Letters

To the Editor: We read with interest the letter by Withers and colleagues1 highlighting the research waste generated by the inclusion of research in the selection criteria for specialist training in Australia and Aotearoa New Zealand. We agree the inclusion of research for selection or completion of specialty training produces unintended incentives that contribute to poor quality studies and research waste. We also support the notion that “research utilisation, research training, or participation in large research teams” should be prioritised by colleges. Selection criteria for medical specialty training should incentivise contribution to high quality projects and the development of research literacy and skills, rather than publishing many small, low impact articles. Student‐ and trainee‐led collaborative research groups, such as the TASMAN (Trials and Audit in Surgery by Medical Students in Australia and New Zealand) Collaborative, provide a promising solution to this research waste.2,3 Similar groups have emerged locally and internationally and have successfully delivered large randomised controlled trials and cohort studies.2,3,4 These high impact publications have provided practice‐changing results4 as well as training and opportunities for collaborators to develop research skills. Locally, medical students, junior doctors and surgical trainees have contributed to the recent SUNRRiSE (Single Use Negative Pressure Dressing for Reduction in Surgical Site Infection Following Emergency Laparotomy) randomised controlled trial,5 and POSTVenTT (Postoperative Variations in Anaemia Treatment and Transfusions) prospective audit,6 the results of which are eagerly awaited. Despite this, the contribution to collaborative research studies is not currently recognised for selection into most specialty training programs in Australia or Aotearoa New Zealand. We echo the calls of our international counterparts for participation in collaborative research to be accounted for in applications for postgraduate training.7 We applaud the Royal Australasian College of Surgeons and General Surgeons Australia for incorporating collaborative research as part of the selection criteria and the points‐based research requirements during General Surgical Education and Training in 2022,8,9 and hope that other specialties follow suit. We look forward to the emergence of student‐ and trainee‐led collaborative groups from other medical specialties in Australia and Aotearoa New Zealand. Further recognition of collaborative research will improve research skills in medical graduates, reduce research waste and, most importantly, generate meaningful data to improve patient outcomes.

TASMAN Collaborative

To the Editor: We present a case of hypotensive shock after insulin pump delivery interruption during a diabetes technology trial, and offer a systematic insulin pump troubleshooting approach.1 A 61‐year‐old woman with long‐duration type 1 diabetes using an insulin pump, coronary artery disease and hypertension presented to an emergency department with hyperglycaemia, hypotension and presyncope. She had self‐identified insulin pump delivery interruption but felt too unwell to troubleshoot independently. Initial subcutaneous insulin bolus dose administration via the pump as directed by the hospital doctor was ineffective and diabetic ketoacidosis developed. In the context of cardiovascular comorbidities and antihypertensives, including transdermal glyceryl trinitrate, the patient deteriorated rapidly progressing to hypotensive shock requiring intensive care. Insulin delivery interruption was ultimately attributed to line blockage; earlier identification and prompt administration of insulin via an alternative route may have avoided the development of ketoacidosis. Therapeutic diabetes technology is evolving, with an increasing proportion of people with type 1 diabetes now using insulin pumps to replace lost pancreatic β‐cell function.2 Insulin pumps subcutaneously infuse rapid‐acting insulin alone, providing both background basal and bolus insulin doses. These insulin preparations have peak effect at 1–2 hours and duration of action of about 4 hours. Therefore, for people with type 1 diabetes, insulin pump delivery interruption typically causes relative insulin deficiency within 2 hours and absolute insulin deficiency within 4 hours with consequent risk of rapid development of hyperglycaemia, ketosis and ketoacidosis.3 Pump users are routinely instructed on how to troubleshoot insulin delivery problems, including proactive self‐management of pump‐related issues; however, management may revert to clinicians during acute illness. All clinicians should therefore be familiar with common complications during insulin pump therapy, and when to convert to an alternative route of insulin administration to prevent rapid metabolic deterioration.4 The treatment of hyperglycaemia and ketosis is time‐critical, and decisions regarding insulin pump continuation during hospital admissions should be individualised within the acute context.4,5 Once stabilised, patients should reconnect with their diabetes management team for ongoing education. We present a systematic approach to managing rising blood glucose and/or ketones during insulin pump use (Box). As the vignette illustrates, individuals with medical comorbid conditions are at risk of rapid, life‐threatening deterioration after insulin pump delivery interruption. Box – Clinical practice flowchart of steps to troubleshoot insulin pump delivery‐related problems SGLT‐2 = sodium glucose co‐transporter 2. The potential causes of insulin delivery interruption are presented systematically from the site of insulin delivery to the insulin pump. * Hyperglycaemia generally considered to be glucose level>14mmol/L. † Ketosis generally considered to be ketone level>0.6mmol/L. Clinicians should consider the flowchart to be a general guide; always adapt treatment approach to the clinical presentation and refer to local policies and procedures. Management of hyperglycaemia and ketosis is time‐critical; refer to the Royal Australian College of General Practitioners and Australian Diabetes Society joint clinical position statement for a primary care emergency management algorithm.4 Assess whether it is clinically appropriate for hospitalised patients to continue insulin pump therapy.5 Inset: Schematic diagram of the main components of insulin pump therapy; the insulin “infusion set” comprises the cannula (inserted subcutaneously), delivery line and insulin reservoir.

