Volume 213 Issue 6

We need to burn, but it is not a cure-all and the side effects can be serious

Fay H Johnston

“Now we say Black Lives Matter but … the fact of the matter is, we just Black matter to them”1

If Black lives matter we need to be prepared to examine and address racial violence within the Australian health system My name is Kevin Yow Yeh and today I march for every Black death in custody but I especially march for my grandfather Kevin Yow Yeh Sr. At the age of 34 this man apparently had a heart attack at a Mackay watch house … This last month we've seen plenty of stats, 430 plus Black deaths in custody … and that's only since the Royal Commission, but what about all those deaths that led to that. My grandfather was one of them. Let's humanise these stories. When this man had a heart attack, he left his wife and he left five young children. My grandmother was still having his children when she had to put this man in the ground. That's why we march! Of course we stand in solidarity with our brothers in America. And, of course we stand in solidarity with our sisters in West Papua … but today we stand for our lives here, on stolen land.2 The statistical story of Indigenous health and death, despite how stark, fails to do justice to the violence of racialised health inequities that Aboriginal and Torres Strait Islander peoples continue to experience. This story has been reported on unremarkably in federal parliament for over a decade, as an annual account‐keeping exercise of policy failure and statistical targets not met.3 This story of failure and failing health has been told countless times in health and medical journal publications, and despite growing more frequent in number, these contributions to new knowledge never seem to translate to improved health outcomes. This story of failure does not do justice to the trauma and loss that Aboriginal and Torres Strait Islander communities experience. This story of failure does not do justice to the pain of never meeting the grandfather that you are named after. Tragically, despite the parlous state of Indigenous health, we have not been met here with the kind of urgency that the global Black Lives Matter movement has spurred elsewhere. What we have been presented with, aside from the Health Minister admonishing Black Lives Matter protestors for putting the health of the public at risk,4 has been the triumphal announcement of “research projects”,5 the release of a “landmark report”,6 and a drafting of “refreshed” and “historic targets”.7 All of these supposedly fresh responses were on track before the Black Lives Matter movement hit our shore. Rather than the “new normal” which the threat of coronavirus disease 2019 (COVID‐19) inspired, the Australian health system's Black Lives Matter moment is best characterised as indifferent; a “business as usual” approach that we know from experience betokens failure. When the threat of COVID‐19 loomed, action was swift and the Aboriginal and Torres Strait Islander leadership within and outside of the health system was even swifter in establishing taskforces, lobbying for additional resources for the community controlled sector, instituting special border control measures for remote Indigenous communities, and the development of emergency response plans to protect their communities.8,9 The effective response to the COVID‐19 pandemic sits in sharp contrast to the ongoing pandemic of racism that Indigenous peoples have been fighting since 1788 and which has taken far more Black lives in Australia. Sweet points out: “To date, there is very little sign that senior health policy makers, from the Chief Medical Officer to Health Minister Greg Hunt, will use their authority to name and address the system racism that contributes to poorer healthcare, as it does to overincarceration”.10 While broad attention is often focused on Black deaths in custody, the premature deaths of Indigenous peoples from supposed natural causes inside and outside of custody tell a consistent story of failure and violence that marks the Australian health system and society more broadly. Against the quietude of the Australian health system on racism are the powerful voices of Aboriginal and Torres Strait Islander peoples, on television screens, on public streets and in our spreadsheets, speaking the truth about how little Black lives seem to matter. Both Indigenous clients and clinicians have stories to tell of the violence of racism in the health system, of being cast in the category of less capable, less compliant, less deserving of care and less worthy of the category of human. This then brings us to the coronial inquiry, the endgame of not caring; of neglect. Here, never let us forget the mothers, the children, the cousins and the spouses weeping outside coroner's courts, bearing photos of their loved ones in their hands and on their clothing, simultaneously appealing for care and for justice.11 Moreover, let us not for a second dismiss the anguish of having to fight for the release of recorded footage of your loved one's final moments, to be replayed over and over, in which they too plead vainly, “I can't breathe”.12 So many grieving Indigenous families continue to appeal to the state for care and for justice via coronial inquiries in the hope that their tragedy will not befall another. But the awful truth is that the recommendations of coronial inquiries are not enforceable because the inquest is meant to discover what happened rather than determine responsibility. So again, regardless of the findings, the resulting outcome is business as usual. The coronial inquiry represents a theatre of power where, in the presence of an avoidable Indigenous death, the state declares its benevolence; duly recording the steps taken and policies and procedures adhered to or those requiring review, and the best efforts of police, medical officers or first responders, to deem the death another “unavoidable” tragedy. Gomeroi scholar Whittaker11 notes how the discourse of “natural causes” in coronial inquiries works to render Indigenous peoples as “fated to die” and beyond care because they were “already dead”. The coronial inquiry represents a moment of confluence of the health and legal systems and the state that seek to erase Indigenous existence and affirm the settler trope of a dying race. It represents the theatre of Indigenous health policy writ large. The story of Indigenous health failure, of persisting and alarming health statistics that are routinely attributed to a complex web of social, cultural and economic factors, sustains the notion of the inevitability of Indigenous ill health, of a race destined to die out, despite the best of efforts and intentions. How do we explain an unwavering commitment to a failed Indigenous health policy framework amid a global movement centred around the importance of Black lives, and a National Aboriginal and Torres Strait Islander Health Plan vision of a health system “free of racism” with no strategy for addressing systemic racism?13 How do we further explain the focus on the individual health behaviours or “choices” of Aboriginal and Torres Strait Islander peoples when we know “incessant racial health inequities across nearly every major health index reveal less about what patients have failed to feel and more about what systems have failed to do”.14 As Boyd and colleagues point out, “The solution to racial health inequities is to address racism and its attendant harms and erect a new health care infrastructure that no longer profits from the persistence of inequitable disease”.14 Earlier this year, the National Registration and Accreditation Scheme demonstrated the type of Black Lives Matter moment that the Closing the Gap refresh missed, by launching the Aboriginal and Torres Strait Islander Health and Cultural Safety Strategy 2020‐2025.15 The strategy sets clear directions for the Australian Health Practitioner Regulation Agency, the national boards and accreditation authorities, which regulate Australia's 740 000 registered health practitioners to ensure that patient safety for Aboriginal and Torres Strait Islander peoples is the norm. The landmark strategy embodies ambition and partnership to address racism and culturally safe care; shifting the blame of failure for good health from Black bodies and instead demanding structural and individual health reform of health practitioners and the systems that regulate them. It is this shift of focus that has been central to the calls from Aboriginal and Torres Strait Islander peoples. Black wounds have been laid bare, to reveal the violence of health and legal systems upon Aboriginal and Torres Strait Islander peoples in a desperate appeal for those same systems to care. At 34 years of age my grandfather died, where's his justice? … what about all the other families, what about all the other fathers, brothers, sisters, nephews and nieces …? What about all the other mob? Where's their justice? My name's Kevin Yow Yeh, f*** the system, if you're not with us you're against us! What is needed is an Australian health system that has a steadfast commitment to Black lives: not as in need of saving, but as deserving of care; one that matches the staunchness of grieving Black families marching the streets of our capital cities in the midst of a pandemic. Such a commitment demands that we abandon the failed Indigenous health policy of Closing the Gap16 in favour of a health justice framework,17 which would include, but not be limited to: A foregrounding of Indigenous sovereignty rendering visible the strength, capability and humanity of Aboriginal and Torres Strait Islander peoples, services and communities in all processes of health policy formation and implementation, not as partners but as architects. State and federal government commitment to the recommendations of the coronial inquiries into the deaths of Aboriginal and Torres Strait Islander peoples who have died of preventable or avoidable conditions in the health system, and the establishment of an Indigenous taskforce to oversee implementation. An explicit financial commitment from the National Health and Medical Research Council and the South Australian Health and Medical Research Institute (via the Indigenous Medical Research Future Fund) and the Australian Research Council for research that attends to the nature and function of race in producing the conditions that allow racialised health inequalities to persist, from birth to death, including the embodied consequences of racism. The establishment of awareness‐raising campaigns that make clear the various ways in which Aboriginal and Torres Strait Islander peoples may seek justice when experiencing discrimination within the health system, and commeasurable resourcing of legal services to support Indigenous peoples to take action. Introduction of publication guidelines for health and medical journals requiring research relating to racialised health disparities to foreground institutional racism in its analysis, rather than socio‐economic disadvantage and other social and cultural factors. Development of an interdisciplinary Indigenous health workforce agenda that centres the care of Indigenous people beyond capacity building to include attending to racial violence within workplaces across the Australian health system. We offer these strategies not as a solution, but as some small steps towards a radical reimagining of the Black body within the Australian health system; one which demonstrates a more genuine commitment to the cries of “Black Lives Matter” from Blackfullas in this place right now.

