ABSTRACT
Background: Poor sleep is common in the community (33%–45%) and even more common in people with multiple sclerosis (MS) (60%–70%). Strategies used by people living with MS to get to sleep or stay asleep, and whether they impact sleep quality and daytime sleepiness are not clear. We aimed to identify sleep treatment strategies and evaluate the effect of prescription medications on sleep quality and daytime sleepiness in people with MS, after accounting for why medications were used.
Methods: Australian MS Longitudinal Study participants (n=1590) self-reported treatments used for sleep, sleep quality, and daytime sleepiness. Prescription medications that could influence sleep were identified from administrative data collected <3 months prior to self-report. Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI); daytime sleepiness using the Epworth Sleepiness Scale (ESS). Effects of prescription medications on sleep outcomes were examined using doubly robust inverse probability weighted regression adjustment (IPWRA), to account for confounding by indication.
Results: Using treatment strategies to improve sleep was common, with 72% utilising ≥1 and 47% utilising ≥3 treatments. Poor sleepers were more likely to use any strategy (80% vs 52%), multiple strategies (50% vs 33%), sleep hygiene behaviours (62% vs 42%), and medications (51% vs 13%), compared to good sleepers. Benzodiazepine users had worse sleep quality, selective serotonin reuptake inhibitors users had worse daytime sleepiness, even after adjusting for indications for medication use and factors associated with sleep quality or daytime sleepiness.
Conclusion: Using treatment strategies to get to sleep or stay asleep was common. As expected, those with sleep in the clinically poor range were most likely to use any strategies, and multiple strategies. Most prescription medications had no impact on sleep after adjusting for confounding. Participants using benzodiazepines had poorer sleep quality, and selective serotonin reuptake inhibitors users had poorer daytime sleepiness.

Background: Loneliness and social isolation are associated with poor mental health. Young people with disability are more likely to experience loneliness and social isolation, and have poorer mental health, compared to young people without disability. This project examined whether loneliness, social support and social connection modified the effect of disability on mental health in young people. Methods: We used data from 1,940 young people (15-24 years) obtained from wave 22 of the Household, Income and Labour Dynamics in Australia Survey. People were categorised into three disability groups: (i) no disability; (ii) non-psychological disability; (iii) psychological disability, with or without other types of disability. We used population-weighted linear regression models to estimate mean differences in mental health scores between the disability groups, including interaction terms to examine effect modification. Four effect modifiers were considered: loneliness, social support, frequency of social connection, and having a pet. Lived experience from young people with disability informed this project. Results: Young people with psychological disability had substantially poorer mental health compared to young people without disability or with non-psychological disability. There was some evidence of effect modification by social support, loneliness and frequency of social connection for the impact of disability type on mental health. Although there was no difference in the impact of social connection between people with no disability and non-psychological disability, the estimated mean difference in mental health score between young people with psychological disability and young people without disability was of greater magnitude for people who were lonely, had low levels of social support, and less frequent social interactions. There was no evidence of effect modification by having a pet. Conclusion: These findings highlight the importance of developing strategies to reduce loneliness, increase social support and promote social connection for young people with psychological disability to improve their mental health outcomes.

BACKGROUND
Welding is a common work activity in a wide variety of industries such as mining, manufacturing and construction and many Australian workers are exposed to welding fume in the course of their work. There is little published information available on exposure levels and the use of control measures in regard to occupational exposure to welding fume in Australia. This study aimed to provide information on the level of exposure to welding fume in workplaces in Australia and the use of control measures in these workplaces.

METHODS
In a cross-sectional study, personal and static samples of welding fume particulates and gases were collected at a convenience sample of 20 workplaces in New South Wales whilst welding was being undertaken. Personal samples were taken from the breathing zone of 67 welders. In addition, eighty-two static samples were collected and personal real-time ozone monitoring was undertaken for a welder at each of 16 worksites. Study personnel also recorded observations of the workplace.

RESULTS
The total welding fume particulate concentrations in the breathing zone were above the newly established workplace exposure standard for 61% of the welders. Peak concentrations of ozone were above the short-term exposure limit for most welders (75%) when they had their head over the welding area. Very few workplaces had active ventilation in place and functioning at the time of the worksite visits. The use of respiratory protective equipment was sub-optimal; although 54% of welders used powered air-purifying respirators some of the time, about a quarter (22%) used no respiratory protection at all.