Anindita Chakrabarti · Richard J MacIsaac · Sybil A McAuley

To the Editor: We thank Polkinghorne and Kerr1 for their editorial on the health care needs of people with chronic kidney disease (CKD). We write to highlight the burden of visual loss suffered by people with CKD and its effects on quality of life and mortality. The prevalence of eye diseases associated with visual impairment (combined World Health Organization definitions of blindness and low vision) in people with CKD is about 36%, rising to about 60% in people with end‐stage renal disease.2 The commonest causes of visual impairment are diabetic retinopathy (prevalence in patients with CKD, 19–46%), cataract (prevalence, 33–75%) and age‐related macular degeneration (prevalence, 8–36%).2 Increased incidence of retinal vein occlusions and hypertensive retinopathy are also associated with CKD. Patients receiving haemodialysis have increased risk of additional ocular complications. A recent study of 121 patients found at least one ocular finding in over 89% cases, including conjunctival (32%) and corneal (32%) calcification, and optic atrophy (19%).3 Ectopic calcification has been a common cause of irritated eyes in patients receiving dialysis, but this may be decreasing with modern dialysis methods. Increased risk of optic atrophy may be related to chronic anaemia and an increased risk of non‐arteritic anterior ischaemic optic neuropathy. Uraemic optic neuropathy is now an uncommon event. People with CKD are recognised to have an increased risk of dying from heart disease (up to 20 times that of age and gender matched people for people on haemodialysis).4,5 Vision impairment may incrementally contribute to increased mortality; a recent meta‐analysis demonstrated an all‐cause mortality hazard ratio of 1.43 (95% CI, 1.22–1.68) for visual acuity worse than 6/18.6 Vision‐related quality of life is potentially reduced in people with CKD, affecting their ability to complete activities of daily living and their social, emotional and economic wellbeing. Vision loss is associated with increased risk of falls and increased mental health burden, typically anxiety and depression, and may limit the ability to live independently, including self‐medicating with insulin and performing home dialysis. Multiple studies show people rate losing vision as worse than losing hearing, memory, speech or a limb.7 We agree that supportive care clinics for people electing to not receive kidney replacement therapy are important, and suggest that these clinics include regular eye care services.

Heather G Mack · Deborah J Colville · Judith A Savige

To the Editor: Recently published studies have highlighted concerns that pulse oximeter devices may underestimate hypoxia (overestimate oxygen saturation) in patients with dark skin. This occurs at levels where key decisions are made around supplemental oxygen and hospital admission (arterial oxygen saturation [SaO2] 88–94%). Amid the coronavirus disease 2019 (COVID‐19) pandemic, this important public health issue prompted the Therapeutic Goods Administration (TGA) to publish a medical device safety update.1 Long‐standing concerns about reduced pulse oximeter accuracy in people with dark skin2,3 have evolved into characterisation of significant racial discrepancies. A recent article compared 48097 pairs of measurements by pulse oximetry and arterial blood gas (ABG) in adults receiving oxygen across 179 hospitals in the United States. Among patients saturating >92% on pulse oximetry, hypoxaemia (ABG saturation<88%) was nearly three times more common in black patients than in white patients. Graphs illustrate the median oxygen saturation bias for black patients was 3% in the 89–96% range.4 A recent retrospective cohort study analysed registry SaO2 data in 372 individuals (73.1% with COVID‐19) about to undergo extracorporeal membrane oxygenation for respiratory failure. In patients with pulse oximeter readings of 92–96%, ABG oxygen saturation was <88% in 21.5% of black patients and in 10.2% of white patients.5 In these retrospective audits, patient ethnicity was based on hospital record identification, not skin colour. The oximetry devices used were not specified. COVID‐19 guidelines may incorporate pulse oximeter readings into decisions regarding hospital transfer of home‐care patients. Clinicians and services should arguably have lower thresholds for hospital review and admission of patients with dark skin (including Indigenous Australians and those of African and South Asian descent) with borderline oxygen saturations, while balancing risks of increased ABG and invasive treatment rates. Device manufacturers should develop pulse oximeters that perform accurately across more diverse populations. Further research must identify mechanisms of racial discrepancies in oxygen saturation and address them through device design. Calibration and testing processes for existing devices should be strengthened. The pre‐market approval processes of the US Food and Drug Administration currently require that only 15% of a study population have darker skin, while Australia has no specified requirement. The TGA does not regulate pulse oximeters sold directly to consumers (in stores or online) for general wellness or sporting purposes only. There are significant constraints on the scope for regulatory interventions for medical device pulse oximeters: this essential equipment cannot be excluded from the market or its supply compromised; mandating changes to instructions for use may have limited impact; differentiating between devices is hampered by evidence limitations and any consequent actions would be legally fraught; and mandating accuracy studies would be difficult to enforce. Contemporary evidence demonstrates that dark skin is a risk factor for hypoxia being undetected by pulse oximetry. Clinicians should adjust treatments and guidelines accordingly and consider audits of devices used in their institutions. Failure to address this problem at a design and testing level compromises racial equity in health care outcomes.