Chelsea J Bond · Lisa J Whop · David Singh · Helena Kajlich

Guidelines need to reflect the mounting evidence for airborne transmission of SARS-CoV-2

C Raina MacIntyre · Michelle Ananda‐Rajah · Mark Nicholls · Ashley L Quigley

AI has arrived, with the potential for enormous change in the delivery of health care, but are we ready? Artificial intelligence (AI) is the trigger for the next great transformation of society: the fifth Industrial Revolution. AI has already arrived in health care, but are we ready for the kind of changes that it will introduce? In this article, we map out the current areas where AI has begun to permeate and make predictions about the kind of changes it will make to health care. Definition of AI AI comprises any digital system “that mimics human reasoning capabilities, including pattern recognition, abstract reasoning and planning”.1 It includes the concept of machine learning, where machines are able to learn from experience in ways that mimic human behaviour, but with the ability to assimilate much more data and with potential for greater accuracy and speed. Machine learning is a research field that has seen recent advances due to exponential increases in computing power (a phenomenon known as Moore's law), algorithmic coding that mimics the human cognitive process (deep learning), and access to large, linked sources of big data. The scope of AI can be specific, performing narrowly defined tasks (narrow AI) such as image interpretation, or more general, applying knowledge and skills in different contexts (general AI) such as making a diagnosis and predicting disease outcome. On the other hand, machine learning can also be designated “supervised”, in which a dataset is provided for the algorithm to evaluate its performance, or “unsupervised”, in which the machine is allowed to extract unknown potential features in developing an algorithm. The arrival of AI into current practice AI, machine learning, and deep neural network tools can assist medical decision making and management, and have already permeated into at least three different levels: AI‐assisted image interpretation; AI‐assisted diagnosis; and AI‐assisted prediction and prognostication. From diagnosing retinopathy to cardiac arrhythmias, from screening for skin cancer to breast cancer, from predicting outcome of stroke to self‐management of chronic diseases, AI and machine learning devices can replace many time‐consuming, labour‐intensive, repetitive and mundane tasks of clinicians and give possible suggestions of management plans (Box 1).2,3,4,5,6,7 While the advancement and new capabilities and opportunities are exciting, the responsibility and liability issues of AI‐assisted clinical diagnosis and management need much deliberation. AI‐assisted image interpretation One of the major advances in AI is pattern recognition enhancing image‐based diagnosis in radiology, pathology and endoscopy. AI‐assisted image analysis aids the detection of adenoma and polyps during colonoscopy. It can even provide optical biopsy to determine the nature of lesions with implications of treatment.8 Wireless capsule endoscopy is a groundbreaking advance in medical technology, allowing painless examination of the gut, reaching areas where conventional endoscopes cannot reach. However, reading thousands of images produced by the capsule is extremely time‐consuming. Deep neural network systems trained to read images of capsule endoscopy can scan thousands of pictures within minutes to reduce the burden of time and energy for gastroenterologists and also minimise the chance of missing significant lesions.9 Similarly, systems have been trained to read echocardiographic images to provide physiological measurements within seconds, and to read coronary computed tomography angiography images to determine coronary calcification, coronary stenosis severity, and functional haemodynamic effects of the stenosis. AI‐assisted diagnosis The diagnosis of many conditions (eg, acute or old myocardial infarction) and arrhythmias (eg, atrial fibrillation and ventricular tachycardia) can be made by experts reading electrocardiograms (ECGs) according to well established rules. Application of such rules in algorithms have allowed computers to make these diagnoses automatically in ECG machines for many years, but the diagnoses are subject to verification by physicians using the same rules. AI using machine learning and deep neural network can do the same from raw ECG data, but does not rely on the same rules, and thus can do much more than conventional ECG analysis. In the most basic AI formulation, diagnoses of important cardiac arrhythmias from a single lead rhythm strip or continuous single lead ECG recordings were made by machine learning using deep neural network algorithms with greater accuracy than an individual cardiologist and similar to a consensus panel of cardiologists.10 Where AI excels, however, is in discerning patterns not apparent to the experts, such that current or future paroxysmal atrial fibrillation can be diagnosed from an ECG in sinus rhythm,11 and asymptomatic left ventricular dysfunction can be diagnosed by a 12‐lead ECG.12 AI‐assisted prediction and prognostication AI may predict the occurrence of certain diagnoses and prognosticate clinical outcomes of patients based on clinical datasets, genomic information and medical images. Cardiologists have developed algorithms to assess the risk of cardiovascular disease and claimed that their prediction is superior to existing scoring systems. Gastroenterologists have also developed AI models to predict recurrence of bleeding and requirement of surgery in patients with gastrointestinal bleeding.13 Combining genomic, epigenetic and metagenomic data with biochemical and lifestyle information using machine learning will be a very powerful tool in medicine. However, mechanisms or reasons for reaching the machine decision may not be comprehensible to clinicians. The integration of various datasets in multilayer informatics could take prediction, prognostication and prevention of diseases to new levels that cannot be achieved by conventional statistical models. This capability, if validated in properly designed studies, will offer new dimensions to personalised medicine. Preparing for the future of AI Health disparities, excluded populations and data biases The quality of AI in health care is dependent on the quality of the data on which it is based. Algorithms are being developed and validated on data generated by health care systems where current practices may already be inequitable. A system built on poor quality, biased data will reflect those problems (“garbage in, garbage out”). If a health care system has excluded populations of patients, the structural inequalities of health care will be repeatedly reinforced by the AI. This is a not a new problem and we must do better science and be awake to the limits of data quality and evidence‐based medicine. Data sovereignty and stewardship AI is built on access to big data. Big data in health care is primarily generated by public health systems, funded by the public for the public. Increasingly, claims over the health data generated by these public systems are being contested.14 There was enormous public outcry over the use of British National Health System data by Google‐owned DeepMind, a company creating an AI‐based smartphone application for kidney disease. Many were angry about the private use of public data, when there was little public control over what would happen to the data or what benefit was being provided back to the National Health System.14 Issues of data sovereignty therefore threaten the existence of effective AI. Patient data should not be provided to technology giants without a good governance structure to protect data sovereignty. Changing standards of care An immediate issue for the use of AI is the question of how it will transform standards of care. Common law jurisdictions judge health professions, by and large, by measuring performance against competent professional practice as set by the professions themselves. If AI keeps its promise of benefit and it is integrated more into practice, standards of care must require AI use, and traditional forms of therapeutics will be forced to change. We will see a time when all medicine and allied health work as a team with AI. Those who refuse to partner with AI might be replaced by it. Legal responsibility for AI‐caused injury AI promises to massively reduce the occurrence of iatrogenic harms via increasing the quality of decision making, but the continued existence of AI‐related injury is easy to foresee. As machine algorithms improve themselves without human intervention, making the “black box” more opaque, regulatory agencies such as the Australian Therapeutic Goods Administration and the United States Food and Drug Administration need to refine their regulations. To the extent that AI continues to play a role in assisting clinical management, questions of responsibility for harm should be determined by ordinary rules of product liability. It is likely that courts will determine some of these liability questions by using analogies with vicarious liability — an employer is responsible for the negligence of the staff when the injury occurs in the course of the staff's employment. A doctor using AI should be responsible for AI decisions made in the course of treatment, especially if the doctor retains the power to make the final decision regarding treatment. But as AI takes on more autonomous decision making, it might be argued by some doctors that they should not be responsible for that which they cannot control. Similarly, it seems unfair for doctors to be held responsible for an AI decision when they are unable to deduce how and why that decision was made. Such matters are outside the scope of clinicians’ expertise and best dealt with legally as a product liability claim. A stepwise gradation model of shared responsibility between the human doctor and the machine in diagnosis and clinical management has been proposed15 (Box 2). Conclusions Before AI tools can be put into daily use in medicine, data quality and ownership, transparency in governance, trust‐building in black box medicine, and legal responsibility for mishaps are some of the hurdles that need to be resolved. Much effort is needed to translate algorithms into problem solving tools in clinical settings and demonstrate improvement in clinical outcomes with saving of resources. Box 1 – Examples of artificial intelligence (AI) permeation into clinical practices of different specialties Clinical management AI capability Diabetic retinopathy2 Detection of early changes in fundi of patients with diabetes Reading the retina and blood vessels to identify patients at risk of developing complicated diabetic retinal disease Breast cancer3 Diagnosis of early breast cancer based on mammography Reading mammographic pictures to detect early malignant transformation in breast cancer screening Skin cancer4 Diagnosis of skin cancer by its clinical morphology Identification of skin cancer by pictures and classification of types of skin neoplasia Cerebrovascular disease5 Predicting outcome after a cerebrovascular accident Predicting the outcome (mobility, morbidity and mortality) of stroke 90 days after the event Non‐communicable chronic diseases6 Monitoring of diabetes and heart failure in primary care setting Assisting patients monitoring of blood pressure and blood glucose at home and transmitting information to family medicine clinics Heart failure7 Predicting the clinical outcome of patients with heart failure Predicting in‐hospital mortality among patients with heart disease based on echocardiography Box 2 – Levels of artificial intelligence (AI)‐assisted decision in diagnosis and clinical management and possible share of responsibility between human doctor and machine

