Research

Objective: To compare the characteristics and outcomes of drug‐induced liver injury (DILI) caused by paracetamol and non‐paracetamol medications, particularly herbal and dietary supplements. Design: Retrospective electronic medical record data analysis. Setting, participants: Adults admitted with DILI to the Gastroenterology and Liver Centre at the Royal Prince Alfred Hospital, Sydney (a quaternary referral liver transplantation centre), 2009–2020. Main outcome measures: 90‐day transplant‐free survival; drugs implicated as causal agents in DILI. Results: A total of 115 patients with paracetamol‐related DILI and 69 with non‐paracetamol DILI were admitted to our centre. The most frequently implicated non‐paracetamol medications were antibiotics (19, 28%), herbal and dietary supplements (15, 22%), anti‐tuberculosis medications (six, 9%), and anti‐cancer medications (five, 7%). The number of non‐paracetamol DILI admissions was similar across the study period, but the proportion linked with herbal and dietary supplements increased from 2 of 13 (15%) during 2009–11 to 9 of 19 (47%) during 2018–20 (linear trend: P = 0.011). Despite higher median baseline model for end‐stage liver disease (MELD) scores, 90‐day transplant‐free survival for patients with paracetamol‐related DILI was higher than for patients with non‐paracetamol DILI (86%; 95% CI, 79–93% v 71%; 95% CI, 60–82%) and herbal and dietary supplement‐related cases (59%; 95% CI, 34–85%). MELD score was an independent predictor of poorer 90‐day transplant‐free survival in both paracetamol‐related (per point increase: adjusted hazard ratio [aHR], 1.19; 95% CI, 1.09–3.74) and non‐paracetamol DILI (aHR, 1.24; 95% CI, 1.14–1.36). Conclusion: In our single centre study, the proportion of cases of people hospitalised with DILI linked with herbal and dietary supplements has increased since 2009. Ninety‐day transplant‐free survival for patients with non‐paracetamol DILI, especially those with supplement‐related DILI, is poorer than for those with paracetamol‐related DILI.

Emily Nash · Abdul‐Hamid Sabih · John Chetwood · Georgette Wood · Keval Pandya · Terry Yip · Avik Majumdar · Geoffrey W McCaughan · Simone I Strasser · Ken Liu

Vaccination delivery was generally resilient in a period of unprecedented social and health care disruption

Brynley P Hull · Alexandra J Hendry · Aditi Dey · Kerin Bryant · Catherine Radkowski · Stephen Pellissier · Kristine Macartney · Frank H Beard

Objective: To examine relationships between changing general practitioner after entering residential aged care and overall medicines prescribing (including polypharmacy) and that of psychotropic medicines in particular. Design: Retrospective data linkage study. Setting, participants: 45 and Up Study participants in New South Wales with dementia who were PBS concession card holders and entered permanent residential aged care during January 2010 ‒ June 2014 and were alive six months after entry. Main outcome measures: Inverse probability of treatment‐weighted numbers of medicines dispensed to residents and proportions of residents dispensed antipsychotics, benzodiazepines, and antidepressants in the six months after residential care entry, by most frequent residential care GP category: usual (same as during two years preceding entry), known (another GP, but known to the resident), or new GP. Results: Of 2250 new residents with dementia (mean age, 84.1 years; SD, 7.0 years; 1236 women [55%]), 625 most frequently saw their usual GPs (28%), 645 saw known GPs (29%), and 980 saw new GPs (44%). The increase in mean number of dispensed medicines after residential care entry was larger for residents with new GPs (+1.6 medicines; 95% CI, 1.4‒1.9 medicines) than for those attended by their usual GPs (+0.7 medicines; 95% CI, 0.4‒1.1 medicines; adjusted rate ratio, 2.42; 95% CI, 1.59‒3.70). The odds of being dispensed antipsychotics (adjusted odds ratio [aOR], 1.59; 95% CI, 1.18‒2.12) or benzodiazepines (aOR, 1.69; 95% CI, 1.25‒2.30), but not antidepressants (aOR, 1.32; 95% CI, 0.98‒1.77), were also higher for the new GP group. Differences between the known and usual GP groups were not statistically significant. Conclusions: Increases in medicine use and rates of psychotropic dispensing were higher for people with dementia who changed GP when they entered residential care. Facilitating continuity of GP care for new residents and more structured transfer of GP care may prevent potentially inappropriate initiation of psychotropic medicines.

Heidi J Welberry · Louisa R Jorm · Andrea L Schaffer · Sebastiano Barbieri · Benjumin Hsu · Mark F Harris · John Hall · Henry Brodaty

On 23 May 2019, the Central Queensland Public Health Unit received a confirmed laboratory notification of a dengue virus serotype‐2 (DENV‐2) infection in a Rockhampton resident. On 5 May, a 71‐year‐old man without a history of travel overseas or to Far North Queensland had developed symptoms consistent with a zoonotic disease, and presented later that month to his general practitioner because his symptoms had not abated. Between 23 May and 7 October 2019, 21 locally acquired cases of DENV‐2 were identified in Rockhampton: 13 laboratory‐confirmed cases and eight probable cases detected by active surveillance. This was the first outbreak of locally acquired dengue in Central Queensland for 65 years.1 In 14 cases (67%), the infected persons sought medical attention; two required hospitalisation. A formal outbreak response was initiated by the Central Queensland Public Health Unit on 23 May 2019, including extensive mosquito surveillance and active and passive human surveillance within 200 metres of the residences of each identified infected person. Particular attention was directed to surveying locations that might facilitate increased dengue transmission in the community (such as schools, a plant nursery, and aged care facilities) for artificial and natural containers that could serve as breeding areas for infected mosquitoes (Aedes aegypti). Such containers were either removed or emptied of residual water and treated with pellets of the insect growth regulator (S)‐methoprene, and the premises and buildings were sprayed inside and out with the residual insecticide Temprid 75 (Bayer; includes imidacloprid and β‐cyfluthrin). In addition to the house‐to‐house human surveillance, a novel “lure and kill” approach was adopted for vector control: lethal ovitraps were deployed within 200 metres of the residence of any person with a probable or confirmed infection. Ae. aegypti was found in 105 of 1107 inspected residential premises (9.5%), or more than half of the 205 premises found to contain mosquitoes. Enhanced serological surveillance was undertaken to detect patients with viraemia early, enabling prompt public health and mosquito control interventions. The complete DENV‐2 genome sequence (GenBank accession number, MN982899.1) indicated that the implicated virus was most closely related to Southeast Asian strains of DENV‐2. Given the presence of Ae. aegypti in Central Queensland and the increasing numbers of travellers and visitors returning from countries in which dengue is endemic, it is important that Rockhampton be recognised as a dengue‐receptive area. As locally acquired cases of dengue are being reported outside Far North Queensland, the state map of dengue‐receptive areas2 should be updated; specifically, the broad geographic area from Townsville south to Rockhampton should be considered dengue‐receptive.

Jacina Walker · Alyssa Pyke · Paul Florian · Rachael M Rodney Harris · Gulam Khandaker