CONCLUSION
This study provides the first comprehensive information on exposure to welding fume particulate and gases in Australian workplaces. The findings suggest there is considerable scope for improvement in the control of exposure to welding fume.

Background:
The evolution of data-driven public health research demands robust causal inference skills. Despite its importance in epidemiology and biostatistics education, there are significant challenges in integrating modern causal inference approach into mainstream curricula. Understanding both the current landscape and barriers to modern methodological training is crucial for developing the knowledge and skills of future epidemiologists.
Methods:
We employed a multiphase mixed-methods approach: (1) A systematic scoping review following Arksey and O’Malley’s five-step framework, analysing learning materials from peer-reviewed journals and online platforms using Bigg’s constructive alignment theory; and (2) A knowledge, attitude, and practices (KAP) survey administrated to epidemiologists, biostatisticians, clinical researchers and students across academic institutions to identify barriers in methodological capacity building.
Results:
The scoping review revealed substantial variation in training approaches, with structured courses offering comprehensive coverage but lacking customisation options and cost variations. Tutorial papers provided in-depth methodological focus but required high learner autonomy. The KAP survey (n=29) identified critical barriers to adopting causal inference methods: 53% reported difficulty comprehending mathematical notations, 44% faced time constraints, 31% struggled with method complexity, and 14% cited financial barriers to access training courses. These finding highlighted systemic challenges in current epidemiological capacity building strategies.
Conclusion:
This comprehensive assessment demonstrates the need for standardised approaches to causal inference training and addresses specific barriers in methodological capacity building. Future curriculum development and career support programs should focus on flexible learning formats, time-efficient modules, instructor support, resource accessibility, and a better balance of theoretical-practical integration to strengthen epidemiological training.

The World Health Organization recommends that influenza-vaccine-naïve children <9 years receive two doses, at least one month apart, in the first year of vaccination. This recommendation is based on immunological evidence.
To assess the real-world evidence for this schedule we conducted a systematic review and meta-analysis of studies of influenza vaccine efficacy and effectiveness by dose for children that were previously influenza-vaccine-naïve. To control for intra-seasonal variation, we calculated the pooled absolute difference of directly comparable vaccine efficacy and effectiveness estimates.
Our search of EMBASE, Medline OVID and CINAHL identified 49 papers that met the inclusion criteria. Sixteen papers reported the results of 13 randomised controlled trials (RCTs), 27 were inactivated influenza vaccine (IIV) effectiveness studies and six were monovalent 2009 pandemic effectiveness studies.
All seven of the live attenuated influenza vaccine (LAIV) trials assessed one-dose efficacy and three also assessed two-dose efficacy. Two studies reported directly comparable one and two dose efficacy and the pooled absolute difference between the directly comparable estimates of LAIV efficacy was 10% (95%CI: -1.6% to 21%).
Of the 27 IIV effectiveness studies, 13 were classified as naïve studies (used a population that was previously influenza-vaccine-naïve) and 14 were classified as mixed history studies (only children who received one dose were guaranteed to be previously influenza-vaccine-naïve). Pooled absolute difference between one and two doses of IIV effectiveness from naïve studies was 14%; (95%CI: -3.4% to 31%).
Existing studies are insufficient to assess the additional benefit of a second dose of influenza vaccine in the first year of vaccination. Half the identified VE studies did not adequately control for influenza vaccine history. Further high-quality observational studies or RCTs comparing the impact of one and two doses among influenza-vaccine-naïve children are required to guide immunisation policy.

Background: Researchers worked with the MATES in Construction suicide prevention charity to adapt, implement, and evaluate their blue-collar male-oriented program to the manufacturing sector.

Methods: A two-arm cluster randomised design was used. Ten manufacturing worksites were recruited, with five each randomly assigned to intervention and wait-list control conditions. There was a total of 1245 respondents at baseline (87% response rate) and 648 at final (35% response rate). The Literacy of Suicide Scale (LOSS) score was measured as a process outcome. Help-seeking intentions were the primary outcome, measured using the General Help-Seeking Questionnaire (GHSQ). Secondary outcomes included actual help sought, suicidal thoughts and likelihood of suicide attempt scores, and Kessler-6 distress scale score. Linear mixed models for repeated measures were used in intention-to-treat (ITT) and completers analyses.

