Population Subgroups Research Articles

Sleep health across the life cycle: an unmet and neglected public health priority

Abstract The presentation will set the stage on the central role of sleep health across the life cycle. Despite the large body of evidence on the importance of poor sleep behaviors as a modifiable risk factor for several adverse health outcomes, sleep health is still an unmet and neglected issue in public health policy, especially in the European context. Sleep problems are widespread and potentially affect everyone, from childhood to adult age. In addition, sleep health disparities represent a major source of health inequities across marginalized and high-risk population subgroups, such us people from low socio-economic status, women, refugees and immigrants, shift workers, people living with physical and mental illness. There is an urgent need to promote evidence-based health promotion and public health policy to enhance sleep health both in the general population and high-risk subgroups.

Malnutrition in infants aged 6-23 months in China's poorest rural counties from 2016 to 2021: cross sectional study.

To assess trends and differences in child malnutrition by population subgroups among infants aged 6-23 months in China's poorest rural counties. Six consecutive cross sectional surveys were conducted annually. The study was conducted in 116 counties in 19 provinces from 2016 to 2021, representing China's 832 poorest counties. A total of 210 088 participants were selected through a multistage cluster sampling procedure; all participants were infants aged 6-23 months. Prevalence of anaemia, stunting, wasting, overweight, and growth status in children (measured by length-for-age and weight-for-length z scores). Four main malnutrition forms were prevalent in 2016: anaemia (prevalence 18.3%), stunting (7.5%), wasting (4.7%), and overweight (3.1%). The prevalence of any two coexisting malnutrition forms was low. All four forms of malnutrition decreased from 2016 to 2021. Anaemia decreased by more than half, with an annual reduction rate of 9.11% (95% confidence interval (CI) 4.83% to 13.20%). Stunting was reduced by over a third, with an annual reduction rate of 10.44% (7.56% to 13.22%), which is faster than the World Health Organization's target of 3.9%. Differences in child growth by county gross domestic product quarters were small and decreased over time, but growth differences related to education persisted. Infants whose mothers completed education up to primary school level had approximately twice the risk of stunting (adjusted rate ratio 2.29 (95% CI 1.87 to 2.81)) and wasting (1.73 (1.40 to 2.13)) compared with children whose mothers had an education level of a college degree or above. Boys had poorer growth and higher anaemia than did girls. For all outcomes, differences related to sex and education were greatest at 6 months of age. Education related inequalities in growth of infants persists, with these differences particularly affecting children whose mothers completed education only up to primary school level.

Health selection in three cohorts of outmigrants, return migrants and non-migrants

Abstract Background The two main hypotheses behind the healthy migrant effect are selection of healthier migrants and the salmon bias. However, these hypotheses have not been thoroughly and simultaneously investigated in distinct cohorts due to data limitations. Methods Linking national and provincial immigration registers with health care utilization datasets at the Manitoba Centre for Health Policy, we assembled a cohort of 816,185 adults who resided in Manitoba, Canada and were followed up for at least one year between 1985 to 2023 to outmigration, return migration or death. We included three subcohorts of international immigrants (16.4%), interprovincial migrants (10.8%) and all other Manitobans (AOM) (72.8%). Within each subcohort, we matched ‘stayers’ who never migrated, outmigrants and returnees on sex, birth year and place of residence and compared their hospitalization rates and Charlson and Elixhauser comorbidity scores 1-year before outmigration and 1-year after return migration. Results Outmigrants had lower hospitalization rates than stayers among AOM [Adjusted Relative Rate (RR): 0.80; 95% confidence interval (CI): 0.78, 0.82] and in the other two subcohorts. Comorbidity scores were also consistently lower among outmigrants compared to stayers in all three subcohorts, even after restricting to hospitalized cases. Outmigrants whose destination was another country were healthier than those who migrated to other provinces. Returnees had better health status than stayers upon return in the AOM subcohort only but lower than those who did not return at the time of outmigration in all three subcohorts. Conclusions Contrary to common belief, health selection is not restricted to international immigrants but also among interprovincial migrants and the local population. Migration dynamics may bias comparisons between population subgroups. Key messages • Migration is not only associated with positive health selection among international immigrants but also among interprovincial migrants and the local population. • Selective migration may represent a source of bias in health-related population-based studies.