Measuring prostate‐specific antigen (PSA) levels is widely used for screening for the early diagnosis of prostate cancer.1 However, the value of PSA testing for reducing prostate cancer‐specific or all‐cause mortality in asymptomatic men is uncertain.2 The Prostate Cancer Foundation of Australia and the Cancer Council of Australia recommend that men aged 50–69 years be offered biennial PSA testing if they make an informed decision to be tested.1 We investigated age‐specific patterns of PSA testing and PSA levels in men aged 40 years or more without symptoms of prostate cancer. We analysed routinely collected data from 180 Victorian general practices, pooled by the data custodian, Outcome Health.3 We included all men who had attended the same practice at least three times during October 2016 – September 2018. We identified tested men and testing frequency from recorded PSA test results. PSA testing prevalence was defined as the number of men tested at least once divided by the total number of men in an age group. We evaluated the relationship between log‐transformed PSA levels and age as a continuous variable in a linear regression model. The Royal Australian College of General Practitioners National Research and Evaluation Ethics Committee (17‐008) and the Macquarie University Human Research Ethics Committee (5201700872) approved our study. A total of 142 016 male patients were identified. The proportion who underwent PSA testing at least once (prevalence) or at least twice during the study period increased with age: prevalence peaked with the 65–69 year age group (8132 men tested, 54% of age group), and the proportion of men tested at least twice with the 70–74 year age group (3159 men, 46% of age group) (Box 1). A total of 78 818 PSA test results were recorded during the study period; about one‐third were for men aged 60–69 years (25 496 tests, 32%). The median PSA level increased from 0.7 ng/mL (interquartile range [IQR], 0.5–0.9 ng/mL; 95th percentile, 1.7 ng/mL) for men aged 40–44 years to 1.9 ng/mL (IQR, 0.8–4.5 ng/mL; 95th percentile, 11.7 ng/mL) for men aged 85 years or more (Box 2). The median PSA level increased by 3.2% per year of age (95% confidence interval, 3.1–3.3%). We found that PSA testing prevalence, the proportion of men tested more than once within 24 months, and median PSA levels each increased with age. For men over 69 years of age, this could lead to further invasive tests and treatments, some of which may be unnecessary or cause harm. The high PSA testing prevalence among older men was similar to previous Australian reports (48% of men aged 70 years or more reported they had PSA tests in the preceding two years4), and the PSA levels matched current age‐specific 95th percentile reference ranges.5 Why patterns of testing are different to those recommended (ie, more frequently than biennial and beyond 69 years of age) cannot be explained by general practice data, but reasons may include patient expectations, fragmentation of care, and the persistence of old guideline recommendations. We have reported the most comprehensive snapshot of PSA testing in Australia based on electronic general practice data since the release of the current guidelines for PSA testing of asymptomatic men. Our study also showed that such data can be used to establish benchmarks for designing quality improvement activities and to promote evidence‐based decision‐making in general practice. Box 1 – Prostate‐specific antigen (PSA) testing for 180 Victorian general practices, October 2016 – September 2018, by age group* * Proportion of male patients in age group who were tested. It is recommended that men aged 50–69 years be tested once every two years. Numbers of patients in each age group are provided in the online Supporting Information. Box 2 – Number of prostate‐specific antigen (PSA) tests and median PSA test result levels, by age group table#t2 tbody td:nth-child(n+2) P. Pleft { text-align: center; } Age group (years) Number of tests PSA level (ng/mL) Median (IQR) 95th percentile 40–44 2685 (3.4%) 0.7 (0.5–0.9) 1.7 45–49 5894 (7.5%) 0.7 (0.5–1.0) 2.1 50–54 9544 (12.1%) 0.8 (0.5–1.2) 2.6 55–59 12 359 (15.7%) 0.9 (0.6–1.5) 3.7 60–64 12 944 (16.4%) 1.1 (0.6–2.0) 4.7 65–69 12 551 (15.9%) 1.3 (0.7–2.4) 5.5 70–74 10 999 (14.0%) 1.5 (0.8–2.8) 6.4 75–79 6440 (8.2%) 1.6 (0.8–3.3) 8.0 80–84 3327 (4.2%) 1.8 (0.8–3.6) 9.2 ≥ 85 2075 (2.6%) 1.9 (0.8–4.5) 11.7 All ages groups 78 818 1.0 (0.6–2.0) 5.4 IQR = interquartile range.

Guilherme S Franco · Rae‐Anne Hardie · Ling Li · Chisato Imai · Gorkem Sezgin · Julie Li · Adam McLeod · Christopher Pearce · Andrew Georgiou

The social determinants of incidence should be addressed, and other relevant host, pathogen, and health system factors investigated

Cameron M Wright · Rachael Moorin · Glenn Pearson · John R Dyer · Jonathan R Carapetis · Laurens Manning

When SARS-CoV-2 prevalence is low, a risk-based approach to screening could improve testing efficiency and reduce resource needs

Eloise Williams · Katherine Bond · Deborah A Williamson

Objective: To identify functional performance trajectories and the characteristics of people who receive inpatient geriatric rehabilitation after hospital admissions. Design, setting, participants: REStORing health of acutely unwell adulTs (RESORT) is an observational, prospective, longitudinal inception cohort study of consecutive patients admitted to geriatric rehabilitation wards at the Royal Melbourne Hospital. Recruitment commenced on 15 October 2017. Main outcome measures: Functional performance, assessed with the Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL) scales two weeks before acute hospitalisation, on admission to and discharge from geriatric rehabilitation, and three months after discharge from geriatric rehabilitation. Results: A total of 618 rehabilitation patients were included in our analysis. For each of the two scales, three distinct functional performance trajectories were identified by latent class growth modelling: poor at baseline and 3‐month follow‐up (remained poor: ADL, 6.6% of patients; IADL, 42%), good at baseline but poor recovery (deteriorated: ADL, 33%; IADL, 20%), and good at baseline and good recovery (recovered: ADL, 60%; IADL, 35%). Higher Clinical Frailty Scale (CFS) score (v recovered, per point: odds ratio [OR], 2.51; 95% CI, 1.64–3.84) and cognitive impairment (OR, 6.33; 95% CI, 2.09–19.1) were associated with greater likelihood of remaining poor in ADL, and also with deterioration (CFS score: OR, 1.76; 95% CI, 1.45–2.13; cognitive impairment: OR, 1.87; 95% CI, 1.24–2.82). Higher CFS score (OR, 1.64; 95% CI, 1.37–1.97) and cognitive impairment (OR, 3.60; 95% CI, 2.31–5.61) were associated with remaining poor in IADL, and higher CFS score was also associated with deterioration (OR, 1.63; 95% CI, 1.33–1.99). Conclusions: Based on ADL assessments, most people who underwent inpatient geriatric rehabilitation regained their baseline functional performance. As higher CFS score and cognitive impairment were associated with poorer functional recovery, assessing frailty and cognition at hospital admission could assist intervention and discharge planning.

Cheng Hwee Soh · Esmee M Reijnierse · Camilla Tuttle · Celia Marston · Rose Goonan · Wen Kwang Lim · Andrea B Maier

Objective: To compare the frequency and outcomes of cardiac arrests in general practice clinics with those of paramedic‐witnessed cardiac arrests. Design, setting: Retrospective study; analysis of Victorian Ambulance Cardiac Arrest Registry data, 1 January 2000 – 30 December 2019. Participants: Patients with non‐traumatic cardiac arrests whom emergency medical services staff attempted to resuscitate. Main outcome measures: Survival to hospital discharge. Results: 6363 cases of cardiac arrest were identified: 216 in general practice clinics (3.4%) and 6147 witnessed by paramedics (96.6%). The proportion of patients presenting with initial shockable rhythms was larger in clinic (126 patients, 58.3%) than paramedic‐witnessed cases (1929, 31.4%). The proportion of general practice clinic cases in which defibrillation was provided in the clinic increased from 2 of 37 in 2000–2003 (5%) to 19 of 57 patients in 2016–2019 (33%); survival increased from 7 of 37 (19%) to 23 of 57 patients (40%). For patients with initial shockable rhythms, 57 of 126 in clinic cases (45%) and 1221 of 1929 people in paramedic‐witnessed cases (63.3%) survived to hospital discharge; of 47 general practice patients defibrillated by clinic staff, 27 survived (57%). For patients with initial shockable rhythms, the odds of survival were greater following paramedic‐witnessed events (adjusted odds ratio [aOR], 3.39; 95% CI, 2.08–5.54) or general clinic arrests with defibrillation by clinic staff (aOR, 2.23; 95% CI, 1.03–4.83) than for general practice clinic arrests in which arriving paramedics provided defibrillation. Conclusion: Emergency medical services should be alerted as soon as possible after people experience heart attack warning symptoms. Automated external defibrillators should be standard equipment in general practice clinics, enabling prompt defibrillation, which may substantially reduce the risk of death for people in cardiac arrest.

