Population Datasets Research Articles

Abstract 4129397: Cardiovascular Risk Models Using Large-Scale Physical Examination Data

Background: Physical examination (PE) data offer more potential key risk factors that can enhance the development of risk models. Additionally, the latest CVD risk model, PREVENT, developed from an American population, necessitates validation within China. Objectives: This study aimed to develop CVD risk models based on large-scale PE records and compare them with PREVENT model in Chinese adults. Methods: Data were collected from a China national database (2016-2024), including 929682 participants aged 30-79 years without previous CVD (Table I). The variables covered demographics, comorbidities, and PE items, resulting in 59 features as input. The endpoint events were defined as total CVD and its subtypes, ASCVD and heart failure (HF). Sex-specific survival analysis was developed based on age-scale, utilizing Cox Proportional Hazards and Random Survival Forests (RSF). Feature importance was assessed using the average minimal depth in RSF. The performance was evaluated by the Harrell C-statistic in the validation set. Results: The results showed that our RSF model achieved the best performance in total CVD, with the highest C-index of 0.82 (Table II). The top 15 predictive risk factors for each gender are presented in Figure 1. Additionally, validation of the PREVENT model in our Chinese population dataset showed its great generalizability, particularly for HF events. Conclusions: Overall, this research underscores the importance of leveraging large-scale PE data and advanced modeling techniques to improve CVD risk prediction. Beyond traditional risk factors, greater emphasis should be placed on more demographic characteristics, comorbidities, urinalysis, and ECG features.

Space-for-time substitutions exaggerate urban bird-habitat ecological relationships.

North American bird abundance has declined by 29% over the last 50 years. These continental population dynamics interact with local landscape changes to affect local bird diversity. Mitigating local declines in cities is particularly significant because (a) such declines greatly impact human-bird relationships since most people live in cities and (b) cities provide levers to create bird-friendly habitat, such as managing yards and gardens, street trees, and urban parks. Yet, the potential for cities to modify habitats to mitigate broader bird declines remains unclear. Studies have been stymied by the difficulty of assembling mutidecadal habitat-bird population datasets. Instead, studies have substituted space for time (e.g. used habitat associations across space at one time point to project future species abundance due to changing land use), but this method may fail amidst nonstationary environments of the Anthropocene. Here, we test the validity of space-for-time substitutions for explaining changes in bird abundance in a North American city over the past two decades by examining the degree to which these changes are explainable by changes in local landcover at multiple spatial scales. Specifically, we use longitudinal urban bird surveys of Metro Vancouver, BC, Canada from 1997 and 2020; deep learning models of remote sensing data to classify contemporaneous landcover; out-of-sample prediction and boosted regression trees to identify multiple spatial scales of landcover that best explained bird abundance (i.e. optimal scale of effect for each species by each habitat); and Bayesian multispecies abundance models in Stan to determine relationships between changes in landcover and bird abundance. We found that total bird abundance declined by 26% over the last two decades. Landcover measured at both 50 m and optimal scales explained spatial variation in bird abundance, but only landcover at the optimal scale explained temporal changes, and only partially. These results suggest that space-for-time substitutions overemphasize habitat-bird ecological relationships, urban habitats only partially determine bird abundance, and measuring habitat at the appropriate scale is important for capturing the most relevant changes in landscapes.

Grading Habitats for Ticks by Mapping a Suitability Index based on Remotely Sensed Data and Meta® population dataset in Aosta Valley, NW Italy.

Ticks represent a reservoir of zoonotic pathogens, and their numbers are increasing largely in wildlife. This work is aimed at producing maps of suitable habitats for ticks in Aosta Valley, Italy based on multitemporal EO data and veterinary datasets (tick species and distribution in wild hosts). EO data were processed in Google Earth Engine considering the following inputs: A) Growing Degree Ticks (GDT), B) NDVI from MOD09GA, C) NDVI entropy, D) distance from water bodies, E) topography, F) rainfalls from CHIRPS as monthly composites along the 2020, 2021 and 2022 years. Ticks were collected from hunted, injured, and found-dead wild animals ( Sus scrofa, Capreolus capreolus, Rupicapra rupicapra, Cervus elaphus); they were labeled at species level using taxonomic keys. Between September 2020 and December 2022, a total of 90 ticks were collected from 89 wild animals. Ixodes ricinus was the most prevalent tick species, followed by Dermacentor marginatusand Dermacentor spp. Molecular analyses demonstrated the presence of Anaplasma spp., B. burgdorferi sensu lato and Rickettsia spp. pathogens in infected ticks. To assess human population potential exposure to tick Meta® population dataset was used. In conclusion this study shows the potentialities of Remote sensing improving the technological transfer to the veterinarian sector.