Burden of disease due to road traffic noise in Flanders, Belgium

Abstract This study estimates the disease burden of ischemic heart disease (IHD), stroke and diabetes mellitus attributable to road traffic noise in the Flemish Region (Belgium) for the year 2020, expressed in disability-adjusted life years (DALYs). The research proposes an approach to stratify population exposure to noise according to age and sex, which allows to examine internal differences in noise burden related to both varying risk and baseline burden. Stratification of population exposure to noise is achieved by (1) linking modelled noise exposure to age- and sex-stratified population data at small area-level, (2) dividing the prevalence in each exposure class proportionally over the strata, and (3) aggregating the small areas to a Flemish total per stratum. Annual day-evening-night level exposure is translated into the fraction of attributable burden relying on relative risks (RR) from the WHO’s Environmental Noise Guidelines and more recent studies. Baseline burden for the three outcomes is sourced from the Belgian National Burden of Disease Study. The lower and upper bound of the RR’s 95% confidence interval (CI) are used to calculate the uncertainty range on the burden estimates. The total noise DALYs estimated are 1092 (95% CI: 135, 2061) for IHD, 761 (379, 1017) for stroke, and 671 (486, 918) for diabetes. Noise burden rates rise steeply with age for all outcomes, although the wide CIs on the IHD estimates prevent to ascertain a significant trend. The IHD burden is twice as high in men as in women, while both sexes are almost equal for the other outcomes. The variations between age and sex groups are largely explained by differences in baseline burden, as exposure is fairly constant across strata. This study estimates age- and sex-stratified disease burden attributable to noise traffic in Flanders, Belgium. As differential exposure between the population subgroups is limited, any variation in noise DALYs is mainly related to stratum-specific baseline burden. Key messages • Road traffic noise is responsible for hundreds of healthy life years lost in Flanders, Belgium. • The noise burden for stroke and diabetes rises sharply with age, and the burden for ischemic heart disease is twice as high in men as in women.

Past-year mental health issues: discrepant survey & administrative data

Abstract Introduction Accurately measuring mental health disorder prevalence is crucial for public health planning. Administrative and survey data are common methods, but often yield differing results. This study compares case identification of mental health issues across these data sources in Slovenia. Methods We linked the 2019 European Health Interview Survey (EHIS) with three Slovenian health administrative databases: the National Hospital Health Care Statistics Database, Outpatient Prescription Drugs Database, and Absence From Work Database. Case identification utilized self-report of any mental health issue in the past 12 months and healthcare utilization records in 12 months preceding participation in EHIS. Multinomial logistic regression was used to examine the association between socio-demographic factors and case identification across data sources. Results Significant differences were found in 12 month prevalence of any mental health issue estimates between data sources. Only 45.9% of self-reported cases were identified in administrative data, and 36.6% of administrative cases self-reported mental health issues (Kappa = 0.302). Socio-demographics, including age, gender, education, and employment status, were significantly associated with the likelihood of identification in specific data sources. Discussion Our findings underscore the importance of data source and assessment tool selection on mental health prevalence estimates. Variations in identification across population subgroups could reflect differences in healthcare access and self-disclosure biases. Linking data sources and considering their inherent limitations is crucial for accurate burden estimation. Key messages • Data source and assessment methods significantly impact how we identify individuals with mental health issues in the past year. • Combining data sources and understanding their limitations is essential for improving mental health burden estimation.

Patient characteristics and incidence of cardiac amyloidosis: an observational study in the UK clinical setting using electronic health records

