Green space exposure benefits mental health, but existing studies indicate that evidence on the independent contributions of accessibility and utilisation, and on whether utilisation mediates the effects of accessibility, is limited and inconsistent due to data limitations. The objectives of this study were to (i) quantify the independent associations of green space accessibility and utilisation with common mental health difficulties (CMD) prevalence; (ii) test whether utilisation mediates the effect of accessibility; and (iii) investigate whether accessibility and utilisation moderate one another. We integrated anonymised mobile phone data, high-resolution environmental metrics, and 2.6 million primary care records in Cheshire and Merseyside, England to quantify these pathways in 2024. Results show that accessibility and utilisation are independent, protective predictors of CMD prevalence. Residential tree exposure was the most consistent accessibility predictor (RR = 0.936), while walking time in green space showed the strongest utilisation effect (RR = 0.900). Crucially, mediation analysis revealed that utilisation does not mediate the benefits of accessibility, suggesting passive exposure holds intrinsic value. However, moderation models identified a compensatory interaction such that the mental health returns of active engagement were greatest where environmental access was lowest. Our findings demonstrate that accessibility and utilisation jointly shape CMD prevalence, underscoring the need for urban policies that promote both proximity and actual visits to green space.

The effects of green and blue spaces on newborn health are unclear with most studies using cross-sectional methods. Our retrospective longitudinal study aimed to investigate the effects of green and blue spaces on incidence of low birthweight and pre-term birth, and changes in birthweight (g) and gestational age (weeks). Our cohort was limited to births in Wales (2008 - 2019). Exposures were greenness at a 300m buffer (Enhanced Vegetation Index) and distances to blue spaces (1600m buffer), centred around a mother's home. Annual metrics were derived for the year pre-pregnancy and during-pregnancy. We utilised generalised linear models and controlled for individual-level confounders (e.g. mother's educational attainment) using census data. We conducted moderation analyses using mother's employment status and area-level deprivation. Our cohort comprised 176 459 births. Ten percent greater greenness pre-pregnancy was associated with an increase of 3.04g (95% CI 0.33 - 5.76g, p = 0.03) in birthweight. Further, 10% greater greenness pre-pregnancy was associated with a 1.0-week (p = 0.01) increase in gestational age. There were no greenness effects on pre-term birth or low birthweight. Increasing during-pregnancy distance between home and blue space was associated with increased gestational age (β = 1.0-week, p = 0.01). Greater greenness and living in moderately deprived areas were associated with increased birthweights. Our results indicate that greenness is associated with increases in birthweight and gestational age, while larger distances to blue space are associated with gestational age. The effect of green spaces on birth outcomes is not homogeneous based on area-level deprivation.

Background:The National Health Service uses formulae based on historic patterns of service use to distribute funding locally. These formulae are adjusted using avoidable mortality rates, to account for unmet needs and address health inequalities. We do not know whether this approach accurately reflects variation in unmet need between areas. Methods:We define need for National Health Service resources as the expected expenditure required to provide all standard National Health Services from which a population can benefit. We clarify the two components required to measure need: the relevant characteristics of the population living in each area and the cost weights indicating the expected needed expenditure associated with each of these characteristics. Four public advisor workshops were conducted to support the plain English description of the aims and methods used in the National Health Service resource allocation process. These led to the production of an animation to support public understanding of National Health Service resource allocation procedures. We developed four adjustments to the current National Health Service allocation formulae to better account for unmet need using electronic health records for England from 2010 and 2018 and linked primary care, hospital and mortality data from 2008 to 2020 from the Clinical Practice Research Datalink. Findings:Our responsiveness adjustment provides a method for more accurately estimating cost weights from regression models by only using data from those geographical areas that are better at aligning resources with patient need. Our longitudinal adjustment provides a new method for improving the way the utilisation formula accounts for supply-induced variation in utilisation, by using longitudinal data on people as they move between geographical areas. The primary care diagnosis adjustment shows how using linked data on primary care diagnosis improves the prediction of secondary care utilisation. Using data on people who have died from conditions, but who were not previously diagnosed, we estimate the proportion of people with 11 chronic diseases who were undiagnosed in each geographical area in England, providing an approach for adjusting National Health Service resource allocation formulae for unmet need due to underdiagnosis. Finally using instrumental variable methods we show that healthcare expenditure has larger effects on mortality in the middle deprivation groups, and smaller impacts in the top and bottom deprivation groups. These estimates indicate that some of our adjustments for unmet need may increase health inequalities. This highlights the need to maintain a 'health inequalities adjustment' that allocates more resources to socioeconomic groups with poor health outcomes to ensure that the National Health Service continues to help reduce health inequalities. Limitations:Limitations in the data available and a lack of coherence between adjustments for unmet need, derived from this research, indicate that a great deal of uncertainty remains. Further work is needed to enhance the availability of linked electronic health records and survey-based symptom and biomarker data to support the modelling of unmet needs. Conclusions:This programme of research has developed four adjustments targeting separate facets of unmet needs that can be implemented with the National Health Service resource allocation process as well as providing a tool that can be used to model the impact on health inequalities of any adjustment. Funding:This synopsis presents independent research funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme as award number NIHR130258.