Jeffrey J Brownscombe · Heather Loane · Bridget Honan

To the Editor: The rapid uptake of telehealth has been a cornerstone of the response to the coronavirus disease 2019 (COVID‐19) pandemic, and has ensured the provision of essential health care despite restrictions and lockdowns. Although not new technology, telehealth has dramatically increased in prominence and received broad acceptance by doctors and patients alike. Given its success, the Australian Government has confirmed the permanent retention of multiple telehealth item numbers within the Medicare Benefits Schedule (MBS).1 However, it is concerning that this announcement also contained the fine print that the long‐standing MBS incentive for providing telepsychiatry consultations to rural and remote patients will be abolished. This is despite patients in rural and remote communities experiencing well established difficulties accessing health care and having poorer outcomes than their metropolitan counterparts.2 Telehealth consultations have occurred in psychiatry since well before the COVID‐19 pandemic, and have filled an important gap in the workforce by increasing services available in rural and remote areas.3 Video‐based consultations are particularly suited to psychiatry as the key skills of history taking, mental state examination, and psychotherapy do not require physical proximity. Delivering diagnostic assessment and psychological treatment via telehealth have long been demonstrated to be effective and tolerable.4,5 The MBS item number 288 was introduced in 2011 as an adjunct billing code that attracted a 50% loading for psychiatric consultations conducted via telehealth for patients located in a rural or remote setting, aged care facility, or Aboriginal health service. This loading incentivised bulk‐billing of these telehealth assessments. The deletion of this item number from 1 January 2022 will likely result in two adverse consequences: i) fewer telepsychiatry consultations to rural and remote locations will be bulk billed, and ii) telepsychiatry appointments that previously were only available for rural and remote patients will increasingly be offered to metropolitan patients. This will ensure fewer and less affordable options. The cessation of the rural loading for telehealth assessments is a retrograde step that is likely to further entrench long‐standing inequities in both access to care and patient outcomes for psychiatric patients who do not live in the cities. The 288 item number should be reinstated or replaced with an alternative funding mechanism to ensure bulk billed consultations continue to be available for rural and remote patients.

Michael J Weightman

To the Editor: The socio‐economic modelling of the impact of a hypothetical disease‐modifying treatment (DMT) for Alzheimer disease by the National Centre for Social and Economic Modelling (NATSEM)1 is an interesting contribution to what is a critical question for policymakers: how effective does a new antidementia treatment need to be to justify a given cost to the community? Unfortunately, the report does not address this, and several internal deficits call into question the conclusions: • The disease progression pathway lacks backwards transitions. It is commonly understood that mild cognitive impairment is an unstable diagnostic state; individuals are at higher risk for transition to dementia, but a predictable proportion also spontaneously revert back to cognitive normality.2 • The durability of DMT efficacy is unrealistic. It is implausible to assume that a 12‐month treatment with an anti‐amyloid will deliver lifelong cognitive benefits after cessation. In the EMERGE and ENGAGE trials,3 the treatment was for the duration of the trials (18 months), and it is generally accepted that this form of treatment will require infusions for years. • The clinical efficacy of DMT is unfounded. Scientific opinion is divided as to whether modification of cerebral amyloid burden has clinical benefits on cognition or daily function. Further, there is no good reason to assume that a 23% relative difference on the continuous measure of cognitive decline observed in EMERGE — incidentally, not replicated in the identically designed ENGAGE trial3 — will translate to a relative difference in categorical transitions between mild cognitive impairment and mild or moderate dementia. Given this is the main driver of projected cost savings, assuming a 25% reduction in such transitions is unrealistic4 and, surprisingly, not subject to sensitivity analysis. • The adverse costs of DMTs are not modelled. It is not appropriate to model presumed clinical benefits of a hypothetical DMT without accounting for the personal, medical and social cost of their documented adverse effects, including cerebral oedema, brain haemorrhage, and falls.2,5 Given that the list price of any such DMT to the health system was also deliberately not modelled, NATSEM is encouraged to address these concerns in a revised report.