Abstract Background Cardiac amyloidosis (CA) is associated with a high mortality rate due to sudden or progressive heart failure. It is often misdiagnosed due to non-specific symptoms, and its true incidence and prevalence in the United Kingdom (UK) remain uncertain. Purpose To describe patient characteristics and incidence of CA in the UK clinical setting using electronic health records. Methods In this retrospective observational study, anonymised, linked primary and secondary care data (Clinical Practice Research Datalink, CPRD and Hospital Episode Statistics, HES) were used to describe individuals diagnosed with CA. The CPRD database contains de-identified patient data sourced from a sample of general practitioner (GP) practices in the UK. Patients aged 18 years and older with a recorded Read or SNOMED-CT diagnosis code for CA or record of treatment with tafamidis (a licensed but not reimbursed treatment for transthyretin amyloid cardiomyopathy, ATTR-CM in the UK) in the period between January 2004 to December 2021 were included. The index date was defined as the earliest recorded date of CA diagnosis or tafamidis treatment. Demographic and clinical data were summarised using descriptive statistics. Incidence rates were reported per 100,000 person-years (PY). Results There were a total of 3,788 patients with CA (mean age: 67 years; 61% males) of which only 277 were coded as having wild-type transthyretin (ATTRwt) CA in mostly older patients (mean age: 79 years). The low patient counts of ATTRwt amyloidosis could be attributed to the lack of specificity in CA types and the low utilisation of ATTR amyloidosis codes. Heart Failure, essential hypertension and dyspnoea were the most common clinical presentations recorded in the 12 months to first diagnosis code or tafamidis prescription for CA (Table). The incidence of CA varied from 0.75 in 2004 to a peak of 1.96 in 2021, per 100,000 PY; a 2.9-fold increase in newly coded diagnoses in CA. There was a slight decline in the rates in 2020 and 2021 (Figure). Incidence rates in males and females increased over the 17-year period, with males consistently having higher rates than females. The incidence rate in females started from a lower base at 0.62/100,000 PY in 2004, increased by a factor of 1.9 to 1.1/100,000 PY, whereas in males it started at 0.88/100,000 PY and rose by a factor of 2.9 to 2.5 /100,000 PY. Incidence rates in subgroup populations have also been analysed. Conclusions In the UK clinical setting, the incidence rate of newly coded CA increased overall between 2004 and 2021, as well as in both male and female subgroups. The data suggest CA is becoming more common, or that awareness has improved, and there is now an urgent need for enhanced methods of detection, characterisation, and the need for thorough data capture to shape future pathway advancements.

STI/HIV care in times of COVID-19 according to Dutch healthcare professionals: A mixed methods study

Abstract Background Surveillance data from the Netherlands show that STI/HIV testing decreased at the start of the COVID-19 pandemic, suggesting barriers in access to STI/HIV care. However, the impact of the pandemic on STI/HIV care may be more complex, and key populations could be differentially affected. The aim of this study was to gain more insight into the impact of COVID-19 on STI/HIV care in the Netherlands from the perspective of STI/HIV care providers. Methods We investigated whether professionals in STI/HIV care experienced changes compared to pre-COVID in access to care, demand and provision of care, shifts online care, and quality of care. An online survey was completed by 192 STI/HIV care professionals. Additionally, semi-structured interviews were held with 23 STI/HIV care professionals. Results According to participants, specific subgroups of priority populations may have had difficulties accessing STI/HIV care during the pandemic especially during lockdowns and at public sexual health services. Professionals were concerned to have lost sight of subpopulations in vulnerable circumstances and that the pandemic may have magnified inequalities in access to STI/HIV care, which may be related to different levels of health and digital literacy required for telephone and video consultations. Furthermore, participants found online care provision was not up to standards and were concerned about follow-up care for patients using private online providers of remote tests. Conclusions Professionals experienced that COVID-19 measures may have compounded existing disparities. To minimize inequalities in access to STI/HIV care during a pandemic, it is recommended to prioritize services for people in vulnerable circumstances such as recent migrants and people with low health or digital literacy. More collaboration between and within sectors may help in better organizing healthcare and prevent overburdening of specific sectors. Key messages • According to STI/HIV care providers, the COVID-19 pandemic differentially impacted different STI/HIV care sectors, but also increased existing inequalities in access to care for vulnerable groups. • Professionals experienced that COVID-19 measures impacted the provision of STI/HIV care, access to care for vulnerable populations, and quality of care.

Association between high versus low cardiovascular risk in individuals with corresponding unexpected absent versus severe subclinical coronary atherosclerosis, a nationwide, multicenter study