Joseph JY Sung · Cameron L Stewart · Ben Freedman

The challenge now is how to implement artificial intelligence technology safely into clinical practice Artificial intelligence is a branch of computer science that, in broad terms, deals with either decision making or classification. The aim of artificial intelligence is to surpass human cognitive functioning such that automated decisions can be made. Machine learning — an application of artificial intelligence — is commonly used in image recognition. In general, the machine, or algorithm, learns from exposure to a large dataset. Once learning has taken place, the algorithm can be applied to unseen data. The potential advantages of this approach in health care are clear: machines can learn from very large datasets in relatively short time frames and can apply themselves to new data without fatigue or intra‐observer replication error. Machine learning has recently demonstrated remarkable performance in image‐based diagnosis across various medical fields, including ophthalmology, radiology, pathology and dermatology. In dermatology, the primary focus has been on developing machine learning systems that facilitate classification and decision support for skin cancer management. Skin cancer (including melanocytic and keratinocytic malignancy) is the most common cancer in Australia and among Caucasian populations worldwide. Melanoma is responsible for the majority of skin cancer deaths in Australia and has various presentations.1,2 While dermoscopy has improved the accuracy of melanoma diagnosis, significant variability occurs and is largely a function of clinical expertise. Recent studies show that machine learning algorithms have the potential to surpass the diagnostic performance of experts, and the challenge now is how to implement this new technology safely into clinical practice. Although there are a number of machine learning algorithms that could be used in the dermatology setting, convolutional neural networks (CNNs) are the most promising. This is largely because they learn from data without any feature specification, and they are known to exhibit superior performance for image recognition in comparison with other machine learning algorithms.3 The aim of the CNN is to generalise its previously learned knowledge on unseen images beyond the training dataset. There are numerous parameters within a CNN that can be tweaked to maximise algorithm performance. Most of these parameters are adjusted automatically by the algorithm, without user input. Therefore, very little can be known, in principle, about why and how the algorithm reaches any particular decision. Currently, there are efforts underway to reduce the “black box” effect of CNNs. Some commercial software programs coupled to imaging devices will provide the user with comparable lesions to justify the algorithm's output and improve transparency. However, this retrieval system may fail for rare or unseen cases and does not provide a decision‐making process. While the black box phenomenon remains, there are two potentially negative implications for clinical practice: first, clinicians may have difficulty upskilling by following the algorithms’ outputs; and second, there exists the potential for deskilling and underperforming due to an over‐reliance on technology.4,5 The effect of a faulty system has been explored by manipulating a previously trusted algorithm to generate incorrect classifications and found that doctors of all experience levels were susceptible to being misled by the recommendation.5 Algorithm performance is dependent on both the size and quality of the training image dataset and on whether the algorithm is used in situations for which it was intended. Depending on the training set, the device may be limited in its ability to diagnose specific lesions (eg, non‐pigmented), or lesions in certain skin types (eg, darker skin) or sites (eg, scalp or acral). Retrospective image databases used to train algorithms may be associated with bias. In addition, artefacts (eg, hair, dermoscopic gel, air bubbles, rulers, pen markings, reflections) can distract from key features. However, if a CNN is trained on a large enough cohort, it can learn to deal with potential artefacts. Nonetheless, unbiased lesion selection and standardised image capture would invariably improve algorithm performance, and recent advances in three‐dimensional (3D) imaging modalities will enable this.6 Several studies have now shown that CNNs trained on retrospective image data collected at a single time point are capable of classifying skin cancer with sensitivities and specificities equal or superior to that of dermatologists (Box 1),5,7,8,9,11 and clinicians with less experience gain most from AI support under experimental conditions.5 Hypomelanotic and acral melanoma can be more challenging to diagnose clinically,1 and this could potentially present a challenge for automated classification. However, CNNs have achieved greater accuracy for hypopigmented and acral lesions in comparison with human experts, at least in silica.9,11 In addition to clinical images, CNNs have been applied to histopathological images of melanoma and benign naevi with promising results.10 The ground truth for lesion diagnosis The gold standard for melanoma diagnosis is histopathological assessment. However, there exists significant inter‐ and intra‐observer variability in histological diagnostic labels attributed to atypical melanocytic lesions.12 The existence of such variability in diagnoses poses the dilemma of whether the CNN has learnt from the correct set of diagnoses. Consensus diagnoses, if practical, may help overcome this problem. Molecular biomarkers may assist in establishing a diagnosis13 and identifying high risk biology,14 but they require extensive validation before clinical use. Pathologists and clinicians also rely on metadata (age, personal and family history, lesion symptoms, recent change), which may influence diagnostic likelihoods. Importantly, it is possible to incorporate different data types, including metadata, sequential image data coupled with histopathology, to train future CNN algorithms and improve diagnostic discrimination of borderline lesions (Box 2). Use of artificial intelligence for melanoma screening It is well known that the incidence of invasive melanoma in Australia has increased over the past 40 years. In addition, there has been a striking increase in incidence of in situ melanoma over the past decade, from 32 cases per 100 000 population in 2004 to 80 per 100 000 population in 2019, with age‐standardised mortality remaining fairly stable.2 The potential causes for the increase in incidence are complex, and involve a true increase, driven by poor sun exposure practices of individuals born before the SunSmart era, combined with increased awareness, excessive screening, and overdiagnosis. It has recently been estimated that 54% of melanomas (15% of invasive melanomas) are overdiagnosed.15 Artificial intelligence‐assisted targeted screening of high risk individuals is likely to be a more effective strategy to save lives than the current opportunistic approach. With sequential whole‐body image datasets linked to metadata, molecular biomarkers and clinical outcomes, our ability to identify lesions associated with sinister biological potential will improve (Box 2), thereby reducing unnecessary biopsies, minimising overdiagnosis and other potential harms associated with screening. Use of artificial intelligence in clinical practice There are advantages and disadvantages of introducing artificial intelligence at different points in the patient care pathway.16 An artificial intelligence system used as a triaging tool before clinician assessment would enable automated risk stratification of individuals and/or lesions (Box 2). This approach could dramatically improve clinician workload and timely access to specialist care for people requiring urgent attention. Alternatively, artificial intelligence consulted following an examination by the clinician may act as a second opinion to improve diagnostic sensitivity and reduce unnecessary biopsies.5 The latter is more closely aligned with current clinical workflows and therefore likely to be preferred while the field matures. There is potential for over‐reliance on artificial intelligence systems in both scenarios. A secondary support system may provide the clinician with a diagnosis or a management decision. Doctors are more likely to change their minds if they are uncertain of a diagnosis and an algorithm provides a conflicting result.5 It is thus important to consider how an algorithm might convey uncertainty to avoid false guidance. For example, a decision‐support output (eg, excise, monitor or reassure) avoids the diagnostic dilemma of differentiating between melanoma and dysplastic naevi. However, the problem is complex and arguments exist as to why, in many situations, a diagnostic probability output might be more desirable. Safe implementation of new technologies The Therapeutic Goods Administration (TGA) has developed an action plan to improve the processes by which new devices are approved for use in Australia, strengthen monitoring and follow‐up, and provide more information to consumers about the devices they use.17 International collaborations also exist with groups, such as the International Medical Device Regulators Forum, to establish better processes for medical device regulation globally. If software is classified as a medical device (ie, it is intended for diagnosis, prevention, monitoring, treatment or alleviation of disease), it must be registered on the Australian Register of Therapeutic Goods following TGA approval and before distribution within Australia. Consumers and clinicians need to be aware of the intended use of an application or device. There are several smartphone applications available to the general public, with functionality ranging from education to monitoring and tracking to skin lesion classification. Some of these provide skin lesion risk assessment, although they may state that they are not intended to be used as a diagnostic device. There is concern that, if this is not immediately obvious to the consumer, unregistered applications may be used in lieu of seeking medical advice. Unsupervised consumer‐operated diagnostic devices would require careful testing before they can be recommended. Conclusion As clinicians, we need to be aware of the limitations of any diagnostic tool and interpret outputs accordingly. Although the performance of artificial intelligence to date is promising, it remains to be seen how diagnostic devices in dermatology will influence decision making in the clinic and affect patient outcomes. Regardless of the specialty, any new technologies need to be rigorously tested before implementation and monitored after implementation. Ultimately, responsibility for patient care remains with the clinician and, as such, a high level of clinical acumen must be maintained. Nonetheless, artificial intelligence in dermatology is primed to become a powerful tool in skin cancer assessment. Box 1 – Comparison of skin cancer classification tasks by artificial intelligence (AI) systems and dermatologists/pathologists Study AI architecture Images Classification task Training dataset size Test dataset size AI Dermatologists/pathologists Sensitivity Specificity AUC/overall accuracy Sensitivity Specificity AUC/overall accuracy Tschandl5 ResNet34 CNN Clinical (dermoscopic) Benign v malignant v non‐neoplastic skin lesions 10 015 1412 0.81 (0.79–0.83)* 0.92 (0.90–0.93)* 0.73†(0.70–0.76)* 0.80 (0.78–0.83)* 0.80 (0.77‐0.82)* 0.60† (0.57–0.63)*,‡ 0.86 (0.84–0.88§)* 0.88 (0.87–0.90§)* 0.74† (0.71–0.77§)* Esteva7 GoogleNet Inception v3 CNN Clinical (macroscopic, dermoscopic) Benign v malignant v non‐neoplastic skin lesions 129 450 1942 na na 72.1%¶ ± 0.9% na na 66.0%¶ Haenssle8 GoogleNet Inception v4 CNN Clinical (macroscopic, dermoscopic) Benign melanocytic naevi v melanoma > 100 000 100 86.6%** 82.5%** 0.86** 86.6%** 71.3%** 0.79** 88.9%†† 82.5%†† 0.86†† 88.9%†† 75.7%†† 0.82†† Tschandl9 GoogleNet Inception v3 CNN Clinical (macroscopic, dermoscopic) Benign v malignant hypo‐pigmented lesions 13 724 2072 81% 53.5% 0.73 78% 51.3% 0.68 Hekler10 ResNet50 CNN Histopathology Benign naevus v melanoma 595 100 76% 60% na 51.8%‡‡ 66.5%‡‡ na Fujisawa11 GoogleLeNet DCNN Clinical (macroscopic) Benign v malignant skin lesions§§ 4867 1142 96.3% 89.5% 92.4%¶ ± 2.1% na na 85.3%¶ ± 3.7% AUC = area under the curve; na = not applicable. * 95% CI. † Youden statistic. ‡ Clinicians with varied experience and training. § Clinician accuracy with multiclass probabilistic AI support. ¶ Overall accuracy. ** Level I: AI and human readers provided with dermoscopic images only. †† Level II: AI provided with dermoscopic images only, human readers provided with dermoscopic images, macroscopic images and additional clinical information. ‡‡ Pathologist. §§ 52.6% of melanomas in this study were acral. Box 2 – Incorporation of different data types to train future convolutional neural network (CNN) algorithms and improve diagnostic discrimination of borderline lesions AI = artificial intelligence.