Background:Many countries use geographical funding formulae to distribute public funds for health care to local planning areas in proportion to need. In England, these aim to distribute resources in proportion to all healthcare needs regardless as to whether these are currently met or unmet. The National Health Service also has an additional objective to allocate resources to reduce health inequalities (i.e. differences in health between socioeconomic groups). Adjusting for unmet needs could help achieve this second objective, if a greater proportion of needs are unmet in disadvantaged socioeconomic groups with poorer health compared to more advantaged socioeconomic groups. Alternatively, if there are greater unmet needs for relatively expensive conditions that tend to affect older age groups (e.g. cancer), this could lead to a greater proportion of needs being unmet in more advantaged socioeconomic groups, who will tend to be older due to greater life expectancy. Adjusting for unmet needs would then lead to allocation of a greater share of resources to these more affluent populations with better health, potentially increasing health inequalities. It is, however, unclear how met and unmet healthcare needs should be measured in these formulae and how better accounting for unmet needs influences health inequalities. Aim:We outline a framework for estimating the relative need in geographical healthcare resource allocation and show how the distribution of needed resources between local health planning areas in England changes when accounting for unmet needs due to underdiagnosis for 11 long-term conditions. Design:We derive a synthetic data set for all people aged ≥ 30 years in England, in 2018, including age, sex, socioeconomic deprivation, region, local health planning area and whether people have diagnosed or undiagnosed long-term conditions. We calculated the annual primary and secondary care costs for each condition using linked electronic healthcare record data, then estimated needed expenditure for each health planning area for two scenarios: (1) when only accounting for diagnosed cases and (2) including all cases (diagnosed and undiagnosed). We examine how the distribution of need between places changes between these scenarios and the consequences of this for health inequalities. Results:Based on the estimates of underdiagnosis used, areas with the lowest overall needs tended to have a greater proportion of their needs unmet. Adjusting resource allocation by accounting for these unmet needs due to underdiagnosis would move resources from areas with the highest level of needs to areas with lower overall needs. Moving to this 'fair share distribution' would tend to benefit less deprived areas more than more deprived areas, potentially widening health inequalities. Conclusion:We show how accounting for unmet needs due to underdiagnosis in allocating resources could widen health differences between more and less deprived areas when underdiagnosis and treatment costs increase with age. Further research is needed to confirm our provisional estimates, but we provide a useful framework for improving assessments of relative need for healthcare resource allocation. Alternative approaches are likely to be needed where resource allocation policy additionally aims to reduce health inequalities. Funding:This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health and Social Care Delivery Research programme as award number NIHR130258.