Abstract Background It remains unclear as to why some individuals at low risk of cardiovascular disease (CVD) develop severe coronary atherosclerosis, while those at high CVD risk may remain free from coronary atherosclerosis. Purpose Using Coronary Computed Tomography Angiography (CCTA), we aimed to identify factors associated with absence and severe coronary atherosclerosis in subjects from the general population with very high and low-moderate CVD risk, respectively. Methods We assessed cross-sectionally a nationwide, multicenter population-based cohort, consisting of 30,154 individuals randomly recruited from the general population (age range 50-64 years, 51.4% women). Coronary calcified plaque burden was estimated by coronary artery calcium score (CACS) assessed by CCTA. Absent coronary atherosclerosis was defined as CACS 0 and severe coronary atherosclerosis was defined as CACS ≥301. We investigated only individuals with low-moderate (<5%) and very high CVD risk (≥10%) according to SCORE2 were included. Individuals with CACS 0 and prevalent CVD were excluded, as well as those with moderate CACS (1-300) or SCORE2 risk 5-9%. Out of 26 722 subjects with complete data on CACS and SCORE2, 13 075 were included. Logistic regression was performed adjusted for age and sex. Results The overall distribution of CACS in relation to SCORE2 CVD risk is illustrated in the Figure. Study characteristics are displayed in Table 1. Severe-extensive CACS in combination with low-moderate CVD risk was present in 2.3% (n=296). A further 2.3% (n=311) had very high CVD risk but CACS 0. Factors associated with CACS 0 despite a very-high estimated CVD risk included: lower age, being female, being born in Sweden, not having hypertension or hyperlipidemia, fewer pack-years of smoking, and lower waist circumference in men. On the contrary, severe-extensive CACS despite a low-moderate estimated CVD risk was associated with higher age, being male, being an ex-smoker, reporting more pack-years of smoking, having higher HbA1c, higher BMI, higher waist circumference, and prevalent hyperlipidemia and hypertension (Table 2). Conclusions We identified extreme cardiovascular risk phenotypes (subgroups) in the general population that either lacked coronary calcifications despite being at very high estimated CVD risk or had severe subclinical coronary atherosclerosis despite being at estimated low-moderate CVD risk. Further studies are warranted to elucidate putative enhancing or protective mechanisms explaining this unexpected phenomenon.

Gambling Disorder among Croatian University Students

Abstract Background Gambling has become a routine form of entertainment for many young people studying in universities. This study aimed to investigate the prevalence and characteristics of gambling disorder among Croatian university students. Methods A cross-sectional study using an anonymous self-administered questionnaire with sociodemographic and socioeconomic questions along with the Croatian version of the South Oaks Gambling Screen questionnaire (SOGS-CRO) was conducted in a representative sample of Croatian university students from Osijek, Eastern Croatia during January, and February 2020. Results A total of 844 students participated in the study, 45.6% males and 54.4% females, with a median age of 19.00 years (interquartile range 19.00-23.00 years). Gambling activities were performed by 66.2% of the students, most commonly lotteries, sports bets, and bingo. The final scores of the SOGS-CRO suggested that 21.1% of students may have a gambling disorder, with 16.4% being “probable pathological gamblers” and 4.7% being “problem gamblers”. Playing cards for money, animal bets, stocks bets, slot machines bets, pull-tabs, and other gambling activities such as online or videogames or mobile games present a higher risk of gambling disorder (p < 0.001, p < 0.001, p < 0.001, p < 0.001, p < 0.001, p < 0.001, respectively). Being male, repeating the year of study, being a student in interdisciplinary or technical field of study, and having a mother with gambling disorder (p < 0.001, p = 0.029, p < 0.001, p < 0.001, respectively) were the revealed contributing factors to the development of gambling disorder in study participants. Conclusions The study points to the fact that gambling disorder is quite a prevalent public health issue in the Croatian university student population. Specific public health preventive programs need to be designed and implemented in the study population to successfully prevent the development of investigated behavioral addiction. Key messages • Public health interventions directed toward the prevention of gambling must involve younger population subgroups, such as university students. • When designing the preventive programs for the successful prevention of gambling within the university student population selected sociodemographic variables should be considered.

Quantifying the health burden of environmental stressors

Abstract The Burden of Disease (BoD) concept has gained increasing importance for health risk assessments, driven by the World Health Organization (WHO) and the Global Burden of Disease (GBD) studies. BoD utilizes the disability-adjusted life year (DALY) to express the population health impact of a disease, injury or risk factor. This information is essential for priority-setting in health, monitoring and evaluation of preventive interventions. However, calculating DALYs, particularly DALYs attributable to risk factors, requires a range of choices with regards to input data and methodological aspects. In this presentation we will focus on comparative risk assessment (CRA), a frequently used method to calculate the attribution of DALYs to risk factors. CRA consists of determining outcomes that are associated with a certain risk factor and relative risks of the identified outcomes as a function of exposure. This information is usually derived from systematic literature review(s). Next, information on the level of exposure in a population is required, usually by age, sex, location and/or year, and usually using a range of data sources on level of exposure. Additionally, the level of exposure associated with a minimum risk is assessed. Population attributable fractions (PAFs) are then multiplied by the relevant outcome quantity for each population subgroup to calculate the number of DALYs attributable to risk factors. Finally, mediation of risk factors through other risk factors can be taken into account. Knowledge of and uniformity in methods used to perform CRAs will facilitate the coherence and usability of estimates of the health impacts of risk factors and strengthen evidence-based decision-making in health policy. In this presentation, we will introduce the CRA approach and outline a number of innovations that are being driven by the BEST-COST project, including new methods for correlated outdoor exposures and for outdoor exposures penetrating to indoor environments.

