Andreas Höhn’s research while affiliated with University of Glasgow and other places

Understanding causality, over mere association, is vital for researchers wishing to inform policy and decision making – for example, when seeking to improve population health outcomes. Yet, contemporary causal inference methods have not fully tackled the complexity of data hierarchies, such as the clustering of people within households, neighbourhoods, cities, or regions. However, complex data hierarchies are the rule rather than the exception. Gaining an understanding of these hierarchies is important for complex population outcomes, such as non-communicable disease, which is impacted by various social determinants at different levels of the data hierarchy. The alternative of analysing aggregated data could introduce well-known biases, such as the ecological fallacy or the modifiable areal unit problem. We devise a hierarchical causal diagram that encodes the multilevel data generating mechanism anticipated when evaluating non-communicable diseases in a population. The causal diagram informs data simulation. We also provide a flexible tool to generate synthetic population data that captures all multilevel causal structures, including a cross-level effect due to cluster size. For the very first time, we can then quantify the ecological fallacy within a formal causal framework to show that individual-level data are essential to assess causal relationships that affect the individual. This study also illustrates the importance of causally structured synthetic data for use with other methods, such as Agent Based Modelling or Microsimulation Modelling. Many methodological challenges remain for robust causal evaluation of multilevel data, but this study provides a foundation to investigate these.

Background ‘Twenty-minute neighbourhoods’ (or variations, such as 15-minute cities) are receiving increasing policy attention with anticipated impacts on population health (inequalities) outcomes alongside sustainability improvements. Yet, factors contributing to possible health impacts are not well understood. This scoping review aimed to identify proposed and evidenced pathways to health (inequality) outcomes from international policy plans. Methods We first identified relevant academic literature, searching Scopus, (Ovid) Medline and Embase databases. A second search aimed to identify local or national planning or policy documents on government websites and related organisations. We followed a snowball search strategy to retrieve examples identified from the academic literature search and from the C40 cities network. These policy documents were our primary target for extraction, and we extracted and analysed by individual place. Pathways to health and health inequality outcomes identified in these documents were inductively coded thematically. We used Sankey diagrams to visually aggregate the thematic codes for each place relating to pathways to health outcomes and social determinants (mechanisms). Results In total, 36 places across 17 countries were included, described across 96 academic articles, policy plans and reports. While different health improvement outcomes were included as a goal in nearly all policy plans, most frequently references were to health in general rather than specific health outcomes. Pathways to health were discussed in numerous policy plans across three overarching themes: proximity, place redesign, and environmental action. Proximity pathways were most frequently outlined as the means to achieve health outcomes, with active travel acting through increased physical activity/reduced obesity as the most frequent individual pathway. However, few plans specified what would actually be implemented in practice to achieve the increased proximity to services. Health inequalities were only mentioned by six places specifically, although nearly half of all places mentioned broader inequality aims (e.g., poverty reduction). Possible unintended consequences to health inequalities also received some attention, for example through displacement of residents. Discussion Pathways to assumed health (inequality) outcomes require better specification and evidence. Health inequalities are particularly under-explored, and scenario modelling might provide a means to explore the dynamic aspects necessary to examine these important outcomes pre-implementation.