Quantifying residents' exposure to agricultural pesticides using new geospatial approaches

A better understanding of environmental exposure to agricultural pesticides is crucial for public health, regulatory and management purposes. Residents in close vicinity to agricultural fields are likely to be more exposed to pesticides. In that context, an innovative geospatial approach for mapping estimates of agricultural pesticide exposure was developed in this study.Data on pesticide application rates, high-resolution annual datasets of the geographic distribution of crops, and the 100x100m grid population dataset were utilized to complete this analysis in Wallonia (Belgium) over the period 2015-2019. Pesticide exposure metrics were estimated using a buffer-based exposure model by conducting neighborhood analysis within a 1000m buffer radius. Subsequently, a population weighted method was used to ‘up-scale’ the exposure data to administrative levels. In a limited validation effort, model estimates were compared with pesticide measurements in air collected at 12 stations during the period 2015-2016.The results present the first modeling map of environmental exposure to agricultural pesticides for the whole of the Walloon region. The northern part of the Sambre-Meuse axis demonstrates more intensive agriculture and the highest pesticide weighted exposure indices. The majority of the population resides near agricultural areas, with only 4% living in regions that lack crops within a 1000m radius, primarily in the central areas of large cities. A positive trend association between pesticide measures in air and the index was observed at the different stations, nevertheless further validation efforts are needed to accurately compare the same active ingredients.This work gives a valuable basis for research and environmental health actions in Belgium. Maps highlight areas where human biomonitoring and epidemiological studies should be implemented to investigate the impact of potential environmental exposure to pesticides. This information helps policy-makers to take measures to control and reduce the load of agricultural pesticides in the environment.

Assessment of sleep patterns in dementia and general population cohorts using passive in-home monitoring technologies

BackgroundNocturnal disturbances are a common symptom experienced by People Living with Dementia (PLWD), and these often present prior to diagnosis. Whilst sleep anomalies have been frequently reported, most studies have been conducted in lab environments, which are expensive, invasive and not natural sleeping environments. In this study, we investigate the use of in-home nocturnal monitoring technologies, which enable passive data collection, at low cost, in real-world environments, and without requiring a change in routine.MethodsClustering analysis of passively collected sleep data in the natural sleep environment can help identify distinct sub-groups based on sleep patterns. The analysis uses sleep activity data from; (1) the Minder study, collecting in-home data from PLWD and (2) a general population dataset (combined n = 100, >9500 person-nights).ResultsUnsupervised clustering and profiling analysis identifies three distinct clusters. One cluster is predominantly PLWD relative to the two other groups (72% ± 3.22, p = 6.4 × 10−7, p = 1.2 × 10−2) and has the highest mean age (77.96 ± 0.93, p = 6.8 × 10−4 and p = 6.4 × 10−7). This cluster is defined by increases in light and wake after sleep onset (p = 1.5 × 10−22, p = 1.4 × 10−7 and p = 1.7 × 10−22, p = 1.4 × 10−23) and decreases in rapid eye movement (p = 5.5 × 10−12, p = 5.9 × 10−7) and non-rapid eye movement sleep duration (p = 1.7 × 10−4, p = 3.8 × 10−11), in comparison to the general population.ConclusionsIn line with current clinical knowledge, these results suggest detectable dementia sleep phenotypes, highlighting the potential for using passive digital technologies in PLWD, and for detecting architectural sleep changes more generally. This study indicates the feasibility of leveraging passive in-home technologies for disease monitoring.