Brian Haskins · Ziad Nehme · Peter A Cameron · Karen Smith

Objectives: To evaluate the implementation and acceptability of the self‐collection cervical screening pathway since commencement of the renewed National Cervical Screening Program (rNCSP), from the perspectives of screening participants and primary care practitioners. Design, setting, participants: Qualitative study; individual semi‐structured interviews with 45 screening participants and 18 primary care practitioners in Victoria who had engaged with the self‐collection pathway during the first 17 months of the rNCSP (1 December 2017 ‒ 30 April 2019). Results: The self‐collection pathway was highly acceptable as an alternative cervical screening pathway for most participating screening participants and practitioners. Some screening participants indicated that they would not have been screened had the pathway not been available. Acceptability was lower among those who had tested positive for HPV types not 16/18, a result that requires additional testing of a clinician‐collected cervical sample. Use of the self‐collection pathway is driven more by practitioners than their patients. Interpretations of the self‐collection guidelines varied between practices. Barriers to expanding promotion of the pathway by practitioners included difficulties with identifying eligible participants. Conclusions: Increasing the accessibility of the self‐collection pathway to under‐ and never screened women could reduce inequities in cervical cancer outcomes for those not participating in the main screening pathway. Practitioners should be provided resources to integrate self‐collection into routine practice and to efficiently implement the entire self‐collection pathway, in order to maximise its use and to optimise the experience for screening participants.

Nicola S Creagh · Claire Zammit · Julia ML Brotherton · Marion Saville · Tracey McDermott · Claire Nightingale · Margaret Kelaher

The need to treat chronic pain and the increasing number of patients requiring opioid therapy are among the factors that have led to prescription opioid‐related harm. Several professional society guidelines aim to reduce the frequency of opioid‐related problems by modifying prescriber behaviour.1,2 We analysed Monitoring of Drugs of Dependence System (MODDS) data for adult Queensland residents (18 years or older) for whom opioids were dispensed during 1 January 1997 ‒ 31 December 2018. Our aim was to determine the most frequently prescribed doses of opioids, and to examine associations between the mean number of prescribers per patient and the opioid doses dispensed. The database included data for 228 861 opioid prescribers during the study period and 18 798 942 dispensed opioid prescriptions. We undertook multivariate mixed effects Poisson regression, with random intercept by patient identification to account for within‐patient correlations and between‐patient characteristics. We adjusted for socio‐economic status of patient residence (by postcode),3 as an association between lower socio‐economic status and greater opioid prescribing has been reported.4 We also adjusted our analysis for duration of opioid use, dose in oral morphine milligram equivalents per day (MME/day), and opioid formulation, as in our previous studies.5,6 The study was approved by the Royal Brisbane and Women’s Hospital Human Research Ethics Committee (HREC/17/QRBW/669), the University of Queensland Human Research Ethics Committee (2018000623/HREC/QRBW/669), and by the Department of Health under the Public Health Act 2005 (Qld) (RD007970). The number of patients for whom opioids were dispensed increased from 28 299 in 1997 to 322 307 in 2018; the number of Queensland medical practitioners who prescribed opioids increased from 4537 to 20 226 (online Supporting Information, table). The proportion of opioid prescribers who prescribed opioids at doses of 50 to less than 100 MME/day peaked in 2003 (813 of 5530, 14.7%), as did the proportion prescribing 100 MME/day or more (1077, 19.5%), before declining in 2018 to 1294 (6.4%) and 1064 of 22 941 (5.3%) respectively (Box 1, A; Supporting Information, figure). During 1997‒2018, 1 916 842 people (67.8% of people dispensed opioids) were dispensed doses of less than 20 MME/day, 624 114 were dispensed 20 to less than 50 MME/day (22.1%), 159 448 were dispensed 50 to less than 100 MME/day (5.6%), and 127 323 were dispensed 100 MME/day or more (4.5%) (Box 1, B). For modified release opioids, the largest proportions of prescribers and patients were for the 20 to less than 50 MME/day dose category (Box 1, C, D). The mean number of prescribers per patient per year increased with the dose of opioids dispensed (v < 20 MME/day: 20 to < 50 MME/day, incidence rate ratio [IRR],1.04; 95% confidence interval [CI], 1.03‒1.04; 50 to < 100 MME/day, IRR, 1.09; 95% CI, 1.08‒1.09; ≥ 100 MME/day, IRR, 1.22; 95% CI, 1.21‒1.22) (Box 2). However, the mean number of prescribers per patient per year increased as the duration of opioid dispensing decreased (patients dispensed opioids for one month v for 7‒12 months: IRR, 1.36; 95% CI, 1.35‒1.36). This finding is consistent with a report that people using opioids for shorter periods more frequently visit multiple prescribers.4,5 Our findings indicate that most Queensland medical practitioners prescribe lower opioid doses, and that the proportion prescribing lower doses has increased since 2004. The proportion of people dispensed doses of opioids associated with increased risk of accidental overdose (from 50 MME/day) was small and has declined over time. Box 1 – Proportions of patients prescribed opioid medications and of prescribers, by opioid dose prescribed* * Opioids included were morphine, oxycodone, hydromorphone, tapentadol, fentanyl, buprenorphine, and codeine; methadone was excluded because we could not distinguish between prescribing for pain management and opioid replacement therapy. All formulations except lozenges, suppositories and parenteral formulations were included, apart from tablet and sublingual buprenorphine formulations (used to treat gastrointestinal problems and opiate dependence) and fentanyl (approved for palliative care). For complete inclusion and exclusion criteria, see Supplementary Material 3 in reference 6. Box 2 – Mean numbers of prescribers per patient per year, by daily opioid dose, duration of dispensing, and socio‐economic status table#t2 tbody td:nth-child(n+2) P. Pleft { text-align: center; } Mean prescribers per patient per year (95% CI) Incidence rate ratio (95% CI) Unadjusted Adjusted* Prescribed dose (MME/day) < 20 1.62 (1.61–1.62) 1 1 20 to < 50 1.76 (1.75–1.76) 1.08 (1.08–1.09) 1.04 (1.03–1.04) 50 to < 100 2.05 (2.04–2.05) 1.26 (1.26–1.27) 1.09 (1.08–1.09) ≥ 100 2.90 (2.89–2.92) 1.79 (1.79–1.80) 1.22 (1.21–1.22) Duration of use (months) ≤ 1 1.81 (1.81–1.82) 1.41 (1.41–1.42) 1.36 (1.35–1.36) 2‒3 1.50 (1.49–1.50) 1.17 (1.17–1.18) 1.16 (1.16–1.17) 4‒6 1.40 (1.39–1.40) 1.09 (1.09–1.10) 1.09 (1.09–1.10) 7‒12 1.28 (1.27–1.28) 1 1 Socio‐economic status (quintile)† 1 (lowest) 1.92 (1.92–1.93) 1.11 (1.11–1.12) 1.07 (1.07–1.08) 2 1.93 (1.92–1.93) 1.11 (1.11–1.12) 1.05 (1.05–1.06) 3 1.80 (1.79–1.80) 1.04 (1.04–1.05) 1.03 (1.02–1.03) 4 1.78 (1.78–1.79) 1.03 (1.03–1.04) 1.01 (1.01–1.02) 5 (highest) 1.72 (1.71–1.72) 1 1 Formulation Immediate release 1.66 (1.65–1.66) 1 1 Modified release 1.97 (1.97–1.98) 1.19 (1.18–1.19) 0.98 (0.99–1.00) MME = morphine milligram equivalent; CI = confidence interval. * Multivariate mixed effects Poisson regression adjusted for socio‐economic status, duration of opioid dispensing, opioid dose, and opioid formulation. † Socio‐Economic Indexes for Areas (SEIFA Index of Relative Socio‐Economic Advantage and Disadvantage (IRSAD).3

Adeleke D Adewumi · Joemer C. Maravilla · Rosa Alati · Samantha A Hollingworth · Xuelei Hu · Bill Loveday · Jason Connor

Objectives: To investigate the perspectives of doctors involved with voluntary assisted dying in Victoria regarding the Voluntary Assisted Dying Act 2017 (Vic) and its operation. Design, setting, participants: Qualitative study; semi‐structured interviews with 32 doctors who had participated in the voluntary assisted dying system during its first year of operation (commenced 19 June 2019). Doctors were interviewed during April‒July 2020. Results: Three major themes related to problems during the first year of operation of the Act were identified: the statutory prohibition of health professionals initiating discussions with their patients about voluntary assisted dying; the Department of Health and Human Services guidance requirement that all doctor‒patient, doctor‒pharmacist, and pharmacist‒patient interactions be face‐to‐face; and aspects of implementation, including problems with the voluntary assisted dying online portal, obtaining documentary evidence to establish eligibility, and inadequate resourcing of the Statewide Pharmacy Service. Conclusions: Doctors reported only limited concerns about the Victorian voluntary assisted dying legislation, but have had some problems with its operation, including implications for the accessibility of voluntary assisted dying to eligible patients. While legislative change may resolve some of these concerns, most can be ameliorated by improving the processes and systems.