Results: None of the five intervention sites fully implemented the intervention as planned. There was a significant improvement in LOSS score within the intervention condition (0.49, 95% CI 0.13 – 0.49), but the mean difference in change between intervention and control included the null (0.34, 95% CI -0.10 – 0.80) despite showing a decent effect size (standardised mean difference [SMD] of 0.50). There was no difference in change between the control and intervention conditions for the GHSQ primary outcome with a mean difference in change of 1.52 (95% CI -0.69 – 3.74; SMD = 0.06). None of the secondary outcomes significantly improved relative to control, in either ITT or completers analyses. An exploratory analysis of disaggregated GHSQ help sources showed significantly greater improvement in mean difference in change for the main MATES message of seeking help from trained MATES Connectors.

Conclusion: The intervention, as implemented, was not effective at achieving the primary or secondary outcomes. Both intervention implementation and evaluation were compromised by COVID-19 restrictions, industrial actions, and other disruptions.

Introduction/Background
Healthcare in NSW is responsible for 6.6% of the state’s carbon emissions and 8% of waste, with 25% coming from NSW hospitals. Currently only 60% of healthcare care is high value care, with 40% considered wasteful, low value or harmful. As each test or procedure produces a carbon footprint, low value care creates an avoidable environmental cost, in addition to patient risks and financial costs. To achieve high value, low carbon healthcare, health systems must decarbonize, and cull low value care.

We describe the first significant, strategic investment by a state health department towards achieving net zero healthcare in a partnership between academic researchers and NSW Health’s Climate Risk and Net Zero Unit.

Method
The Net Zero Partnership supports ten projects across metropolitan and regional NSW public hospitals. Projects are led by clinical champions who act as change agents while also undertaking projects that may be scaled up for potential roll-out across the state. The key principles for this partnership include a multidisciplinary approach, codesigned interventions and evaluations, team-based support, and the development of research capability.

Results/Findings
Examples: 1) The Gloves Off project – a staff educational intervention to reduce unnecessary use of non-sterile gloves (nursing); 2) Optimising Pharmaceutical Waste Management – assessing a ward-based education program on appropriate segregation of pharmaceutical waste (pharmacy); 3) Comparison of single use versus reusable surgical equipment and linen (surgery); 4) Reducing energy consumption by turning off a computed tomography (CT) scanner when not in use (medical imaging); 5) The Rational Investigations Program: reducing unnecessary pathology testing (Emergency Dept) 6) Reducing use of volatile anaesthesia (anaesthesia).

Implications/Key Message
The Net Zero Partnership projects have demonstrated carbon emissions reductions, waste reductions, electricity savings and cost savings whilst maintaining or improving the quality of healthcare. Results will be used for scale-up across hospitals and interstate.

Background
High heat has been proposed to induce adverse pregnancy outcomes in high income countries such as low birth weight, preterm births and stillbirth. The occurrence of these effects in low to middle-income countries, including India is uncertain.
Methods
We examined birth outcomes in the Indian National Family Health Survey of 209,266 participants which provides information on household and individual socio-demographic characteristics of representative households across 640 districts, covering 29 states and six union territories of India. We examined the daily WBGT during the entire pregnancy period. Birth outcomes included z-scores of birthweights, preterm birth and stillbirth. Relationships were adjusted for a range of variables, including birth order and sex of the child, BMI, age, anaemia, smoking status and education of mother, wealth index and cooking fuel of the child/mother household.
Results
Days in the 90th percentile were classified as sweltering. Increased heat saw an increased risk of adverse pregnancy outcomes with adjusted odds ratios (aOR) 1.020 (95% CI: 1.018, 1.022) for low birthweight and 1.038 (95% CI: 1.016, 1.061) for stillbirth. Stratification on SES did not change the relationship between heat and LBW outcome. Stratification by the Koppen climate zone saw the tropical wet climate exhibit the more significant effect (aOR 1.07 (95% CI: 1.06, 1.09). Seven consecutive days in the sweltering range saw the aOR climb to 1.17 (95% CI 1.1, 1.17).
Conclusion
Increased temperatures and continuity of heatwaves may have negative impacts on pregnancy in countries with high summer heat, along with humidity and with few adaption options (e.g. air conditioners). Tailored heat action plans and effect early warning systems may present a pathway to adaption in low-middle income countries.