Miki Wada · ZongYuan Ge · Stephen J Gilmore · Victoria J Mar

Enhancing clinical trial recruitment through eConsent has potential but needs more evidence of use Consent for medical interventions or clinical research participation currently relies on the use of printed information combined with a conversation with a health care professional, which is largely undocumented. Studies have shown that few participants are truly informed at all using these traditional means, and have demonstrated that recall or comprehension of what was disclosed is poor.1,2,3 Attempts to develop standardised participant information and consent forms (PICFs) that meet ethical requirements have often resulted in longer and more complex documents. While consumers have been engaged to assist with these programs, the purpose of PICFs is still too heavily weighted toward satisfying regulatory requirements rather than patient information needs. Unsurprisingly, data show that, as PICFs get longer, they are less well understood,4,5 and there is evidence that this is one of the reasons why patients do not agree to participate in clinical research.6 eConsent is not simply a conversion of a paper PICF into an electronically delivered version. It also holds the promise of improving participant engagement in clinical trials through a variety of features that include: the use of multimedia tools to enhance comprehension; ready conversion into multiple languages; a means to track consent in a highly portable manner; and the opportunity to provide information in a more convenient way to persons with an inability to attend clinics. The use of eConsent does not replace the opportunity for participants to ask direct questions to their doctor or the investigators. Moreover, in most instances, participants will still be required to make a physical visit to a clinic to receive their treatment, whereupon they can ask questions and confirm their willingness to participate. There are relatively few studies using eConsent. In an early randomised controlled study, there was a preference for eConsent as well as improved comprehension and intention to participate in people assigned to use computer terminals rather than paper to receive information.7 In a more recent study involving people infected with human immunodeficiency virus,8 eConsent was found to be acceptable and had some advantages over paper information sheets. There were a majority of males included in the study (75%), and more than half were African American, with a mix of sexual orientation. Health literacy of participants was the only factor that emerged as having an impact on comprehension; however, the number of participants (n = 20) is too small to draw statistically sound conclusions. A 2013 study tested comprehension and satisfaction when using iPads to deliver information for a neuropathy in chemotherapy study.9 Importantly, the investigators presented the same information in both formats, but the iPad had an initial video outlining the main features of the study. They found that of the 55 patients who took part in the randomised study, there was a statistically significant association with increased comprehension in the group assigned to the iPad. The sample sizes were too small for definitive findings, but of interest was that use of the iPad did not increase likely participation rates (it was slightly lower). All participants advised that the information provided was still too complex regardless of the media used, and that simplified text, diagrams, animations and other ways to enhance comprehension are needed. A recent study reported on the TransCelerate eConsent Initiative, which employed a large survey of 3045 participants and a number of smaller stakeholder consultations.10 While there was general support by potential participants for the use of eConsent, the survey revealed that people living in the European Union had the greatest level of discomfort with it. In this survey, they also found that people were concerned that eConsent might eliminate site/participant discussion regarding participation, even though this is not the case where it has actually been used. In Australia, there has not been widespread use of eConsent to date. To better understand the Australian context, Clinical Trials: Impact and Quality (CT:IQ) — a cooperative funded by MTPConnect, an Australian Government Industry Growth Centres Initiative, using funds from the federal government's Medical Research Future Fund (MRFF) — set out to investigate stakeholder perceptions of eConsent and, therefore, to identify potential actionable insights. Chrysalis Advisory developed a survey that was sent via email to the members of CT:IQ for distribution to the wider clinical trial sector in the first quarter of 2019. A total of 179 participants completed the survey and as we used a snowball methodology, there is no denominator of persons polled. In addition, there were 19 semi‐structured interviews conducted drawn from the CT:IQ membership. The majority of respondents (68%) were women, 75% were aged 40 years or over, and 80% had more than 10 years of working in trials, demonstrating considerable experience in the sector. The full report is available on the website,11 with the questions presented on pages 58–59 of the report. The key findings are summarised in the Box. We specifically surveyed those deploying eConsent at this stage and not the end users because we wished to understand what the sector was already doing and what the perceived barriers and opportunities were. Although only 29.2% of respondents indicated that they had any direct experience with eConsent, our survey revealed that they were overall cautiously positive toward the use of eConsent. An important finding was that there was optimism that use of electronic formats would enable participants to drive the information‐seeking process in a way that best suited their needs. The physical infrastructure, particularly in some public hospitals, was widely held as not being adequate to support eConsent uptake. Wi‐Fi blind spots within hospitals were cited as a major reason for this, as well as difficulties achieving infrastructure updates within the public health system. Respondents recommended that approaches to eConsent should employ technologies that do not rely on expensive infrastructure delivered by health services. In addition, respondents indicated that, ideally, there should be a sector‐wide standard for site information technology infrastructure requirements combined with clear guidance for sponsors to standardise their approaches. A number of interviewees who had worked on trials with eConsent where sponsors had provided devices noted that the devices were clunky and prone to malfunction, which increased overall study time and burdened trial staff. Clinical trial sites often experienced sponsors insisting on their own standards, resulting in unnecessary duplication or incompatibility of instrumentation at sites. Many respondents cited that differences in the use of eConsent platforms and inconsistencies between organisations regarding eConsent compliance (eg, whether participants would be required to sign electronically, or would be able to consent by using technologies such as face recognition, fingerprint identification etc) made it difficult to adjust to the use of eConsent. Greater industry engagement and collaboration may mitigate this barrier by providing stakeholders with frameworks and support to implement eConsent. Furthermore, setting some national guidelines will facilitate the design, regulatory approval and implementation of strategies to adopt eConsent. While some stakeholders identified data security as a risk associated with eConsent, others did not believe security threats were any greater than similar threats to existing digital technologies in use throughout clinical trials and the medical field more broadly. They suggested that when appropriate security systems are in place and data governance risks are managed, stakeholders were not likely to be concerned about data governance risks for eConsent. Using eConsent does not automatically mean that participants will have the ability to provide consent offsite, simply that they have access to the information offsite. This is no different from participants providing wet ink signatures offsite in terms of risk and the fact that a person comes to a clinic and accepts the study treatments is a clear demonstration of consent. Two‐factor authentication processes enabled by eConsent may provide a more robust means to authenticate consent than current paper‐based processes. It was not surprising that eConsent was considered to add a cost burden over and above a paper‐based approach. However, few of the respondents considered the cost savings made through enabling prior reading of relevant documentation and, in particular, the major cost savings for the site and for the participants this could potentially deliver. A respondent from a large cancer centre articulated the potential benefits by outlining how participants from anywhere outside of a 50 km radius of the tertiary centre could avoid additional time needed in the clinic through being able to use eConsent. This centre is piloting a tele‐trial model to deliver trials in non‐tertiary settings and recognises that eConsent is pivotal to enabling this model, which promises to reduce the burden on patients through reducing their need to travel and to ensure that clinical trial participation is more available beyond metropolitan centres. It appears from our survey that Australia is willing but only partially ready to implement eConsent. The pathway forward will require proactive planning, leading and managing organisational change with the creation of practical demonstration cases of the development, delivery and use of eConsent in the clinical trial setting vital to support wider adoption. CT:IQ is now looking at a program to undertake these pilot projects as part of its initiatives to enhance clinical trial capability across Australia and in other jurisdictions. Box – Key findings of the eConsent survey Barrier Finding Problems with using paper‐based information sheets and consent forms 38% of respondents thought paper consent forms were not a problem, 71.5% thought they were too long, and 62% found them too complex 37.4% of respondents thought paper‐based consent impaired participant comprehension 67% of respondents believed eConsent would improve comprehension, although they did not believe that this would necessarily translate into greater recruitment 59.2% of respondents believed there was a significant issue with providing adequate information to people from culturally and linguistically diverse populations and saw eConsent as a solution to this Perception that regulators, HRECs and hospital governance offices will not accept eConsent 40.8% of respondents believed that ethics committees would not approve use of eConsent, 26.8% were unsure 90.5% of respondents found it necessary to have guidelines for use by both researchers and HRECs Patients will not be sufficiently proficient with technology or have access to suitable devices Certain demographics (eg, older people) were considered likely to struggle with eConsent eConsent was likely to be well received by younger generations Health services lack the infrastructure to deliver eConsent 82.7% of respondents identified a lack of IT infrastructure as a critical barrier to overcome 59.2% indicated that the current infrastructure was inadequate, particularly within hospital sites Difficulties with authentication of individuals and data security 46.3% of respondents believed there would be issues with data governance, security and privacy, but 29% of respondents disagreed with this 59.2% of respondents felt that they would lose the ability to ensure that the person signing the eConsent was actually the participant, the remainder were undecided or felt this was not a problem Lack of consistent practice across the sector 67% of respondents identified a lack of standardised guidelines as a significant barrier to success 49.2% of respondents indicated that staff were able to manage eConsent despite the lack of training and standardised guidance eConsent will be more expensive 60.3% of respondents believed that there would be a significant initial cost, which might be a barrier to uptake HRECs = human research ethics committees; IT = information technology.