Nationwide trends in the suicide mortality among the disabled in Korea from 2006 to 2017

Abstract Aims The disabled are one of the priority populations at risk of suicide, but there has been no study of the factors influencing suicide mortality among the disabled in South Korea. Methods To analyze age-standardized suicide mortality (ICD-10 Codes, X60-84) for each year according to the presence, severity and type of disability, we linked three population datasets (National Health Insurance Claims Database, Disability Registry, and Deaths by Cause). Our dataset covers the entire population of South Korea, extended over a 12-year period from 2006 to 2017. The analyses were stratified by sex, and hazard ratios for suicide were determined by multivariate logistic regression after adjustment for socio-demographic and clinical variables. Results The suicide rate among the disabled was 2.24 times higher than among the non-disabled in 2017. The highest suicide rates were observed among intellectual or mental disabled, and severe disabled. In the disabled group, suicide rates were higher among those with medical benefits, older age (over 70 years), and depressive disorders in both male and female. Conclusions Although the disabled people’s suicide is often seen as an individual problem, it actually affects their families, communities and society. Therefore, government must strive to emphasize the broad view of the disabled population - not just the health of individuals, and track suicide trends and identifying risk and protective factors for suicidal behavior. From this information, suicide prevention strategies for the disabled people should be evaluated to identify the most effective interventions. Financial support: This research was supported by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education (No. 2022R1I1A3070074, and 2022R1I1A1A01068449), and a grant of Patient-Centered Clinical Research Coordinating Center funded by the Ministry of Health & Welfare, Republic of Korea (No. HC23C016500). Key messages • The suicide rate among the disabled was 2.24 times higher than among the non-disabled in 2017. • In the disabled group, suicide rates were higher among those with medical benefits, older age (over 70 years), and depressive disorders in both male and female.

Fine Tuning ECG Interpretation for Young Athletes: ECG Screening Using Z-score-based Analysis

BackgroundElectrocardiograms (ECGs) in athletes commonly reveal findings related to physiologic adaptations to exercise, that may be difficult to discern from true underlying cardiovascular abnormalities. North American and European societies have published consensus statements for normal, borderline, and abnormal ECG findings for athletes, but these criteria are not based on established correlation with disease states. Additionally, data comparing ECG findings in athletes to non-athlete control subjects are lacking. Our objective was to compare the ECGs of collegiate athletes and non-athlete controls using Z-scores for digital ECG variables to better identify significant differences between the groups and to evaluate the ECG variables in athletes falling outside the normal range.MethodsValues for 102 digital ECG variables on 7206 subjects aged 17–22 years, including 672 athletes, from Hawaii Pacific Health, University of Hawaii, and Rady Children’s Hospital San Diego were obtained through retrospective review. Age and sex-specific Z-scores for ECG variables were derived from normal subjects and used to assess the range of values for specific ECG variables in young athletes. Athletes with abnormal ECGs were referred to cardiology consultation and/or echocardiogram.ResultsAthletes had slower heart rate, longer PR interval, more rightward QRS axis, longer QRS duration but shorter QTc duration, larger amplitude and area of T waves, prevalent R’ waves in V1, and higher values of variables traditionally associated with left ventricular hypertrophy (LVH): amplitudes of S waves (leads V1-V2), Q waves (V6, III) and R waves (II, V5, V6). Z-scores of these ECG variables in 558 (83%) of the athletes fell within − 2.5 and 2.5 range derived from the normal population dataset, and 60 (8.9%) athletes had a Z-score outside the − 3 to 3 range. While 191 (28.4%) athletes met traditional voltage criteria for diagnosis of LVH on ECG, only 53 athletes (7.9%) had Z-scores outside the range of -2.5 to 2.5 for both S amplitude in leads V1-V2 and R amplitude in leads V5-6. Only one athlete was diagnosed with hypertrophic cardiomyopathy with a Z-score of R wave in V6 of 2.34 and T wave in V6 of -5.94.ConclusionThe use of Z-scores derived from a normal population may provide more precise screening to define cardiac abnormalities in young athletes and reduce unnecessary secondary testing, restrictions and concern.