Background: The Master of Philosophy in Applied Epidemiology (MAE) is the only Australian Field Epidemiology Training Program (FETP) and contributes to the Epidemiology workforce through partnerships with industry. This presentation will outline the structure of the program, its outcomes and how it remains relevant.
Methods: We audited the program administrative documentation and the 2021 and 2022 program outputs to summarise student learning and program stakeholders and processes.
Results: MAE program partners with field placements to provide on-the-job training accompanied by coursework delivered as four intensives across 22 months. Each scholar is supported by academic and field supervisors to undertake projects and is involved in day-to-day applied epidemiology work within their field placement. Field placements have previously included health departments, national and international research centres, and local public health services.
In 2024 the program was internationally reaccredited with distinction and merit by the global FETP accrediting body which highlights its success in training applied epidemiologists and supporting public health decision-making. The program undertook curriculum improvement projects and evolved based on the changing needs of Australia’s public health.
More than 300 applied epidemiologists have graduated from the program since its establishment in 1991. Approximately 15% of graduates identified as First Nations. MAE scholars have been involved in the establishment and evaluation of a variety of public health surveillance systems. Scholars have investigated various public health emergencies including respiratory, foodborne, vectorborne, and vaccine-preventable disease outbreaks. They have also conducted research on infectious diseases, non-communicable diseases, antimicrobial resistance, environmental health and other public health issues.
Conclusion: The MAE program is a significant contributor to the applied epidemiology workforce. The partnership provides opportunities for the training of job-ready applied epidemiologists. It has also created an environment for collaboration and partnership between academia and public health services to address public health needs as they arise.

Background
Hysterectomy is one of the most common gynaecological procedures worldwide, particularly in high-income countries. It may reduce oestrogen production, potentially affecting long-term health outcomes, including bone health. This study examined the association between hysterectomy, with and without oophorectomy, and fracture, osteoporosis risk, and changes in bone mineral density (BMD).

Methods
We conducted a systematic review and meta-analysis, searching PubMed, Embase, Cochrane Central, and CINAHL from inception to November 2024. Studies assessing the relationship between hysterectomy and/or oophorectomy and fracture, osteoporosis, or BMD in women aged ≥18 years were included. Meta-analysis was performed using random-effects models.

Results
Of 15,305 screened articles, 29 met inclusion criteria, with 19 included in meta-analyses. Bilateral oophorectomy (hysterectomy status unspecified) was associated with increased fracture risk (pooled relative risk [RR] 1.17, 95% CI 1.03–1.33; 10 studies) and osteoporosis (RR 1.46, 95% CI 0.94–2.27; 4 studies) compared to no oophorectomy. Hysterectomy with bilateral oophorectomy was not associated with higher fracture risk compared to hysterectomy alone (RR 0.99, 95% CI 0.83–1.18; 4 studies). However, hysterectomy without oophorectomy was linked to an increased risk of osteoporosis compared to no surgery (RR 1.45, 95% CI 1.37–1.53; 2 studies). Heterogeneity ranged from moderate to high, and most studies did not adjust for age at surgery, menopausal hormone therapy, or surgical indications.

Conclusion
Our findings reinforce the negative impact of bilateral oophorectomy on bone health and highlight an increased osteoporosis risk following hysterectomy alone—an issue that has received less attention but may affect a larger population. Further research is needed to refine risk stratification and guide long-term bone health management, particularly through well-designed longitudinal studies with detailed surgical and medication histories.

Background: Gestational diabetes mellitus (GDM) risk varies by area and is potentially associated with social disparity and variation in the built environment. There remains limited evidence examining the association between socio-environmental factors and GDM risk. This study examined the association between socio-environmental factors and GDM risk at the local area level over time in Australia.

Methods: An ecological study was undertaken using multisource national data at Statistical Area Level 2 (SA2: medium-sized geographical unit) in Australia's largest cities from 2016-22. SA2-level GDM cases, births, and socio-environmental data were obtained from national diabetes, birth, and environmental data registries. Spatiotemporal ecological regression was undertaken, and an adjusted risk ratio with 95% credible intervals (CrIs) was estimated. Effect modification analysis by socioeconomic and country of birth was performed.