Background Projections suggest that the number of adults living with multimorbidity will continue growing in the coming decades. Little is known, however, about the potential impact of prevention policies on multimorbidity. Methods & findings We applied a validated microsimulation model of multimorbidity accumulation to simulate theoretical scenarios of health improvement and inequality reduction in England over 30 years (2019-2049), compared to a baseline scenario of continuing patterns in accumulation. Four theoretical scenarios were based on Benach et al.’s typology of health policies: 1) targeted intervention on the worst-off; 2) universal policy + additional focus on the gap; 3) redistributive policy; 4) proportionate universalism; plus an idealistic fifth scenario completely removing socioeconomic inequality in transition times between states. We selected a target of 3% reduction in mortality for scenarios 1-4, based on reductions seen from tobacco control policies. Outputs compared were: difference in 2049 projected prevalence and numbers compared to baseline, total cases prevented/postponed compared to baseline, and expected years lived without multimorbidity at age 30. Our results suggest that gains in levelling socioeconomic inequalities in health would prevent/postpone multimorbidity cases and reduce relative health inequalities among those aged <65. However, this would also likely lead to increased absolute numbers living with multimorbidity overall. Conclusions Our theoretical modelling suggests effective and equitable policies have potential to reduce the population-level burden of multimorbidity, postponing a substantial number of multimorbidity cases, particularly before age 65. This is, however, likely to lead to greater absolute numbers of multimorbidity cases as individuals live for longer.

BACKGROUND:Childhood height is an important indicator of child health. Stunting is associated with poor health outcomes, with consistent socioeconomic inequalities seen. This study aims to generate hypotheses for causal mechanisms through which socioeconomic deprivation leads to stunting. METHODS:A multiyear cross-sectional study using National Childhood Measurement Programme data from 2013/2014 to 2022/2023 in the Metropolitan Borough of Sefton, and ecological exposures for hypothesised mediators. We used logistic regression models to estimate the effect of socioeconomic deprivation on stunting rates and generate hypotheses for causal pathways. RESULTS:There was a significantly higher odds ratio for stunting in the most deprived quintile compared to the least deprived in children aged 4-5 years (1.60; 95% CI 1.10-2.35) and 10-11 years (2.02; 95% CI 1.28-3.25) after adjustment for confounders. In children aged 10-11 years 75.3% of this effect was attenuated when hypothesised mediators low birth weight, preterm birth, breastfeeding rates, food insecurity and healthcare access, were included; this result was not statistically significant. No attenuation was demonstrated in children aged 4-5 years. CONCLUSIONS:This research shows an association between socioeconomic deprivation and higher stunting rates in Sefton, which was non-significantly attenuated after adjustment for hypothesised mediators in children aged 10-11 years.

Background:Recent stagnation or worsening trends in cardiovascular disease (CVD) risk factors, including low-density lipoprotein cholesterol (LDL-c) and obesity, might slow the decline in Japan's CVD burden. We aimed to quantify the impact of national changes in CVD risk factor distributions on Japan's CVD burden from 2001 to 2019. Methods:We conducted a microsimulation study with counterfactual analysis using IMPACTNCD-JPN, a validated model based on real-world data. It simulated a synthetic Japanese population (ages 30-99) from 2001 to 2019 using life-course data on seven CVD risk factors, estimating CVD incidence, mortality, and healthcare economics for synthetic individuals. The base-case reflected observed trends; counterfactual scenarios assumed 2001 levels persisted. Primary outcome was national CVD incidence (stroke and coronary heart disease). Findings:From 2001 to 2019, systolic blood pressure (SBP) and smoking declined markedly (men/women) by 6·8/7·2 mmHg and 18·4/6·8%, respectively, while LDL-c, HbA1c, body mass index (BMI), physical activity (PA), and fruit/vegetable (FV) consumption showed smaller or adverse trends. Under the base-case and counterfactual scenarios, IMPACTNCD-JPN estimated CVD incidence and quantified the differences between the scenarios. The changes in the CVD risk factors prevented or postponed 840,000 (95% uncertainty interval: 540,000-1,300,000) national CVD cases, cumulative from 2001 to 2019. Individual contributions were: SBP 540,000; smoking 280,000; LDL-c 27,000; HbA1c 7900; BMI -15,000; PA -16,000; and FV consumption -11,000. Interpretation:SBP and smoking reductions drove most CVD burden declines in Japan (2001-2019). Modest benefits came from LDL-c and HbA1c, while rising BMI, and low PA and FV intake partly offset these benefits. Funding:JSPS KAKENHIJP22K17821, JP25K02863; the Ministry of Health, Labour and Welfare Comprehensive Research on Life-Style Related 22FA1015, 24FA1015.