Lindy Willmott · Ben P White · Marcus Sellars · Patsy M Yates

Objectives: To investigate the prevalence of non‐alcoholic fatty liver disease (NAFLD) and its risk factors in regional Victoria. Design: Prospective cross‐sectional observational study (sub‐study to CrossRoads II health study in Shepparton and Mooroopna). Setting: Four towns (populations, 6300‒49 800) in the Goulburn Valley of Victoria. Participants: Randomly selected from households selected from residential address lists provided by local government organisations for participation in the CrossRoads II study. Main outcome measures: Age‐ and sex‐adjusted estimates of NAFLD prevalence, defined by a fatty liver index score of 60 or more in people without excessive alcohol intake or viral hepatitis. Results: A total of 705 invited adults completed all required clinical, laboratory and questionnaire evaluations of alcohol use (participation rate, 37%); 392 were women (56%), and their mean age was 59.1 years (SD, 16.1 years). Of the 705 participants, 274 met the fatty liver index criterion for NAFLD (crude prevalence, 38.9%; age‐ and sex‐standardised prevalence, 35.7%). The mean age of participants with NAFLD (61 years; SD, 15 years) was higher than for those without NAFLD (58 years; SD, 16 years); a larger proportion of people with NAFLD were men (50% v 41%). Metabolic risk factors more frequent among participants with NAFLD included obesity (69% v 15%), hypertension (66% v 48%), diabetes (19% v 8%), and dyslipidaemia (63% v 33%). Mean serum alanine aminotransferase levels were higher (29 U/L; SD, 17 U/L v 24 U/L; SD, 14 U/L) and mean median liver stiffness greater (6.5 kPa; SD, 5.6 kPa v 5.3kPa; SD, 2.0 kPa) in participants with NAFLD. Conclusion: The prevalence of NAFLD among adults in regional Victoria is high. Metabolic risk factors are more common among people with NAFLD, as are elevated markers of liver injury.

Stuart K Roberts · Ammar Majeed · Kristen Glenister · Dianna Magliano · John S Lubel · Lisa Bourke · David Simmons · William W Kemp

Intensive metabolic control reduces the incidence and progression of diabetes‐related micro‐ and macrovascular complications.1,2 Nevertheless, the risk of developing cardiovascular disease is higher for people with diabetes,3 although cardiovascular disease incidence rates are generally declining more rapidly for people with diabetes than for other people.4,5 We analysed hospital discharge data from the Victorian Admitted Episode Dataset6 for 1 January 1999 – 31 December 2016. We identified incident cases of three cardiovascular disease complications (acute myocardial infarction [AMI], stroke, and heart failure) by International Statistical Classification of Diseases, tenth revision, Australian modification (ICD‐10‐AM) codes. Data for 1999‒2003 were examined to ensure that admissions during the observation period (2004‒2016) were index admissions for the specific complication, but were not included in our main analysis. Admission rates were separately calculated for people with type 1 or type 2 diabetes (numbers of people with diagnosed diabetes, by year, were obtained from the National Diabetes Services Scheme, which captures 80–90% of diabetes diagnoses7) and for people without diabetes (derived from Australian Bureau of Statistics census data8). We analysed changes in admission rates by Joinpoint regression (https://surveillance.cancer.gov/joinpoint); points at which changes in the direction or magnitude of linear trends were statistically significant (P < 0.05) were determined with permutation tests. Each trend segment was described by an annual percentage change (APC), and the change for the entire study period as the mean APC (further details: online Supporting Information). The study was approved by the St Vincent’s Hospital Melbourne Human Research Ethics Committee (HREC/18/SVHM/146). A total of 382 107 patients were admitted to Victorian hospitals during 2004–2016 with cardiovascular complications: 278 991 without diabetes (73%), 3645 with type 1 diabetes (1%), and 99 471 with type 2 diabetes (26%). AMI admission rates declined during this period for people with type 1 (mean APC, –7.7%; 95% confidence interval [CI], –13.4% to –1.5%) or type 2 diabetes (mean APC, –11.4%; 95% CI, –13.0% to –9.9%), as well as for people without diabetes (mean APC, –5.0%; 95% CI, –6.7% to –3.4%) (Box 1, Box 2). Stroke admission rates declined significantly during 2004–2016 for people with type 1 diabetes (mean APC, –7.2%; 95% CI, –12.2% to –1.9%); for people with type 2 diabetes, rates declined during 2005–2011 and 2014–2016, but not during 2011–2014 (overall change: –11.9%; 95% CI, –17.0% to –6.5%). For patients without diabetes, the decline during 2005–2014 was significant (mean APC, –4.1%; 95% CI, –5.8% to –2.3%), but not during 2015–2016 (Box 1, Box 2). Admissions for heart failure declined during 2004–2016 for people with type 1 diabetes (mean APC, –10.3%; 95% CI, –14.1% to –6.4%) or type 2 diabetes (mean APC, –9.2%; 95% CI, –11.0% to –7.3%), and also for people without diabetes (mean APC, –2.8%; 95% CI, –4.1% to –1.5%) (Box 1, Box 2). As hospital discharge coding data do not provide information on metabolic control or medication use, we could not assess whether cardiovascular risk factor modification and use of specific medications were associated with changes in admission rates. We also lacked information on disease duration for patients with hospital‐coded diabetes. Further, we have counted admissions of any patients who had presented with complications before 1998 (ie, outside our 5‐year clearance period) as incident admissions; these patients would be at very high risk of further admissions, and their inclusion may have inflated the admission rates we report for the observation period of our study. Few recent studies have assessed outcomes for all three cardiovascular complications in a single investigation. Cardiovascular complication‐related admissions to Victorian hospitals declined during 2004–2016 more rapidly for people with diabetes than for those without diabetes. The relatively greater absolute decline in the numbers of admissions of people with diabetes may be related to the fact that they are considered to be at high risk for cardiovascular disease and are therefore treated more aggressively; the scope for reducing risk with multifactorial target‐driven interventions is greater in these patients. Nevertheless, admission rates for cardiovascular complications of people with diabetes remain relatively high. Box 1 – Age‐ and sex‐adjusted admission rates for cardiovascular complications (with 95% confidence intervals), Victoria, 2004–2016, by diabetes status of patients Box 2 – Annual percentage change (APC) in admissions for cardiovascular complications, Victoria, 2004–2016, by diabetes status Change in event rate, 2004–2016* Change in event rate, by period* Cardiovascular complication and diabetes status Admissions Overall change (95% CI) Mean APC (95% CI%) Mean APC (95% CI) Acute myocardial infarction No diabetes 114 965 –24.8% (–24.9% to –24.7%) –5.0% (–6.7% to –3.4%) — Type 1 diabetes 1272 –7.7% (–8.8% to –6.7%) –7.7% (–13.4% to –1.5%) 1. 2005–2009: +7.0% (–9.7% to +22.8%) 2. 2009–2016: –15.1% (–21.3% to –8.7%) Type 2 diabetes 15 278 –69.0% (–69.0% to –68.8%) –11.4% (–13.0% to –9.9%) — Stroke No diabetes 52 320 –10.9% (–13.6% to –10.6%) –1.7% (–4.9% to +1.5%) 1. 2005–2014: –4.1% (–5.8% to –2.3%) 2. 2014–2016: +9.6% (–10.2% to +33.8%) Type 1 diabetes 504 –44.4% (–50.0% to –41.4%) –7.2% (–12.2% to –1.9%) — Type 2 diabetes 17 440 –68.0% (–68.0% to –67.9%) –11.9% (–17.0% to –6.5%) 1. 2005–2011: –14.7% (–17.6% to –11.7%) 2. 2011–2014: +5.8% (–19.0% to +38.2%) 3. 2014–2016: –26.1% (–39.8% to –9.2%) Heart failure No diabetes 135 524 –22.2% (–22.3% to –22.2%) –2.8% (–4.1% to –1.5%) — Type 1 diabetes 1393 –55.1% (–58.6% to –52.4%) –10.3% (–14.1% to –6.4%) — Type 2 diabetes 52 831 –67.3% (–67.4% to –67.3%) –9.2% (–11.0% to –7.3%) — * Adjusted for age and sex. Event rates for 2004 and 2016 are included in the expanded version of this table in the online Supporting Information.