Nikolajs Zeps · Nicholas Northcott · Leanne Weekes

Telehealth can be used to deliver clinical trials, improve access to novel therapies and develop clinical networks Australia is a vast country. Nearly 32% of Australians reside outside the major capital cities, while 95% of medical specialists practise in cities.1 People living in rural and regional areas consistently experience poorer health outcomes.2 Cancer is a considerable health issue, with 395 new cancer diagnoses per day.3 The regional mortality gap in cancer remains.4 Between 2000 and 2010, patients in regional and rural Australia had a 7% higher cancer mortality compared with those in metropolitan centres, equating to 9000 additional regional and rural cancer deaths.3,5 Barriers to better regional cancer care include travel requirements to metropolitan centres, limited access to expert diagnostics and therapeutics, and less access to clinical trials.6 As well as geographical issues, recruitment and retention of qualified health professionals in regional areas can be difficult, due to professional isolation and a perceived or actual lack of career opportunities.7 These issues relate not only to regional Australia but to many regional populations worldwide.4,8 In the past decade, there has been considerable investment by federal and state governments in the development of regional cancer centres, enabling increased research opportunities.9 Clinical trials remain a gateway to accessing cutting edge therapies and technology. Currently, less than 5% of regional cancer patients participate in any clinical trial; barriers include travel distance to a metropolitan site, a lack of trials available locally, and costs involved for patients and carers such as travel and accommodation and loss of earnings.10 While there are no set targets for participation rates, there has been a correlation between trial participation rates and improved cancer survival, such that a higher rate is desirable.11 In 2017, there were 432 actively recruiting cancer clinical trials in Victoria, totalling 1605 participants. Of these, 426 participants were living in a regional or rural area (27%); however, most participants were enrolled at a metropolitan site, with just 81 (5% of all trial participants) recruited to local clinical trials (personal communication, Christie Allan, Cancer Trials Management Scheme, Cancer Council Victoria, April 2019). Telehealth strategies Telehealth strategies have gained acceptance across many aspects of health care to enable delivery for patients closer to home, including anti‐cancer therapies.12 A logical extension is integration into clinical trial models. Such an approach has many benefits for patients, their families, regional health care, as well as potential economic savings by reducing the need to travel for care. Although this model is a change from usual care, patient safety and quality of care is maintained. The Victorian Comprehensive Cancer Centre (VCCC) is an alliance of ten leading research, clinical and academic institutions in Victoria. The VCCC established a teletrials program to build relationships between regional/rural Victoria and metropolitan centres, using telehealth to provide patients with the opportunity to access clinical trials closer to home. Teletrial framework development In developing a teletrial implementation framework, it was important to consider patient safety, ethical and regulatory requirements. In addition, so that the model would allow for differences across clinical trial requirements and capabilities at individual trial sites, we scoped potential barriers and enablers, to ensure its success. The Clinical Oncology Society of Australia model10 was used as a foundation template for the structure and relational concepts (Box). Importantly, the model recognises the potential for heterogeneity across trials and sites, rather than taking a one‐size‐fits‐all approach. Different sites may perform different roles in different trials; for example, taking blood samples, delivering chemotherapy or medication, trial documentation, or imaging. The model has been used in several teletrials enrolling across Australia.13 An important element was the development of standard operating procedures. Initially developed by Queensland Health, these were modified not only for use in Victoria but for consideration as the basis for national standard operating procedures for teletrials. In developing the teletrial framework, input and feedback were sought from stakeholders in cancer clinical trials. These included contract research organisations; the biopharmaceutical industry; principal investigators; Victorian regional sites through the Regional Trials Network; Human Research Ethics Committees (HRECs); local government through the Victorian Department of Health and Human Services; funding bodies; and consumers. Teletrial supervision plan The teletrial supervision plan (https://www.viccompcancerctr.org/what-we-do/clinical-trials-expansion/teletrials/resources/) contains detailed documentation regarding specific trial conduct and responsibilities, in particular the specific responsibilities of investigators at each site within the trial cluster, and which elements of the trial, imaging and drug delivery are performed at each site. Some trials may have all elements delivered at the local site, others may have most delivered locally but specialist services (eg, radionuclide therapy) at the central site. The supervision plan is site‐, trial‐ and time‐specific. It also includes standard operating procedures, Good Clinical Practice training, monitoring, HREC submissions and oversight, trial‐specific indemnity and contracts, plans for safety reporting, investigational product storage and delivery logistics, and details on joint consultations using telehealth, payments, data entry and document management. The supervision plan is generated in agreement with the principal investigators at the metropolitan and regional sites before the study, but with regular review and modifications as required to allow refinement as needed. Indemnity and legal coverage Teletrial indemnity and legal coverage for trial activities are frequently raised concerns. This can be documented in detail in the supervision plan but is no different for a teletrial over other models. The VCCC commissioned a draft clinical trial activity agreement for investigator‐initiated studies including a teletrial component (https://www.viccompcancerctr.org/what-we-do/clinical-trials-expansion/teletrials/resources/). Governance and ethics approval As with any clinical trial, ethics approval is required, usually through a human research ethics application. Local research governance office requirements will not vary, with local assessment of trial capability, including managing potential toxicities. The principal investigator remains responsible for ethics submissions and communication with HRECs. Each site will obtain local governance approval and be listed on the clinical trial notification form. The process for reporting on safety events remains as per standard of care. Proof of concept Using the framework described, a teletrial has commenced between a metropolitan site and two regional sites in Victoria. The first teletrial site patient was recruited in November 2018 and at 24 July 2020, 91 patients had been successfully recruited in regional centres, with all their trial activity delivered locally. Metropolitan and teletrial sites have successfully undergone study monitoring and further model evaluation is underway. Model evaluation Although the teletrial model is not an intervention in itself, merely a method of trial delivery, it is important to its widespread adoption at a new standard of care that there are benefits to all stakeholders. An ongoing health economic evaluation will evaluate costs associated with the teletrial (and potential savings), patient time and travel estimates, and qualitative assessment of patient and clinician participation in a teletrial to detail possible benefits. In addition, consumer and clinician perspectives studies are planned. A leading contract research organisation was commissioned to undertake an independent process review of the first teletrial to evaluate the model. No major protocol deviations were found in comparison to a conventional site in this pilot study. Potential benefits of a teletrial Teletrials provide a mechanism to enable disadvantaged patients to participate in clinical trials. They may also provide wider benefits14 beyond those experienced by individual participants, including: improved recruitment: as trials have a wider reach, they may recruit faster, translating new interventions to patients faster in a real‐world setting; improved retention: making trial access easier may improve participant retention, reduce missing data and accelerate trial objectives; increased diversity: teletrials may allow for easier access to the increasingly specific and rare subsets of cancer trial populations; professional development: partnerships developed from the trial network may translate into improved routine clinical care delivery and opportunities; and trial cost‐savings: while teletrial costs will be evaluated, the resources required to open a teletrial may be reduced, as much of the trial data will be retained at the primary site. Potential or perceived risks Some of the possible risks raised with the authors by stakeholders have been addressed above, including indemnity, legal and governance issues. Others may include: Clinical safety of new treatments in a regional setting: while a trial may involve a novel therapy, toxicities are often managed on a patient's return home to their regional site. Involving local clinicians in the trial may actually reduce this risk through better education regarding managing novel therapies. Clinical trial expertise: most regional sites already have extensive experience in clinical trials, and Good Clinical Practice training is standard. Trial monitoring challenges: with rapidly increased use of secure digital platforms, monitoring is increasingly becoming a remote activity, so location is not a barrier. We acknowledge that this model represents a change to usual process and therefore requires assessment, transparency and strong support and advocacy to overcome barriers to clinical trial participation.15 Teletrials do more than just meet trial metrics. They develop synchronous partnering between regional and metropolitan centres, allowing regional equity of access to cutting edge diagnostics and therapeutics while maintaining patients’ care delivery closer to home, thereby avoiding disruption to family, work and social interactions. Box – Teletrial model