High-resolution population mapping based on SDGSAT-1 glimmer imagery and deep learning: a case study of the Guangdong-Hong Kong-Marco Greater Bay Area

ABSTRACT Accurate population mapping is crucial for disaster management, urban planning, etc. However, current methods using nighttime light (NTL) and gridded population datasets are limited by low spatial resolution and insufficient training data for complex models such as deep learning. These models do not adequately utilize spatial information in population mapping. To address these limitations, this study proposes a high-resolution population mapping method using the Sustainable Development Goals Science Satellite 1 (SDGSAT-1) glimmer imager data and deep learning. The method includes a sample generation strategy with multiple regression and multilevel screening to provide sufficient, high-quality samples for deep learning. A Fine Population mapping network (FinePop-net) is also developed to train regression models using image samples, capturing multi-scale features for model training. When applied to the Guangdong-Hong Kong-Macao Greater Bay Area with 40-meter resolution SDGSAT 1 glimmer imagery, the method significantly reduced the average absolute error and root-mean-square error by 9.35% and 11.44%, respectively, compared with those of the pixel-level learning methods. It also outperformed other population spatialization datasets and NTL data by over 30% and 10%, respectively, in terms of error reduction. The results highlight the method’s effectiveness and the value of SDGSAT-1 glimmer imagery for fine population spatialization.

Quantifying constraint in the human mitochondrial genome.

Mitochondrial DNA (mtDNA) has an important yet often overlooked role in health and disease. Constraint models quantify the removal of deleterious variation from the population by selection and represent powerful tools for identifying genetic variation that underlies human phenotypes1-4. However, nuclear constraint models are not applicable to mtDNA, owing to its distinct features. Here we describe the development of a mitochondrial genome constraint model and its application to the Genome Aggregation Database (gnomAD), a large-scale population dataset that reports mtDNA variation across 56,434 human participants5. Specifically, we analyse constraint by comparing the observed variation in gnomAD to that expected under neutrality, which was calculated using a mtDNA mutational model and observed maximum heteroplasmy-level data. Our results highlight strong depletion of expected variation, which suggests that many deleterious mtDNA variants remain undetected. To aid their discovery, we compute constraint metrics for every mitochondrial protein, tRNA and rRNA gene, which revealed a range of intolerance to variation. We further characterize the most constrained regions within genes through regional constraint and identify the most constrained sites within the entire mitochondrial genome through local constraint, which showed enrichment of pathogenic variation. Constraint also clustered in three-dimensional structures, which provided insight into functionally important domains and their disease relevance. Notably, we identify constraint at often overlooked sites, including in rRNA and noncoding regions. Last, we demonstrate that these metrics can improve the discovery of deleterious variation that underlies rare and common phenotypes.

Insights from a population grid of South Africa: An applied spatial satellite data analysis

The present study explores the reliability and accuracy of various spatial mapping methodologies in estimating and presenting the spatial characteristic and dynamics (location, distribution, density, and size) of the population in South Africa. As a basic underlying concept, the study first explores spatial heterogeneity, that is, that every location is related to every other location, and those nearby are related stronger. This study, therefore, illustrates the spatial relationships between locations and the spatial pattern of the population in South Africa. Analyzing the spatial images determines the extent of such influence and the nature of the spatial patterns. To this end, a granular gridded population dataset was derived using satellite image data, and the NASA’s Socioeconomic Data and Applications Center gridded population of the world version 4 population images and datasets were used. Several spatial data models and geostatistical applications were applied to study the spatial characteristics and dynamics of the population of South Africa from 2000 to 2020. Spatial analysis was performed using R-Studio, QGIS, and GeoDa. Among others, the results point to the fact that the South African population is very densely located that population density decreases marginally outward and suggests that the underlying process for the population distribution is stationary. This study proposes that it is indeed possible to reliably and accurately estimate and present gridded population images and datasets using spatial and geostatistical methodologies.