Background The absence of a centralised and comprehensive register-based system limits opportunities for studying the interaction of aspects such as health, employment, benefit payments, and housing at the micro-level in Great Britain (GB). In some cases, surveys can provide a swiftly available alternative. However, survey data do typically not allow for a detailed spatial resolution. While area-level linkages of surveys can enable a more granular spatial resolution, sampling strategies are often not representative for sub-national levels and results of aggregations might not be meaningful due to small sample sizes. Survey-based full-scale synthetic population datasets can help to bypass these highlighted limitations of surveys. By providing attribute-rich data for individuals and households, synthetic population datasets can enable both: Representativeness and statistical power at a granular spatial resolution. Methods We present the Synthetic Population for Individuals in Great Britain 2019 – 2021 (SPIGB), a survey-based full-scale synthetic population dataset developed by the System Science in Public Health and Health Economics Research (SIPHER) consortium, and provide details on its creation, validation, limitations, and applications. The SPIGB dataset was created via a combinatorial optimisation algorithm (simulated annealing) and combines individual-level data from the Understanding Society survey (wave 11, ‘k’) with aggregate-level population statistics obtained from the UK Census and population projections for Lower layer Super Output Areas and Data Zones. Results The SPIGB dataset is representative with respect to 8 characteristics; age/sex, highest qualification, ethnicity, marital status, economic activity, general health, household tenure, and household type at a small-area level. Results of external and internal validation suggest that the dataset makes for a well-suited resource across different applications examining health and socioeconomic outcomes across small areas. Ongoing and completed projects have utilised the SPIGB dataset to obtain insights into spatial patterning of alcohol consumption across Greater Manchester, to construct an interactive R-shiny dashboard for policy stakeholders, as an input in microsimulation models exploring the population health impact of the Scottish Child Payment, and to explore the dataset's potential for the creation of synthetic linked administrative data in Scotland's safe havens. Conclusion The SPIGB is a well-suited dataset for exploring health and socioeconomic domains at a granular spatial resolution across a range of different applications. At the same time, care is required when seeking to disentangle causal multilevel structures or individual-level characteristics for which the association with the utilised constraint variables has not been evaluated or is unknown.

Background Government bodies, private enterprises, and researchers increasingly use ‘big data’ to monitor, evaluate interventions, make future predictions, and seek causal understanding. Such data are often complex in structure (i.e., hierarchical), which creates challenges for methods that work for a single homogeneous population, but which mislead if applied to data with substructure. If causal insights are sought, this usually pertains to the individual, yet most datasets are aggregated due to issues surrounding sensitive personal information, which is why it is common to encounter simulation approaches, such as agent-based modelling (ABM), or ecological analyses that evaluate only marginal (i.e., clustered) information. Contemporary causal inference methods are yet to tackle the full complexities of multilevel data structure, beyond longitudinal repeated measures. There is thus a gap in our understanding and methods capabilities surrounding causal analysis of structured data, which this study examines. Methods 1) devise a hierarchical causal diagram that encodes a multilevel data generating mechanism with prespecified cross-level causal relationships; 2) simulate multilevel data from the causal diagram and obtain aggregated data; 3) contrast multilevel and ecological estimates of a simulated individual-level causal effect, to assess the presence and extent of potential biases. Results Unlike a multilevel analysis of the full data, ecological analyses of cluster-level data do not generally yield robust causal effect estimates. While it is known that ecological analyses invoke the ‘ecological fallacy’ (i.e., where attributing features of clusters to units within clusters may mislead), this study quantifies this for the first time within a formal causal framework. An algorithm to simulate causally structured multilevel data is also demonstrated. Conclusion Insights into the limitations of common analytical practices were made possible by simulating causally structured hierarchical data, demonstrating the value of causal diagrams in both simulation and causal analysis. Methodological challenges remain for robust causal evaluation of big data, but this study shows how to investigate these challenges. Results reveal the need for individual-level data with application of multilevel analyses to achieve robust causal inquiry; ecological analyses do not generally provide sound causal effect estimation. If individual-level data are unavailable, synthetic data (informed by available marginal data) becomes necessary to answer causal questions and this study provides a tool to generate synthetic population data that reflects multilevel causal structures, which in turn will then better inform the use of methods such as ABMs. This study has enormous implications for the use of big data when seeking causal insights.

Immigrants and their descendants increasingly shape fertility patterns in European societies. While childbearing among immigrants is well explored, less is known with respect to their descendants. Using Swedish register data, we studied differences in fertility outcomes between first- and second-generation individuals in Sweden and compared with the native Swedish population. We studied men and women separately, distinguished between high- and low-fertility backgrounds, and differentiated whether the descendants of immigrants were offspring from endogamous or exogamous relationships. For most migrants who arrived in Sweden as adults, we found elevated first birth rates shortly after arrival. First birth rates among the second generation were generally close to but lower than the rates observed among native Swedes. Male offspring from exogamous unions with a Swedish-born mother tended to have less depressed rates of first birth than other second-generation individuals. Second birth rates were very similar across population subgroups but generally lower among immigrants and their descendants compared to native Swedes. Third birth rates were often polarized into high- and low-fertility backgrounds, when compared to native Swedes. While fertility patterns among the second generation appeared to drift away from patterns of the first generation, the second generation remained a heterogeneous population subgroup. Nevertheless, and as childbearing patterns of the descendants with one immigrant parent increasingly resembled patterns of native Swedes, exogamous partnerships can likely be considered an important factor behind this gradual family-demographic assimilation process.