Background:A large proportion of chronic conditions are undiagnosed, preventing early treatment, and leading to poorer outcomes. Understanding how levels of underdiagnosis vary between diseases and population groups over time is crucial for effectively allocating resources and targeting interventions to increase diagnosis rates. Methods:We used two annual national surveys: the Health Survey for England (cross-sectional) and the UK Household Longitudinal Survey, to identify people with diabetes, hypertension and depression. Diagnosed cases were defined as a self-report of being told by a nurse or doctor as having a condition; undiagnosed cases were defined as those where screening tools used in the survey identified clinical signs of the condition but the individual did not self-report a diagnosis. We used logistic regression to estimate the proportion of people with these three conditions who are undiagnosed for 540 population segments defined by age group, sex, deprivation quintile and region between 2011 and 2019. These predicted probabilities were applied to population estimates using microsimulation to model the proportion undiagnosed for each disease in each Clinical Commissioning Group (local health planning areas) in England. Results:The proportion of people with diabetes and depression who were undiagnosed reduced between 2011 and 2019, with no change in the proportion of hypertensives undiagnosed. For hypertension, people in more deprived areas were less likely to be undiagnosed than those in less deprived areas. The opposite was true for depression. Younger men with hypertension or diabetes were less likely to be diagnosed than older men. Both those aged under 30 and those over 70 with depression were less likely to be diagnosed compared with those aged 30-70. Conclusion:Strategies aiming to improve undiagnosed hypertension case finding need to understand the reasons for little progress over the past decade. For depression, strategies to increase early diagnosis should prioritise deprived areas. Case finding for all three diseases would benefit from targeting younger age groups.

Multimorbidity is projected to continue increasing in England and many other countries. Here, we use a validated microsimulation model to quantify the potential impact of improving exposure levels of eight risk factors on the burden of major illness among adults aged 30+ in England between 2023-2043. We find that the biggest contributors to incident major illness are body mass index, smoking, systolic blood pressure, and physical inactivity. Theoretical minimum risk exposure levels of all risk factors could reduce 2043 major illness prevalence by 2 percentage points (95% uncertainty intervals: 1.3, 2.7) compared to the continuing trends (base-case) scenario; under a 10% improvement in all risk factors, we project a 0.3 percentage points (0.2, 0.4) reduction in major illness. The impact on health inequalities is mixed. Our findings show that large improvements in risk factors are unlikely to substantially reduce the major illness burden by 2043 due to population ageing.

INTRODUCTION:Undiagnosed chronic disease has serious health consequences, and variation in rates of underdiagnosis between populations can contribute to health inequalities. We aimed to estimate the level of undiagnosed disease of 11 common conditions and its variation across sociodemographic characteristics and regions in England. METHODS:We used linked primary care, hospital and mortality data on approximately 1.3 million patients registered at a GP practice for more than one year from 01/04/2008-31/03/2020 from Clinical Practice Research Datalink. We created a dynamic state model with six states based on the diagnosis and mortality of 11 conditions: coronary heart disease (CHD), stroke, hypertension, chronic obstructive pulmonary disease, type 2 diabetes, dementia, breast cancer, prostate cancer, lung cancer, colorectal cancer, and depression/anxiety. Undiagnosed disease was conceptualised as those who died with a condition but were not previously diagnosed. This was combined with observed data on the incidence of diagnosis, the case fatality rate in the diagnosed, and an assumption about how that rate varies with diagnosis to estimate the number of undiagnosed disease cases over the total number of disease cases (underdiagnosis) in each population group. We estimated underdiagnosis by year, sex, 10-year age group, relative deprivation, and administrative region. We then applied small-area estimation techniques to derive underdiagnosis estimates for health planning areas (CCGs). RESULTS:Levels of underdiagnosis varied between 16% for stroke and 69% for prostate cancer in 2018. For all diseases, the level of underdiagnosis declined over time. Underdiagnosis was not consistently concentrated in areas with high deprivation. For depression/anxiety and stroke, underdiagnosis was estimated to be higher in less deprived CCGs, whilst for CHD and T2DM, it was estimated to be higher in more deprived CCGs, with no apparent relationships for other conditions. We found no uniform spatial patterns of underdiagnosis across all diseases, and the relationship between age, deprivation and the probability of being undiagnosed varied greatly between diseases. DISCUSSION:Our findings suggest that underdiagnosis is not consistently concentrated in areas with high deprivation, nor is there a uniform spatial underdiagnosis pattern across diseases. This novel method for estimating the burden of underdiagnosis within England depends on the quality of routinely collected data, but it suggests that more research is needed to understand the key drivers of underdiagnosis.