Influence of future climate scenarios using CMIP 5 data on malaria transmission in India

BackgroundVector-borne diseases, such as malaria, pose a significant global threat, and climatological factors greatly influence their intensity. Tropical countries, like India, are particularly vulnerable to such diseases, making accurate estimation of malaria risk crucial.MethodsThis study utilized the well-known Vector-borne Disease Community Model, VECTRI, developed by the International Centre for Theoretical Physics in Trieste. The model was implemented to estimate malaria’s Entomological Inoculation Rate (EIR). Future climatic prediction datasets, including CMIP 5 and population data sets, were used as inputs for the analysis. Three RCP scenarios are considered (Representative Concentration Pathways are climate change scenarios that project radiative forcing to 2100 due to future greenhouse gas concentrations). The projections covered the period from 1 Jan, 2020, to 31 Dec, 2029.ResultsThe estimated mean EIR for the years 2020–2029 ranged, and a significant decline in malaria risk was observed with all RCP 2.6, 4.5, and 8.5 scenarios. Each year 0.3 to 2.6 [min–max] EIR/person/day decline is observed with a strong decline in man rainfall ranging from 5 to 17 [min–max] mm/year and associated high temperatures ranging from 0.03 to 0.06 [min–max] °C/year. During the post-monsoon period, August to November were identified as highly prone to malaria transmission. Spatial analysis revealed that the east coast of India faced a higher vulnerability to malaria risk, which kept increasing through RCP scenarios. Thus, it is essential to exercise caution, especially in areas with heavy rainfall.ConclusionThis research provides valuable insights for policy-makers, highlighting the need to implement future strategies to mitigate malaria risk effectively. By utilizing these findings, appropriate measures can be taken to combat the threat posed by malaria and protect public health.

A global dataset of 7 billion individuals with socio-economic characteristics

In global impact modeling, there is a need to address the heterogeneous characteristics of households and individuals that drive different behavioral responses to, for example, environmental risk, socio-economic policy changes and spread of diseases. In this research, we present GLOPOP-S, the first global synthetic population dataset with 1,999,227,130 households and 7,335,881,094 individuals for the year 2015, consistent with population statistics at an administrative unit 1 level. GLOPOS-S contains the following attributes: age, education, gender, income/wealth, settlement type (urban/rural), household size, household type, and for selected countries in the Global South, ownership of agricultural land and dwelling characteristics. To generate GLOPOP-S, we use microdata from the Luxembourg Income Study (LIS) and Demographic and Health Surveys (DHS) and apply synthetic reconstruction techniques to fit national survey data to regional statistics, thereby accounting for spatial differences within and across countries. Additionally, we have developed methods to generate data for countries without available microdata. The dataset can be downloaded per region or country. GLOPOP-S is open source and can be extended with other attributes.

A Study on the Job Characteristics of Police Officers and the Status of Musculoskeletal Disorders.

Musculoskeletal disorders occur frequently in industrial settings, and police officers in particular are at high risk of developing musculoskeletal disorders due to the nature of their work, but research on this is lacking. Therefore, the purpose of this study was to examine the prevalence and risk factors of musculoskeletal disorders according to the job functions of police officers. A survey was conducted, targeting active-duty police officers, and data were collected from 4268 respondents who answered all items on the questionnaire. As a control group, we used the general population dataset from the 2020 National Health and Nutrition Examination Survey conducted in Korea. The survey included general information such as gender, age, working hours, and type of work. It also investigated medical utilization, including the frequency of medical visits and hospitalizations. The assessment of musculoskeletal disorders was analyzed using pain NRS, VAS, SPADI, and ODI. The working characteristics of police officers were found to be more irregular in terms of working hours and involved a higher intensity of work activities compared with the general population. However, there was a lack of precise diagnoses and continuous treatment for these disorders. These results imply that police officers' occupational characteristics are related to musculoskeletal disorders. Considering the direct relationship between police officers' health and public safety, systematic evaluation and management of their musculoskeletal disorders are needed.