Katerina V Kiburg · Andrew I MacIsaac · Andrew Wilson · Vijaya Sundararajan · Richard J MacIsaac

Onsite “buddies” are not always available to monitor the donning and doffing of personal protective equipment (PPE) in hospitals, especially during a pandemic, potentially leading to poor PPE compliance and increased risk of health care infections.1,2 We therefore compared monitoring of PPE donning/doffing procedures in a standard critical care setting3,4 by remote buddies with monitoring by onsite buddies. We designed 30 procedural scenarios (15 donning, 15 doffing) that included random errors in some procedural steps (online Supporting Information). Four buddies (two onsite, two remote), unaware of the number and type of errors in each scenario, concurrently viewed and assessed each step. The remote buddies viewed the procedures via videoconferencing on their computers. The camera of the transmitting laptop computer was positioned so that the entire body of the person donning or doffing PPE could be seen. Procedures were live‐streamed to the remote buddies via the hospital Wi‐Fi network. The buddies were not permitted to communicate with each other or with the person donning or doffing PPE. The study was approved by the Melbourne Health Human Research Ethics Committee (QA2020104). Sensitivity (correctly identifying correct procedure) was 100% for both onsite and remote buddies; specificity (correctly identifying incorrect procedure) was 98.9% for onsite buddies and 94.5% for remote buddies; overall accuracy was respectively 99.7% and 98.7% (Box). Concordance between assessments by onsite and remote buddies (κ = 0.95), by the two onsite buddies (κ = 0.97), and by the two remote buddies (κ = 0.98) was very good. The most frequent error was remote buddies missing chin exposure below the mask, probably because of the two‐dimensional view provided by the camera. Paying specific attention to the mask position when the donner turns side on in front of the camera might prevent this error. Practical considerations for remote buddies include the need for reliable hospital network and internet connections, or a wired hardware system, to avoid disruption of monitoring. As the remote buddy is unable to physically intervene when they identify an error, clear verbal communication is important. The psychological effect of having an onsite buddy was not characterised, but may influence user acceptability of remote buddies. All buddies were very experienced in providing observation feedback, but we did not assess their proficiency. Their accuracy may also have reflected greater vigilance while being observed (the Hawthorne effect). Finally, we did not weight the donning and doffing steps according to their importance for safety. Having a trained observer monitor PPE compliance is important for health care safety. The high level of accuracy and the agreement between onsite and remote buddies were encouraging. Apart from identifying errors, remote buddies could also provide step‐by‐step instruction in donning and doffing procedures, which could improve compliance and minimise contamination.5 Using remote buddies may help preserve PPE supplies and ensure reliable access to monitoring, even when PPE supply or onsite staff numbers are limited, while also reducing the infection exposure risk for the monitoring observers. Box – Personal protective equipment (PPE) monitoring assessment accuracy by onsite and remote buddies Scenario outcome* Buddy outcome* Pass Fail Onsite buddies (390 tests) Pass 298 1 PPV, 99.7% Fail 0 91 NPV, 100% Sensitivity, 100% Specificity, 98.9% Overall accuracy, 99.7% Remote buddies (383 tests†) Pass 292 5 PPV, 98.3% Fail 0 86 NPV, 100% Sensitivity, 100% Specificity, 94.5% Overall accuracy, 98.7% PPV = positive predictive value; NPV = negative predictive value. * For each step of each PPE donning/doffing procedure: pass = correctly performed; fail = not correctly performed. † Seven assessments were missing because of internet interruptions.

Reny Segal · William PL Bradley · Daryl Williams · Romulo Correa de Araujo Nunes · Irene Ng

Objective: To determine rates of potentially preventable hospitalisation of people with intellectual disability in New South Wales, and compare them with those for the NSW population. Design: Retrospective cohort study. Setting: Potentially preventable hospitalisations in NSW, as defined by the National Healthcare Agreement progress indicator 18, 1 July 2001 ‒ 30 June 2015. Participants: Data collected in a retrospective data linkage study of 92 542 people with intellectual disability in NSW; potentially preventable hospitalisations data for NSW published by HealthStats NSW. Main outcome measures: Age‐adjusted rates of potentially preventable hospitalisation by group (people with intellectual disability, NSW population), medical condition type (acute, chronic, vaccine‐preventable), and medical condition. Results: The annual age‐standardised rate for people with intellectual disability ranged between 5286 and 6301 per 100 000 persons, and for the NSW population between 1278 and 1511 per 100 000 persons; the rate ratio (RR) ranged between 3.5 (95% CI, 3.3–3.7) in 2014–15 and 4.5 (95% CI, 4.2–4.9) in 2002–03. The difference was greatest for admissions with acute (RR range: 5.3 [95% CI, 4.9–5.7] in 2014–15 to 8.1 [95% CI, 7.4–8.8] in 2002–03) and vaccine‐preventable conditions (RR range: 2.1 [95% CI, 1.6–3.0] in 2007–08 to 3.4 [95% CI, 2.2–5.2] in 2004–05). By specific condition, the highest age‐standardised rate was for admissions with convulsions and epilepsy (all years, 2567 per 100 000 population; v NSW population: RR, 22.2; 95% CI, 21.3–23.1). Conclusion: Age‐standardised rates of potentially preventable hospitalisation are higher for people with intellectual disability than for the general population. The reasons for these differences should be investigated, and strategies for averting potentially preventable hospitalisation developed.

Janelle C Weise · Preeyaporn Srasuebkul · Julian N Trollor

People in hotel quarantine need Increased psychological support and other mental health services

Michael Dinh · Owen Hutchings · Kendall Bein · James Shaw · Aditi Raut · Saartje Berendsen Russell · Dianna Jagers · Rajip Thapa · Gregory Fox · James Edwards · Timothy Wand · Warwick Britton