> Timeframes and frequency are important considerations when defining chronic conditions for multimorbidity research Heterogeneity in definitions of multimorbidity—the coexistence of several chronic disorders[1][1]—varies widely and is a recognised problem, affecting the transferability and

This report explores how patterns of diagnosed ill health vary by socioeconomic deprivation in England and projects these patterns to 2040.

ObjectiveThe existence of wide inequalities in self-reported health across England is well-documented. Our research adds to this evidence by describing current patterns and projecting future patterns of inequality in diagnosed illness by deprivation. ApproachWe used a microsimulation model which combines individual-level data on demographics, health and mortality from linked data for primary and secondary care with survey responses on leading modifiable risk factors and epidemiological evidence on the associations between risk factors and chronic illness. We used the Cambridge Multimorbidity Score to measure multimorbidity. This assigns a weight to 20 common long-term conditions based on individuals’ healthcare use and their likelihood of death. We further focus on “major illness” which corresponds to a score greater than 1.5. ResultsWe project that health inequalities are not projected to improve between 2019 and 2040. In 2040, we project the difference in the average time spent without major illness between the most and least deprived 10% of areas in England to be 10.7 years. This is largely unchanged from 10.4 years in 2019. In 2040, we project there will be more working age adults living with major illness in the most deprived areas, more than double the rate in the least deprived areas (15.2% versus 6.8%). These rates remain largely unchanged from 2019 at 14.6% and 6.3% respectively. Conclusion and ImplicationsOn current trends, health inequalities are projected to persist into the future. This has significant implications not just for population health but for labour supply and wider economic growth.

BackgroundThere are socioeconomic inequalities in the prevalence of multimorbidity and its accumulation across the life course. Estimates of multimorbidity prevalence in English primary care increased by more than two-thirds from 2004 to 2019. We developed a microsimulation model to quantify current and projected multimorbidity inequalities in the English adult population.MethodsWe used primary care data for adults in England from the Clinical Practice Research Datalink Aurum database between 2004 and 2019, linked to the 2015 English Index of Multiple Deprivation (IMD), to model time individuals spent in four health states (healthy, one chronic condition, basic multimorbidity [two or more chronic conditions], and complex multimorbidity [three or more chronic conditions affecting three or more body systems]) by sex, age, IMD quintile, birth cohort, and region. We applied these transition times in a stochastic dynamic continuous-time microsimulation model to Office for National Statistics population estimates for adults aged 30–90 years. We calculated projected prevalence and cumulative incident cases from 2019 to 2049 by IMD quintile, age group (younger than 65 years vs 65 years and older), and years to be lived without multimorbidity at age 30 years.FindingsUnder the assumption that all chronic conditions were lifelong, and that once diagnosed there was no recovery, we projected prevalence of multimorbidity (basic or complex) increases by 34% from 53·8% in 2019 to 71·9% (95% uncertainty interval 71·8–72·0) in 2049. This rise equates to an 84% increase in the number of people with multimorbidity: from 19·2 million in 2019 to 35·3 million in 2049 (35·3 million to 35·4 million). This projected increase is greatest in the most deprived quintile, with an excess 1·07 million (1·04 million to 1·10 million) cumulative incident basic multimorbidity cases and 0·70 million (0·67 million to 0·74 million) complex multimorbidity cases over and above the projected cases for the least deprived quintile, largely driven by inequalities in those younger than 65 years. The median expected number of years to be lived without multimorbidity at age 30 years in 2019 is 15·12 years (14·62–16·01) in the least deprived IMD quintile and 12·15 years (11·61–12·60) in the most deprived IMD quintile.InterpretationThe number of people living with multimorbidity will probably increase substantially in the next 30 years, a continuation of past observed increases partly driven by changing population size and age structure. Inequalities in the multimorbidity burden increase at each stage of disease accumulation, and are projected to widen, particularly among the working-age population. Substantial action is needed now to address population health and to prepare health-care and social-care systems for coming decades.FundingUniversity of Liverpool and National Institute for Health and Care Research School for Public Health Research.