Results: Across 1,486 SA2s, 207,163 GDM cases occurred from 1,719,396 births. In areas with a high migrant women concentration (location quotient ≥1), GDM was 1.10 (95% CrI; 1.07, 1.13) compared to low concentration. GDM risk was associated with a lower level of socioeconomic position, with least (vs most) advantaged areas having an 18% greater risk of GDM (1.18[95% CrI; 1.10, 1.26]). Residing in the least-moderate walkable areas (second quartile) was associated with higher GDM risk (1.08[95% CrI; 1.02, 1.14]) than most walkable areas. Compared to areas with the least access to fast-food outlets (fourth quartile), other quartiles were associated with increased GDM risk, with the greatest in the highest access (first quartile) (1.12 [95% CrI; 1.04,1.20]). The access to fast food-GDM association was more amplified in the most socioeconomically disadvantaged areas (1.38 [1.17, 1.63]).
Conclusions: GDM risk was associated with socioeconomic disadvantage, high concentrations of at-risk migrant women, least-moderate walkability, and highest fast-food access. Tailored population-level interventions targeting socioeconomic, ethnic, and built environment attributes may improve preventive health efforts

Background: Discordant sibling analyses are often used to assess potential residual confounding in perinatal epidemiological studies. Methodological studies highlight that while the selection process in this design controls for shared unmeasured confounding, it amplifies the effect of non-shared unmeasured confounding. This study explores the strengths and limitations of discordant sibling analyses through a real-life dataset and research question.

Methods: We analysed birth and hospital registry data in Denmark (1997-2019) and Norway (2008-2018). Two perinatal exposures, prenatal antibiotic use and mode of birth, were considered to provide contrast in the anticipated level of non-shared confounding. The outcome was hospitalisation for infections the first 5 years of life. We identified sibling pairs who were the first two singleton births in the dataset for each mother. To explore non-shared confounding, we compared correlations and differences in important perinatal factors (e.g., gestational age, birthweight) according to whether sibling pairs were exposure concordant or discordant. We analysed associations between the two exposures and hospitalisation for infections in (1) the full cohort, (2) sibling pairs, (3) sibling pairs controlling for familial confounding.

Results: Prenatal antibiotic discordant sibling pairs were similar on correlations/differences in perinatal factors to antibiotic concordant pairs. Mode of birth discordant pairs showed larger differences on these factors compared with concordant pairs and sibling pairs overall. Both exposures were associated with increased risk of hospitalised infections in full cohort analyses with similar estimates in sibling pairs. Associations attenuated to the null for prenatal antibiotic exposure and to a lesser extent for mode of birth in the analysis of sibling pairs controlling for familial confounding.

Conclusion: Conflicting findings were observed for the association between prenatal antibiotic exposure and hospitalised infections comparing the full cohort analysis with a discordant sibling analysis, while the association for mode of birth and hospitalised infections remained present in both analyses.

Background: Maternal perinatal depression may increase the risk of neurodevelopmental disorders such as attention deficit hyperactivity disorder (ADHD), in children, either directly or through indirect pathways involving adverse birth outcomes. This study assesses the risk of ADHD in offspring born to mothers with perinatal depressive disorders, examining both the direct and indirect pathways through adverse birth outcomes.
Methods: The study employed a retrospective cohort design, utilising administrative-linked health data from New South Wales. Maternal perinatal depressive disorders and offspring ADHD were identified using the International Classification of Diseases (ICD-10) codes. A generalised linear model with a binomial distribution and a log link function was applied to estimate the direct association. Additionally, a mediation analysis examined the mediational effect of low birth weight, low Apgar scores, and preterm birth on the association between maternal antenatal depressive disorder and ADHD.
Results: After adjusting for potential confounders, offspring of mothers with antenatal, postnatal, and perinatal depressive disorders are 2.10 (RR = 2.10, 95% CI = 1.46–3.03), 1.80 (RR = 1.80, 95% CI = 1.11–2.92), and 2.16 (RR = 2.16, 95% CI = 1.57–2.97) times more likely to have ADHD compared to their counterparts, respectively. The impact of maternal antenatal depressive disorder on offspring ADHD was mediated by preterm birth, but not by low birth weight or low Apgar scores. The proportion of the total effect mediated by preterm birth was only 0.73%, indicating this mediation effect was very minimal, about 45 times smaller than the direct effect.
Conclusion: Our study revealed that maternal perinatal depressive disorders are associated with an increased risk of offspring ADHD, with very minimal or no mediating effects from adverse birth outcomes. Therefore, implementing early intervention strategies aimed at improving maternal mental health is crucial to reducing the risk of ADHD in children.