Australians have around 19 000 hip fractures each year,1 and the estimated cost to the health care system was $445 million in 2015–16.2 Surgery within 48 hours of initial presentation to hospital is widely accepted as a clinically meaningful indicator of best practice care, and is supported by the Australian Hip Fracture Care Clinical Care Standard when there are no clinical contraindications.3 However, timely access to emergency orthopaedic hip fracture surgery is difficult in a country as large and geographically diverse as Australia; patients admitted to remote or regional hospitals that do not provide orthopaedic surgery must be transferred to larger regional centres. In a retrospective population study, we evaluated the impact of pre‐surgery hospital transfer and time to surgery on 30‐day mortality for people aged 65 years or more who underwent surgical interventions for fall‐related hip fractures in NSW public hospitals during 1 January 2011 – 31 December 2018. Hospitalisation data from the NSW Admitted Patient Data Collection and deaths data from the NSW Registry of Births, Deaths and Marriages were linked to provide person‐level records. Time to surgery (in calendar days) was estimated from the date of admission for the first episode of care to the date of surgery. Comorbid conditions during the preceding year were identified with the Charlson Comorbidity Index (CCI). Multilevel multivariable logistic regression models were fitted to assess the influence of patient‐level factors (age, sex, comorbidity) and process factors (transfer status, time to surgery) on 30‐day mortality. Operating hospitals were included as a random effect to account for variation between hospitals. Adjusted odds ratios (aORs) with 95% confidence intervals (CIs) were calculated and residual variation (variance partition coefficient) assessed. All analyses were performed in SAS Enterprise Guide 7.1 and MLwiN 3.02 (http://www.bristol.ac.uk/cmm/software/mlwin). The NSW Population and Health Services Research Ethics Committee approved the study (HREC/17/CIPHS/45). Of 36 956 patients who underwent hip fracture repair procedures in 36 hospitals, 3916 (10.6%) were transferred from peripheral hospitals to operating hospitals for surgery; 1579 were transferred on the day of presentation (40.3%), 1875 the following day (47.9%), and 462 patients (11.8%) spent at least two days at the admitting hospital before being transferred. Larger proportions of transferred patients than of patients admitted directly to operating hospitals were men (29.4% v 27.8%), under 85 years of age (50.9% v 48.4%), or had CCI scores of 1 or more (60.2% v 56.3%). The proportion of transferred patients who underwent surgery within 48 hours of presentation was smaller than for directly admitted patients (53.9% v 72.4%) (Box). In multilevel models adjusted for inter‐hospital variation, transfer was associated with higher risk of 30‐day mortality than direct admission (aOR, 1.15; 95% CI, 1.01–1.32), but after adjusting for age, sex, and comorbidity, neither transfer (aOR, 1.10; 95% CI, 0.95–1.28) nor delayed surgery (> 2 days v ≤ 2 days: aOR, 0.99; 95% CI, 0.89–1.11) significantly influenced mortality. The most influential factor was comorbidity (CCI ≥ 3 v CCI < 3: aOR, 4.89; 95% CI, 4.32–5.54). The discrimination of our fully adjusted model was adequate (area under the curve, 0.73), and 1.8% of residual variation in 30‐day mortality was attributable to differences between hospitals. In our large study of NSW people with hip fractures, we found that transfer from non‐operating to operating hospitals, after adjusting for patient and hospital characteristics, was not associated with higher 30‐day mortality, despite increasing the time between initial presentation and surgery. This is contrary to the findings of earlier, single centre studies in Australia.4,5,6 However, our study was the first to control for several key person‐level factors that increase the risk of death, and our findings suggest that time to surgery may be less important for health outcomes than these factors when other dimensions of care quality are equal. More research is required to understand the interplay between the effects of patient demographic characteristics, pre‐injury health status, and the quality of hip fracture care on 30‐day mortality for patients. Box – Characteristics of patients with hip fractures, by pre‐surgery transfer, New South Wales, 2011–2018* table#t1 tbody td:nth-child(n+2) P. Pleft { text-align: center; } Not transferred Transferred Number of people 33 040 (89.4%) 3916 (10.6%) Sex Women 23 866 (72.2%) 2766 (70.6%) Men 9174 (27.8%) 1150 (29.4%) Age at admission (years) 65–74 4684 (14.2%) 535 (13.7%) 75–84 11 311 (34.2%) 1458 (37.2%) ≥ 85 17 045 (51.6%) 1923 (49.1%) Weighted Charlson Comorbidity Index score 0 14 437 (43.7%) 1556 (39.7%) 1–2 12 667 (38.3%) 1595 (40.7%) ≥ 3 5936 (18.0%) 765 (19.5%) Time to transfer (days) 0 1579 (40.3%) 1 1875 (47.9%) ≥ 2 462 (11.8%) Time to surgery (days) 0 12 991 (39.3%) 739 (18.9%) 1 10 939 (33.1%) 1370 (35.0%) ≥ 2 9110 (27.6%) 1807 (46.1%) Length of stay (days), mean (SD) Total 27.5 (21.9) 26.8 (20.5) Acute care 11.9 (8.5) 12.8 (9.0) 30‐day deaths 2172 (6.6%) 288 (7.4%) SD = standard deviation. * Linked hospitalisation and deaths data.

Lara A Harvey · Ian A Harris · Rebecca J Mitchell · Adrian Webster · Ian D Cameron · Louisa R Jorm · Hannah Seymour · Pooria Sarrami · Jacqueline CT Close

Objective: To assess long term survival and patient characteristics associated with survival following acute myocardial infarction (AMI) in Australia and New Zealand. Design: Cohort study. Setting, participants: All patients admitted with AMI (ICD‐10‐AM codes I21.0‒I21.4) to all public and most private hospitals in Australia and New Zealand during 2009‒2015. Main outcome measure: All‐cause mortality up to seven years after an AMI. Results: 239 402 initial admissions with AMI were identified; the mean age of the patients was 69.3 years (SD, 14.3 years), 154 287 were men (64.5%), and 64 335 had ST‐elevation myocardial infarction (STEMI; 26.9%). 7‐year survival after AMI was 62.3% (STEMI, 70.8%; non‐ST‐elevation myocardial infarction [NSTEMI], 59.2%); survival exceeded 85% for people under 65 years of age, but was 17.4% for those aged 85 years or more. 120 155 patients (50.2%) underwent revascularisation (STEMI, 72.2%; NSTEMI, 42.1%); 7‐year survival exceeded 80% for patients in each group who underwent revascularisation, and was lower than 45% for those who did not. Being older (85 years or older v 18–54 years: adjusted hazard ratio [aHR], 10.6; 95% CI, 10.1–11.1) or a woman (aHR, 1.15; 95% CI, 1.13–1.17) were each associated with greater long term mortality during the study period, as was prior heart failure (aHR, 1.79; 95% CI, 1.76‒1.83). Several non‐cardiac conditions and geriatric syndromes common in these patients were independently associated with lower long term survival, including major and metastatic cancer, cirrhosis and end‐stage liver disease, and dementia. Conclusion: AMI care in Australia and New Zealand is associated with high rates of long term survival; 7‐year rates exceed 80% for patients under 65 years of age and for those who undergo revascularisation. Efforts to further improve survival should target patients with NSTEMI, who are often older and have several comorbid conditions, for whom revascularisation rates are low and survival after AMI poor.

Bora Nadlacki · Dennis Horton · Sadia Hossain · Saranya Hariharaputhiran · Linh Ngo · Anna Ali · Bernadette Aliprandi‐Costa · Chris J Ellis · Robert JT Adams · Renuka Visvanathan · Isuru Ranasinghe

Objectives: To examine associations between community cultural connectedness indicators and suicide mortality rates for young Aboriginal and Torres Strait Islander people. Study design: Retrospective mortality study. Setting, participants: Suicide deaths of people aged 10‒19 years recorded by the Queensland Suicide Register, 2001‒2015. Main outcome measures: Age‐standardised suicide death rates, by Indigenous status, sex, and age group; age‐standardised suicide death rates for young First Nations people by area level remoteness and Index of Relative Socioeconomic Advantage and Disadvantage, and by cultural connectedness indicators (at statistical area level 2): cultural social capital index score, community Indigenous language use, and reported discrimination. Results: The age‐specific suicide rate was 21.1 deaths per 100 000 persons/year for First Nations young people and 5.0 deaths per 100 000 persons/year for non‐Indigenous young people (rate ratio [RR], 4.3; 95% CI, 3.5‒5.1). The rate for Aboriginal and Torres Strait Islander young people was higher in areas with low levels of cultural social capital (greater participation of community members in cultural events, ceremonies, organisations, and community activities) than in areas classified as having high levels (RR, 1.8; 95% CI, 1.2‒2.7), and also in communities with high levels of reported discrimination (RR, 2.7; 95% CI, 1.7‒4.3). Associations with proportions of Indigenous language speakers and area level socio‐economic resource levels were not statistically significant. Conclusion: We found that suicide mortality rates for Aboriginal and Torres Strait Islander young people in Queensland were influenced by community level culturally specific risk and protective factors. Our findings suggest that strategies for increasing community cultural connectedness at the community level and reducing institutional and personal discrimination could reduce suicide rates.