Abstract Background Understanding the prevalence of diseases and where it is detected and recorded in healthcare settings is important for planning effective prevention and care provision. We examined inequalities in the prevalence of 205 chronic conditions and in the care setting where the related diagnoses were recorded in the English National Health Service. Methods We used data from the Clinical Practice Research Datalink Aurum linked with Hospital Episode Statistics for 12.8 million patients registered with 1406 general practices in 2018. We mapped diagnoses recorded in primary and secondary care in the previous 12 years. We used linear regressions to assess associations of ethnicity, deprivation, and general practice with a diagnosis being recorded in primary care only, secondary care only, or both settings. Results 72.65% of patients had at least one diagnosis recorded in any care setting. Most diagnoses were reported only in primary care (62.56%) and a minority only in secondary care (15.24%) or in both settings (22.18%). Black (− 0.08 percentage points (pp)), Asian (− 0.08 pp), mixed (− 0.13 pp), and other ethnicity patients (− 0.31 pp) were less likely than White patients to have a condition recorded. Patients in most deprived areas were 0.27 pp more likely to have a condition recorded (+ 0.07 pp in secondary care only, + 0.10 pp in both primary and secondary care, and + 0.10 pp in primary care only). Differences in prevalence by ethnicity were driven by diagnostic recording in primary care. Higher recording of diagnoses in more deprived areas was consistent across care settings. There were large differences in prevalence and diagnostic recording between general practices after adjusting for patient characteristics. Conclusions Linked primary and secondary care records support the identification of disease prevalence more comprehensively. There are inequalities in the prevalence and setting of diagnostic recording by ethnicity, deprivation, and providers on average across conditions. Further research should examine inequalities for each specific condition and whether they reflect also differences in access or recording as well as disease burden. Improving recording where needed and making national linked records accessible for research are key to understanding and reducing inequalities in disease prevention and management.

Background Multimorbidity prevalence has been increasing, with earlier onset, and persistent socioeconomic inequalities. Previous projection modelling has focused on multimorbidity after age 65, despite greater numbers of individuals aged under 65 living with multimorbidity. This study aimed to project the multimorbidity burden among adults, and differences by quintile of socioeconomic deprivation between 2019–2049 in England. Methods We developed a microsimulation model using data from a random sample of 1 m adults (18+) from the Clinical Practice Research Datalink Aurum database registered at GP practices within England between 2004 and 2019, linked to quintiles of the 2015 English Index of Multiple Deprivation (IMD) as a measure of area-level socioeconomic deprivation. We used parametric survival analysis methods to model the time individuals spent in four health states of interest: healthy, one chronic condition, basic multimorbidity (2 or more chronic conditions), complex multimorbidity (3 or more chronic conditions affecting 3 or more body systems) by sociodemographic characteristics: sex, age, IMD quintile, birth cohort, and region. These transition times were applied to a 1% sample (N = 562,880) of the 2019 Office for National Statistics (ONS) population estimates for adults aged 30–90 in 2019, and projected for 30 years. We calculated annual projected prevalence and cumulative incident cases by IMD quintile and age-group (<65/65+), and years lived without multimorbidity at age 30 by IMD quintile. Results are the median (with 95% uncertainty intervals) from a preliminary 10 model runs, scaled up to the ONS population. Analyses were conducted using R v4.2.2. Results Projected crude prevalence of basic multimorbidity and complex multimorbidity increases by 60%, from 45% [45%-45%] with basic multimorbidity in 2019 to 71% [71%-73%] in 2049. Between 2019–2049, 470,000 [460,000–500,000] more incident basic multimorbidity cases, and 470,000 [460,000–480,000] more complex multimorbidity cases, are projected in the most compared to the least deprived IMD quintile. This is driven by inequalities in the working-age population: a projected 1 million [1 m-1.1 m] more incident basic multimorbidity cases, and 1.5 m [1.4 m-1.5 m] more complex multimorbidity cases, among under 65s from the most deprived IMD quintile compared to the least deprived. The median expected number of years lived without multimorbidity at age 30 was 15 years [15–16] in the least deprived IMD quintile, and 13 [12–13] in the most deprived. Conclusion Continuing trends in multimorbidity accumulation will likely lead to substantial increases in multimorbidity prevalence by 2049, with growing inequalities, particularly pronounced among the working-age population. Equitable policies are needed to postpone or prevent multimorbidity onset.