Michael Valenzuela

To the Editor: We congratulate Schutte and colleagues1 for their call to action for improved management of blood pressure in Australia, highlighting that 68% of people have uncontrolled high blood pressure. The burden of high blood pressure is unevenly distributed, with Aboriginal and Torres Strait Islander (hereafter referred to respectfully as Indigenous) people reportedly having a higher rate of high blood pressure than non‐Indigenous Australians in every age group.2 Reducing the prevalence of high blood pressure is one of the most important means of reducing serious circulatory diseases, which are among the leading causes of death for Indigenous Australians.3 Hence, we want to extend the call to action and report on what is happening in primary health care settings with the control of blood pressure for Indigenous people. During 2012–13, we analysed blood pressure screening and follow‐up for patients diagnosed with hypertension (n = 6523) from 123 primary health care centres across Australia using continuous quality improvement data from audits of adherence to best practice chronic illness care.4 Given there are no recently available data on follow‐up actions after an abnormal blood pressure reading at this geographic scale, and as blood pressure continues to be relatively uncontrolled, these data continue to provide unique insight. The data, aggregated at primary health care centre level (Box), tells a story of clinical inertia. Regular blood pressure screening was done well — centres on average completed blood pressure screening for about 90% of patients within the past 6 and 12 months. In this cohort, about 65% of patients recorded abnormal blood pressure (n = 4240). Most primary health care centres had documentation of a follow‐up plan for more than 70% of patients, but there was wide variation (range, 0–100%; Box). Dealing with the low levels of medication reviews and adjustments (mean, ~15%; range, 0–100%) is a vital early step in limiting the contribution of uncontrolled blood pressure to adverse health outcomes for Indigenous people. These data support the need for training on strategies to overcome clinical inertia, which was identified as a top priority by over 200 Indigenous primary health care practitioners, managers and policymakers.5 We add to the call for more attention on prevention of cardiovascular disease and suggest additional investment in evidence‐based interventions appropriate to Indigenous Australian culture and needs. The time for system‐wide action has come. Box – Boxplots showing a record of scheduled services received by patients with hypertension and follow‐up of abnormal findings within the last 12 months of audit (unless otherwise indicated) at primary health centres during 2012–13 x = mean value. More information on how to interpret box plots is available in Matthews et al.4

Jodie Bailie · Veronica Matthews · Ross S Bailie

To the Editor: Fraile Navarro and colleagues1 recently published 20 recommendations for the treatment of coronavirus disease 2019 (COVID‐19) in children and adolescents from the National COVID‐19 Clinical Evidence Taskforce. For the paediatric inflammatory multisystem syndrome (PIMS‐TS) recommendations, the Taskforce convened an expert advisory group.1 In the absence of clinical trials, the panel considered peer‐reviewed guidelines and cohort studies to formulate consensus recommendations.1 However, they deferred providing any guidance to help clinicians prevent thromboembolism. We suggest the Taskforce consider the same approach for paediatric anticoagulation guidance. COVID‐19 is associated with marked coagulation activation and hypercoagulability in children.2,3 Life‐threatening pulmonary embolus requiring thrombolysis has been encountered in Australian adolescents hospitalised with COVID‐19. A retrospective cohort study published in 2021 found that 2.1% of children hospitalised with symptomatic COVID‐19 infection and 6.5% of those with PIMS‐TS developed thrombosis.4 Thrombosis occurred more frequently in children aged 12years and over who had central lines, PIMS‐TS, or an underlying oncological diagnosis. A D‐dimer of more than five times the upper limit of normal was significantly associated with thrombosis.4 The authors refer to “paediatric guidelines published in the US”, which are published on behalf of the Pediatric/Neonatal Hemostasis and Thrombosis Subcommittee of the International Society of Thrombosis and Haemostasis; these adapt current consensus prophylaxis guidelines to include COVID‐19‐specific features.5 In deferring making specific recommendations, the authors suggested using existing local thromboprophylaxis guidelines. The Royal Children’s Hospital, Melbourne and the Sydney Children’s Hospital, Randwick have both independently developed COVID‐19‐specific thromboprophylaxis guidelines (that are very closely aligned),6,7 as have many other centres globally because previous local thromboprophylaxis guidelines are inadequate for COVID‐19‐associated thrombotic coagulopathy. The Melbourne/Sydney guidelines advise baseline coagulation testing in hospitalised children with COVID‐19, incorporating D‐dimer to assist risk assessment, twice‐daily enoxaparin and anti‐Xa monitoring/dose titration.6,7 These could be provided as supplemental material in these living guidelines. The COVID‐19 anticoagulation in Children–Thromboprophylaxis (COVAC‐TP) trial — a phase 2 single‐arm study looking at 40 children who will receive monitored, low dose, twice‐daily enoxaparin (ClinicalTrials.gov Identifier NCT04354155) — will not change the level of evidence, so waiting for completion of this trial does not seem appropriate.