Mandy Gibson · Jaimee Stuart · Stuart Leske · Raelene Ward · Robert Tanton

Characterising the care requirements of patients with coronavirus disease 2019 (COVID‐19) is essential for resource allocation.1 Knowledge of care needs is based predominantly on experience in regions where health care capacity has been strained, and may not reflect ideal practice.2,3 We therefore examined COVID‐19 testing data for South Australia, the care requirements of patients with confirmed COVID‐19, and the disposition of people with potential COVID‐19 who presented to the designated COVID‐19 hospital for SA, the Royal Adelaide Hospital (RAH), during a period of low COVID‐19 prevalence and limited community transmission (30 January – 26 April 2020). We analysed SA Pathology data on tests for severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) and other respiratory pathogens, and clinical data from hospital electronic health records (further details: online Supporting Information). The Central Adelaide Local Health Network Human Research Ethics Committee approved the study, and waived the requirement for patient consent (reference, 13091). Of 52 883 people tested in SA for SARS‐CoV‐2, 438 had polymerase chain reaction (PCR)‐confirmed infections (0.8%); their median age was 54 years (interquartile range [IQR], 31–64 years; range, 1–94 years), and 211 were female (48%). The median age of screened people with negative results was 43 years (IQR, 28–60 years; range, 0–104 years), of whom 30 380 were female (58%). Seventeen people with confirmed COVID‐19 (3.9%) and 7588 of those without COVID‐19 (14%) were also positive for another respiratory pathogen (Supporting Information, table 1). There were no cases of COVID‐19 among people in high care nursing facilities or prisons, nor among homeless people; one infection of a health care worker caring for people with COVID‐19 was recorded in Adelaide. The number of patients admitted to the RAH with COVID‐19 broadly paralleled that of new cases in SA, but intensive care unit (ICU) occupancy peaked (6/7 April) and the number of people screened in the RAH emergency department for COVID‐19 declined (from 17 April) after the peak in new cases (21 March) (Box 1). The median time from diagnosis to viral clearance (according to national guidelines4) was 15 days (IQR, 12–19 days); it was lower for people managed in the community (14 days; IQR, 11–17 days) than for those admitted to hospital (17 days; IQR, 13–22 days), and there were no sex‐ or age‐related differences (data not shown). A total of 285 people with confirmed COVID‐19 (227 aged 18–65 years; 58 over 65 years) were managed entirely in the community (Box 2); their age and sex distributions were similar to those of all SARS‐CoV‐2‐positive people (data not shown). Of 18 228 patients who presented to the RAH emergency department, 2327 (12.8%) met screening criteria for potential COVID‐19, of whom 120 (5.2%) proved to be SARS‐CoV‐2‐positive (new diagnoses in 19 people) (Supporting Information, figure). Among people who met the criteria for potential COVID‐19, a larger proportion of people with positive results than of those with negative results arrived by private vehicle (59 [49%] v 797 [36%]), and smaller proportions required resuscitation (one [0.8%] v 72 [3%]) or had conditions deemed imminently life‐threatening (14 [12%]) v 706 [32%]); among people over 65 years, two of 37 SARS‐CoV‐2‐positive people (5%) and 406 of 939 SARS‐CoV‐2‐negative people (43%) required resuscitation or emergency review. Most people with confirmed infections were admitted to the inpatient COVID‐19 unit (90 [75%] v 419 with negative results [19%]), while three SARS‐CoV‐2‐positive (2%) and 79 SARS‐CoV‐2‐negative people (4%) were admitted from the emergency department to the ICU (Supporting Information, table 2). One of 18 228 people who presented to the ED did not meet screening criteria for potential COVID‐19 but subsequently tested positive (screening failure rate, 0.005%). A total of 536 patients were admitted to the inpatient COVID‐19 unit, including 117 who were SARS‐CoV‐2‐positive (22%). The proportion of SARS‐CoV‐2‐positive patients aged 18–65 years was larger than for other patients in the COVID‐19 unit (84 [72%] v 188 patients [45%]); the proportions of women were similar (53 [45%] v 186 patients [44%]). Median length of stay was longer for SARS‐CoV‐2‐positive than for SARS‐CoV‐2‐negative patients over 65 years of age (182 h; IQR, 87–285 h v 96 h; IQR, 48–158 h), but was similar for all patients aged 18–65 years. Six SARS‐CoV‐2‐positive (18%) and five SARS‐CoV‐2‐negative patients over 65 (2%) were transferred from the COVID‐19 unit to the ICU (Supporting Information, table 3). Seventeen patients hospitalised with COVID‐19 (14%) were admitted to the ICU. The median time from hospital to ICU admission was 2.4 days (IQR, 1.8–3.4 days) for the eight patients over 65, and 5.2 days (IQR, 1.0–6.1 days) for the nine aged 18–65 years; the median ICU stay was 17.3 days (IQR, 3.0–29.3 days) for those over 65, and 2.2 days (IQR, 1.6–4.4 days) for those aged 18–65 years. Four patients died (24%), the only COVID‐19‐related deaths in South Australia (overall case fatality, 0.9%; 18–65 years, 0.3%; over 65 years, 3.2%) (Supporting Information, table 4). Over the past 14 years, 15% of RAH patients with viral pneumonia in intensive care died, with a medium length of stay of 6.2 days (18–65 years, 7.4 days; over 65 years, 4.8 days) (unpublished data). More modest, but persistent, prevalence of COVID‐19 is expected to follow the major pandemic wave of 2020. Our data, gathered in an environment of low community transmission and a health care system with considerably greater capacity than demand, reflects the COVID‐19‐related resource burden that might be anticipated as we prepare for living with COVID‐19. Box 1 – New confirmed cases of COVID‐19 in South Australia, numbers of inpatients with COVID‐19 in the Royal Adelaide Hospital, and numbers of people presenting with potential COVID‐19 infection to the Royal Adelaide Hospital emergency department, 30 January – 26 April 2020 COVID‐19 = coronavirus disease 2019. Box 2 – Care requirements of people with confirmed COVID‐19 admitted to the Royal Adelaide Hospital, 30 January – 26 April 2020 COVID‐19 = coronavirus disease 2019; SARS‐CoV‐2 = severe acute respiratory syndrome coronavirus 2. * Includes one patient initially admitted under a non‐COVID‐19 inpatient team. † Fourteen SARS‐CoV‐2‐positive patients admitted under the COVID‐19 inpatient team required transfer to the intensive care unit, 11 of whom returned to the COVID‐19 inpatient team, as did two of three patients admitted to the intensive care unit from the emergency department. These patients are counted in both intensive care unit and COVID‐19 inpatient team numbers. ‡ includes three intensive care unit patients admitted directly from the emergency department then transferred to the inpatient team, and one patient who was still an inpatient at the end of the study.

Daniel Haustead · Dylan J Toh · Benjamin Reddi · Emily Kirkpatrick · Emily Rowe · Pamela Outhwaite · Elizabett Harnack · Michael Cusack · Megan Brooks

Objectives: To assess the quality of care for patients with diabetes in Queensland hospitals, including blood glucose control, rates of hospital‐acquired harm, the incidence of insulin prescription and management errors, and appropriate foot and peri‐operative care. Design, setting: Cross‐sectional audit of 27 public hospitals in Queensland: four of five tertiary/quaternary referral centres, four of seven large regional or outer metropolitan hospitals, seven of 13 smaller outer metropolitan or small regional hospitals, and 12 of 88 hospitals in rural or remote locations. Participants: 850 adult inpatients with diabetes mellitus in medical, surgical, mental health, high dependency, or intensive care wards. Results: Twenty‐seven of 115 public hospitals that admit acute inpatients participated in the audit, including 4175 of 6652 eligible acute hospital beds in Queensland. A total of 1003 patients had diabetes (24%), and data were collected for 850 (85%). Their mean age was 65.9 years (SD, 15.1 years), 357 were women (42%), and their mean HbA1c level was 66 mmol/mol (SD, 26 mmol/mol). Rates of good diabetes days (appropriate monitoring, no more than one blood glucose measurement greater than 10 mmol/L, and none below 5 mmol/L) were low in patients with type 1 diabetes (22.1 per 100 patient‐days) or type 2 diabetes treated with insulin (40.1 per 100 patient‐days); hypoglycaemia rates were high for patients with type 1 diabetes mellitus (24.1 episodes per 100 patient‐days). One or more medication errors were identified for 201 patients (32%), including insulin prescribing errors for 127 patients (39%). Four patients with type 1 diabetes experienced diabetic ketoacidosis in hospital (8%); 121 patients (14%) met the criteria for review by a specialist diabetes team but were not reviewed by any diabetes specialist (medical, nursing, allied health). Conclusions: We identified several deficits in inpatient diabetes management in Queensland, including high rates of medication error and hospital‐acquired harm and low rates of appropriate glycaemic control, particularly for patients treated with insulin. These deficits require attention, and ongoing evaluation of outcomes is necessary.

Peter Donovan · Jade Eccles-Smith · Nicola Hinton · Clare Cutmore · Kerry Porter · Jennifer Abel · Lee Allam · Alexis Dermedgoglou · Gaurav Puri

Objectives: To estimate initial levels of symptoms of depression and anxiety, and their changes during the early months of the COVID‐19 pandemic in Australia; to identify trajectories of symptoms of depression and anxiety; to identify factors associated with these trajectories. Design, setting, participants: Longitudinal cohort study; seven fortnightly online surveys of a representative sample of 1296 Australian adults from the beginning of COVID‐19‐related restrictions in late March 2020 to mid‐June 2020. Main outcome measures: Symptoms of depression and anxiety, measured with the Patient Health Questionnaire (PHQ‐9) depression and Generalised Anxiety Disorder (GAD‐7) scales; trajectories of symptom change. Results: Younger age, being female, greater COVID‐19‐related work and social impairment, COVID‐19‐related financial distress, having a neurological or mental illness diagnosis, and recent adversity were each significantly associated with higher baseline depression and anxiety scores. Growth mixture models identified three latent trajectories for depression symptoms (low throughout the study, 81% of participants; moderate throughout the study, 10%; initially severe then declining, 9%) and four for anxiety symptoms (low throughout the study, 77%; initially moderate then increasing, 10%; initially moderate then declining, 5%; initially mild then increasing before again declining, 8%). Factors statistically associated with not having a low symptom trajectory included mental disorder diagnoses, COVID‐19‐related financial distress and social and work impairment, and bushfire exposure. Conclusion: Our longitudinal data enabled identification of distinct symptom trajectories during the first three months of the COVID‐19 pandemic in Australia. Early intervention to ensure that vulnerable people are clinically and socially supported during a pandemic should be a priority.