Ian M Collins · Kate Burbury · Craig R Underhill

Patel C, Chiu CK, Beard FH, et al. One disease, two vaccines: challenges in prevention of meningococcal disease. Med J Aust 2020; 212: 453–456.e1. https://doi.org/10.5694/mja2.50567. In this Perspective article, in Box 1 on page 454, on the x‐axis, where it says: “<12 months, 12–23 months, 2–4 months, 5–9 months, 10–14 months, 15–19 months, 20–24 months, 25–44 months, 45–64 months, ≥65 months”, it should read: “<12 months, 12–23 months, 2–4 years, 5–9 years, 10–14 years, 15–19 years, 20–24 years, 25–44 years, 45–64 years, ≥65 years”.

A 5-year-old girl sustained partial thickness facial burns from oleander leaves that she applied as pretend make-up

Sarah Oudtshoorn · Sanchita Gera · Parshotam Gera

Air pollution poses global health, equity, and environmental problems with short and long term consequences

Julie M Marchant · Anne B Chang

Patience and well designed studies are important for balancing opportunity and risk in uncertain times

Michael Wagels · Dietmar W Hutmacher

Objective: To evaluate associations between exposure during early life to mine fire smoke and parent‐reported indicators of respiratory and atopic illness 2–4 years later. Design, setting: The Hazelwood coalmine fire exposed a regional Australian community to markedly increased air pollution during February – March 2014. During June 2016 – October 2018 we conducted a prospective cohort study of children from the Latrobe Valley. Participants: Seventy‐nine children exposed to smoke in utero, 81 exposed during early childhood (0–2 years of age), and 129 children conceived after the fire (ie, unexposed). Exposure: Individualised mean daily and peak 24‐hour fire‐attributable fine particulate matter (PM2.5) exposure during the fire period, based on modelled air quality and time‐activity data. Main outcome measures: Parent‐reported symptoms, medications use, and contacts with medical professionals, collected in monthly online diaries for 29 months, 2–4 years after the fire. Results: In the in utero exposure analysis (2678 monthly diaries for 160 children exposed in utero or unexposed), each 10 μg/m3 increase in mean daily PM2.5 exposure was associated with increased reports of runny nose/cough (relative risk [RR], 1.09; 95% CI, 1.02–1.17), wheeze (RR, 1.56; 95% CI, 1.18–2.07), seeking health professional advice (RR, 1.17; 95% CI 1.06–1.29), and doctor diagnoses of upper respiratory tract infections, cold or flu (RR, 1.35; 95% CI, 1.14–1.60). Associations with peak 24‐hour PM2.5 exposure were similar. In the early childhood exposure analysis (3290 diaries for 210 children exposed during early childhood, or unexposed), each 100 μg/m3 increase in peak 24‐hour PM2.5 exposure was associated with increased use of asthma inhalers (RR, 1.26; 95% CI, 1.01–1.58). Conclusions: Exposure to mine fire smoke in utero was associated with increased reports by parents of respiratory infections and wheeze in their children 2–4 years later.

Gabriela A Willis · Kate Chappell · Stephanie Williams · Shannon M Melody · Amanda Wheeler · Marita Dalton · Shyamali C Dharmage · Graeme R Zosky · Fay H Johnston

Collecting nasal samples with 3D-printed swabs is feasible, acceptable to patients and health carers, and convenient

Eloise Williams · Katherine Bond · Nicole Isles · Brian Chong · Douglas Johnson · Julian Druce · Tuyet Hoang · Susan A Ballard · Victoria Hall · Stephen Muhi · Kirsty L Buising · Seok Lim · Dick Strugnell · Mike Catton · Louis B Irving · Benjamin P Howden · Eric Bert · Deborah A Williamson

Ambient air quality in Australia is regulated by the National Environment Protection Measure (NEPM), which sets a maximum 24‐hour mean concentration of 50 μg/m3 for particulate matter less than 10 μm in diameter (PM10) and 25 μg/m3 for PM2.5. Each state and territory is required by the NEPM to annually report all breaches of this standard, including the sources of pollution.1 We analysed NEPM reports for Western Australia to identify days during 1 January 2002 – 31 December 2017 on which atmospheric particulate matter levels exceeded air quality standard levels, and classified them according to the most frequently reported sources of pollution: prescribed burns, wildfires, and other (crustal particles such as dust, wood smoke, and indeterminate). During 2008–2013, exceedances caused by smoke from prescribed burns, wildfires, and wood smoke were all recorded by the WA Department of Environment Regulation as “smoke haze”. For this period, we therefore applied a random forest algorithm, a machine learning method that uses a random sample of observations for known classifications to predict the classifications for new data.2 We included the variables month, day of the week, temperature, and pollution level as model predictors. To estimate background PM2.5 level, we obtained historical hourly values for PM10 and PM2.5 from the WA Department of Water and Environmental Regulation3 and calculated historical monthly means, excluding days on which particle levels exceeded the air quality standard. We estimated daily PM2.5 concentrations attributable to smoke events by subtracting the background PM2.5 level from measured daily values. Applying standard methods for assessing the health impact of air pollution,4 we estimated the numbers of premature deaths, hospitalisations for cardiovascular and respiratory problems, and emergency department presentations with asthma attributable to elevated PM2.5 levels. We used the value of statistical life (VSL)5 to estimate costs associated with premature mortality. The VSL is based on the willingness to pay for reduced risk of premature mortality, and does not take into account underlying health status, age, or life expectancy of individuals. Deaths associated with acute exposure to increased air pollution are more likely among people at greater risk because of advanced age or chronic illness.6 We estimated hospital service costs according to the mean cost of each episode of care as reported in the Independent Hospital Pricing Authority national cost data collection report7 and the Health Policy Analysis emergency care costing report.8 We also undertook a sensitivity analysis in which we excluded data for 2008–2013, when exceedances caused by smoke from prescribed burns, wildfires, and wood smoke were all recorded in NEPM reports as “smoke haze”. Further details on our methods, including underlying assumptions and limitations, are included in the online Supporting Information. During 2002–2017, particulate air pollution exceeded the national standard on 271 of 5844 days (4.6%), including 197 days (73%) attributable to prescribed burns or wildfires. We estimated that 41 (95% confidence interval [CI], 15–68) premature deaths, 99 (95% CI, 19–182) hospitalisations for cardiovascular problems and 174 (95% CI, 0–373) for respiratory conditions, and 123 (95% CI, 70–179) emergency department visits with asthma were attributable to elevated PM2.5 concentration (Box 1). Total estimated health costs were $188.8 million (95% CI, $68.1–311.1 million); $97.1 million (51%) was attributable to prescribed burns and $77.7 million (41%) to wildfires. Mean estimated health costs were lower on days affected by smoke from prescribed burns ($703 984; 95% CI, $254 064–$1.2 million) than those affected by wildfire smoke ($1.3 million; 95% CI, $475 000–$2.2 million), although more days were affected by prescribed burns (138) than by wildfires (59). The estimated smoke‐related costs of wildfires were highest in 2012 ($24.8 million); in many years, prescribed fires often accounted for most health‐related costs, peaking in 2017 ($24.1 million) (Box 2). In our sensitivity analysis excluding the period 2008–2013, the relative costs by source were similar (prescribed burns, 53% [$58.4 million]; wildfires, 38% [$41.6 million]; Supporting Information). Particulate matter in fire smoke is associated with adverse health outcomes,9 even at relatively low concentrations.10 Landscape fire smoke was the greatest contributor to excessive atmospheric particulate matter levels in WA during 2002–2017 and was associated with substantial health costs. Our estimates of the health impacts may be conservative, as we included only days when PM2.5 concentrations exceeded the national standard, excluding smoky days on which the air quality standard was not breached. Further, our selection of health outcomes did not encompass the total health burden attributable to smoke exposure. Our study highlights the different smoke‐related health effects and costs of infrequent severe wildfire and regular prescribed burning. While prescribed burning reduces the risk of wildfire, better understanding and incorporation into control strategies of the full health impacts of each type of fire are needed for sustainable fire management.11 Box 1 – Estimated health burden attributable to elevated PM2.5 concentrations, Western Australia, 2002–2017, by particulate matter source Outcome Estimated number of cases (95% confidence interval) Prescribed burns Wildfires Other Total Excess deaths (any cause) 21 (8–35) 17 (6–28) 3 (1–5) 41 (15–68) Hospital admissions, cardiovascular 51 (10–94) 41 (8–75) 7 (1–13) 99 (19–182) Hospital admissions, respiratory 89 (0–192) 72 (0–154) 13 (0–27) 174 (0–373) Emergency department attendances, asthma 63 (36–91) 51 (29–75) 9 (5–13) 123 (70–179) Box 2 – Estimated health costs tributable to elevated PM2.5 concentrations, Western Australia, 2002–2017, by particulate matter source