Gemma L Crighton · Anthea Greenway · Susan Russell

To the editor: Clinicians are faced with requests for mask exemptions but guidance remains limited. In keeping with the Australasian College of Dermatologists’ guidelines,1 we believe skin problems are rarely severe enough to warrant exemption. The Department of Health and Human Services states people with “a serious skin condition of the face” are eligible for mask exemption,2 but this statement is open to interpretation. Mask exemptions for skin conditions are provided by numerous clinicians and not limited to dermatologists. Regardless of immunisation status, cases that may warrant exemption include severe dermatitis with crusting or weeping; severe infections such as impetigo or eczema herpeticum; bullous dermatoses, ectodermal dysplasias and other rare conditions featuring facial skin fragility; and post‐surgical procedures involving grafts or flaps where masks may impede healing. In addition, treatments for actinic damage such as 5‐fluorouracil, imiquimod or photodynamic therapy may cause severe inflammation.3 We suggest if exemptions are warranted, duration should be minimised, which may be before resolution of the dermatoses (eg, 2weeks followed by a review). This is essential given masks have been key in reducing severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) transmission.4 The development of an assessment pathway for facial dermatoses impeding mask use may be beneficial and should differentiate between health care workers, who wear fit‐tested masks, and the general public, guiding prompt treatment and follow‐up to facilitate a return to mask use. From our experience, facial masks may irritate the skin from pressure, sweating and humidity, and commonly aggravate underlying dermatoses, such as seborrheic dermatitis, acne or rosacea. Facial masks may rarely cause allergic contact dermatitis,5 and these cases should involve a contact dermatitis expert. Education regarding skin care is vital, in particular regular cleansing and reducing the number of products used which may aggravate acne. When utilising reusable masks, it is important to opt for an appropriate material such as light‐coloured cotton and maintain mask hygiene, which includes daily mask changes, regular washing, not sharing masks, and taking regular breaks from mask wearing. In summary, clinicians should remain vigilant when writing mask exemptions, aiming to minimise the duration by treating underlying skin problems and providing patient education.

Kajal Patel · Rosemary L Nixon

To the Editor: Larcombe and colleagues1 report a range of potentially harmful chemicals in locally purchased e‐liquids, highlighting the need for rigorous Australian standards and regulation of vaping liquids. However, the biological and clinical significance of these findings is unclear. Simply detecting the presence of a chemical with a potential risk has little meaning. The risk to human health depends on the level of exposure, under a fundamental principle of toxicology that “the dose makes the poison”.2 Apart from flavouring chemicals, most of the chemicals in the e‐liquids were at low or very low levels. Low doses of chemicals are ubiquitous in the environment, including arsenic in tap water and acrylamide in coffee.3 These cause little harm. To properly assess risk, the dose should be referenced to some standard, such as an established occupational or environmental exposure standard. No such comparison was made; therefore, no conclusions should be drawn. Nicotine was found at trace levels in six out of 65 samples. The authors imply that this has “implications for health and addiction”. However, nicotine was detected at a dose of 0.29 mg/L, which is around 1000 times lower than the lowest concentration of nicotine (3 mg/mL) in commercial e‐liquids and is not of any biological or clinical significance. Furthermore, as the authors note, predicting the risk to health should be based on the analysis of vapour rather than e‐liquid. This is a far more useful guide to the exposure of chemicals to the user. Most importantly, any risk should be compared with the risk from smoking. As frequent vaping is largely confined to smokers, minor harm from vaping is justified if it allows switching from the far more harmful behaviour of smoking.4 A Public Health England review found that most toxins responsible for health damage from smoking are absent in vapour and those that are present are at much lower levels (below 5% and mostly below 1%) than in tobacco smoke.5 It is disappointing that the authors declare no relevant disclosures when three of the four funding organisations have established antivaping positions (the Minderoo Foundation, Lung Foundation Australia, Cancer Council Western Australia). This should have been declared in the competing interests statement.

Colin P Mendelsohn · Alex D Wodak

To the Editor: We recently reviewed the first studies of the real‐world effectiveness of coronavirus disease 2019 (COVID‐19) vaccines.1 We found evidence of protection against serious illness and death but noted the difficulties in performing such studies in Australia. This was because of the (then) low national case numbers and lack of ready access to the necessary linked health data. Since then, the literature on vaccine effectiveness has expanded dramatically. By February 2022 the Johns Hopkins Bloomberg School of Public Health and partners had generated a database of 181 studies conducted in 26 countries.2 The most studied vaccines were Pfizer (117 studies), Moderna (47), AstraZeneca (43), Janssen (17) and Sinovac (8). Twenty‐three studies investigated booster doses, and seven studies mentioned analysis of the Omicron variant. Study outcomes included infections (117 studies), hospitalisations (69), deaths (30), and viral transmission (5). Study designs varied, with 32 mentioning cohort analysis and 16 mentioning test negative analysis in their titles. Most studies were performed in the United States (56 studies), followed by Israel (32), the United Kingdom (29), Qatar (9), Canada (8), Brazil (6) and Denmark (4). Not one of the listed studies was conducted in Australia. We should be asking why. Australia no longer lacks the case numbers to make estimates of vaccine effectiveness. We collect good data on vaccination status, infections (including viral variants), hospitalisations and deaths, plus the information needed to adjust for confounding of the associations between vaccine exposure and outcomes. However, authorities have not linked these datasets at individual level and made them available for detailed analysis. This situation should not continue. There are well established principles for protecting the privacy of individuals who are included in routinely collected data.3 The Commonwealth and state governments and relevant agencies seem unable or unwilling to link and properly analyse these data. Consequently, they should ensure that regularly updated comprehensively linked de‐identified datasets can be accessed by qualified researchers. Stephen Duckett has recently called for an Australian review of lessons from the COVID‐19 pandemic using a systems rather than a punitive lens.4 We agree. Better linkage, access and analysis of our health system data should be high on the list.