A sarcoptic mange outbreak drives long-term patterns of a fitness trait in an exotic ungulate

Species phenotypes are driven by persistent selective pressures that condition their adaptation to the environment. However, stochastic critical events, such as disease outbreaks, may also act as selective pressures because of drastic population decline. External structures like horns of ungulate species may reflect the consequences of these selective pressures on animal fitness. Using a generalized linear model approach, we analysed how a massive mortality event caused by a sarcoptic mange outbreak in 1992 affected the horn size of an Iberian aoudad population (dataset from 1980 to 2014) by comparing horn length before and after the outbreak. We found that this episodic, albeit massive, outbreak changed long-term trends in age-size horn relations. Both females and males presented shorter horns after the sarcoptic mange outbreak, although more marked in males. Even when animals reached mature ages, horn length did not increase with age to the same degree as before the outbreak. We consider that two processes may have acted on the population: a bottleneck favoured “shorter horns” phenotype and a post-mange enzootic period limited horn growth, as infected animals may invest more resources in the immune system. Overall, our results show a top-down regulation of a fitness trait for an ungulate population mediated by a parasite. This disease outbreak showed long-lasting effects on the ungulate population, with ecological and management implications.Graphical

Comparing Genetic Ne Reconstructions Over Time With Long-Time Wolf Monitoring Data in Two Populations.

Many methods are now available to calculate N e , but their performance varies depending on assumptions. Although simulated data are useful to discover certain types of bias, real empirical data supported by detailed known population histories allow us to discern how well methods perform with actual messy and complex data. Here, we focus on two genomic data sets of grey wolf populations for which population size changes of the past 40-120 years are well documented. We use this background to explore in what detail we can retrieve the known population history from these populations, in the light of pitfalls relating to population history, sampling design and the change in the spatial scale at which N e is estimated as we go further back in time. The Scandinavian wolf population was founded in the early 1980s from a few individuals and has gradually expanded up to 510 wolves. Although the founder event of the Scandinavian population was detected by GONE, the founding effective population size was strongly overestimated when the most recent samples were used, but less so when older samples were considered. Nevertheless, the present-day N e corresponds to theoretical expectations. The western Great Lakes wolf population of Minnesota is the only population in the contiguous United States that persisted throughout the 20th century, surviving intense persecution. We found a good concordance between the estimated N e and trends in census size data, but the reconstruction of N e clearly highlights the difficulty of interpreting results in spatially structured populations that underwent demographic fluctuations.

Transforming personalized chronic pain management with artificial intelligence: A commentary on the current landscape and future directions

Artificial intelligence (AI) has the potential to revolutionize chronic pain management by guiding the development of effective treatment strategies that are tailored to individual patient needs. This potential comes from AI's ability to analyze large and heterogeneous datasets to identify hidden patterns. When applied to clinical datasets of a particular patient population, AI can be used to identify pain subtypes among patients, predict treatment responses, and guide the clinical decision-making process. However, integrating AI into the clinical practice requires overcoming challenges such as data quality, the complexity of human pain physiology, and validation against diverse patient populations. Targeted, collaborative efforts among clinicians, researchers, and AI specialists will be needed to maximize AI's capabilities and advance current management and treatment of chronic pain conditions.

Comparison of Explainable AI Models for MRI-based Alzheimer's Disease Classification.

Deep learning models based on convolutional neural networks (CNNs) have been used to classify Alzheimer's disease or infer dementia severity from 3D T1-weighted brain MRI scans. Here, we examine the value of adding occlusion sensitivity analysis (OSA) and gradient-weighted class activation mapping (Grad-CAM) to these models to make the results more interpretable. Much research in this area focuses on specific datasets such as the Alzheimer's Disease Neuroimaging Initiative (ADNI) or National Alzheimer's Coordinating Center (NACC), which assess people of North American, predominantly European ancestry, so we examine how well models trained on these data generalize to a new population dataset from India (NIMHANS cohort). We also evaluate the benefit of using a combined dataset to train the CNN models. Our experiments show feature localization consistent with knowledge of AD from other methods. OSA and Grad-CAM resolve features at different scales to help interpret diagnostic inferences made by CNNs.

CellChat for systematic analysis of cell-cell communication from single-cell transcriptomics.

