家总体统计数据始终反映出，与其他澳大利亚儿童相比，原住民儿童的疾病负担过重。³文献的重点主要是描述缺陷，而不是试图理解它们，因此对土著儿科健康劣势的复杂途径往往了解甚少。

对于许多原住民父母和照顾者来说，医院是一种疏远的环境，让人想起殖民时期的痛苦历史，当时儿童被强行带走，原住民家庭被视为二等公民。^{4, 5}为了最大限度地提高原住民儿童的健康和福祉，重要的是要认识到他们及其家人与当今卫生系统互动的方式往往受到这些过去（有时是持续的）经历的影响。

当患者（患者为儿童时的父母或照顾者）在临床团队正式出院之前离开住院设施或急诊室时，就会发生不顾医疗建议出院 (DAMA)；这可能会增加患者再次入院或健康状况不佳的风险。⁶复杂的人、医院和环境因素促成 DAMA。在澳大利亚，DAMA 在原住民患者中发生的频率更高，DAMA 率被视为衡量医院对原住民患者需求反应程度的指标。这包括患者在接受治疗时感受到的文化安全水平，以及系统如何满足或不满足他们的需求。^{6 家}澳大利亚医院定期接受 DAMA 费率审计。迄今为止，大多数现有研究都集中在成年人身上。

在最新一期的《儿科和围产期流行病学》中，Christensen 及其同事⁷利用了其“克服困难”研究中关联的西澳大利亚州 (WA) 行政数据。他们调查了住院和急诊科 (ED) 环境中的 DAMA 与 5 岁以下原住民出生队列中的儿童、家庭和服务事件特征之间的关联。全州范围内的 ED DAMA（主要是“不等待”）和医院 DAMA 分别进行了分析，考虑到大多数对比结果，这是一个谨慎的决定。

不出所料，DAMA 在急诊科 (4.7%) 中比在住院患者中 (1.6%) 更常见。调整相关协变量后，围产期和母亲因素通常与住院期间的 DAMA 无关——尽管出生体重 <2500 g 的患者的 DAMA 风险呈临界增加（调整后的优势比 [aOR] 1.31，95% 置信区间 [CI] 1.00） ，1.72）。相比之下，ED DAMA 与多胎 (aOR 0.80) 和 3 次或以上生育 (aOR 0.92) 呈负相关，而母亲出生年龄 <20 岁与 DAMA 风险增加相关 (aOR 1.14)。因此，ED DAMA 更多地与母亲及其处境有关。

入院时 1 岁以下儿童比 1-4 岁儿童更有可能 (aOR 1.24) 从医院接受 DAMA，但从 ED 接受 DAMA 的可能性较小 (aOR 0.74)。正如预期的那样，急诊入院患者入院 DAMA 的可能性是候补名单上择期入院的 3.4 倍，并且 ED 分诊类别的紧急程度存在梯度，半紧急和非紧急与紧急就诊的 aOR 约为 11 。

地理环境被证明是 DAMA 的重要决定因素。这并不奇怪：西澳涵盖广阔且各不相同的地理区域，在服务的可及性、人口密度、气候和社会经济地位方面。对于住院治疗，出生时居住的所有非大都市地区的 DAMA 风险均增加。有证据表明，随着偏远地区的增加，存在梯度效应，居住在非常偏远地区的儿童患 DAMA 的风险是珀斯大都市的 4.7 倍。医院地点也是如此（乡村医院 aOR 6.99）。然而，对于 ED 的演示，非常偏远的居住地区与 DAMA 的几率降低40% 相关，而与其他偏远程度没有关联。农村急诊室的赔率低 12%（aOR 0.88），主要反映了居住在偏远地区的儿童的初级医疗保健选择有限。

一个重要的发现是研究期间住院 DAMA 减少，但 ED DAMA 增加。这表明原住民儿童的住院系统已得到实质性改善。这可能是由于对西澳卫生系统⁸的工作人员引入了强制性在线文化安全培训，并辅以有针对性的策略和监测来降低 DAMA 率。这些相同的策略似乎不足以影响 ED 率。事实上，随着时间的推移，急诊 DAMA 的增加表明了医院系统之外更广泛的背景因素，特别是在 2019-2020 年 COVID-19 大流行之前，获得初级卫生保健的机会减少，从而导致急诊就诊人数每年增加 3.2%。⁹研究期间发生了许多变化，包括减少非工作时间的全科医生服务、原住民医疗服务部门缺乏非工作时间服务、服务成本增加导致自付费用增加（由于一般医疗机构减少了“批量计费”）费用完全由政府补贴的做法）和增加的生活成本。¹⁰

患者 DAMA 重要吗？由于仅关注调整后的结果，该论文没有充分强调医院 DAMA 中未经调整的再入院几率比其他患者高 40%，即使调整后的估计表明几率有所增加。这意味着在所有条件相同的情况下，医院 DAMA 值得关注——即使有一系列因素可以解释 DAMA。ED DAMA 中未调整的再入院几率降低支持了以下观点：ED 中的 DAMA 患者通常可以在初级保健中得到充分治疗。

这项研究在技术上具有示范意义。使用覆盖整个州的基于人口的出生队列来实现全面且公正的队列，并通过多个数据源提供丰富的背景信息。除了最佳实践算法之外，还使用家庭关系数据严格进行土著识别。在大多数拥有大量少数民族土著人口的其他国家，这种识别方法很少可行。对解释模型的关注使得原住民内部的比较能够确定优势和风险。复杂的统计分析还考虑了嵌套的数据结构——多层次的服务事件、儿童和母亲。有向无环图明确用于评估潜在的混杂关系。这允许生成最小的充分调整集，以减少每次暴露的混杂，尽管没有明确提供特定集的基本原理，并且在分析中可能过度或调整不足。

有几个问题需要考虑。首先，DAMA 的记录涉及基于临床经验的判断、理解父母背景/需求的努力以及编码人员的解释。其次，关于家长的经历、原因、态度和系统改进建议的地方级定性信息可以为理解影响 DAMA 决策的力量增加价值。对系统进行实用的、共同设计的变更、加强沟通和支持可以在减少 DAMA 方面发挥重要作用。此外，它们还有助于改善原住民儿童及其家人的医院体验，以及卫生系统工作人员的士气。

这项与政策相关的研究是由强大的原住民和非原住民流行病学团队进行的，致力于改善原住民儿童的健康。这项研究的结果为儿科原住民住院患者中的 DAMA 提供了积极的启示：这种做法相对较少，但自 2002 年至 2005 年以来已得到显着和实质性的改善。这些改进与对员工培训的重视和“缩小差距”政策的推出相一致，该政策旨在大幅提高澳大利亚原住民的预期寿命。²虽然缺乏文化安全可能会加剧医院和 ED DAMA 发生的情况，但社会层面和更广泛的系统层面的挑战可能是 ED DAMA 的主要促成因素，而不仅仅是医院系统和工作人员。改善获得文化上有保障的初级保健的机会、更好地为乡村医院提供资源以及招聘合适的工作人员并进行文化安全培训，有可能减少 DAMA，特别是在服务需求和 DAMA 不断增加的急诊室环境中。