David A Henry · Mark A Jones · Paulina Stehlik · Paul P Glasziou

To the Editor: In their recent article, Larcombe and colleagues1 describe the analysis of 65 electronic cigarette fluid samples. The gas chromatography mass spectrometry method, as described in the Supporting Information, used “the ratio between the peak area corresponding to the fragment with the highest signal‐to‐noise ratio … and the peak area of the internal standard” for quantification.1 Direct comparison of the peaks for the analyte of interest and the internal standard does not take into account differences in response factor and/or ionisation efficiency for the different molecules. For accurate quantification, individual calibration curves should be prepared for all molecules of interest. If the authors prepared calibration curves but did not include this information, then the methods section should be modified to reflect this, and additional validation should be provided for all analytes, including limits of detection, limits of quantification, and coefficients of determination for all curves. In addition, all samples should be analysed in triplicate and a standard deviation should be provided to further validate the analysis. Larcombe et al1 note in their article that the determined concentrations of benzaldehyde, menthol, 2‐chlorophenol and benzyl alcohol exceed inhalational LC50 (the median lethal concentrations that kill 50% of a test animal population) values for these compounds. This assertion, as written, is incorrect and should have been reworded to a less confident statement. As the authors themselves acknowledge, the e‐liquid concentration and the inhalational LC50 values are not comparable. Furthermore, the authors did not provide the LC50 values they were using nor a citation to their source for these values. For a more accurate comparison, the volume of e‐liquid vaporised per litre of inhaled vapour would need to be calculated. Assuming the consumption of 9.47 mg of e‐liquid per puff and a puff volume of 55 mL,2 the maximum concentration observed for both benzaldehyde (2.58 mg/m3) and menthol (30.5 mg/m3) would fall below their respective derived no‐effect level (DNEL) values of 9.8 mg/m3 and 132 mg/m3.3,4 We were unable to locate an inhalational DNEL or LC50 value for 2‐chlorophenol. The maximum concentration for benzyl alcohol (251 mg/m3) would exceed its DNEL value of 110 mg/m3 for acute exposure.5 This, along with the high prevalence of benzyl alcohol in e‐liquid samples, requires further investigation.

Jody Morgan · Alison Jones · Celine Kelso

To the Editor: Chan and colleagues1 recently reported a case of putative e‐cigarette or vaping product use‐associated lung injury (EVALI) in a 15 year‐old girl who was a low level user of vaporised nicotine (without adulterants). We believe that, rather than EVALI, her presentation is better explained by urosepsis‐related acute lung injury. Current guidance from the United States Centers for Disease Control and Prevention (CDC)2 emphasises the role of adulterants, especially vitamin E acetate, in EVALI. In a US study completed before the widespread adoption of e‐cigarettes, the incidence of acute lung injury in 15–19‐year‐olds was 16 per 100 000 patient‐years, with many cases stemming from non‐pulmonary sepsis.3 In February 2020, only 2807 cases of vaping lung injury had been reported in the US, representing an incidence of well under one case per 100 000 patient‐years.1 Given these rates, as well as the patient’s prominent dysuria, polyuria, back pain and worsening pyrexia, we think urosepsis triggered the acute lung injury in this case. The authors say that sepsis was ruled out due to negative blood and urine cultures. However, if samples were collected after the initiation of antibiotics, false negative cultures are common in sepsis. The patient met the accepted criteria for sepsis, with suspected infection, a systemic inflammatory response syndrome and acute end‐organ failure,4 and was treated for this condition with antibiotics and corticosteroids for the acute lung injury. The CDC criteria for EVALI emphasise that the diagnosis should only be made where there is “no evidence in [the] medical record of alternative plausible diagnoses”.5 Dysuria, polyuria and back pain are not known symptoms of EVALI, and the authors have not explained how EVALI could account for this aspect of her presentation nor why these symptoms preceded the respiratory symptoms. In conclusion, the evidence to support a diagnosis of EVALI is insufficient in this case, and an alternative explanation is far more likely. Therefore, this case report should not be regarded as evidence for a case of EVALI occurring in Australia.