Recent advances in single-cell sequencing technologies offer an opportunity to explore cell-cell communication in tissues systematically and with reduced bias. A key challenge is integrating known molecular interactions and measurements into a framework to identify and analyze complex cell-cell communication networks. Previously, we developed a computational tool, named CellChat, that infers and analyzes cell-cell communication networks from single-cell transcriptomic data within an easily interpretable framework. CellChat quantifies the signaling communication probability between two cell groups using a simplified mass-action-based model, which incorporates the core interaction between ligands and receptors with multisubunit structure along with modulation by cofactors. Importantly, CellChat performs a systematic and comparative analysis of cell-cell communication using a variety of quantitative metrics and machine-learning approaches. CellChat v2 is an updated version that includes additional comparison functionalities, an expanded database of ligand-receptor pairs along with rich functional annotations, and an Interactive CellChat Explorer. Here we provide a step-by-step protocol for using CellChat v2 on single-cell transcriptomic data, including inference and analysis of cell-cell communication from one dataset and identification of altered intercellular communication, signals and cell populations from different datasets across biological conditions. The R implementation of CellChat v2 toolkit and its tutorials together with the graphic outputs are available at https://github.com/jinworks/CellChat . This protocol typically takes ~5 min depending on dataset size and requires a basic understanding of R and single-cell data analysis but no specialized bioinformatics training for its implementation.

Using publicly available UK datasets to identify recruitment sites to maximise inclusion of under-served groups: three case studies

Background There is strong evidence that those recruited into studies are not always representative of the population for whom the research is most relevant. Development of the study design and funding decisions are points in the research process where considerations about inclusion of under-served populations may usefully be made. Current practical guidance focuses on designing and modifying participant recruitment and retention approaches but an area that has not been addressed is recruitment site selection. Methods We present case studies of three NIHR funded trials to demonstrate how publicly available UK population datasets can be used to facilitate the identification of under-served communities for inclusion in trials. The trials have different designs, address different needs and demonstrate recruitment planning across Trauma centres, NHS Trusts and special educational settings. We describe our use of national freely available datasets, such as those provided by NHS Digital and the Office for National Statistics, to identify potential recruitment sites with consideration of health status, socio-economic status and ethnicity as well as clinical and risk factors to support inclusivity. For all three studies, we produced lists of potential recruitment sites in excess of the number anticipated as necessary to meet the recruitment targets. Discussion We reflect on the challenges to our approach and some potential future developments. The datasets used are all free to use but each has their limitations. Agreeing search parameters, acceptable proxies and identifying the appropriate datasets, then cross referencing between datasets takes considerable time and particular expertise. The case studies are trials, but the methods are generalisable for various other study types. Conclusion Through these exemplars, we aim to build on the NIHR INCLUDE project, by providing trialists with a much needed practical approach to embedding EDI into trial design at the grant application stage.

Collaborating across government agencies to combine linkage processes and algorithms to create an Australian National Linkage Spine (NLS) to facilitate cross-jurisdictional linkage

The Australian Bureau of Statistics (ABS) and the Australian Institute of Health and Welfare (AIHW) are co-technical lead agencies for developing the National Disability Data Asset (NDDA) which will bring together data across jurisdictions to build a more complete picture of the lives of people with disability. ABS and AIHW are collaboratively building a National Linkage Spine (NLS), a high-quality high-coverage population dataset to be shared under new legislation to facilitate cross-jurisdictional linkage and build the NDDA. ABS and AIHW each have extensive data integration experience, however our linkage methods have evolved differently in response to the needs of our respective researcher client bases and the datasets we work with. An ABS-AIHW cross-agency team was established to align terminology and share data processing methods, linkage approaches, and locally built linkage tools. The team collaborated on the design of an NLS that combines the best aspects of each of our approaches. The NLS build methodology delivers aligned processing of input data and applies a combination of ABS’s deterministic linkage algorithm and AIHW’s probabilistic linkage and machine learning algorithms. The NLS was peer reviewed by its future jurisdictional users and is considered fit for purpose for delivering the NDDA. The open sharing of knowledge and tools has significantly benefited both agencies and the broader NDDA project. Takeaways from this project extend beyond it, with improved data process approaches being considered by each agency in other work, and a growing common agreement of linkage best practice.