Nicolas Borchers Arriagada · Andrew J Palmer · David MJS Bowman · Fay H Johnston

Weather conditions conducive to extreme bushfires are becoming more frequent as a consequence of climate change.1 Such fires have substantial social, ecological, and economic effects, including the effects on public health associated with smoke, such as premature mortality and exacerbation of cardio‐respiratory conditions.2,3 During the final quarter of 2019 and the first of 2020, bushfires burned in many forested regions of Australia, and smoke affected large numbers of people in New South Wales, Queensland, the Australian Capital Territory and Victoria. The scale and duration of these bushfires was unprecedented in Australia. We undertook a preliminary evaluation of the health burden attributable to air pollution generated by bushfires during this period. Using standard methods for assessing the health impact of air pollution,4 we estimated the numbers of excess deaths, hospitalisations for cardiovascular and respiratory problems, and emergency department presentations with asthma in NSW, Queensland, the ACT and Victoria between 1 October 2019 and 10 February 2020 that could be attributed to bushfire smoke exposure. We estimated population exposure to particulate matter less than 2.5 μm in diameter (PM2.5) for the regions of NSW, Queensland, the ACT and Victoria for which publicly available air quality monitoring data were available (for about 90% of the total population of these states). Data were obtained from the NSW Department of Planning, Industry and Environment,5 the Queensland Department of Science,6 ACT Health,7 and the Environmental Protection Agency Victoria.8 We defined bushfire smoke‐affected days as days on which the 24‐hour mean PM2.5 concentration exceeded the 95th percentile of historical daily mean values for individual air quality stations. We estimated daily mean PM2.5 levels by Statistical Area Level 2 (SA2), using station level data whenever at least one monitoring station was within 100 km of the SA2 centroid, and applying inverse distance weighting.9 Published population and health data from the Australian Bureau of Statistics,10,11 the Australian Institute of Health and Welfare,12,13,14,15 and the NSW Ministry of Health were used.16 We quantified health outcomes by combining baseline incidence rates12,13,14,15 for each health outcome with daily exposure data and applying the relevant exposure–response risk coefficients for each outcome.17,18 We also conducted sensitivity analyses with different PM2.5 thresholds for defining bushfire smoke‐affected days. Further methodological details, including underlying assumptions and limitations, are included in the online Supporting Information. Our analysis of publicly available aggregated data did not require ethics approval. During the study period, PM2.5 concentrations exceeding the 95th percentile of historical daily mean values were recorded by at least one monitoring station in the study area on 125 of 133 days (Box 1). We estimated that bushfire smoke was responsible for 417 (95% CI, 153–680) excess deaths, 1124 (95% CI, 211–2047) hospitalisations for cardiovascular problems and 2027 (95% CI, 0–4252) for respiratory problems, and 1305 (95% CI, 705–1908) presentations to emergency departments with asthma (Box 2). Applying lower thresholds for defining bushfire smoke‐affected days (no threshold, 90th percentile of historical values) did not markedly alter our findings; a higher threshold (99th percentile) reduced the estimates by about 20%. The highest population‐weighted PM2.5 exposure level, 98.5 μg/m3 on 14 January 2020 (Box 1), exceeded the national air quality 24‐hour standard (25 μg/m3)19 and was more than fourteen times the historical population‐weighted mean 24‐hour PM2.5 value of 6.8 μg/m3. We have estimated the excess health burden during 19 weeks’ continuous fire activity in the states most severely affected by smoke. Our estimates are based on air quality data from monitoring stations in the four eastern states — that is, we did not include data for smoke from all extreme fires in Australia during the study period — and we did not attempt to estimate health effects for which exposure–response relationships are less well characterised, such as primary health care attendances and ambulance calls. Detailed epidemiological analysis of more comprehensive exposure estimation and empirical health data will provide more complete information about the harms attributable to the severe air pollution associated with these unprecedented fires, but our findings indicate that the smoke‐related health impact was substantial. Smoke is just one of many problems that will intensify with the increasing frequency and severity of major bushfires associated with climate change. Expanded and diversified approaches to bushfire mitigation and adaptation to living in an increasingly hot and fire‐prone country are urgently needed.20 Box 1 – Population‐weighted PM2.5 levels, New South Wales, Queensland, the Australian Capital Territory and Victoria, 1 October 2019 – 10 February 2020* * Data by state are included in the online Supporting Information. Box 2 – Estimated health burden attributable to bushfire smoke, Queensland, New South Wales, the Australian Capital Territory and Victoria, 1 October 2019 – 10 February 2020 Outcome Estimated number of cases (95% confidence intervals) Queensland New South Wales Australian Capital Territory Victoria Total Excess deaths (any cause) 47 (17–77) 219 (81–357) 31 (12–51) 120 (44–195) 417 (153–680) Hospital admissions, cardiovascular 135 (25–246) 577 (108–1050) 82 (15–149) 331 (62–602) 1124 (211–2047) Hospital admissions, respiratory 245 (0–513) 1050 (0–2204) 147 (0–308) 585 (0–1227) 2027 (0–4252) Emergency department attendances, asthma 113 (61–165) 702 (379–1026) 89 (48–131) 401 (217–586) 1305 (705–1908)

Nicolas Borchers Arriagada · Andrew J Palmer · David MJS Bowman · Geoffrey G Morgan · Bin B Jalaludin · Fay H Johnston

To the Editor: The incidence of bushfires, forest fires and wildfires, is increasing globally. Epidemiology shows that individuals with chronic respiratory diseases are most affected with increased hospitalisations. However, the impacts or safe exposure levels of bushfire smoke are not well known.1 We were recently awarded the Medical Research Future Fund's Bushfire Impact Research grant 2020 and in this project we will address the following questions: How does bushfire smoke exposure affect respiratory health? How does it exacerbate chronic respiratory diseases and affect different age groups? What are the impacts on cells, tissues and molecular pathways? How can we target the effects therapeutically? Bushfire smoke is a complex mix of inspirable particles, volatile organics, aldehydes, carbon monoxide, and particulate matter (PM).2 Although extensive research evaluating the effects of bushfire smoke has not been carried out, studies utilising cigarette smoke or vehicular PM10−2.5 show that exposure to these insults induces lung inflammation and oxidative stress, and promotes the progression of chronic respiratory diseases.3,4,5 Further, in vitro studies with healthy human fibroblasts and bronchoepithelial cells show that bushfire smoke affects pathways including oxidative stress, barrier function, innate defence, and autophagy.6 Accordingly, we plan to expose mice to the different PM particles from bushfire smoke and will elucidate the acute and prolonged effects on lung inflammation, airway remodelling and lung function. In addition, by using our mouse model of chronic respiratory diseases (chronic obstructive pulmonary disease, asthma) and mice at different ages (pregnant, infant, aged), we will assess the impact of bushfire smoke on predisposition, pathogenesis and progression of chronic respiratory diseases. We will use advanced molecular and multi‐omics (single cell/tissue sequencing, proteomics, epigenetics) technology to elucidate cell and tissue responses. Furthermore, we will define therapeutic avenues for prevention and treatment (antioxidants, metabolic modulators) (Box). The outcomes of this project will inform the development of safe exposure guidelines and define preventive/treatment measures. Moreover, we will address evidence gaps related to harmful health effects of hazardous bushfire smoke exposure which we hope will aid government and health agencies to design appropriate policies, prevention measures, and treatment strategies to deal with future bushfire smoke events. Box – Methodology for evaluating the impact of bushfire smoke COPD = chronic obstructive pulmonary disease; PM = particulate matter.