Introduction
The impact of prenatal pain medicines, including opioids, on long-term child development remains unclear. To address this knowledge gap, we aimed to investigate third-grade academic performance among children with prenatal exposure to opioid analgesics using real-world data.
Methods
We conducted a population-based cohort study by linking childbirth records, medicine dispensing data, and the National Assessment Program – Literacy and Numeracy (NAPLAN) standardised education test scores among all concessional beneficiary pregnancies resulting in a livebirth in NSW, Australia (2003-2011). Opioid exposure was defined as ≥1 dispensing during pregnancy (primary analysis) or ≥2 dispensings (sensitivity analysis). Analyses were stratified by exposure timing during pregnancy (early, late, both periods), dose (low/high oral morphine equivalents [OME]), and monotherapy type (codeine, oxycodone, tramadol). Z-scores for reading and numeracy tests were calculated and compared between exposure groups using linear mixed effects models incorporating propensity-score weights of baseline covariates. Our sensitivity analyses addressed missing data and potential confounding through applying inverse probability of selection weights, discontinuer, and sibling comparisons, respectively.
Results
Among 87,026 eligible children, 75,506 (86.8%) had available test scores, with 7,664 (10.2%) exposed to opioid analgesics prenatally. Compared with unexposed children, reading and numeracy scores were lowest among children with opioid exposure in both early and late pregnancy (reading weighted β [wβ]-0.17, 95% -0.23 to -0.11; numeracy wβ -0.17, 95% -0.23 to -0.11), and tramadol exposure (wβ -0.21, 95% -0.28 to -0.15; wβ -0.23, 95% -0.30 to -0.17). While sensitivity analyses accounting for ≥2 dispensings and missing test data yielded results consistent with our primary analysis, conversely, the discontinuer and sibling analysis suggested no difference in academic performance between exposure groups.
Conclusions
The observed differences in reading and numeracy scores among children with prenatal opioid exposure were small and not clinically meaningful. Our findings may help inform clinical decision-making regarding pain management during pregnancy.

Background: Age at menarche, a critical milestone in female reproductive development, is associated with various cancers and disorders and shows a declining trend globally. This study aimed to investigate the underlying biological mechanisms of menarche related to DNA methylation.
Methods: This study included adult women: 479 from the Australian Mammographic Density Twins and Sisters Study and 2,614 from the Melbourne Collaborative Cohort Study as the discovery sample, and 336 from the European Prospective Investigation into Cancer and Nutrition-Italy for replication. Genome-wide DNA methylation patterns were measured using blood samples (mean age at blood collection 56 years; standard deviation=8 years) and the HumanMethylation450 BeadChip array. We conducted analyses for differentially methylated cytosine-guanine dinucleotide sites across genome and differentially methylated regions (DMRs). Mendelian randomisation analyses were performed to investigate the causal relationships between the identified methylation patterns and age at menarche. To explore the regulatory potential of the identified DMRs, public databases were queried to assess if there was overlap with binding sites for known transcription factors and enhancers.
Results: We identified 63 DMRs in the discovery sample and two of these associations were replicated (P<0.05). The two DMRs were mapped to HOXA5/HOXA6 and TRIM61, which are involved in developmental and cellular regulatory processes. Within the TRIM61 DMR, we found 346 transcription factor binding sites but no enhancers. We found evidence for methylation in the TRIM61 DMR to have a causal effect on age at menarche (P=0.036), and evidence for causal effects of methylation in the TRIM61 DMR on the expression level of TRIM61, RP11-366M4.11, and FAM218A (P<0.001).
Conclusions: Our study identified DMRs, positioned at TRIM61, that potentially influence reproductive development through regulatory mechanisms related to cell proliferation and survival. These results provide new insights into the epigenetic regulation of puberty timing and its broader implications for health and disease.