The efficacy of vaccination against respiratory syncytial virus across priority population subgroups

Abstract Background Respiratory syncytial virus (RSV) has recently become a vaccine-preventable disease as two vaccines have been approved for use and a third one is likely to become available by the end of the year. National Immunization Technical Advisory Groups as well as scientific societies across the world have issued recommendations on RSV vaccination that differ in respect to the identification of the target group. The latter has been identified in respect to age or underlying medical conditions mostly according to epidemiological data. In order to strengthen the evidence basis for future decisions, we performed a stratified meta-analysis of efficacy data considering all the three RSV vaccines currently or soon available. Methods We searched Randomized Controlled Trials (RCTs) assessing the efficacy of RSV vaccines in terms of reduction of RSV-related Lower Respiratory Tract Infection (LRTI). The research was conducted on PubMed and clinicaltrial.gov. The extraction of relevant data was stratified by pre-defined age groups (60-69, 70-79, 80+) and presence of underlying conditions increasing the risk for RSV. Results Three RCTs, one for each of the three vaccines, were considered for the quantitative analysis. Considering the overall 60+ population, the efficacy of RSV vaccines was 78% (95%CI 76-86%) in preventing LRTI. Considering the different age groups (60-69, 70-79, 80+), efficacy was respectively 72% (95%CI 53-84%), 90% (95%CI 66-97%) and 57% (95%CI 0-93%). The efficacy in healthy people and those with at least one underlying condition was similar, namely 77% (95%CI 60-87%) and 78% (95%CI 50-90%). Conclusions The available evidence shows the efficacy of RSV vaccines in reducing LRTI with no differences across different priority groups. Further studies may contribute to better understanding if the efficacy may change according to individual characteristics, that could be highly relevant in the decision-making process. Key messages • The available evidence from RCTs shows the efficacy of RSV vaccines in reducing LRTI with no differences across different priority groups. • National recommendations on RSV vaccination can be further informed for future studies investigating RSV vaccines efficacy across different subgroups.

Driving policy changes: RISE-Vac lessons learnt and recommendations

Abstract The RISE-Vac project consortium applied evidence-based public health methods to collate, assess, and analyze all the evidence generated through project activities, including literature review and prospective data collection on quantitative and qualitative aspects of vaccine services implementation in prison settings. A multi-sectorial and inter-disciplinary group of experts was set-up including health professionals, statisticians, social scientists and people with lived experience from civil society organisations, health services, prison administrations, academia and international agencies. A set of recommendations, advice, and policy options on how to increase vaccine uptake, targeted to people in prison and using a citizen-centred approach, were developed and tailored to specific needs profiles of different prison population subgroups. Vaccinations to be considered in prison settings were grouped according to target population and/or public health objective into: Seasonal vaccinations and vaccination during outbreak; Childhood life-course vaccinations (booster doses); Vaccinations for cancer prevention; Vaccination for older population groups and people with chronic conditions. In consideration of the diversity of the countries participating in the project, their epidemiological context and available resources and infrastructures, a multi-tiered approach for the development and implementation of models for vaccination services implementation in prison settings was devised, encompassing core vaccinations, such as HBV, Flu, COVID-19 and tetanus, as well as expanded vaccination offer to respond to enhanced health needs and to promote equity of health outcomes, including e.g. HPV vaccination.