Vivek Dharwal · Keshav R Paudel · Philip M Hansbro

To the Editor: In Australia, 2020 began with raging bushfires, and we now confront the coronavirus disease 2019 (COVID‐19) pandemic. While health protection is currently at the top of the public policy agenda, can we rise from these huge ruptures and “build back better”? The full health costs of the bushfires, including the mental health toll, are yet to be quantified. No sooner had the bushfires abated than the battle against the COVID‐19 pandemic began. The immediate public health response has been well managed in Australia.1 Although Victoria is currently grappling with a second wave of infections, by international comparisons the number of cases and deaths around the country has remained low.2 Government leaders have listened to health experts and acted on evidence, including the need for strict physical distancing in the absence of a vaccine, supplemented by universal masking in Victoria. As governments move to revitalise the economy with financial stimulus, what guidance can health experts provide to inform this stimulus? One clear priority is that stimulus accelerates the decarbonisation of the Australian economy. Climate change is a recognised health issue. Published as the bushfires erupted, the 2019 MJA–Lancet Countdown on health and climate change report3 found that Australia is extremely vulnerable to the impacts of climate change on health. There are also health co‐benefits from action on climate change. The clearest example is the transition to renewable energy generation. Globally, in 2015 alone, more than 460 000 preventable deaths were attributable to coal burning.4 An urgent transition to renewable energy would be an evidence‐based public policy response to these deaths and assist a global green recovery from the pandemic which is called for by the World Health Organization.5 Australia is well placed to lead such a recovery as indicated in a recent report by ClimateWorks Australia,6 which provides a blueprint to achieve net zero emissions by 2050 through accelerated uptake of mature zero emission technologies and the rapid development and commercialisation of emerging zero emission technologies in harder to abate sectors (Box). Beyond stimulus for decarbonisation, investments in affordable housing, mass transit infrastructure, safe routes for walking and cycling, regeneration of degraded ecosystems and infrastructure to support working from home would also benefit health through reduced homelessness, improved levels of physical activity, and improved urban air quality. Australia has, thus far, avoided the high COVID‐19 case numbers and death rates seen in some other countries because of evidence‐based decision making. It is essential that decisions about the stimulus for economic recovery are similarly grounded in evidence. The health and wellbeing of current and future generations of Australians depend on it. Box – Summary table of key emissions‐reduction solutions by sector CCS = carbon capture and storage.

Selina N Lo · Anna Skarbek · Anthony Capon

To the Editor: We read with interest the article by Koff and Lyons1 and agree that there is a need to develop, implement and evaluate health systems around patient needs and wishes. Implementing value‐based health care is an excellent initiative to address sustainability and patient‐centred care.2 Genuine reform requires a transition away from volume‐based service contracting towards a multidisciplinary approach focused on evidence of improved outcomes.1,2 This would reward doctors and the system for keeping patients healthy and independent in their own homes, with community support, for as long as possible.2 The Leading Better Value Care initiative (2016–2020)1 may be misinterpreted as another set of top‐down policies. It may also have unintended consequences such as reinforcing the silo approach to disease states, diverting finite hospital and local health district resources, such as staff and expertise, to these 13 policy‐driven priority projects. In our work in perioperative health care, we have identified some concerns. First, the sustainability of our health systems is tested by patients who are frailer, who have chronic diseases, and who present for high risk surgery.1,3,4 Second, these patients have a higher incidence of post‐operative complications3,4 and are more likely to be discharged to a higher care facility, rather than back to their home.4 Third, performing surgery on these patients is associated with higher costs and hospital readmissions.4 Fourth, our research has found that past policy for surgical patients5 has led to today's “wicked problem”; that is, frontline perioperative clinicians and managers are dealing with lack of time, increased demand for precision, fragmentation of care, lack of coordination across an episode of care, bed block, complexity of care, and unclear patient outcome measures. In this context, work is required to empower patients and staff in shared decision making to understand the true complexity of risks and outcomes associated with high risk surgery. In conclusion, implementing statewide value‐based care is timely and can be transformational. The high risk surgical patient cohort and the staff providing their care are likely to benefit from, and should be included in, this important reform agenda.

Su‐Jen Yap · Roberto Forero · David Greenfield · Kenneth M Hillman

To the Editor: Poor oral health profoundly affects a person's ability to eat, speak, socialise, work and learn.1 It has an impact on social and emotional wellbeing, productivity in the workplace, and quality of life. Pain from dental caries is a common experience. In children, dental caries may require treatment under a hospital‐based general anaesthetic — at considerable cost and itself not without risk.2 Poor oral health in childhood is the leading cause of poor adult oral health.1 A higher proportion of Australians who are socially disadvantaged have dental caries. In the 2012–2014 National Child Oral Health Survey, the mean number of deciduous teeth with dental caries in Indigenous children aged 5–10 years was 6.3 (95% CI, 5.2–7.4) compared with 2.9 (95% CI, 2.7–3.1) among non‐Indigenous children.3 In the 2004–2006 National Survey of Adult Oral Health, almost 60% of Indigenous adults had untreated dental caries compared with 25% of non‐Indigenous Australians.4 In the interests of equity, it is desirable for water fluoridation to provide a greater benefit to groups carrying the highest burden of disease. In Australia, this is the Indigenous population. Community water fluoridation is one of the most effective public health interventions of the 20th century. Its success has been attributed to wide population coverage with no concurrent behaviour change required. Evidence in Australia demonstrates that community water fluoridation has decreased both the prevalence (proportion of population) and severity (amount per person) of tooth decay by 44% in children and 27% in adults.5 However, nearly 3 million Australians (11% of the population) cannot access a fluoridated water supply.5 Access to fluoridated water in Australia varies. In Queensland before 2008, access was limited to 5% of the population.5 At that time, there were higher rates of untreated dental caries in non‐fluoridated than in fluoridated communities. In 2008, the Queensland Government mandated water fluoridation for all community water supplies that serviced communities of more than 1000 people; 134 water supplies were identified. Within 4 years, 90% of Queenslanders had access to fluoridated water and rates of dental caries declined.6 After the 2012 Queensland election, the new government overturned mandatory water fluoridation, with the decision to fluoridate community water reverting to water supply authorities. The subsequent deactivation of water fluoridation plants in 18 local government areas reduced the population coverage to around 76%. This had a disproportionate impact on Indigenous Australians, who are more likely to reside in areas where water fluoridation ceased after 2012 or in areas where it was never implemented. The consequence is that only 50% of the Indigenous population in Queensland have access to fluoridated water compared with 76% of non‐Indigenous Queenslanders.7 The denial of access to fluoridated drinking water for Indigenous Australians is of great concern. We urge the Commonwealth government, through current negotiations for funding agreements for public dental care, to mandate that all states and territories maintain a minimum standard of 90% population access to fluoridated water. Water fluoridation would then be an effective as well as socially equitable public health intervention to reduce the oral health inequalities between Indigenous and non‐Indigenous Australians.

Andrew McAuliffe · Chris Bourke · Lisa M Jamieson

To the Editor: We thank Davidson and colleagues1 for their comprehensive narrative review on skin infections in Australian Aboriginal children. A significant factor in both individual and mass drug administration therapy of scabies is the uncertainty regarding the safety of oral ivermectin in small children and during pregnancy. Australian guidelines state ivermectin should not be used in children aged under 5 years or who weigh less than 15 kg or in pregnant women.2 A retrospective cohort study of 170 children aged 1–64 months (median age, 15 months) or weighing under 15 kg treated with ivermectin (mean dose, 223 μg/kg) found only minor self‐limiting adverse effects in seven patients (4%).3 A review of previous literature found 60 children aged under 5 years or weighing less than 15 kg who had been treated with ivermectin at a dose range of 150–200 μg/kg for whom safety data were available.4 Only four of 60 children (7%) developed an adverse reaction, all of which were benign and transient, with no long term sequelae. A recent study of oral ivermectin (dose 400 μg/kg) in the treatment of head lice revealed no adverse effects in 54 children aged under 5 years.5 The Ivermectin Exposure in Small Children Study Group expected to commence the analysis in late 2019 of data collected from 2017 to 2019.6 Three studies totalling 363 women with inadvertent maternal exposure to ivermectin 150 μg/kg (76–85% in first trimester) for filariasis and onchocerciasis found no increased risk of congenital malformations, miscarriage or stillbirth.7 A study of 199 pregnancies with maternal treatment in the second trimester with ivermectin and albendazole, and 198 with ivermectin alone in the management of helminth infections, found no increased risk of adverse pregnancy outcomes.8 In France, the use of oral ivermectin is permitted during pregnancy and in children weighing less than 15 kg when topical therapy has failed.9 Further published data regarding the safety of ivermectin in these populations would be useful, particularly with respect to mass drug administration programs.

Sarah K Morton · Adam Morton

In reply

Lucy Davidson · Asha C Bowen