Objectives Quantifying area-level inequalities in population health can help to inform policy responses. We describe an approach for estimating quality-adjusted life expectancy (QALE), a comprehensive health expectancy measure, for local authorities (LAs) in Great Britain (GB). To identify potential factors accounting for LA-level QALE inequalities, we examined the association between inclusive economy indicators and QALE. Setting 361/363 LAs in GB (lower tier/district level) within the period 2018–2020. Data and methods We estimated life tables for LAs using official statistics and utility scores from an area-level linkage of the Understanding Society survey. Using the Sullivan method, we estimated QALE at birth in years with corresponding 80% CIs. To examine the association between inclusive economy indicators and QALE, we used an open access data set operationalising the inclusive economy, created by the System Science in Public Health and Health Economics Research consortium. Results Population-weighted QALE estimates across LAs in GB were lowest in Scotland (females/males: 65.1 years/64.9 years) and Wales (65.0 years/65.2 years), while they were highest in England (67.5 years/67.6 years). The range across LAs for females was from 56.3 years (80% CI 45.6 to 67.1) in Mansfield to 77.7 years (80% CI 65.11 to 90.2) in Runnymede. QALE for males ranged from 57.5 years (80% CI 40.2 to 74.7) in Merthyr Tydfil to 77.2 years (80% CI 65.4 to 89.1) in Runnymede. Indicators of the inclusive economy accounted for more than half of the variation in QALE at the LA level (adjusted R² females/males: 50%/57%). Although more inclusivity was generally associated with higher levels of QALE at the LA level, this association was not consistent across all 13 inclusive economy indicators. Conclusions QALE can be estimated for LAs in GB, enabling further research into area-level health inequalities. The associations we identified between inclusive economy indicators and QALE highlight potential policy priorities for improving population health and reducing health inequalities.

Background Monitoring geographical inequalities in health can help national and local public health teams to develop appropriately tailored policy approaches. Within the past years, an increasing number of studies has utilised Quality-Adjusted Life Expectancy (QALE) as an outcome measure to quantify the impact of policy interventions on population health. Quality-Adjusted Life Expectancy (QALE) is a holistic population-level health metric which captures mortality alongside mental and physical health. The greater detail in the way that QALE captures health distinguishes it from other health expectancy metrics such as Healthy Life Expectancy or Disability-Free Life Expectancy. We describe an approach to estimate QALE for local authority districts (LAs) in England, Scotland and Wales and examine the association between economic inclusion and QALE as an exemplary case study. Methods In a first step, we estimated lifetables for females and males in all LAs using TOPALS and Kannisto models. In a second step, we estimated age- and sex-specific health state utility scores using the Understanding Society main stage survey. For this purpose, we mapped Short Form 12 (SF-12) group averages for mental and physical health to utility scores. We then used the Sullivan method to estimate QALE at birth in years. Indicators on various dimensions of economic inclusion were obtained from the Inclusive Economy dataset which as previously created by the System Science in Public Health and Health Economics Research (SIPHER) consortium. Results In 2018–2020, QALE was lower on average in Scotland (females/males: 65.09 y/64.90 y) and Wales (65.13 y/65.35 y) than in England (67.55 y; 67.69 y). For females, QALE ranged from 56.33 y in Mansfield to 77.76 y in Runnymede. Among males, QALE was lowest in Merthyr Tydfil (57.64 y) and highest in Runnymede (77.44 y). We found that several indicators of economic inclusion were associated with QALE, including digital connectivity, access to public transport, affordability of housing, and child poverty. Indicators of economic inclusion accounted for more than half of the variation in QALE at the LA level (Adjusted R-squared females/males: 51%/58%). Conclusion Our study provides an estimation method for QALE for LAs in England, Scotland, and Wales – allowing further research into spatial health inequalities. Our results indicate that differences in QALE are large across local authorities in Great Britain, indicating substantial area-level inequalities in population health. Economic inclusion might be particularly important in explaining these area-level inequalities in population health. Research and policy design, especially within the context of ‘Levelling Up’ strategies, should consider the importance of economic inclusion for health outcomes.