Trends in non-daily cigarette smoking in England, 2006–2024

BackgroundCigarette smoking is incredibly harmful, even for people who do not smoke every day. This study aimed to estimate trends in non-daily smoking in England between 2006 and 2024, how these differed across population subgroups, and to explore changes in the profile of non-daily smokers in terms of their sociodemographic and smoking characteristics and vaping and alcohol consumption.MethodsData were collected monthly between November 2006 and April 2024 as part of a nationally representative, repeat cross-sectional survey of adults (≥ 18 years; n = 353,711). We used logistic regression to estimate associations between survey wave and non-daily smoking and used descriptive statistics to characterise the profile of non-daily smokers across 3-year periods.ResultsThe proportion who smoked non-daily was relatively stable between November 2006 and November 2013, at an average of 10.5% [10.1–10.9%] of cigarette smokers, then increased to 27.2% [26.0–28.4%] of cigarette smokers (4.0% [3.7–4.2%] of adults) by April 2024. This increase was particularly pronounced among younger adults (e.g. reaching 52.8%, 20.4%, and 14.4% of 18-, 45-, and 65-year-old cigarette smokers by April 2024) and those who vape (reaching 34.2% among vapers vs. 23.1% among non-vapers). Over time, there were reductions in non-daily smokers’ mean weekly cigarette consumption (from 34.3 in 2006–2009 to 21.1 in 2021–2024), urges to smoke (e.g. the proportion reporting no urges increased from 29.2 to 38.0%), and motivation to stop smoking (e.g. the proportion highly motivated to quit within the next 3 months decreased from 30.8 to 21.0%).ConclusionsAn increasing proportion of adults in England who smoke cigarettes do not smoke every day, particularly younger adults. Although non-daily smokers report smoking fewer cigarettes and weaker urges to smoke than they used to, which may make it easier for them to stop smoking, they appear to be decreasingly motivated to quit.

Equity-centered adaptive sampling in sub-sewershed wastewater surveillance using census data.

Sub-city, or sub-sewershed, wastewater monitoring for infectious diseases offers a data-driven strategy to inform local public health response and complements city-wide data from centralized wastewater treatment plants. Developing strategies for equitable representation of diverse populations in sub-city wastewater sampling frameworks is complicated by misalignment between demographic data and sampling zones. We address this challenge by: (1) developing a geospatial analysis tool that probabilistically assigns demographic data for subgroups aggregated by race and age from census blocks to sub-city sampling zones; (2) evaluating representativeness of subgroup populations for COVID-19 wastewater-based disease surveillance in Davis, California; and (3) demonstrating scenario planning that prioritizes vulnerable populations. We monitored SARS-CoV-2 in wastewater as a proxy for COVID-19 incidence in Davis (November 2021-September 2022). Daily city-wide sampling and thrice-weekly sub-city sampling from 16 maintenance holes covered nearly the entire city population. Sub-city wastewater data, aggregated as a population-weighted mean, correlated strongly with centralized treatment plant data (Spearman's correlation 0.909). Probabilistic assignment of demographic data can inform decisions when adapting sampling locations to prioritize vulnerable groups. We considered four scenarios that reduced the number of sampling zones from baseline by 25% and 50%, chosen randomly or to prioritize coverage of >65-year-old populations. Prioritizing representation increased coverage of >65-year-olds from 51.1% to 67.2% when removing half the zones, while increasing coverage of Black or African American populations from 67.5% to 76.7%. Downscaling had little effect on correlations between sub-city and centralized data (Spearman's correlations ranged from 0.875 to 0.917), with strongest correlations observed when prioritizing coverage of >65-year-old populations.

Conceptualizing bias in EHR data: A case study in performance disparities by demographic subgroups for a pediatric obesity incidence classifier.

Electronic Health Records (EHRs) are increasingly used to develop machine learning models in predictive medicine. There has been limited research on utilizing machine learning methods to predict childhood obesity and related disparities in classifier performance among vulnerable patient subpopulations. In this work, classification models are developed to recognize pediatric obesity using temporal condition patterns obtained from patient EHR data in a U.S. study population. We trained four machine learning algorithms (Logistic Regression, Random Forest, Gradient Boosted Trees, and Neural Networks) to classify cases and controls as obesity positive or negative, and optimized hyperparameter settings through a bootstrapping methodology. To assess the classifiers for bias, we studied model performance by population subgroups then used permutation analysis to identify the most predictive features for each model and the demographic characteristics of patients with these features. Mean AUC-ROC values were consistent across classifiers, ranging from 0.72-0.80. Some evidence of bias was identified, although this was through the models performing better for minority subgroups (African Americans and patients enrolled in Medicaid). Permutation analysis revealed that patients from vulnerable population subgroups were over-represented among patients with the most predictive diagnostic patterns. We hypothesize that our models performed better on under-represented groups because the features more strongly associated with obesity were more commonly observed among minority patients. These findings highlight the complex ways that bias may arise in machine learning models and can be incorporated into future research to develop a thorough analytical approach to identify and mitigate bias that may arise from features and within EHR datasets when developing more equitable models.