Background Policy simulation models (PSMs) have been used extensively to shape health policies before real-world implementation and evaluate post-implementation impact. This systematic review aimed to examine best practices, identify common pitfalls in tobacco control PSMs and propose a modelling quality assessment framework. Methods We searched five databases to identify eligible publications from July 2013 to August 2019. We additionally included papers from Feirman et al for studies before July 2013. Tobacco control PSMs that project tobacco use and tobacco-related outcomes from smoking policies were included. We extracted model inputs, structure and outputs data for models used in two or more included papers. Using our proposed quality assessment framework, we scored these models on population representativeness, policy effectiveness evidence, simulated smoking histories, included smoking-related diseases, exposure-outcome lag time, transparency, sensitivity analysis, validation and equity. Findings We found 146 eligible papers and 25 distinct models. Most models used population data from public or administrative registries, and all performed sensitivity analysis. However, smoking behaviour was commonly modelled into crude categories of smoking status. Eight models only presented overall changes in mortality rather than explicitly considering smoking-related diseases. Only four models reported impacts on health inequalities, and none offered the source code. Overall, the higher scored models achieved higher citation rates. Conclusions While fragments of good practices were widespread across the reviewed PSMs, only a few included a ‘critical mass’ of the good practices specified in our quality assessment framework. This framework might, therefore, potentially serve as a benchmark and support sharing of good modelling practices.

ObjectiveExamine the association between country-level gender social norms and (1) cardiovascular disease mortality rates; (2) female to male cardiovascular disease mortality ratios; and (3) life expectancy.DesignEcological study with the country as the unit of analysis.SettingGlobal, country-level data.ParticipantsGlobal population of countries with data available on gender social norms as measured by the Gender Social Norms Index (developed by the United Nations Development Programme).Main outcome measuresCountry-level female and male age-standardised cardiovascular disease mortality rates, population age-standardised cardiovascular disease mortality rates, female to male cardiovascular disease mortality ratios, female and male life expectancy at birth. Outcome measure data were retrieved from the WHO and the Institute for Health Metrics and Evaluation. Multivariable linear regression models were fitted to explore the relationship between gender social norms and the outcome variables.ResultsHigher levels of biased gender social norms, as measured by the Gender Social Norms Index, were associated with higher female, male and population cardiovascular disease mortality rates in the multivariable models (β 4.86, 95% CIs 3.18 to 6.54; β 5.28, 95% CIs 3.42 to 7.15; β 4.89, 95% CIs 3.18 to 6.60), and lower female and male life expectancy (β −0.07, 95% CIs −0.11 to −0.03; β −0.05, 95% CIs −0.10 to −0.01). These results included adjustment within the models for potentially confounding country-level factors including gross domestic product per capita, population mean years of schooling, physicians per 1000 population, year of Gender Social Norms Index data collection and maternal mortality ratio.ConclusionsOur analysis suggests that higher levels of biased gender social norms are associated with higher rates of population cardiovascular disease mortality and lower life expectancy for both sexes. Future research should explore this relationship further, to define its causal role and promote public health action.