Cameron RL McKenzie · Joshua Davis · Adrian J Dunlop

To the Editor: People without Medicare coverage cannot access Pharmaceutical Benefits Scheme (PBS)‐subsidised human immunodeficiency virus (HIV) pre‐exposure prophylaxis (PrEP) or associated clinical care. Rates of HIV infection diagnosis are disproportionately higher among overseas‐born gay and bisexual men compared with Australian‐born gay and bisexual men.1 In response, in June 2020, the Alfred Hospital and the Victorian Infectious Diseases Reference Laboratory established the free PrEPMe Clinic for Medicare‐ineligible people. Data were collected using proformas after patients provided verbal consent (Alfred Health Ethics Committee approval No. 656/18). The first 100 consecutive patients were all born overseas (Box). Melbourne’s only public sexual health clinic referred 65 patients. Almost all patients were male, all patients had sex with men and reported a median of three sexual partners in 3 months at baseline; 76 patients inconsistently used condoms for anal sex. Fifty‐eight patients reported previous sexually transmissible infections (STIs); STIs were diagnosed in 12/100 patients at baseline, a rate similar to that found in Medicare‐eligible PrEP users.2 Thirty‐four patients had previously accessed HIV post‐exposure prophylaxis (PEP), and 49 patients had previously unsuccessfully attempted to obtain PrEP. The reported barriers to access mainly included costs of medical appointments and pathology, and difficulties navigating Australia’s health care system. All patients received a non‐PBS PrEP prescription. At 3‐month follow‐up, 87 patients had commenced PrEP. Local pharmacies supplied PrEP at cost price (A$40–55 per month) or free to patients with financial hardship; other patients purchased PrEP online (US$20–30 per month) or obtained free PrEP online using assistance coupons (www.pan.org.au; Box). Most patients who ordered PrEP online experienced delivery delays of 4–6 weeks, leaving them at risk of HIV infection. We report that Medicare‐ineligible gay and bisexual men and transgender women were at high risk of HIV infection, yet faced significant financial barriers to accessing PrEP. PrEP uptake has been associated with significant population‐level declines in incident HIV infection in Australia.3 Australia’s Eighth National HIV Strategy aims for virtual elimination of HIV transmissions by 2022,4 and to achieve this goal, Australia must provide universally subsidised PrEP medication and clinical services, irrespective of Medicare status.5 Medicare‐ineligible gay and bisexual men often already attend publicly funded sexual health clinics for free HIV/STI testing and treatment, as reported here. In a high income country like Australia, the additional cost of providing universally subsidised PrEP care would likely be lower than treating preventable new HIV infections, with an estimated lifetime cost of more than US$350 000 per HIV infection diagnosis.6 Box – Demographic characteristics, immunodeficiency virus (HIV) acquisition risk, and prior efforts to obtain pre‐exposure prophylaxis (PrEP) in the first 100 consecutive patients to attend the PrEPMe HIV prevention clinic at the Alfred Hospital in Melbourne, Australia* Values Total number of patients 100 Demographic characteristics Region of birth Asia 47 Latin America 31 Europe 14 Other 8 Age, years, median (IQR) 28 (26–31) Gender Cisgender male 96 Transgender female 4 Visa status Student visa 62 Working visa 34 Other 4 Referral sources Melbourne Sexual Health Centre 65 Word of mouth 16 Other† 13 Unknown 6 HIV risk at initial clinical assessment Sexual partners (3 months), median (IQR) 3 (1–5) Condom use for anal sex (3 months) Always 24 Mostly or sometimes 60 Never 13 Not applicable 1 Unknown 2 Previous STIs (ever) Yes 58 No 42 Previous STIs (ever, specific STIs) Gonorrhoea 35 Chlamydia 21 Syphilis 21 Other‡ 5 STIs diagnosed at baseline Chlamydia only 6 Other§ 6 Previous attempts at HIV risk reduction Previous use of PEP Yes 34 No 57 Unknown 9 Previous unsuccessful attempts to obtain PrEP Yes 49 No 46 Unknown 5 PrEP commencement by 3‐month follow‐up Commenced PrEP 87 Local pharmacy 65 Online 19 Online order did not arrive, then purchased at pharmacy 3 PrEP not commenced 6 Online order did not arrive 3 Other¶ 3 Lost to follow‐up 7 COVID‐19 = coronavirus disease 2019; IQR = interquartile range; PEP = post‐exposure prophylaxis; STIs = sexually transmissible infections. * Enrolment dates: 1 June 2020 to 26 October 2020. † Includes general practices, internet search, “PrEP Access Now” Facebook page, Alfred Hospital PEP program. ‡ Includes herpes simplex virus, Mycoplasma genitalium, hepatitis B virus. § Includes syphilis, hepatitis B virus, both chlamydia and gonorrhoea. ¶ Includes lost prescription, no sex due to COVID‐19.