Development and assessment of a machine learning tool for predicting emergency admission in Scotland

Emergency admissions (EA), where a patient requires urgent in-hospital care, are a major challenge for healthcare systems. The development of risk prediction models can partly alleviate this problem by supporting primary care interventions and public health planning. Here, we introduce SPARRAv4, a predictive score for EA risk that will be deployed nationwide in Scotland. SPARRAv4 was derived using supervised and unsupervised machine-learning methods applied to routinely collected electronic health records from approximately 4.8M Scottish residents (2013-18). We demonstrate improvements in discrimination and calibration with respect to previous scores deployed in Scotland, as well as stability over a 3-year timeframe. Our analysis also provides insights about the epidemiology of EA risk in Scotland, by studying predictive performance across different population sub-groups and reasons for admission, as well as by quantifying the effect of individual input features. Finally, we discuss broader challenges including reproducibility and how to safely update risk prediction models that are already deployed at population level.

Worldwide prevalence of chagas cardiomyopathy-an analysis from the global burden of disease dataset.

The worldwide prevalence of Chagas Cardiomyopathy (CCM) as well as the trends in the prevalence of CCM over time have not been well characterized. An analysis of the Global Burden of Disease (GBD) data from 1990 to 2019 was conducted to assess the burden of CCM. This study focused on determining the prevalence of CCM, along with its age-standardized prevalence rate (ASR) per 1,00,000 people, considering various patient demographics and geographical regions as defined in the GBD. Additionally, the study examined the temporal trends over this 30-year period by calculating the estimated annual percentage change (EAPC) in CCM prevalence for the global population and specific subgroups. Worldwide, the GBD reported 220,166 individuals with CCM in 1990 and 2,83,236 individuals in 2019, with a decline in the ASR from 5.23 (3.34-7.47) to 3.42 (2.2-4.91) per 1,00,000 individuals during that period. In 2019, the prevalence was highest in individuals over age 70 and in males compared to females. Among available geographic classifications in 2019, Latin American regions had the highest rates (ASR of 39.49-61.15/1,00,000), while high income North American and Western European regions had the lowest rates (ASRs of 0.67 and 0.34/1,00,000, respectively). Between 1990 and 2019, the worldwide prevalence of CCM per 1,00,000 decreased (EAPC of -0.35, -0.37 to -0.32), with similar trends among most regions and subgroups. This analysis of the GBD data reveals both global and country-specific patterns in the prevalence and trends of CCM. Notably, CCM shows the highest prevalence in Latin American countries, although it's also significantly present in regions beyond Latin America. Notably, the global age-standardized rate of CCM is on the decline, suggesting improvements in healthcare strategies or lifestyle changes across the world.

Sex-specific analysis of acute alcohol use in suicides and reporting of alcohol as a contributor to suicide deaths in New Zealand 2007-2020: a cross-sectional study of coronial data.

Acute alcohol use (AAU) can increase suicide risk. It is unknown if this effect differs by population sub-group in New Zealand, and what characteristics are associated with alcohol being coded as contributory to death, when AAU is identified. This study aimed to answer: 1) are the characteristics associated with suicide involving AAU different between females and males, and 2) among suicides that involved AAU, what factors are associated with alcohol being coded as a contributory factor? Secondary analysis was conducted of suicide data from 2007-2020, from the National Coronial Information System. Binomial regression models for females and males were used to estimate sex-specific differences in risk of suicide involving AAU. Poisson regression modelling was used to estimate the relative risk of alcohol being coded as contributory where AAU was identified. Suicide was more likely to involve AAU among Māori females (adjusted risk ratio [ARR] 1.35, 95% confidence interval [CI] 1.08-1.68) and Pacific females (ARR 1.75, 95% CI 1.22-2.51), compared to European females. Compared to males who were employed, all other employment statuses had significantly lower risk of suicide that involved AAU. Those who died by hanging (ARR 0.75, 95% CI 0.62-0.92) or firearms (ARR 0.55, 95% CI 0.38-0.90) were less likely to have alcohol coded as contributory, compared to those who died by poisoning. Targeted public health interventions designed by and for specific demographic groups (particularly Māori and Pacific females) are needed, alongside universal interventions that address social and structural determinants. Data systems and coding must accurately reflect the association between AAU and suicide in New Zealand.

COVID-19 Vaccine Preferences in General Populations in Canada, Germany, the United Kingdom, and the United States: Discrete Choice Experiment.