Background Many complex public health evidence gaps cannot be fully resolved using only conventional public health methods. We aim to familiarise public health researchers with selected systems science methods that may contribute to a better understanding of complex phenomena and lead to more impactful interventions. As a case study, we choose the current cost-of-living crisis, which affects disposable income as a key structural determinant of health. Methods We first outline the potential role of systems science methods for public health research more generally, then provide an overview of the complexity of the cost-of-living crisis as a specific case study. We propose how four systems science methods (soft systems, microsimulation, agent-based and system dynamics models) could be applied to provide more in-depth understanding. For each method, we illustrate its unique knowledge contributions, and set out one or more options for studies that could help inform policy and practice responses. Results Due to its fundamental impact on the determinants of health, while limiting resources for population-level interventions, the cost-of-living crisis presents a complex public health challenge. When confronted with complexity, non-linearity, feedback loops and adaptation processes, systems methods allow a deeper understanding and forecasting of the interactions and spill-over effects common with real-world interventions and policies. Conclusions Systems science methods provide a rich methodological toolbox that complements our traditional public health methods. This toolbox may be particularly useful in early stages of the current cost-of-living crisis: for understanding the situation, developing solutions and sandboxing potential responses to improve population health.

This document outlines the protocol for a scoping review researching the potential impacts of x-min city or neighbourhood concepts on health and health inequalities.

Background We report the first study to estimate the socioeconomic gap in period life expectancy (LE) and life years spent with and without complications in a national cohort of individuals with type 1 diabetes. Methods This retrospective cohort study used linked healthcare records from SCI-Diabetes, the population-based diabetes register of Scotland. We studied all individuals aged 50 and older with a diagnosis of type 1 diabetes who were alive and residing in Scotland on 1 January 2013 (N = 8591). We used the Scottish Index of Multiple Deprivation (SIMD) 2016 as an area-based measure of socioeconomic deprivation. For each individual, we constructed a history of transitions by capturing whether individuals developed retinopathy/maculopathy, cardiovascular disease, chronic kidney disease, and diabetic foot, or died throughout the study period, which lasted until 31 December 2018. Using parametric multistate survival models, we estimated total and state-specific LE at an attained age of 50. Results At age 50, remaining LE was 22.2 years (95% confidence interval (95% CI): 21.6 − 22.8) for males and 25.1 years (95% CI: 24.4 − 25.9) for females. Remaining LE at age 50 was around 8 years lower among the most deprived SIMD quintile when compared with the least deprived SIMD quintile: 18.7 years (95% CI: 17.5 − 19.9) vs. 26.3 years (95% CI: 24.5 − 28.1) among males, and 21.2 years (95% CI: 19.7 − 22.7) vs. 29.3 years (95% CI: 27.5 − 31.1) among females. The gap in life years spent without complications was around 5 years between the most and the least deprived SIMD quintile: 4.9 years (95% CI: 3.6 − 6.1) vs. 9.3 years (95% CI: 7.5 − 11.1) among males, and 5.3 years (95% CI: 3.7 − 6.9) vs. 10.3 years (95% CI: 8.3 − 12.3) among females. SIMD differences in transition rates decreased marginally when controlling for time-updated information on risk factors such as HbA1c, blood pressure, BMI, or smoking. Conclusions In addition to societal interventions, tailored support to reduce the impact of diabetes is needed for individuals from low socioeconomic backgrounds, including access to innovations in management of diabetes and the prevention of complications.