Philip J Batterham · Alison L Calear · Sonia M McCallum · Alyssa R Morse · Michelle Banfield · Louise M Farrer · Amelia Gulliver · Nicolas Cherbuin · Rachael M Rodney Harris · Yiyun Shou · Amy Dawel

During the coronavirus disease 2019 (COVID‐19) epidemic, the continued presence of viral RNA in the upper airways of infected people has been reported.1 Such persistence does not necessarily signify active infection or that the virus can be transmitted.2 In Queensland, repeat testing for severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) in people with an initial positive test result was undertaken until June 2020, providing an opportunity to explore patterns of test positivity, apparent rates of clearance of viral RNA, and the extent to which each varied by the age and sex of the infected person. We analysed de‐identified data for people who underwent swab tests for SARS‐CoV‐2 processed in Queensland Health public laboratories between 10 January and 4 June 2020. SARS‐CoV‐2 RNA was detected by polymerase chain reaction (PCR). Testing was initially restricted to people with relevant symptoms who had visited high risk areas (Box 1); from April 2020, anyone with relevant symptoms could be tested. We analysed data on PCR test result, age and sex of the tested person, and postcode of the facility that requested the test; clinical information and reasons for testing were not available. People with positive results who subsequently received two consecutive negative test results at least 24 hours apart were defined as achieving “negative status”. Details of dataset structure, analysis and visualisation methods and code have been reported elsewhere.3,4 Our investigation was exempted from formal ethics review by the Gold Coast Health Human Research Ethics Committee (reference, LNR/2020/QGC/63045). We analysed data for 103 984 swabs from 97 476 people during the 146‐day study period. Time to negative status was examined by Kaplan–Meier analysis. Differences by age (under 65 years, 65 years or over) and sex, with adjustment for both, were calculated by Cox regression. Other associations between variables were quantified as unadjusted odds ratios. The timing of sample collection, particularly of repeat swabs, was not standardised, reflecting the exploratory nature of SARS‐CoV‐2 testing early in the pandemic. The median age of tested people was 41 years (interquartile range [IQR], 27–57 years; range, under one to 105 years); 55 708 (57%) were female. Nine hundred and fifty‐eight people (0.98%) were positive for SARS‐CoV‐2; their median age was 45 years (IQR, 29–61 years; range, under one to 88 years), and 496 were female (52%). Compared with people under 16 years of age, the odds of a positive result were higher for people aged 17–64 years (odds ratio [OR], 5.2; 95% confidence interval [CI], 3.4–8.1) and for those aged 65 years or more (OR, 6.0; 95% CI, 4.0–9.5); the odds of a positive test were lower for females than for males (OR, 0.80; 95% CI, 0.70–0.91). The numbers of people tested and of those positive for SARS‐CoV‐2 both peaked in the second half of March 2020, after which testing rates declined until late April before climbing again, while positivity rates remained low (Box 1). Of the 958 people with positive test results, 317 (33.1%) had repeat tests. Of the 243 people with initial positive results and at least two repeat tests, 147 (60.5%) achieved negative status. The median age of those who achieved negative status was 45 years (IQR, 30–59 years; range 20–84 years); 94 were women, 53 men (OR, 1.7; 95% CI, 1.0–2.9). Of the 243 people who underwent two or more repeat tests, 224 (92.2%) had positive results beyond 10 days and up to 72 days after their initial tests (Box 2). Seven of 147 people who achieved negative status (5%) subsequently had positive test results, including six men. For the 147 positive patients who achieved negative status, median time to clearance was 31 days (IQR, 18–47 days), and was unaffected by sex (women, 30 days; IQR 16–45 days; men: 31 days; IQR 20–49 days; hazard ratio [HR], 0.93; 95% CI, 0.66–1.3). Clearance was more rapid in people under 65 years of age (median, 29 days; IQR, 17–45 days) than in people aged 65 years or more (median, 43 days; IQR, 25–62 days; HR, 1.82; 95% CI, 1.17–2.93) (Box 3). We found that positive PCR test results often persisted for ten or more days after an initial positive result, in one case for 72 days. Such persistence does not indicate continued viral replication.2,5 From 21 March 2020, patients in Queensland, other than workers at high risk, were released from isolation on the basis of their symptoms and illness duration (ie, without further testing), and local transmission declined to zero (Box 1). Our finding of lower infection rates in younger people is consistent with previous reports,6 as is our finding that infection rates were higher for males than females.7 After adjusting for age, the viral clearance rate was similar for males and females. Clearance was greater for people under 65 years of age than for those aged 65 or more, as noted previously.8 This effect may have clinical significance; rates of hospitalisation, admission to intensive care, and death from COVID‐19 are higher among older people. Box 1 – Numbers of SARS‐CoV‐2 tests processed by Queensland Health public laboratories and of people with positive results, 10 January – 4 June 2020, with trend lines and indications for testing* * Repeat tests after first positive result are not included. Test trend line based on a generalised additive model for “all tests”; positive result trend line based on local polynomial regression fitting. Box 2 – Categorical heat map of SARS‐CoV‐2 tests for people with initial positive results who had at least two subsequent tests Box 3 – Kaplan–Meier analyses of virus clearance in 958 people who were initially positive for SARS‐CoV‐2, by age and sex* * Confidence bands generated by Cox proportional hazards regression, with Efron approximation (coxph function in R 3.6.3).

Paulina Stehlik · Kylie Alcorn · Anna Jones · Sanmarie Schlebusch · Andre Wattiaux · David A Henry

Objectives: To determine whether a multifaceted primary health care intervention better controlled cardiovascular disease (CVD) risk factors in patients with high risk of CVD than usual care. Design, setting: Parallel arm, cluster randomised trial in 71 Australian general practices, 5 December 2016 – 13 September 2019. Participants: General practices that predominantly used an electronic medical record system compatible with the HealthTracker electronic decision support tool, and willing to implement all components of the INTEGRATE intervention. Intervention: Electronic point‐of‐care decision support for general practices; combination cardiovascular medications (polypills); and a pharmacy‐based medication adherence program. Main outcome measures: Proportion of patients with high CVD risk not on an optimal preventive medication regimen at baseline who had achieved both blood pressure and low‐density lipoprotein (LDL) cholesterol goals at study end. Results: After a median 15 months’ follow‐up, primary outcome data were available for 4477 of 7165 patients in the primary outcome cohort (62%). The proportion of patients who achieved both treatment targets was similar in the intervention (423 of 2156; 19.6%) and control groups (466 of 2321; 20.1%; relative risk, 1.06; 95% CI, 0.85–1.32). Further, no statistically significant differences were found for a number of secondary outcomes, including risk factor screening, preventive medication prescribing, and risk factor levels. Use of intervention components was low; it was highest for HealthTracker, used at least once for 347 of 3236 undertreated patients with high CVD risk (10.7%). Conclusions: Despite evidence for the efficacy of its individual components, the INTEGRATE intervention was not broadly implemented and did not improve CVD risk management in participating Australian general practices. Trial registration: Australian New Zealand Clinical Trials Registry, ACTRN12616000233426 (prospective).

Ruth Webster · Tim Usherwood · Rohina Joshi · Bandana Saini · Carol Armour · Sue Critchley · Gian Luca Di Tanna · Shane Galgey · Charlotte M Hespe · Stephen Jan · Ajay Karia · Baldeep Kaur · Ines Krass · Tracey‐Lea Laba · Qiang Li · Serigne Lo · David P Peiris · Christopher Reid · Anthony Rodgers · Louise Shiel · Jessica Strathdee · Nuria Zamora · Anushka Patel