Despite strong evidence supporting COVID-19 vaccine efficacy and safety, a proportion of the population remains hesitant to receive immunization. Discrete choice experiments (DCEs) can help assess preferences and decision-making drivers. We aim to (1) elicit preferences for COVID-19 vaccines in Canada, Germany, the United Kingdom, and the United States; (2) understand which vaccine attributes people there value; and (3) gain insight into the choices that different population subgroups make regarding COVID-19 vaccines. Participants in the 2019nCoV-408 study were aged ≥18 years; self-reported antivaccinationists were excluded. A DCE with a series of 2 hypothetical vaccine options was embedded into a survey to determine participant treatment preferences (primary objective). Survey questions covered vaccine preference, previous COVID-19 experiences, and demographics, which were summarized using descriptive statistics to understand the study participants' backgrounds. In the DCE, participants were provided choice pairs: 1 set with and 1 without an "opt-out" option. Each participant viewed 11 unique vaccine profiles. Vaccine attributes consisted of type (messenger RNA or protein), level of protection against any or severe COVID-19, risk of side effects (common and serious), and potential coadministration of COVID-19 and influenza vaccines. Attribute level selections were included for protection and safety (degree of effectiveness and side effect risk, respectively). Participants were stratified by vaccination status (unvaccinated, or partially or fully vaccinated) and disease risk group (high-risk or non-high-risk). A conditional logit model was used to analyze DCE data to estimate preferences of vaccine attributes, with the percentage relative importance calculated to allow for its ranking. Each model was run twice to account for sets with and without the opt-out options. The mean age of participants (N=2000) was 48 (SD 18.8) years, and 51.25% (1025/2000) were male. The DCE revealed that the most important COVID-19 vaccine attributes were protection against severe COVID-19 or any severity of COVID-19 and common side effects. Protection against severe COVID-19 was the most important attribute for fully vaccinated participants, which significantly differed from the unvaccinated or partially vaccinated subgroup (relative importance 34.8% vs 30.6%; P=.049). Avoiding serious vaccine side effects was a significantly higher priority for the unvaccinated or partially versus fully vaccinated subgroup (relative importance 10.7% vs 8.2%; P=.044). Attributes with significant differences in the relative importance between the high-risk versus non-high-risk subgroups were protection against severe COVID-19 (38.2% vs 31.5%; P<.000), avoiding common vaccine side effects (12% vs 20.5%; P<.000), and avoiding serious vaccine side effects (9.7% vs 7.5%; P=.002). This DCE identified COVID-19 vaccine attributes, such as protection against severe COVID-19, that may influence preference and drive choice and can inform vaccine strategies. The high ranking of common and serious vaccine side effects suggests that, when the efficacy of 2 vaccines is comparable, safety is a key decision-making factor.

Relationship between metabolically healthy overweight/obesity and risk of type 2 diabetes in different ethnicity: a prospective cohort study in southwest China

BackgroundTo estimate and compare the association between metabolically healthy overweight/obesity (MHO) and the risk of type 2 diabetes (T2DM) among different ethnic groups in southwest China, while also exploring possible ethnic differences.MethodsThis is a prospective cohort study of 6,820 participants in Southwest China. MHO was defined as body mass index (BMI) ≥ 24 kg/m2 and the presence of ≤ 1 component of metabolic syndrome. Cox proportional risk models were utilized to analyze the association between MHO and the risk of T2DM.ResultsThe median follow-up time was 6.58 years, during which 708 new cases of T2DM were diagnosed. In the total population, after adjusting for confounding factors, MHO was found to increase the risk of T2DM compared to metabolically healthy normal weight (MHNW) individuals (HR = 1.49, 95% CI: 1.15–1.93). Subgroup analysis by ethnicity revealed that, MHO increased the risk of T2DM in the Han population (HR = 1.64, 95% CI: 1.21–2.23), however, the difference was not statistically significant in the ethnic minority population.The results of sensitivity analysis further supported the robustness of these findings. Meanwhile, stratified by sex, age, and urban/rural, it was found that ethnic differences in the association between MHO and T2DM still existed, however, it is important to note that the association between MHO and T2DM was not statistically significant in the Han population subgroup aged ≥ 45 years (p > 0.05).ConclusionMHO was associated with an increased risk of T2DM compared to MHNW, and there are ethnic differences. Future interventions need to be strengthened for Han Chinese key populations to reduce the risk of T2DM.