Vincent J Cornelisse · Jude Armishaw · Mike Catton · Dean Murphy · Edwina J Wright

To the Editor: The coronavirus disease 2019 (COVID‐19) pandemic has strained health systems in New South Wales, and hospitals have rapidly adapted to care for inpatients with COVID‐19. In the 4 weeks leading up to 9 September 2021, 9330 locally acquired cases were diagnosed in Western Sydney alone.1 The management of large numbers of COVID‐19 inpatients with diabetes has been challenging. People with diabetes are a vulnerable population who are at risk of adverse outcomes from COVID‐19, with a two‐ to threefold likelihood of death compared with people without diabetes.2 Hyperglycaemia is associated with higher risk;3 hence, good glucose management is desirable. Exacerbation of diabetes by dexamethasone therapy, used to treat patients with COVID‐19, and the development of steroid‐induced hyperglycaemia in non‐diabetic patients present further challenges. Traditional models of care relying on referrals from parent teams to an endocrinologist (or registrar), who then reviews the patient daily to chart insulin, are inefficient and impractical for this situation. We have developed a virtual inpatient diabetes management service (vIDMS) as a means for a small diabetes team to manage COVID‐19 inpatients with diabetes. The success of this model has revolved around an electronic medical record, electronic inpatient prescribing, a diabetes dashboard, and videoconferencing communications. The recording of all glucose measurements (including point of care) within the electronic medical record has enabled the systematic capture and display of hospital‐wide glucose data on a diabetes dashboard (Box). This also allows viewing and filtering by any variable in the electronic medical record, such as ward, age, biochemistry (including formal laboratory glucose and glycated haemoglobin), development of hypoglycaemia, prescribed medications (including corticosteroids), and COVID‐19 status. Therefore, COVID‐19 patients with diabetes or hyperglycaemia are easily identified. The vIDMS, comprising of a consultant, a registrar and a diabetes educator, reviewed patients with COVID‐19 and hyperglycaemia on a daily basis, using the dashboard and electronic medical record, by sharing a screen on a videoconferencing platform. Remote management was undertaken through the electronic medical record, including medication and insulin dose adjustments. Communication with ward staff and patients with COVID‐19 through the electronic medical record, or by telephone or video, was undertaken when needed, including for diabetes education. Entry into the COVID‐19 wards and usage of personal protective equipment was not required. In the 6 weeks to 5 September 2021, 112 COVID‐19 patients with diabetes were thus managed in Westmead Hospital (median age, 62 years; range, 23–91 years), with up to 40 patients reviewed per day. Necessitated by COVID‐19, the future of inpatient diabetes management is now here. With one‐quarter of patients in metropolitan hospitals having self‐reported diabetes4 but insufficient specialised diabetes staff to provide individual management, the vIDMS will become a significant part of the wider model of diabetes care for large hospitals.5 While initial and intermittent face‐to‐face contact remains valuable to build a relationship and discuss relevant issues, and careful review of medical records is necessary to understand perturbations in glucose levels (eg, fasting, missed medication), the vIDMS enables daily specialist care for large numbers of patients with diabetes by a small team. The health system needs to facilitate its wider application for the management of both COVID‐19 and non‐COVID‐19 patients with diabetes in hospital. Box – Diabetes dashboard showing hospital‐wide glucose data for patients with coronavirus disease 2019 (COVID‐19)

N Wah Cheung · Amanda Hor · Tien‐Ming Hng

To the Editor: We must clarify that the findings of Iorfino and colleagues1 do not apply to headspace clients. This understandable misperception comes from the article’s title, “early intervention mental health services”, and participants coming from “clinics” that “provide both primary care services (headspace) and more specialised services”. Quotes from an InSight+ article2 and the accompanying editorial3 infer that the findings generalise to headspace, but they do not. Although some participants in the Iorfino study came from two headspace‐branded centres, they also included young people accessing specialised services. This is evident in the limitations: “our sample was restricted to young people who remained in contact with the service for at least two years … biasing our sample towards people who required ongoing care and were accordingly more likely to have poorer outcomes”. Generally, headspace does not provide care over a period of two years or more; centres typically provide brief episodes of care, befitting young people with mild to moderate presentations for common mental health problems. The average number of sessions is 4.4 (standard deviation [SD], 6.2); 98% of clients receive 20 sessions or fewer. A negligible 0.4% of clients are still receiving care at two years. Average time between the first and last session is 73.7 days (SD, 120), about 2.5 months (headspace 2015–2021, national unpublished data). Clarification that the findings are not representative of, nor generalisable to, headspace clients is essential; the headspace initiative is not targeted at young people who need more sustained mental health care. The results are inconsistent with reported outcomes for 24 034 headspace clients from 55 fully established centres.4 Significant improvement in the Social and Occupational Functioning Assessment Scale (SOFAS) scores was evident for 37.1% of headspace clients, 43.4% had no significant change, and 19.5% significantly deteriorated. Updated outcomes are forthcoming. Importantly, multiple outcomes must be considered for headspace clients; these are young people with diverse early intervention needs — the single‐item, clinician‐rated SOFAS is insufficient to ascertain meaningful outcomes. Despite the inability to generalise from Iorfino’s study to the national headspace centre network, we agree that young people with complex and persisting mental health conditions require more resource‐intensive responses than headspace primary care services were designed for.

Debra J Rickwood · Jason Trethowan · Annette Carruthers